key: cord-0979440-1s7yfoy2
authors: nan
title: Research forum abstracts
date: 2003-10-31
journal: Annals of Emergency Medicine
DOI: 10.1016/s0196-0644(03)00821-7
sha: e384fc58c3ce9ec16698d34474cc97c7dc9721ab
doc_id: 979440
cord_uid: 1s7yfoy2

nan

Study objective: We examine whether patients with minor traumatic head injury and coagulopathy are at increased risk of significant intracranial injury.

Methods: We performed a secondary analysis of data collected prospectively as part of the NEXUS II head computed tomography (CT) derivation study of all patients who had a head CT scan after blunt head trauma. Minor head injury was defined as Glasgow Coma Score 15, no neurologic deficit, and no evidence of significant skull fracture. Coagulopathy was defined as any impairment in blood clotting. Significant intracranial injury was defined by predetermined CT criteria according to findings that would likely change the patient's management. χ 2 Was used to assess differences in the prevalence of significant intracranial injury between coagulopathy and noncoagulopathy patients.

Results: Thirteen thousand seven hundred twenty-eight patients were enrolled, of whom 493 had coagulopathy and 1,193 had significant intracranial injury. In the coagulopathy group, 62 (12.6%; 95% confidence interval [CI] 9.8 to 15.8) had significant intracranial injury versus 1,131 (8.5%; 95% CI 8.1 to 9.0) in the noncoagulopathy group (χ 2 =9.8; P=.002). Among 74 patients with "inconsequential" intracranial injuries, 5 of 7 coagulopathy patients (1 with normal initial CT; 71.4%; 95% CI 29.0 to 96.3) required directed intervention (ie, intubation, intracranial pressure monitoring, craniotomy). In comparison, 3 of 67 noncoagulopathy patients (4.4%; 95% CI 0.9 to 12.5) required intervention (χ 2 =29; P<.001).

Conclusion: Coagulopathy patients are at greater risk of significant intracranial injury from minor head trauma. Coagulopathy is associated with a substantially increased risk of delayed morbidity despite the presence of minimal intracranial injuries.

Lyon M, Blaivas M, Brannam L/Medical College of Georgia, Augusta, GA Study objective: Detecting and monitoring blood loss in trauma patients can often be challenging when an obvious source of hemorrhage is not readily seen. We provide a noninvasive measurement of circulating blood volume and of drop therein by measuring the change in the inferior vena cava diameter in relationship to blood loss.

Methods: This was a prospective observational study on blood donors at a donation center. The inferior vena cava diameter, both during inspiration (IVCi) and during expiration (IVCe) was measured in volunteers both before and after blood donation of 450 mL. The study was approved by the authors' human assurance committee and the medical director of the blood donation center, and written consent was obtained before ultrasonographic examination. All actual blood donors older than 17 years were eligible for enrollment. Persons who were younger than 18 years, who declined to participate in the study, or who did not meet blood center criteria for blood donation were excluded. All examinations were performed in the supine position, with the ultrasonographic transducer placed in a subxiphoid location. Sagittal sections of the inferior vena cava diameter behind the liver were imaged, and the maximal diameter of the inferior vena cava in quiet expiration and the minimal diameter in quiet inspiration were measured. Statistical analysis included test for normality, paired t test, and correlation analysis.

Results: A total of 31 volunteers (18 men) with a mean age of 49.5 years (range 18 to 73 years) were studied. The mean IVCe before blood donation was 17.4 mm (95% confidence interval [CI] 15.2 to 19.7 mm) and after blood donation was 11.9 mm (95% CI 10.3 to 13.6 mm). The mean IVCi before blood donation was 13.3 mm (95% CI 11.3 to 15.3 mm) but after blood donation was 8.13 mm (95% CI 6.7 to 9.6 mm). The difference between IVCe before and after blood donation was 5.5 mm (95% CI; 4.3 to 6.3 mm) yielding a P less than .0001. The difference between IVCi before and after donation was 5.16 mm (95% CI 4.2 to 5.9 mm) yielding a P less than .0001. The difference between IVCe and the difference between IVCi before and after blood donation were closely correlated (r=0.83). Similarly, the pre-IVCe correlated well to the post-IVCe (r=.74) and the pre-IVCi correlated well to the post-IVCi (r=.75).

Conclusion: Our data indicate that the measurement of the inferior vena cava diameter is a reliable indicator of blood loss, even in small amounts of 450 mL. On average, there was about a 5-mm decrease in the IVCe and IVCi after donation of 450

arrest caused by pulseless ventricular tachycardia or ventricular fibrillation treated with lidocaine, amiodarone, or amiodarone plus lidocaine in a naturalistic setting. Methods: One hundred ninety-four adult patients hospitalized in 3 institutions between August 1, 2000, and July 31, 2002, who received amiodarone, lidocaine, or both for treatment of ventricular tachycardia or ventricular fibrillation during a cardiac arrest were evaluated. Patients were identified retrospectively by using cardiopulmonary resuscitation forms, International Classification of Diseases, Ninth Revision codes for ventricular tachycardia or ventricular fibrillation, and pharmacy charge codes for study drugs. Information was obtained by medical record review. Data were collected according to the Utstein style. One-way analysis of variance was used to compare means for time data. Inhospital survival rates for all patients were analyzed with χ 2 . Survival curves were plotted for patients by using the Kaplan-Meier procedure. Equality of survival curves was tested by the log-rank method. Cox regression was used to predict mortality by using age, cardiac history, initial rhythm, witnessed event, location of code, and antidysrhythmic agent received.

Results: For the testing equality of 3 survival curves, log-rank equaled 3.02 and P equaled .22. For the pairwise comparison of lidocaine and amiodarone, log-rank equaled 2.51 and P equaled .11; and for lidocaine and a combination of lidocaine plus amiodarone, log-rank equaled 2.07 and P equaled .15. Furthermore, Cox regression indicated that the patient's initial rhythm and antidysrhythmic agent received significantly predicted mortality (amiodarone versus lidocaine: hazard rate 2.1; P=.01; Tables 1 and 2).

Conclusion: Use of lidocaine leads to better survival rates and longer estimated mean survival time compared with amiodarone by 14% and 418 hours, respectively. Kaplan-Meier survival analysis did not demonstrate a statistically significant difference among the 3 drug groups. However, Cox regression indicated that patients who received lidocaine are twice as likely to survive than those who received amiodarone. These results call into question the current trend of using amiodarone for first-line treatment of inhospital cardiac arrests. Larger trials are needed to confirm these findings. ED presentation into the Web-based follow-up system, and later (after researching the patients' subsequent course) entered final follow-up information into the system. Before follow-up entry completion, the resident was asked to enter whether he or she believed that the act of performing follow-up led to an improvement in that patient's care (yes/no). The completed entry was then routed electronically to the faculty member who originally supervised the care of that patient. Blinded to the resident's response about patient care improvement, the faculty member reviewed the completed follow-up entry and independently evaluated whether he or she thought that the patient's care was improved through the process of resident followup (yes/no). Data entered into the Web-based follow-up system were then summarized as percentages.

Results: During the 1-year study period, 846 patient follow-up encounters (535 admitted; 311 discharged) were completed by 18 emergency medicine residents and 29 emergency medicine faculty. In 21.9% of cases, either the resident or the faculty member believed that the act of performing follow-up improved patient care. Faculty members were more likely than residents to think that follow-up enhanced patient care, citing improvements in 19.7% of cases (versus 3.0%). Residents thought that follow-up improved patient care in 5.5% of discharged patients versus 1.5% of the admitted, compared with faculty assessment of 25.1% and 16.6%, respectively.

Conclusion: Emergency medicine residents and faculty think that the follow-up of resident-selected ED patients infrequently leads to patient care improvements but suggest that if a benefit does exist, it may be in discharged patients. 

Hunter GC, Burbank-Schmitt ER, Brown DF, Camargo CA/Massachusetts General Hospital, Boston, MA Study objective: We evaluate the effect of an enhanced computer information system on emergency department (ED) length of stay.

Methods: On February 11, 2003, a Boston ED implemented an enhanced computer information system. Previously, the ED kept track of patients by combining a computer registry and white boards in each treatment area. The new system was interactive, obviated the boards, and highlighted important information (eg, assigned Results: A total of 302 (14.1%) of the 2,148 patients experienced 30-day adverse outcome. ROC curve area of the Bayesian network probability score for presence of 30day adverse outcome was 0.78 (95% confidence interval [CI] 0.77 to 0.80). The optimal probability score for detection of 30-day adverse outcome was more than 0.099 (sensitivity 79.5%, specificity 63.3%; positive likelihood ratio 2.2; negative likelihood ratio 0.32). Patients with a Bayesian network probability score more than 0.099 had 5.7 times increased odds of acute myocardial infarction (95% CI 3.8 to 8.6; P<.0001), 7.5 times increased odds of 30-day percutaneous coronary intervention (95% CI 4.8 to 11.9; P<.0001), 5.5 times increased odds of 30-day coronary artery bypass grafting (95% CI 3.1 to 10.0; P<.0001), 5.5 times increased odds of 30-day life-threatening complication (95% CI 2.0 to 14.6; P<.0001), and 5.4 times increased risk of death (95% CI 1.1 to 25.5; P=.02) compared with patients with probability score no higher than 0.099.

Conclusion: A Bayesian network representation of the joint probabilities of a model using key data elements obtained on initial patient triage effectively risk stratifies patients for 30-day adverse outcome. Studies are under way that incorporate key clinical examination findings, ECG findings, and laboratory findings into improving the predictive ability of our Bayesian network. Future studies are needed to determine whether this information can be used to influence management decisions in the real-time evaluation of chest pain patients.

Fesmire FM, Adams DE, Wali GO/University of Tennessee College of Medicine, Chattanooga, TN Study objectives: We determine predictive value of the simplified Ho clinical score, which uses only 5 clinical variables compared with the more complex Morise and Framingham clinical scores for predicting 30-day adverse outcome in emergency department (ED) chest pain patients with suspected acute coronary syndromes and history of ischemic heart disease.

Methods: This was a retrospective analysis of a previous prospectively acquired database of 769 consecutive ED chest pain patients with a history of ischemic heart disease and with absence of injury on initial ECG who underwent a standardized chest pain evaluation protocol. All patients were followed up for 30-day adverse outcome that was defined as acute myocardial infarction, percutaneous coronary intervention, coronary artery bypass grafting, life-threatening complication, or cardiac death within 30 days of ED presentation. History of ischemic heart disease was defined as previous acute myocardial infarction, percutaneous coronary intervention, or coronary artery bypass grafting. The Ho clinical score was calculated as follows: 1 point for each decade of patient's age older than 40 years, 1 point for male sex, 1 point for previous myocardial infarction (excluded from present study), 1 point for typical chest pain, 1 point for non-insulin dependent diabetes, and 2 points for insulin-dependent diabetes. The Morise clinical scores and the Framingham clinical scores were calculated according to a previously defined scoring systems by using a combination of typical risk factors and select clinical findings.

Results: Thirty-day adverse outcome occurred in 154 patients (20.0%). Area under the receiver operating characteristic (ROC) curve for 30-day adverse outcome for the Ho score, the Morise score, and Framingham score was 0.58±0.03, 0.60±0.03, and 0.54±0.03, respectively. There were no significant differences in areas between any of the 3 scoring systems. At the most accurate cutoff value, none of the 3 scoring systems had sufficient positive (+) or negative (-) likelihood ratios (LR) to influence clinical decisionmaking: Ho score more than 3 (sensitivity 86.4%, specificity 28.9%, +LR 1.2; -LR 0.47); Morise score more than 11 (sensitivity 82.5%; specificity 32.4%; +LR 1.2; -LR 0.54); Framingham more than 14 (sensitivity 75.3%, specificity 35.0%, +LR 1.2; -LR 0.71).

Conclusion: Clinical scoring systems using standard clinical risk factors and select clinical findings perform poorly in chest pain patients with preexisting ischemic heart disease for prediction of adverse outcome. Physicians must maintain a high index of suspicion in chest pain patients with previous acute myocardial infarction, percutaneous coronary intervention, or coronary artery bypass grafting, regardless of presence or absence of typical cardiac risk factors or clinical presentation. results were disclosed, and physicians were asked what (if any) decisions they would change.

Results: With physicians blinded to BNP results, 24 of 88 (27%) patients were given a primary diagnosis of heart failure, and 18 (20%) patients were given a secondary diagnosis of heart failure. For the former group, disclosure of BNP results (median 709; range 97 to >1,300) resulted in no changes in diagnosis or treatment. For the latter group, disclosure of BNP results (median 495; range 21 to >1,300) caused heart failure to become the primary diagnosis in 4 patients (22%) and led to 5 changes in medical management (28%). Among the 46 patients initially given neither a primary nor secondary diagnosis of heart failure, disclosure of BNP results (median 77; range 1 to >1,300) caused heart failure to become the primary diagnosis in 1 (2%) patient and a secondary diagnosis in 4 (9%) patients and led to 5 changes in medical management (11%).

Conclusion: Among older ED patients who were given a primary clinical diagnosis of heart failure, BNP testing had little impact on medical decisionmaking. However, among other older ED patients with dyspnea, BNP levels were frequently elevated, often influencing medical decisionmaking, particularly when heart failure was in the differential diagnosis.

Elderly Patients Admitted for Sepsis? Lee CC, Crupi RS/Flushing Hospital Medical Center, Flushing, NY Background: It is believed that the incidence of sepsis has continued to rise throughout the past decade. More than 500,000 patients are reported to develop sepsis in the United States each year. Sepsis is defined as the systemic response to infection manifested by 2 or more of the following conditions as a result of infections:

(1) temperature more than 38°C (>100.4°F) or less than 36°C (<96.8°F); (2) pulse rate more than 90 beats/min; (3) respiratory rate more than 20 breaths/min; and (4) WBC count more than 12,000/mm 3 , less than 4,000/mm 3 , or more than 10% immature band forms. Septic shock is defined as sepsis-induced hypotension despite adequate fluid resuscitation, along with the presence of perfusion abnormalities that may include, but are not limited to, lactic acidosis, oliguria, and an acute alteration in mental status.

Study objectives: This study was conducted to determine whether thrombocytopenia is a predictor of future septic shock among patients admitted for sepsis.

Methods: This study is a retrospective medical record review from October 2002 to December 2002 for patients who were admitted with a primary diagnosis of sepsis from the emergency department according to International Classification of Diseases, Ninth Revision codes. A medical record review was done to see which septic patients developed septic shock. A total of 63 patients were identified to have sepsis. All patients received appropriate antibiotics and other supportive measures. Average age was 78.2 years. Male to female patient ratio was 1.4:1.

Results: Twenty-one (33%) of 63 septic patients developed septic shock during hospitalization despite appropriate treatments. Among these 21 patients, 10 (48%) patients were found to have thrombocytopenia. The average platelet counts for these thrombocytopenic patients was 87,000/µL. Lower platelet counts were associated with higher mortality rates. Seven of 10 (70%) patients died despite appropriate treatments.

Conclusion: In this preliminary study, we observed thrombocytopenia as a predictor of future septic shock among elderly patients admitted for sepsis, despite appropriate treatment. Also, we observed that lower platelet counts were associated with higher mortality among septic shock patients.

Morgan CJ, Pyne GJ, Bhadri P, Beyette FR, Majumder A, Caffrey JJ, Wagner KR, Zuccarello M, Clark JF, Jauch ED/University of Cincinnati, Cincinnati, OH Study objectives: The diagnostic paradigm is incomplete for patients presenting with a suspected ruptured cerebral aneurysm. Although the clinical and radiographic characteristics of aneurysmal subarachnoid hemorrhage (SAH) are reasonably defined, no consensus exists for the definition of lumbar puncture findings consistent with SAH. Bilirubin, a marker of erythrocyte catabolism, is produced by the induced heat shock protein heme oxygenase-1. Within 12 hours after SAH, patients develop elevated levels of bilirubin in their cerebrospinal fluid. Quantification of bilirubin in cerebrospinal fluid provides a sensitive, potentially rapid mechanism to detect an SAH and differentiate it from a traumatic spinal tap. An advanced point-of-circadian variation in the incidence of myocardial infarction and death, with peak rates in the 4 hours from noon to 4 PM. Assuming that congestive heart failure (CHF) patients most likely to be intubated are those who are more severely ill, we hypothesized that intubation rates should also peak between noon and 4 PM.

Methods: This was a retrospective analysis of a computerized billing database of ED visits in 15 New Jersey hospital emergency departments (EDs) for consecutive patients treated by emergency physicians from January 1, 1996 , to March 31, 2003 . We compared rates of intubation in 1-hour intervals by using the χ 2 test, with P less than .05 taken as statistically significant. To relate our study to the previous one, we also assessed 4-hour intervals.

Results: There were a total of 3.6 million patients in the database, of whom 50,155 had an ED diagnosis of CHF. Among these patients, we found significant circadian variation in intubation rates (4.0% peak at midnight, 1.1% trough at noon; P<.0001). The 4-hour period between noon and 4 PM had the lowest rates (1.1%) compared with the rest of the day (P<.0001).

Conclusion: We found significant circadian variation in intubation rates, with the peak at midnight. A previous study showed the peak incidence of myocardial infarction and death from noon to 4 PM, the time of lowest intubation rates. Patient characteristics that predispose to intubation may be different from those that heighten the risk of myocardial infarction and death.

Shah K, McGillicuddy D, Friedberg R, Nathanson L, Edlow J/St. Luke's-Roosevelt Hospital, New York, NY; Beth Israel Deaconess Medical Center, Boston, MA Study objectives: We determine the sensitivity of the current standard for synovial fluid leukocytosis analysis as defined by the American Rheumatism Association (WBC count >50,000/mm 3 ) in diagnosing infectious arthritis.

Methods: This was a retrospective study conducted at an urban, university, tertiary care referral center with 50,000 annual adult emergency department visits. The study population includes all patients with infectious arthritis confirmed by synovial fluid cultures with growth of a pathogenic organism. The study period was from January 1996 to December 2002. The synovial fluid culture, Gram's stain, and leukocyte count test results and the discharge diagnosis were obtained. We used a WBC count of 50,000/mm 3 to set the cutoff of low versus high WBC count in infectious arthritis. All positive cultures that were contaminants were eliminated from the study. Fisher's exact test was used to compare proportions, and sensitivity and means were reported with 95% confidence intervals (CIs).

Results: There were 49 culture-positive synovial fluid aspirates in the 6-year study period. Although an aspirate with WBC count less than 50,000/mm 3 is not considered to represent an infection, there were 19 (39%) synovial aspirates with a WBC count less than 50,000/mm 3 that grew a pathogenic organism on culture. Thus, a WBC count more than 50, 000/mm 3 was only 61% (95% CI 48% to 75%) sensitive in diagnosing septic arthritis. Among the 49 patients, 41 (88%) had an underlying medical condition, including previous surgery, diabetes mellitus, and dialysis. When specific medical conditions were compared between groups with low versus high WBC count, there was no statistically significant relationship.

Conclusion: In our study, a WBC count less than 50,000/mm 3 lacked the sensitivity required to be clinically useful in ruling out infectious arthritis. The study also showed a high incidence of underlying medical conditions in patients with infectious arthritis. This study shows that the WBC count is a poor test to determine infectious arthritis and that a prospective study to find better predictors of septic arthritis may be valuable for clinical decisionmaking.

likely overestimating the incidence of negative hematuria. A prospective study would be better equipped to answer the question of the incidence of hematuria in renal colic. However, using renal protocol spiral CT as the criterion standard for diagnosis would require subjecting certain patients to unwarranted radiation (ie, patients noted to be excluded by our study, the young patients with "typical" recurrent pain, who come to the ED for pain control and antiemetics). Moreover, there is not a uniform agreement on what constitutes a normal number of RBCs in microscopic analysis or how the urine sample should be handled before being received in the laboratory. Observational differences between dip readings are well documented as well.

Conclusion: Urinalysis does not appear to be a suitable screening tool for diagnosis of renal colic.

Shenbagamurthi S, McGraw P, O'Connor R/Christiana Care Hospital, Newark, DE Study objectives: Patients' blood is routinely drawn for cultures before initiation of antimicrobial therapy in the emergency department (ED). We conducted this study to determine the diagnostic yield of blood cultures drawn in the ED and whether culture results led to alteration in selection of empiric antimicrobial agent.

Methods: Patients presenting to the ED with acute febrile illness (temperature >38.5°C [>101.3°F]) were eligible for the study. Patients with blood drawn for cultures in the ED were enrolled, and all patients had 2 sets of blood drawn. Blood cultures were classified as positive or negative according to whether a bacterial pathogen was identified. Contaminated specimens were excluded. Patient records were reviewed to ascertain whether the empirically selected antimicrobial agent was changed because of culture results. Statistical analysis was performed using 95% confidence intervals (CIs).

Results: A total of 330 patients had blood drawn for culture, with 39 (12%; 95% CI 10% to 14%) cultures showing bacterial pathogen growth. Age range was 25 days to 96 years. A mean of 2.9 days elapsed between the time the culture was obtained until it was reported. A total of 15 of the positive cultures (7%; 95% CI 5% to 9%) resulted in change in antibiotic selection. In 5 cases, antibiotics were changed because of poor clinical response and before culture results were known. The remainder resulted in no change in therapy. The most commonly recovered pathogen was Staphylococcus aureus (in 10 cases), followed by Streptococcus. Three of the 10 S aureus strains recovered were methicillin resistant. Average charge per culture was $125. Cost per positive culture was $2,115, and cost for each positive culture resulting in change in therapy was $5,500.

Conclusion: Blood drawn in the ED for cultures has relatively low yield, with less than 40% of positive cultures resulting in change in antimicrobial therapy. We recommend that the practice of routinely obtaining blood for cultures before initiation of antibiotics be called into question.

Brown MD, Kline JA, Nelson RD, Lau J/Carolinas Medical Center, Charlotte, NC; Michigan State University, Grand Rapids, MI; Tufts-New England Medical Center, Boston, MA Study objective: We evaluate the diagnostic performance characteristics of the latex turbidimetric D-dimer test in the diagnosis of pulmonary embolism (PE) in the adult emergency department (ED) population.

Methods: A search of MEDLINE, EMBASE, and bibliographies of previous systematic reviews was conducted with no language restriction. Experts in the field of PE research were contacted to identify unpublished studies. Prospective investigations involving a predominantly outpatient population with suspected PE and using a turbidimetric D-dimer test were included. Two authors extracted data independently and assessed study quality according to the composition of the patient spectrum and the reference standard used. Consensus was reached by conference. The analysis was based on a summary receiver operating characteristic curve and combining sensitivity and specificity independently across studies by using a random-effects model.

Results: The search yielded 264 publications and 2 unpublished studies. Nine studies met the inclusion criteria and provided a sample of 1,901 subjects. Eight of the 9 studies were homogeneous in terms of sensitivity and specificity. One study has similar sensitivity but higher specificity. Combining the studies yielded an overall sensitivity of 0.93 (95% confidence interval [CI] 0.89 to 0.96) and an overall specificity of 0.51 (95% CI 0.42 to 0.59).

Conclusion: The turbidimetric D-dimer test is sensitive but nonspecific for the detection of PE in the ED setting. D-Dimer tests using latex turbidimetric methods care spectrophotometer is being developed that will measure bilirubin at the patient's bedside.

Methods: In this porcine model, we injected whole arterial blood into the cisterna magna to induce an SAH, which was volumetrically equivalent to a typical human sentinel SAH. Serial samples of cerebrospinal fluid from the cisterna magna and lumbar cistern were collected hourly during 24 hours and assayed for bilirubin. Four animals underwent an SAH, and 3 control animals were injected with normal saline solution volumes equal to the SAH. Using a diazochemical assay to measure bilirubin levels, we compared the assay values with the spectrophotometer's prediction of bilirubin levels after application of our algorithm.

Results: After a 12-hour lag, bilirubin levels in the cisterna magna and lumbar cistern of the SAH animals showed consistent and reproducible elevations. At 24 hours after injection, cisterna magna bilirubin levels averaged (±SD) 7.29±1.33 µmol/L in the SAH animals versus 1.33±0.14 µmol/L in control animals. Lumbar puncture bilirubin levels (±SD) at 24 hours were 4.38±1.04 µmol/L in SAH animals and 1.02±0.05 µmol/L in control animals. Concentrations of bilirubin reported in the chemical assay were comparable with results supplied by the spectrophotometric algorithm.

Conclusion: Our results suggest bilirubin levels are significantly elevated 12 hours after occurrence of an SAH. This time lag likely results from the requisite induction of heme oxygenase-1 and is not problematic clinically because it occurs during the ultrasensitive phase of computed tomography. Bilirubin's production lag is precisely the value of this diagnostic modality. Thus, blood introduced during a traumatic tap will not have sufficient time for the catabolic conversion to bilirubin. Assimilation of a rapid, quantitative analysis for bilirubin as a point-of-care technique will complete the diagnostic paradigm for SAH and allow emergency physicians to better direct patient care.

Osgood JG, Aubin C/Barnes-Jewish Hospital, St. Louis, MO Study objective: Flank pain caused by urinary tract stones is a common presentation to the emergency department (ED). Newer-generation spiral computed tomography (CT) has replaced intravenous pyelogram in the investigation of renal colic. Although several recent studies seem to suggest that hematuria is not a good predictor of ureterolithiasis, we decided to examine the incidence of hematuria at our institution by using renal protocol spiral CT as the criterion standard for stone detection.

Method: This was a retrospective medical record review. We searched radiology computer files for "stone" and "lithiasis" and then reviewed records of patients with positive scan results; we then recorded and tabulated the urinalysis results. Institutional review board approval was obtained before data collection.

Results: This search technique during a 6-month period yielded 91 patients with a positive CT scan and a recorded urinalysis, including a dip or urine microscopy. One patient had no microscopy recorded, and 2 patients had no dip recorded. Overall, of 89 patients with microscopy recorded, 10.1% (9 of 89) had 0 RBCs per high-powered field. Twenty-seven percent of patients had 3 or fewer RBCs per high-powered field. When urine dip was used, 23.3% of patients were recorded as negative (21 of 90).

Discussion: Renal stones are thought to affect roughly 2% to 10% of people in industrialized nations. In the United States, about 5% of women and 10% of men are affected in their lifetimes, with recurrence rates approaching 50% in 10 years. Annual costs exceed 1.8 billion dollars. Although most stones pass spontaneously, some patients require hospitalization for parenteral pain control, intravenous hydration, severe obstruction, or infection. For emergency physicians, management is usually less an issue than diagnosis because the differential for flank pain is diverse and includes other entities with high morbidity and mortality, such as aortic aneurysms and dissections, pancreatitis, and colitis including ischemia, appendicitis, and pyelonephritis. As technology has advanced, spiral computed axial tomography scan has replaced other modalities for diagnosing urinary tract stones because of its high sensitivity and its ability to visualize other intra-abdominal pathology. The radiation dose is not trivial, however, and given the recurrence rates of nephrolithiasis, many practitioners continue to use urinalysis as a screening tool for kidney stones. Researchers have recently published studies showing that hematuria is not a positive predictor for urinary tract stones, and our data appear congruent with their findings. This study is limited by several factors. The sample size is small but the results are clear and consistent with other studies. Selection bias is a problem too. In this study, a retrospective medical record review was done, selecting for positive scan results then including patients with a documented urinalysis, which excluded some patients, especially younger patients presenting with "typical" recurrent pain and hematuria, thus Study objectives: To the authors' knowledge, this is the first assessment of a urine trypsinogen bedside test in the diagnosis of pancreatitis in the emergency department (ED) using a non-Scandinavian population.

Methods: This was a prospective, double-blind study of patients presenting with abdominal pain or symptoms suspicious for pancreatitis. A urine trypsinogen bedside test (Actim Pancreatitis), which takes 3 to 5 minutes, was compared with serum lipase measurements and final clinical diagnoses. Clinicians were blinded to the results of the urine trypsinogen test and made the final diagnosis according to a combination of laboratory test results, radiology results, and the patient's hospital course.

Results: Of 192 patients included in this study, 17 patients were diagnosed with either acute or chronic pancreatitis, all of whom had positive urine trypsinogen test results. Another 7 patients had positive urine trypsinogen test results but were subsequently determined not to have pancreatitis. The sensitivity, specificity, positive predictive value, and negative predictive value of the urine trypsinogen test for the final diagnosis of pancreatitis were 100% (95% CI 77.1% to 100%), 96% (95% CI 91.6% to 98.2%), 70.8% (95% CI 48.8% to 86.6%), and 100% (95% CI 97.2% to 100%), respectively.

The predictive value for a positive urine trypsinogen test and the final diagnosis of pancreatitis was 0.81 (95% CI 0.67 to 0.95) compared with 0.61 (95% CI 0.41 to 0.81) for a serum lipase measurement of at least 120 µ/L. Four (24%) of the 17 patients with pancreatitis had diagnostic findings on computed tomography and positive urine trypsinogen tests but normal serum lipase levels.

Conclusion: The urine trypsinogen bedside test is quick and likely to be helpful in making an accurate diagnosis of pancreatitis in the ED.

Han SK, Song KJ, Song HG, Jeong YK/Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Korea Study objectives: In emergency situations, it is crucial to confirm the position of the endotracheal tube. Guidelines 2000 recommends end-tidal CO 2 and the Esophageal Detector Device as secondary confirmatory methods. Therefore, we studied the availability of portable sonography in secondary confirmation of the position of the endotracheal tube.

Methods: We used portable sonography (SonoSite 180 PLUS) to evaluate 15 patients who were admitted to the emergency department in September 2002 and were intubated endotracheally, which was confirmed by chest radiograph and other secondary confirmation methods. A linear probe (10-5 MHz) was placed on the cricothyroid membrane to search for the vocal cord and confirm the passage of endotracheal tube. Three deceased patients were intubated esophageally and used as a control group by comparing the images.

Results: We were able to obtain a double meniscus sign across the vocal cords in 12 of the 15 patients who had endotracheal intubation confirmed by chest radiograph and other secondary confirmation. Three patients did not produce a definite image. We were not able to obtain a double meniscus sign from 3 patients who were intubated esophageally.

Conclusion: The advantages of sonography as a method of secondary confirmation of endotracheal intubation are visualization of the endotracheal tube and accessibility of follow-up evaluation. Therefore, if further study is conducted, usage of sonography to confirm the position of endotracheal intubation may be added to the group of secondary confirmation methods used today.

critical patient requiring an airway. Most of the currently accepted methods to confirm proper tube placement have limitations in their use, particularly in patients who are in cardiopulmonary arrest. We describe a method of determining correct endotracheal tube placement by using ultrasonography in a pig model.

Methods: After a 30-minute demonstration of ultrasonography use and landmarks used to confirm endotracheal tube placement, novice users determined endotracheal tube placement in 8 newly killed pigs randomized to tracheal or esophageal placement of a 6.0 endotracheal tube. The animals were rerandomized for each user. Intubations were performed by the authors and confirmed by direct visualization and an esophageal detector device. For ultrasonography determination of endotracheal tube location, we used a GE Logic Pro 400 ultrasonography machine with a 6.6-MHz linear probe. We recorded endotracheal tube location, time, level of training (medical student, postgraduate year), users' confidence level, and the ultrasonographic signs that influenced their interpretation.

Results: Thirty-three users completed the course and completed 89 ultrasonographic examinations. They correctly identified the endotracheal tube placement 92% of the time and incorrectly 8% of the time (82 of 89 correct). Sensitivity and specificity of ultrasonography for endotracheal tube placement was 96% and 88%. The mean time to interpretation was 32.4 seconds in the correct group and 60 seconds in the incorrect group (P<.05). Level of training and confidence of interpretation did not correlate with proper interpretation (P=not significant). There was no single ultrasonographic sign identified by the users that predicted successful interpretation.

Conclusion: Novice users can reliably predict the correct placement of an endotracheal tube by using ultrasonography in a porcine model. Those who successfully identified endotracheal tube placement spent significantly less time making their decision.

Chen YY, Kim MC, Ko YG, Baik HH, Cho YH/KyungHee University, Seoul, Korea Study objectives: It is now well recognized that reperfusion of ischemic tissues initiates a complex series of reactions that can paradoxically injure tissues. Apoptosis occurs in select cell populations during morphologic development and during cellular injury, including oxygen radical exposure, ischemia-reperfusion, and sepsis. Thus, in this study, we examine the relation of the melatonin effect to the injection time and the dose, and we examine the role of melatonin in apoptosis.

Methods: Intestinal ischemia-reperfusion injury was induced in rats by clamping the superior mesenteric artery for 30 minutes. After reperfusion injury for 30 minutes, the experimental group was administered melatonin (10 mg/kg) intraperitoneally, and the control group received saline solution and ethanol. At 30 minutes, 60 minutes, and 90 minutes, pulmonary histologic assessments (interstitial polymorphonucleocytes/10 high-power fields and lung [alveolar] injury score), alveolar microvascular permeability assessments (wet-weight to dry-weight ratio and lipid peroxidation activity, malondialdehyde, malondialdehyde), and Western blotting assessments (p53, p21, Bax, and Bcl-2) were made. For comparison, long-term (60-minute) reperfusion and double-dosage melatonin (20 mg/kg) were also studied.

Results: The lung injury score was 1.00±0 in the melatonin group at 90 minutes and 3.28±0.30 in the saline solution group (P<.01). The number of sequestered neutrophils was significantly higher in the control group at 90 minutes (34.38±16.76/10 HPFs) than in the melatonin-treated group (5.63±2.73/10 HPFs; P<.01). In the melatonin group at 90 minutes, the wet-weight to dry-weight ratio was 4.69±0.16, and in the saline solution group, the ratio was 4.78±0.17 (P>.05). A marked difference was found between the ischemia-reperfusion control group and the experimental group at 90 minutes for lipid peroxidation activity (malondialdehyde, 16.45±0.19 µmol/L versus 10.93±0.11 µmol/L; P<.01). In the melatonin group, p21, p51, bcl-2, and Bax expressions were found to be much higher than in the control group.

Conclusion: Melatonin injection within 60 minutes after reperfusion may promote recovery of reperfusion injury, but double-dose melatonin injection is inefficacious. Also, melatonin inhibits apoptosis by p21 expression.

Lewin MR/University of California San Francisco-Fresno, Fresno, CA Study objectives: Nociception is the physiologic response to incipient or actual tissue damage and is a critical element of the subjective experience of pain. Nociceptive sensitization is an enhanced sensory response after noxious stimulation. Biochemical

Methods: An observational study was conducted in an urban emergency department (ED) from February to April 2001, with a convenience sample of ambulatory volunteers (ED patients and ED staff) who had no history of urologic problems. Three emergency medicine residents received 0.5 hour of training: imaging and measuring the bladder with ultrasonography on supine patients by using a 2.5-MHz annular array probe. Anteroposterior and lateral bladder dimensions were measured with electronic calipers from the largest transverse image obtained. Patients and volunteers then voided into a calibrated urinal, and this volume was recorded. Bladders were reimaged to confirm complete emptying. Bladder volume was estimated from the transverse image only by using the following derived equation: anteroposterior dimension (centimeters) × lateral dimension (centimeters) × 6. Estimated volume was then compared with the measured volume.

Results: Fifty-nine patients were enrolled by 3 emergency medicine residents. Patients were randomized to derivation (N=31) and validation (N=28). Median voided volume was 300 mL (range 75 to 1150 mL). Comparing the estimated measurements from the derived formula with actual voided volume resulted in a correlation coefficient of 0.93 and a difference between voided and estimated volumes with a mean of 41.1 mL (95% confidence interval -11.7 to 93.9 mL).

Conclusion: The simplified formula derived from regression analysis is easily learned. Estimated bladder volumes were accurate within clinically relevant parameters.

Reynolds B, Blaivas M, Kuhn W, Brannam L/Medical College of Georgia, Augusta, GA Study objective: Physicians practicing outdoors in remote areas are typically significantly limited in their choice of diagnostic tools. We determine whether the use of a portable ultrasonography device on select patients in a remote setting would alter physician diagnosis and management.

Methods: This was a prospective observational study of the effects ultrasonography use has on physician decisionmaking in this setting. A battery-operated Sonosite 180 Plus with 2 interchangeable transducers (4-to 7-MHz broadband intercavitary transducer and 2-to 5-MHz broadband abdominal transducer) was transported to a remote field clinic deep in the Amazon jungle. The patient population consisted of Nomatzieguenga and Ashaninca tribal people in the Zona Amazona (Amazon jungle). Two of the physicians on the team performed ultrasonographic examinations. Physicians requesting ultrasonographic examinations filled out a survey before and after the ultrasonographic examination. Before the ultrasonographic examination, the referring physician filled out a survey describing the patient's initial complaint, pertinent medical history and physical findings, and an initial (pre-ultrasonographic examination) differential diagnosis and planned treatment with expected disposition. After the results of the ultrasonographic examination were reviewed with the referring physician, the referring physician was asked to fill out the remainder of the survey, allowing comparison of pre-and post-ultrasonographic examination differential diagnosis, treatment plan, and disposition. Ultrasonography performance by the team is standard on these missions.

Results: A total of 10 ultrasonographic studies were performed, including a trauma ultrasonographic scan, hepatobiliary study, transabdominal pelvic study, and transvaginal pelvic study. The monitor on the ultrasonography unit experienced a rare failure shortly after being used at 17,000 feet above sea level and then 10 times at sea level, and no further ultrasonographic scans could be performed. Results of the trauma ultrasonographic scan and a transvaginal pelvic study drastically altered the disposition of 2 patients by avoiding a more than 1-day hike to the nearest permanent clinic.

Conclusion: Even when used in a remote location, portable ultrasonography provides a significant benefit that can drastically alter disposition and treatment, at least in trauma and obstetric patients. Study objectives: In fiscal year 1995, 148,000 US citizens died and 2.6 million were hospitalized for more than 1 week, at a loss of $260 billion, from trauma-, sepsis-, and burn-related syndromes. This mortality and morbidity rate is dramatically reduced from 30 years ago because of improved patient transportation and resuscitation but includes no reduction in morbidity and mortality of patients surviving the initial resuscitation rate. These patients are still developing systemic inflammatory response syndrome (SIRS) and multiple organ dysfunction syndrome (MODS) at the same rate, despite multiple clinical ICU-based trials of single-agent biologic response modifiers. To date, therapy directed to prevent or improve MODS/SIRS by modulating the host's endogenous inflammatory response has had only limited success. Only a finite window of time may exist in which treatments for prevention of posttrauma MODS will be therapeutic. Postinjury monocyte (Mφ) activation and increased production of proinflammatory cytokines (tumor necrosis factor-α, interleukin-6) and other mediators are repeatedly linked to development of MODS. Concomitantly, to these aberrant inflammatory responses, depressed Mφ immunostimulatory cytokine production and diminished antigen presenting function occur in MODS. This seemingly paradoxic overactivation and depressed activation of trauma patients' Mφ results from an unbalanced Mφ differentiation toward inflammatory macrophage (Mac) and away from immunostimulatory dendritic cells (DC) . Alterations in Mφ differentiation capacity have been linked to altered posttrauma Mφ receptor expression and consequent altered signal transduction, as well as aberrant Mφ gene activation. In particular, endogenous prostaglandin (EP) E 2 (PGE 2 ) stimulation alters Mφ receptors. There is differential expression and disparate activation of the EP 1/3 versus EP 2/4 subtypes of PGE 2 receptors on Mφ. High and low PGE 2 levels differentially activate EP 1/3 versus EP 2/4 . These EP receptors' different signaling pathways could differentially affect Mφ functions, differentiation patterns, and receptor expression. Evidence suggests that altered EP receptor signaling may affect the expression of toll-like receptors (TLR2/4), which control Mφ responses to bacterial, fungal, and viral pathogens. We hypothesized that early increases in posttrauma PGE 2 levels alter Mφ EP receptor expression and Mφ TLR expression, as well as Mφ differentiation receptor expression, leading to increased inflammatory Mac responses and decreased differentiation to immunostimulatory DC. We further postulated that these EP receptor changes may occur in the first hours after injury when patients are in the emergency department.

Methods: Trauma patients with an Injury Severity Score more than 15 gave informed consent, and 30 mL of blood was drawn. Peripheral blood leukocytes were isolated by Ficoll-Hypaque separation, and Mφ was separated by magnetic bead negative isolation. Normal control peripheral blood samples collected from consenting laboratory and staff volunteers were simultaneously processed as a systems control. Mac and iDC were generated from fresh Mφ by using macrophage colony-stimulating factor and granulocyte-macrophage colony-stimulating factor/IL-4, respectively. These Mφ, Mac, and iDC were also stimulated with high-dose (10 -6 ) and low-dose (10 -8 ) PGE 2 . Isolated Mφ/Mac/immunostimulatory dendritic cells (iDC) and EP 1-4 and TLR2/4 receptor expression was then assessed by 4-color flow cytometry on a dual-laser BD flow cytometer. The following markers were used for the various cell subtypes: Mφ (CD66b, CD11c, CD33, and CD14), Mac (CD64, CD163, CD14 and MRP14/8), and iDC (CD86, CD209, CD14, CD1a) by using the fluorescent labels fluorescein isothiocyanate, phycoerythrin, peridin chlorophyll protein, and antigen presenting cell, which each emit in a distant spectrum. The cells were concomitantly stained with the aforementioned EP 1-4 and TLR 2/4 receptors by using a biotinstrepavidin indirect label to amplify the marker expression and to tag the antibodies not easily directly fluorochrome labeled. Appropriate isotype antibodies served as controls. EP 1-4 and TLR 2/4 receptor protein expression was also assessed by singleand double-label immunocytochemistry/immunofluorescence, and messenger RNA was evaluated by reverse transcriptase-polymerase chain reaction (RT-PCR).

Results: Because of a presumably low abundance of the prostaglandin E2 receptor only EP4 expression was detectable on monocytes by flow cytometry. However, single-and double-label immunocytochemistry/immunofluorescence revealed coexpression of the EP3R and TLR4, predominantly in the Mac population. iDC expressed the EP1 and EP2 receptors. Mφ from patients and normal individuals were also assessed for modulation of TLR2/4 by low or high concentrations of PGE 2 . High mechanisms regulating nociceptive sensitization may be highly conserved in evolution. Thus, invertebrates may have utility for the study of pain mechanisms important to vertebrates. The objective of this study was to assess the viability of an invertebrate model system to probe fundamental biochemical and physiologic systems underlying short-and long-term responses to noxious stimuli.

Methods: Larval Manduca sexta (aka, tomato hornworm, endemic throughout the United States) were raised from eggs, on artificial diet, in aerated 10-and 20-qt Rubbermaid containers. Baseline sensitivity of the fifth-instar larval M sexta prolegs were assayed with calibrated von Frey hairs. After a standardized noxious stimulation protocol, changes in the threshold for the larval head-strike (defensive) reflex were assayed after 5 and 10 minutes and then after 10 and 24 hours. Baseline characteristics of untreated animals studied to optimize the assay included growth versus baseline sensitivity curves as a function of time. All animals were used for at least 2 time points. Data are expressed as mean±SD, and comparisons of prestimulation and poststimulation striking thresholds were made by using paired, 2-tailed t tests. P less than .05 was considered sufficient to reject the null hypothesis. All experiments were conducted in the author's kitchen, and every Manduca larva was released into the tomato garden after its last test.

Results: The mean prestimulation threshold was 2.3 g±1.8 g (n=38 animals). Poststimulation thresholds were as follows: at 5 minutes 0.06 g±0.06 g (n=14); 10 minutes 0.05 g±0.05 g (n=10); 20 minutes 1.1 g±1.7 g (n=26); 10 hours 1.0 g±1.1 g (n=25); and 24 hours 1.5 g±1.2 g (n=7), with P=.43 at 24 hours. However, P values for the 5-, 10-, and 20-minute time points and that at 10 hours were all less than .001.

Conclusion: Larval M sexta can be sensitized for at least 10 hours. Average strike thresholds were reduced up to 50-fold in the short-term experiments and by more than twofold in the long-term experiments. These findings, along with the ease of assaying the nociceptive response (defensive head strike), strongly suggest that this animal can be used to probe hypotheses underlying physiologic responses to noxious stimuli and, potentially, to screen novel analgesic agents. Vertebrate animal models used to investigate nociception and analgesia are exceedingly expensive and highly regulated and require large, specialized facilities. Caterpillars are not expensive, are easy to raise, and require little space. In addition, their use in the laboratory is not regulated by the National Institutes of Health. Thus, they may develop into a convenient and powerful, yet inexpensive, tool to aid our understanding of critical aspects of the most common complaint in the emergency department: pain.

Dickson EW, Bird SB, Gaspari RJ/University of Massachusetts Medical School, Worcester, MA Study objectives: Mortality from severe organophosphate poisoning is the result of acute respiratory failure. Present evidence suggests that 2 major mechanisms are involved in this process: bronchoconstriction or bronchorrhea-induced airway obstruction (peripheral cholinergic effects) and central nervous system-induced respiratory depression. The objective of the current study was to determine which mechanism predominates in the setting of acute organophosphate poisoning with dichlorvos.

Methods: Respiratory effort (negative pleural pressure generated×respiratory rate) was measured by placement of a pleural catheter in Wistar rats (n=9). After mean baseline respiratory effort was established, animals received pretreatment with either intraperitoneal normal saline solution (controls) or the peripherally acting anticholinergic glycopyrrolate (1 mg/kg). Five minutes after pretreatment, all animals were poisoned with 100 mg/kg intramuscular dichlorvos. Changes in respiratory effort (percentage of baseline) at 2 and 4 minutes after poisoning were then compared by 2-way analysis of variance (P≤.05) Results: Baseline respiratory effort was similar between the 2 groups. Two minutes after poisoning, there was no significant change in respiratory effort compared with baseline for either cohort. By 4 minutes, all control animals exhibited profound respiratory depression, with respiratory effort measuring only 21%±18% of the baseline values. Although treatment with glycopyrrolate prevented the decrease in heart rate and mean blood pressure seen in controls, it did not attenuate the observed decrease in respiratory effort (3%±1%). The decrease in respiratory effort occurred despite adequate perfusion (mean blood pressure 90 mm Hg at 4 minutes) and oxygenation (mean PaO 2 313 mm Hg on 100% oxygen at 4 minutes).

Conclusion: Central respiratory depression is the predominant mechanism of mortality in acute poisoning with the organophosphate dichlorvos. temperature at 37°C (98.6°F) rectal. Sham animals underwent instrumentation but did not undergo blood withdrawal or carotid clamping and were maintained under anesthesia for 30 minutes before being killed. Treatment-group animals received intravenous insulin (20 U/kg) on removal of carotid clamps. Blood glucose level was maintained at 100 to 200 mg/dL. Brain sections (20 micron) were taken to immunofluorescence by using antibody against Akt-Ser 473(P). Photomicrographs were taken on a Leica DMLB fluorescence microscope with a Nikon 990 digital camera.

Results: Sham-operated animals demonstrated high levels of phosphorylated Akt in neurons of the cortex and dentate gyrus, with lower levels in the vulnerable cornu ammonis (CA) 1 and hilus regions of the hippocampus. By 30 minutes of reperfusion, there were no changes of phosphorylated Akt in CA1 neurons compared with that of sham-operated controls, but there were much lower levels in cortex. Animals treated with high-dose insulin demonstrated increased cytoplasmic and nuclear staining for phosphorylated Akt in cortex and CA1. These effects of insulin were greatly attenuated by the PI3K inhibitor wortmannin.

Conclusion: Our results indicate that vulnerable neurons of the CA1 have low baseline levels of phosphorylated Akt compared with other neuronal populations, that insulin induces phosphorylation of Akt in reperfused neurons by 30 minutes, and that this response is PI3K dependent. Further study will be needed to delineate which downstream substrates of Akt play a role in the salutary effects of insulin in the setting of cerebral ischemia.

Chen G, Han J, Li J, Yang H, Wang H, Ward MF, Sama AE, Tracey KJ, Wang H/New York University School of Medicine, New York, NY; North Shore University Hospital, Manhasset, NY; Long Island Jewish Research Institute, Manhasset, NY Background: Gram-negative bacterial infection and systemic inflammation are widespread problems in critically ill patients. A product of gram-negative bacteria, endotoxin (lipopolysaccharide [LPS] ), stimulates macrophages to release excess amounts of proinflammatory cytokines (such as tumor necrosis factor [TNF] and interleukin-1) during overwhelming infection. If delivered early enough, anti-TNF therapy can be effective in animal models of systemic inflammation, but early treatment is difficult to achieve in humans. An alternative therapeutic strategy would be to identify "late" macrophage mediators that may be clinically more accessible. We recently discovered that a ubiquitous protein, HMGB1, is released late by activated macrophages and functions as a late mediator of endotoxin lethality. Fetuin (fetus protein in Greek) was first identified by Pederson as a major fetal plasma protein more than 50 years ago and subsequently characterized as a liver-derived negative acute phase protein.

Study objectives: We elucidate the roles of fetuin in the regulation of innate immune response, and we examine the effect of fetuin on endotoxin-induced HMGB1 release in cultures of murine macrophage-like RAW 264.7 cells and human peripheral blood mononuclear cells.

Methods: Macrophage-like RAW 264.7 cells or human peripheral blood mononuclear cells were cultured in low-serum OPTI-MEM I medium until 80% to 90% confluence and stimulated with LPS (100 ng/mL) either alone or in the presence of fetuin at various concentrations (0, 10, 25 , and 100 µg/mL). At 16 hours after LPS stimulation, the levels of TNF, HMGB1, and nitric oxide in the culture medium were determined.

Results: Fetuin dose-dependently attenuated LPS-induced release of HMGB1, TNF, and nitric oxide by macrophage cultures, with a maximal suppression by 45%±10%, 65%±12%, and more than 95%±5% for TNF, nitric oxide, and HMGB1, respectively. Similarly, it promoted a dose-dependent inhibition of LPS-induced release of TNF and HMGB1 in cultures of human peripheral blood mononuclear cells.

Conclusion: Fetuin occupies an important role in the regulation of the innate immune response in endotoxemia. Supplementation of exogenous fetuin holds potential as a therapeutically effective treatment of lethal systemic inflammation (eg, endotoxemia, sepsis).

Younger JG, Valencia GA, Shankar-Sinha S/University of Michigan, Ann Arbor, MI Study objectives: Bacterial surface carbohydrates are important pathogenic factors during gram-negative pneumonia. Among these, O-antigen has been reported to protect pathogens against complement-mediated killing.

Methods: To examine further the role of O-antigen, we insertionally inactivated

PGE 2 concentrations decreased Mφ to iDC TLR2/4 expression. However, low PGE 2 concentrations failed to modulate TLR4 expression on patients' Mφ, which had an Mφ-to-iDC differentiation defect. In addition, RT-PCR for the EP1-4 and the TLR2/4 was performed in a patient who initially had no Mφ-to-iDC defect but later developed this defect, thereby succumbing to MODS/SIRS. In this manner, genetic differences in receptor expression between individuals were eliminated. Data revealed a significant increase in EP4 and TLR4 messenger RNA when this patient developed a Mφ-to-iDC defect.

Conclusion: This study is unique in its linking of postinjury altered EP receptors to altered effects of PGE 2 on other Mφ receptors and to aberrant Mφ differentiation. One of this study's primary purposes is to determine whether a finite early window of opportunity for posttrauma therapeutic intervention for MODS exists in the emergency department. The experiments proposed could lead to a definition of mechanisms triggering posttrauma MODS, to designing early modulatory therapy, and to a possible means of rapid flow cytometric identification of patients at risk of developing MODS.

Traylor J, Roming M, Hsu CK/The Brooklyn Hospital Center, Weill College of Medicine, Cornell University, New York, NY Background: Ultrasonography is available in the emergency department (ED) to assist in rapid bedside diagnosis. Investigations have shown that bedside ultrasonographic imaging may confirm the diagnosis of a depressed skull fracture (DSF). Imaging was acquired by using ultrasonography gel to fill the cavity created by the DSF.

Study objectives: We determine whether ultrasonography conductive gel is a safe and representative substitute for blood or clot in investigative or training protocols.

Methods: Eight porcine heads were harvested postmortem. The skin of each was reflected. Holes (1.5 cm) were drilled, for a total of 21 holes. Each skull fragment was depressed 1 cm onto the brain. Each site was filled with ultrasonography gel, clotted blood, or unclotted blood, and the skin flap was replaced. Using a 10-MHz probe, an operator familiar with sonography for DSF but blinded to the conductive medium in the DSF cavity obtained the best image described as an echogenic column. Each of the 21 holes was randomly filled with alternate conductive media and then reevaluated with ultrasonography, for a total of 63 images.

Results: Each set of 3 images (N=63) for each DSF was ranked 1, 2, or 3, with 3 being the best image. The average score and SD were calculated as follows: clotted blood 2.00 (SD 0.71); unclotted blood 1.86 (SD 0.854); ultrasonography gel 2.14 (SD 0.91).

Conclusion: Ultrasonography gel placement in the wound cavity of a model for depressed skull fracture appears to be a representative substitute for unclotted or clotted blood without the risk of blood-borne pathogen transmission.

Sullivan JM, Murariu-Dobrin AC, Sanderson TH, Owen CR, Page AB, Krause GS/Wayne State University/Detroit Receiving Hospital, Detroit, MI Study objectives: Growth factors, including insulin, affect salvage of vulnerable neurons in the setting of global brain ischemia. We have shown that high-dose insulin administered after global brain ischemia induces dephosphorylation of eukaryotic initiation factor 2-α (eIF2-α) and restoration of protein synthesis in selectively vulnerable brain neurons. The serine-threonine kinase Akt plays a central role in survival signaling and translational control, and its activation by growth factors is mediated by phosphatidylinositol-3-kinase (PI3K). We used immunofluorescent microscopy to localize phosphorylated Akt and to investigate the effect of wortmannin, a PI3K inhibitor, on this signaling system in the setting of transient global brain ischemia.

Methods: Male Long Evans rats weighing 400 to 500 g were anesthetized with ketamine and xylazine. Intracerebroventricular injection of 5 µL of 10 µmol/L wortmannin in 0.1% dimethyl sulfoxide or vehicle alone was performed. Animals were intubated, instrumented for vascular access and hemodynamic monitoring, and subjected to global brain ischemia by withdrawal of blood to achieve a mean arterial pressure of 30 to 35 mm Hg. The carotid arteries were occluded with aneurysm clips for 10 minutes. Blood was returned to restore normal mean arterial pressure, and the animals were reperfused for 30 minutes. A homeothermic pad maintained body there were no significant differences found in the AstroTurf field with 2.5 cm of padding and the FieldTurf field, there were significant differences in the FieldTurf surface at different sites on the field (F=83.57, df=4,15, P<.0001) . The g force experienced on the FieldTurf field at the 30-yard line was 238, whereas on the old AstroTurf field, the same spot produced a g force of 188. Conclusion: The indoor practice field used by a professional football team was changed from AstroTurf with 2.5 cm of padding to FieldTurf in an attempt to improve impact attenuation. Overall, there was no significant difference in impact attenuation; however, there were areas within the field that were significantly compacted, causing some spots on the field to be much harder than the AstroTurf surface, which is presumably caused by the compacting of the shredded rubber on which the field is based.

Chan TC, Clausen J, Neuman T, Eisele JW, Vilke GM/University of California-San Diego, San Diego, CA; Forensic Pathology and Medicine, San Diego, CA Background: Violent, combative patients often require physical restraint by emergency department, hospital, out-of-hospital, and law enforcement personnel. Concern has been raised that weight force, commonly applied during the restraining process, can compromise respiratory function, placing individuals at risk for asphyxiation.

Study objectives: We sought to determine the impact of weight force on pulmonary and respiratory function in a simulated restrained patient. We hypothesized that weight force would result in changes in pulmonary and respiratory characteristics.

Methods: Ten volunteers completed a randomized crossover, controlled trial in a pulmonary function laboratory. Subjects were placed in the sitting, prone restraint (PR), PR with 25-lb (11.25 kg) sandbag weight (PR+25), and PR with 50-lb (22.5 kg) sandbag weight (PR+50) placed on the back between the scapula. Subjects remained in each position for 5 minutes, during which spirometry was performed at 1 and 5 minutes, and transcutaneous oximetry (O 2 sat) and end-tidal CO 2 levels (ETCO 2 ) were monitored continuously. Data were collected on percentage of predicted forced vital capacity (%predFVC), percentage of predicted forced expiratory volume in 1 second (%predFEV 1 ), O 2 sat, and ETCO 2 and compared by repeated measures analysis of variance with post hoc testing (P<.05 significance, STATA software).

Results: At 1 minute, mean %predFVC decreased from 101% to 87.1% to 84.7% to 84.2% and mean %predFEV 1 decreased from 98.2% to 83.4% to 82.0% to 80.0% (sitting to PR to PR+25 to PR+50, respectively). Similar results were found at 5 minutes into each position (mean %predFVC decreased from 102% to 86.8% to 82.5% to 80.5%; mean %predFEV 1 decreased from 99.3% to 82.2% to 79.5% to 74.7%, respectively). However, mean O 2 sat remained above 95% and mean ETCO 2 levels remained below 45 mm Hg throughout the 5-minute period for all positions.

Conclusion: Prone restraint with weight force resulted in decrements in FVC and FEV 1 consistent with a restrictive pulmonary function pattern but did not result in evidence of hypoxia or hypercapnia.

Collinsworth A, Kass L, Petersen HA, Carmack J, Lyle W, Akbar S, Schay J, Brenner BE/University of Arkansas for Medical Sciences, Little Rock, AR Study objective: The acute medical implications of triathlon participation have not been well described. According to 2 recent triathlons, we report a new constellation of signs and symptoms termed the postextreme endurance syndrome (PEES). The purpose of this study was to describe the syndrome, risk factors, and recommendations for preparations by race medical personnel.

Methods: This was a prospective, observational study of the Ironman Triathlon in Panama City Beach, FL, in October 2001 and of the Ultramax Triathlon in Table- rock, MO, in September 2002 . Demographic data and vital signs were collected on all racers before the race. Physical examination findings, treatments, and outcomes were collected prospectively on all individuals who presented for medical treatment. t Test, χ 2 , and 95% confidence intervals [CIs] are used as appropriate.

Results: Two thousand one hundred thirteen individuals (82% men, median age 37 years) participated in the 2 triathlons. Two hundred seventy-six (13%) racers sought medical attention for symptoms consistent with PEES, including muscle cramps (8%), nausea or vomiting (21%), relative hypothermia (temperature <98°F the gene encoding a galactosyltransferase necessary for O-antigen synthesis from Klebsiella pneumoniae 43816.

Results: Analysis of the mutant lipopolysaccharide (LPS) by sodium dodeclysulfate polyacrylamide gel electrophoresis confirmed the absence of O-antigen. In vitro, there were no detectable differences between wild-type K pneumoniae and the Oantigen deficient mutant in regard to avid binding by murine complement C3 or resistance to serum-or whole blood-mediated killing. Nevertheless, the 72-hour median lethal dose of the wild-type strain was 30-fold greater than that of the mutant (2×10 3 versus 6×10 4 colony-forming units) after intratracheal injection in ICR-strain mice. Despite being less lethal, the mutant organism exhibited comparable intrapulmonary proliferation at 24 hours compared with the wild type. Wholelung chemokine expression (CCL3 and CXCL2) and bronchoalveolar inflammatory cell content was also similar between the 2 infections. However, whereas the wildtype organism produced bacteremia within 24 hours of infection in every instance, bacteremia was not seen in mutant-infected mice. Mutant and wild-type K pneumoniae were equally lethal when administered to C3 knockout mice, indicating a role for complement despite the lack of detectable in vitro differences.

Conclusion: These results suggest that during murine pneumonia with K pneumoniae, O-antigen contributes to lethality by increasing the propensity for bacteremia and not by significantly changing the early course of intrapulmonary infection.

Lawrence EJ, Curtis KM, Neumar RW/University of Pennsylvania School of Medicine, Philadelphia, PA Study objective: Clinical trials have demonstrated that therapeutic hypothermia improves the outcome of comatose cardiac arrest survivors. However, the therapeutic window for initiation of hypothermia remains unclear. This study evaluates the efficacy of delayed hypothermia after simulated brain ischemia in vitro.

Methods: Organotypic hippocampal slice cultures were prepared from P5 Wistar rat pups and cultured for 1 week before analysis. Ischemia was simulated by transferring slices to glucose-free buffer in a chamber purged with humidified 95% N 2 /5% CO 2 at 37°C (98.6°F) for 30 minutes. Continuous hypothermia (33°C [91.4°F]) was induced at 0, 1, 2, or 4 hours after reoxygenation. Cell death was assayed at 24 hours by propidium iodide uptake. Propidium iodide fluorescence was quantified by computerized densitometry and compared by 1-way analysis of variance with Scheffe post hoc analysis.

Results: Simulated ischemia caused significant CA1 neuronal death compared with nonischemic controls. Hypothermia initiated 0, 1, 2, or 4 hours after reoxygenation reduced postischemic CA1 neuronal death by 47%±34% (P=.003), 85%±4% (P<.001), 88%±3% (P<.001), and 89%±5% (P<.001), respectively.

Conclusion: In this model of simulated global brain ischemia, delaying initiation of hypothermia for up to 4 hours after reoxygenation does not reduce therapeutic efficacy.

Naunheim RS, Parrott H, Standeven J/Washington University, St. Louis, MO Study objectives: The purpose of the study is to test a newer form of artificial turf used by a professional football team to determine whether its impact attenuation differs significantly from that of older types of artificial turf.

Methods: An instrumented computerized recording sphere, an impact recording device (Techmark, Inc., Lansing, MI), was dropped 20 times from a height of 48 inches on to 5 types of turf used by a professional football team: FieldTurf used on an inside field, outside grass at 24.5°C (76.1°F), outside grass at 0°C (32°F), Astro-Turf used on an inside field, and AstroTurf used at a domed stadium. FieldTurf is an artificial surface composed of a subsurface of shredded tires. AstroTurf is a grass carpet over a subsurface of foam, which was tested with 2 thicknesses of foam. The surface at a domed stadium consisted of an AstroTurf carpet on top of 1.6 cm of padding over concrete. The second AstroTurf surface was an AstroTurf carpet on top of 2.5 cm of padding over concrete. All measurements were made in g, the acceleration caused by gravity.

Results: A 1-way analysis of variance was done to compare all 5 fields, which were found to be significantly different. (F=220.02, df=4,95, P<.0001) . Duncan's range test shows that the outside grass at 24.5°C and the domed stadium are not different and the FieldTurf and AstroTurf with 2.5 cm of padding are not different. Although

The age range was 8 to 14 years, with a mean age of 10.4 years. Within the study group, 86.0% of patients reported riding a bicycle, and 86.0% reported riding a scooter. For participants who ride a bike, 69.8% indicated that they wear a helmet while riding a bike. However, only 44.2% of scooter riders reported using a helmet while riding their scooter. Parents and children were asked about the existence of a helmet law with a true/false question. Eighty-eight percent of parents correctly answered that this law exists, but only 27.5% reported that their children always wear their gear. Seventy-two percent of the children correctly answered the question, but only 27.5% reported that they always wear their gear. Children were asked to circle the properly fitting helmet (pictures provided by the Consumer Product Safety Commission [CPSC] ). Before viewing the video, 74.0% of children were able to circle the properly fitting helmet. After viewing the video, 100% of children were able to circle the properly fitting helmet. Participants were also asked to identify the recommended safety gear to be worn while riding a scooter. A correct answer required the child to circle all recommended gear (helmet, knee pads, and elbow pads). Before watching the video, 60.0% of participants identified all the recommended gear. After watching the video, this number increased to 82.0% (P=.001). Twenty-one of the fifty participants completed the follow-up questions. Of these, 15 children rode a scooter at the initial survey and at the follow-up. On the initial parent survey, 33.3% of these parents reported that their children always wore their scooter safety gear. This number increased to 86.7% at follow-up.

Conclusion: The CPSC reports that there were more than 84,400 ED visits for scooter-related injuries in 2001 and 16 deaths from scooters in the same year (www.cpsc.gov/PR/prscoot.html). Eighty-five percent of the injuries were in children younger than 15 years. The CPSC has established several recommendations for the safe use of scooters (www.cpsc.gov). Scooter use is high within our study population, but helmet use with scooters is lower than that with bikes. The improvement in correct answers to scooter safety questions after the scooter safety video was viewed indicates that the video is effective in improving scooter safety knowledge in the study population. In addition, the results of the follow-up telephone survey suggest that an increase in scooter safety awareness leads to a change in scooter-riding behavior. Limitations to this study include the small sample size. In addition, non-English speaking parents and children were excluded from the study because the video and study forms were not available in other languages. We did not control for ED diagnosis, and patients with injuries may be more receptive to this subject in the ED. We did not compare it with other educational methods on this subject. Our pilot data showed that the scooter safety video is effective in teaching scooter safety.

Davis-Moon L, Corbin T, Hall R, Lopez B, Riviello RJ/Thomas Jefferson University, Philadelphia, PA Study objectives: Violence is the leading cause of injury in young adults. Identification of factors that would identify high-risk individuals may help prevent future episodes of violence and may assist in proper resource referral. The Jefferson Community Violence Prevention Project (JCVPP) provides support and referral services to high-risk emergency department (ED) victims of intentional violence (VIV) who are aged 14 to 25 years. The purpose of this study is to characterize women who were VIV by using a prospectively collected JCVPP database.

Methods: Women VIV aged 14 to 25 years were identified by the ED and assessed by a JCVPP 24-hour inhospital chaplain who performed a bedside assessment and profile. VIV were offered a community-based referral to an existing network of antiviolence resources located within their communities. Patients' responses were compared with those of an age-matched control group of ED patients presenting with nonviolence-related illness or injury and were analyzed by χ 2 testing.

Results: From July 1, 2000, to December 3, 2002, there were 215 women VIV enrolled. Seventy-eight percent stated involvement in violent incidences within the past 12 months, 44% knew the perpetrator, 7% formed specific plans for retaliation, and 22% were concerned for their safety. One of 4 women also reported being victims of sexual assault. Although not statistically different from controls, 33% of the patients could easily obtain a gun, and 9% had a gun at home. Interestingly, the VIV had a higher rate of participation in after-school or community programs.

Conclusion: An alarmingly high percentage of women VIV had previous violent incidents, with a large number knowing their perpetrator. Future research in this area is needed to identify factors associated with women VIV. Findings may allow the emergency physician a tool to better identify those at risk so that proper resource allocation can be instituted. Other expected potential risk factors, such as professional athlete status (OR 1.7), were not significant (P>.15). Differences in humidity and the ambient air and water temperature were associated with the risk of developing PEES (P<.05). Conclusions: PEES is common among participants in triathlon events. Among this group of highly trained individuals, age and sex can identify groups at higher risk; however, environmental factors are also important considerations for race personnel attempting to predict medical resource needs.

Houry DE, Corneal K, White M, Kellermann A/Emory University, Atlanta, GA Background: Motorcycle helmet laws vary greatly in the United States. Our state does mandate helmet use but does not specify whether these helmets must be standard or regulation helmets. Some reports have suggested that certain populations wear phony motorcycle helmets and that these phony helmets will not protect motorcyclists from head injury.

Study objective: The objective of this pilot study was to determine the frequency of phony helmet use among motorcyclists in a state that mandates motorcycle helmet use.

Methods: An observational study was conducted throughout a 1-month period. Varied locations, times of day, and days of the week were selected to ensure a representative sample. Two trained research assistants used a standardized checklist and observed motorcyclists and patterns of helmet use. Helmets were considered phony according to previously published studies: less than an inch (2.5 cm) thick, whether or not the entire ear was covered, unusual surfaces (canvas, baseball caps), and whether there were attachments to the helmet.

Results: A total of 107 motorcyclists were observed. All motorcyclists (100%) were wearing helmets. Five (4.7%) patients were wearing phony helmets, and another 12 (11.2%) were wearing helmets whose phony or standard status the research assistants were unsure of, so up to 17 (15.9%) motorcyclists could have been wearing substandard helmets.

Conclusion: The majority of motorcyclists did wear standard motorcycle helmets, but there was still a significant minority who wore phony motorcycle helmets. These motorcyclists are at risk for head injury, and future research should be targeted at interventions for this population.

Wittels KA, Mello MJ, Palmisciano L/Harvard Affiliated, Brigham and Women's Hospital, Massachusetts General Hospital, Boston, MA; Brown Medical School, Providence, RI; Rhode Island Hospital, Providence, RI; Hasbro Children's Hospital, Providence, RI Study objectives: Since the introduction of nonmotorized scooters in 1999, emergency departments (EDs) have seen a dramatic rise in injuries related to their use. Taking advantage of the influence that media has on education, the Rhode Island American College of Emergency Physicians (ACEP) chapter, with funding from ACEP, produced a video on scooter safety. The objective of this study is to evaluate the scooter safety knowledge of children aged 8 to 14 years and their parents presenting to an urban pediatric ED and to determine whether the children's knowledge of scooter safety improves after viewing the scooter safety video. A second objective is to determine whether improved knowledge of scooter safety results in parental report of a change in behavior.

Methods: The study population was a convenience sample of children aged 8 to 14 years (patients and their siblings) at an urban pediatric ED waiting area and treatment area during June 2002. A brief survey consisting of demographic information, baseline information on the participants' use of wheeled sporting equipment, and questions on their knowledge of scooter safety assessed before and after viewing the video was given to patients. Parents were also given a short survey. In addition, parents were asked for their consent to receive a follow-up telephone call 5 months after completing the survey to reassess their children's scooter use and safety equipment use. Participants were given a small compensation (value <$3) for completion of the survey and follow-up telephone call. The Rhode Island Hospital institutional review board approved the research protocol.

Results: The study included 50 participants. There were 24 girls and 26 boys. Study objectives: In an attempt to better understand how patients respond to and cope with traumatic experiences, a trial for medication and counseling was developed. Previous attempts to track patients after hospital discharge have met with poor results. This study was designed to attempt immediate consent and study screening after the emergency department (ED) treatment phase to avoid not being able to contact a large number of possible research subjects.

Methods: The data collection was conducted in a prospective manner, with a fulltime research assistant interviewing all ED patients who experienced a traumatic injury as a result of abuse, injury, accident, or disaster. The study requires patients to consent to laboratory and radiologic imaging and a trial of medication for those who test positive on a psychiatric stress testing scale. The University institutional review board approved the study. t Test and Fisher's exact test were used to compare patient demographic and risk factors for patients who consented to study participation compared with the same factors for those who did not consent.

Result: Throughout a 90-day period, 231 consecutive trauma patients presenting to a Level I University Hospital were included in the initial screening for participation. Twenty-two percent (48) of these patients agreed to an interview for study purposes, and a net total of 8.3% actually consented to full study participation. There were higher percentages of patients with head injury with or without loss of consciousness who consented compared with those with positive test results for ethanol or substance abuse (P<.05). Of all trauma causes, interpersonal violence victims were more likely to consent than motor vehicle crash, fall, or accidental injury victims (Fisher's, P<.04 ). An increased rate of discharged patients (3.7:1) favored study participation compared with admitted patients. Age was not found to be a significant factor for giving consent.

Conclusion: The ability to successfully recruit trauma patients for a long-term prospective study remains difficult. Specific injury patterns and types of trauma appear to signify groups of improved compliance. The high percentage of consent from patients with head injury may require further institutional review board review for true comprehension.

Aswegan AL, Burkhart M, Rosenbaum R, Bollinger M, Farley HL, O'Connor R/Christiana Care Health Systems, Newark, DE Study objectives: We conducted this study to determine whether the practice of ordering a routine blood bank type and screen is appropriate in patients meeting criteria for trauma team activation and to identify variables that can classify patients according to anticipated need of blood transfusion.

Methods: All patients presenting to a Level I trauma center and meeting the American College of Surgeons Committee on Trauma criteria for trauma team activation were eligible for enrollment. Patients who died before receiving crossmatched blood were excluded. The following patient data were recorded: age, sex, initial systolic blood pressure (ED SBP) and pulse in the emergency department, Glasgow Coma Scale (GCS) score, Injury Severity Score (ISS), disposition, and number of units of packed RBCs administered. Data were analyzed by using multiple linear regression and the t test.

Results: A total of 2,125 patients requiring trauma team activation were studied. Two hundred eighty-four (13%) patients received transfusion, with a mean of 3.3 units of packed RBCs (range 1 to 84) administered. Mean age of those receiving transfusion was higher (47 versus 40 years), as was the ISS (22 versus 9; P<.001). Three characteristics were associated with transfusion: ED SBP less than 100, GCS less than 15, and initial pulse more than ED SBP. Patients receiving more than 5 units of packed RBCs had a mean ISS of 28 and a mean GCS of 10 but did not differ significantly from those receiving fewer units of packed RBCs with respect to ED SBP and initial pulse. Patients receiving more than 5 units of packed RBCs were also more likely to undergo surgery.

Conclusion: Routinely ordering a type and screen on patients meeting criteria for trauma team activation is an inefficient use of resources because the vast majority of these patients never receive blood. Classifying patients according to anticipated need of transfusion by high ISS, ED SBP less than 100, GCS less than 15, and initial pulse more than ED SBP will help reduce inappropriate ordering of this test. Study objectives: Auto racing is one of the fastest-growing spectator sports. We hypothesize that racing fans have more aggressive driving attitudes. The objective of this study was to identify the driving behaviors and attitudes of automobile racing fans and compare them with that of the general population.

Methods: This study was a retrospective analysis of a survey of spectators attending the Dover 400 NASCAR race on June 2, 2002. An 18-question survey was used to examine participants' attitudes toward traffic enforcement measures and their personal driving behaviors. The 1997 National Survey of Speeding and Other Unsafe Driving Actions (National Highway Traffic Safety Administration) was used as a historical control.

Results: One hundred ninety-eight surveys were completed. Regarding aggressive driving behaviors, in each instance the study group had significantly higher percentages of positive responses (χ 2 P<.001) compared with controls. Racing fans "tend to pass other cars more than other cars tend to pass them" (57% versus 30%); "enjoy the feeling of speed" (83% versus 39%); "drove 10 mph over the speed limit during the past year" (92% versus 60%); "tailgated another vehicle during the past year" (65% versus 23%); "drove when affected by alcohol during the past year" (33% versus 8%); and "raced another driver during the past year" (23% versus 6%). Finally, racing fans were less supportive of measures to enforce traffic safety laws; they "think police should always ticket drivers going 10 mph over the posted limit on a highway with traffic laws" (20% versus 35%); "think it is a good idea to use photo enforcement systems to identify speeders" (49% versus 71%); "think it is very important that something is done to reduce speeding" (40% versus 52%).

Conclusion: Racing fans are more likely to report high-risk driving attitudes and behaviors than the general population and are less supportive of enforcement of traffic safety laws. Given this information, driver safety education programs could target this select population.

Korn CS, Calder K, Hollinger M, Franklin A, Henderson SO/Keck School of Medicine, University of Southern California, Los Angeles, CA Background: Patients with electrical injuries present to emergency departments (EDs) with the potential for burns, associated trauma, and dysrhythmias. These injuries depend in large part on the type of electrical energy source, amount and duration of current flow, parts of the body affected, and resulting trauma, such as falls and blast injuries.

Study objective: We describe our experience with patients presenting to the ED with electrical injuries.

Methods: This was a retrospective review of all patients with electrical injuries who presented to the ED of an urban Level I trauma center throughout a 5-year period (1997 to 2002) . Data included age, sex, mechanism of injury, voltage, relation to work activities, ED diagnosis, associated trauma, cardiac monitoring, 12-lead ECG, cardiac rhythm, and disposition.

Results: Forty-six patients presented to the ED, and all medical records were available for review. Forty-three (93%) of 46 patients were men, with a mean age of 34.4 years (range 2 to 56 years). Thirty (65%) patients had injuries sustained in a work setting, with all but 1 associated with an alternating current discharge. Voltage ranged from 110 to 480. ED diagnosis was either electrical injury (52%) or electrical burns (48%) with extremity burns (22 of 22 [100%]), being the most common injury. All other injuries were associated with falls (10 [22%] of 46) sustained after electrocution. Thirty-four (74%) of 46 of the patients were admitted to the hospital, 16 patients to unmonitored units, 15 to monitored units, and 1 to the operating room, and 2 were transferred. The remaining 12 patients were discharged home. Cardiac rhythms varied, with the majority of patients (25 [54%] of 46) having normal sinus rhythm. Other rhythms included sinus tachycardia (14 [30%] of 46), sinus arrhythmia, first-degree atrioventricular block, bigeminy, and 1 A-flutter after cardiac arrest.

Conclusion: The majority of electrical injuries seen in this population were work related and required admission to a burn unit for extremity burns. Associated trauma was linked to falls sustained after electrocution. Cardiac arrhythmias were varied, with the majority of patients presenting in normal sinus rhythm or sinus tachycardia.

Whitaker K, Kestler A, Kendall J/Denver Health Medical Center, Denver, CO Background: Most centers advocate 23-hour observation after blunt abdominal trauma (BAT) in patients with initially negative studies; our institution uses an 8hour period that has not yet been evaluated.

Study objectives: We assess outcomes of patients observed for 8 hours after BAT and to identify a patient subset eligible for early discharge.

Methods: This was a retrospective 1-year medical record review of all patients admitted to an emergency department (ED) observation unit after BAT. Inclusion criteria were an initial evaluation in the ED that was negative for abdominal injury, no other injuries that required admission, and a mechanism serious enough to warrant obtaining an ED focused abdominal sonogram, serial abdominal examinations, and serial hematocrits throughout 8 hours. Exclusion criteria were age younger than 18 years, pregnancy, coagulopathy, or observation less than 8 hours.

Results: Of 533 patients meeting criteria for admission to an ED observation unit after BAT, 69 were excluded and 464 remained for the final analysis. From the observation unit, 29 (6%) patients were admitted to the hospital, 2 for abdominal injuries detected in the observation unit and 27 for other reasons. None of the 27 admitted for other reasons later was discovered to have abdominal injuries not recognized in the ED observation unit. Of the 435 (94%) patients discharged directly from the observation unit, abdominal injury in 1 was missed. The patient returned with persistent pain and was found to have a spleen injury that was treated nonoperatively. A low-risk subset was defined: ambulatory after ED studies; unintoxicated; no hypotension, tachycardia, abdominal pain, or tenderness; and no distracting injury during the out-of-hospital or ED course. Ninety-three (20%) patients met low-risk criteria. No low-risk patients had unscheduled repeated visits related to their trauma.

Conclusion: Eight-hour observation is sufficient for the evaluation of patients with BAT who have initially negative ED study results. A low-risk subset of patients may be considered for immediate discharge after ED testing regardless of injury mechanism.

Leung M, Thomas T, Steele R/Loma Linda University Medical Center, Loma Linda, CA Background: In the United States, patients with blunt spinal trauma are placed in cervical spine immobilization (CSI) in the out-of-hospital setting according to advanced trauma life support protocol.

Study objective: We determine whether there is a difference in neurologic outcome of patients who receive CSI versus those who do not receive CSI.

Methods: An observational, retrospective comparison study of patients with blunt spinal trauma was performed. Patients treated at a university hospital in China between 1993 and 2002 were compared with patients treated at a US university hospital. No patients taken to the university hospital in China received CSI. Neurologic injuries were assigned to 2 categories: disabling and not disabling. Patients with complete quadriplegia or paraplegia, inability to ambulate without assistance, incontinence or the need for chronic catheterization, and those who died were classified as having disability. Patients with previous neurologic injury or incomplete medical records were excluded.

Results: In the university hospital in China group, 63 patients had spinal injury, 18 (28%) of them were disabled (95% confidence interval [CI] 18% to 41%). Our US control identified 334 patients with spinal injury, 70 (21%) patients with disability (95% CI 17% to 28%). The level of spinal injury in the China group included 61 cervical, 1 thoracic, and 1 lumbar, whereas the US group included 113 cervical, 107 thoracic, and 113 lumbar (χ 2 =85.067). The disability percentages of these 2 groups were compared with a χ 2 test (χ 2 =0.784; df=1; P=.376).

Conclusion: There is no statistical difference in neurologic outcome between CSI patients versus no CSI patients. Background: The most common at-risk diagnosis associated with the development of acute respiratory distress syndrome (ARDS) is sepsis, trauma, and aspiration of gastric contents. In a recent article, sex differences in mortality from ARDS were identified in a retrospective analysis. They demonstrated that mortality ratios were higher in men compared with women.

Study objectives: Men have a significantly different incidence of ARDS than women. Methods: The study design is a prospective observational cohort. From July 2000 to June 2002, we screened trauma patients 7 days a week in the trauma section of the emergency department and the surgical intensive care unit at least every 12 hours. Patients were screened at a major Level I trauma center. Critically ill trauma patients who had an Injury Severity Score (ISS) of at least 16 and were mechanically ventilated were eligible for the study. Patients who were younger than 18 years, had a do-not-resuscitate order, or did not survive at least 3 days were excluded. Multivariate logistic regression analysis was used to assess the correlation between sex and ARDS.

Results: One hundred seventeen patients were eligible; 15 were excluded. There were 102 patients enrolled. In this cohort, the mean age was 34 years, mean ISS was 29, and mean Acute Physiology and Chronic Health Evaluation (APACHE) II score was 20. There were 24% (24 of 102) females. 34% (35 of 102) of the patients developed ARDS. Mortality was 30% (7 of 24) for women and 19% (15 of 78) for men (P=.3031). Forty-six percent (11 of 24) of women developed ARDS, whereas 31% (24 of 78) of men developed ARDS (P=.02). In a multivariate logistic regression analysis model that looked at ISS, patient age, massive transfusion, and sex, the relative risk of women was 3.99 (95% confidence interval 1.40 to 11.34; P=.0094).

Conclusion: The incidence of ARDS is greater in women trauma patients compared with men. The mortality was higher in women but not statistically significant. Further studies are required to validate these findings, as well as explore potential etiologies for the difference.

Windemuth RS/Loma Linda University School of Medicine, Loma Linda, CA Study objective: Recent research has shown that the motor component of the Glasgow Coma Scale (GCS-M) is as reliable when tested between 2 physicians as the total Glasgow Coma Scale (GCS) score. The objective of this study was to assess whether the GCS-M was as accurate as the total GCS at predicting outcomes for emergency department (ED) trauma activations.

Methods: This retrospective study was conducted at a Level I trauma center by using a trauma data bank in which data were collected prospectively for all trauma activations between 1990 and 2001. The patients' GCS total and GCS-M components were evaluated for their ability to accurately predict 5 clinically significant outcome measures: ED intubation, brain injury by computed tomography (CT), mortality during index hospitalization, head Abbreviated Injury Scale (AIS) scores, and Injury Severity Scores (ISS). We compared the predictive value of GCS and GCS-M against the 3 dichotomous outcomes (mortality, intubation, and brain injury on CT) by constructing receiver operating characteristic (ROC) curves and measuring the area under these curves. We compared the predictive value of GCS and GCS-M against the 2 ordinal outcomes (AIS, ISS) by constructing graphic plots of their data and calculating Spearman correlation coefficients. All analyses were performed using Stata 8 software (Stata Corporation, College Station, TX).

Results: Seven thousand eight hundred eighteen records were available for review, of which 7,531 contained the necessary components for evaluation. Areas under ROC curves for GCS and GCS-M for the following were observed: ED intubation (GCS 0.8596; GCS-M 0.8196); mortality (GCS 0.9056; GCS-M 0.8943); and brain injury on CT (GCS 0.6578; GCS-M 0.6252). Spearman coefficients for GCS and GCS-M when AIS and ISS were evaluated were as follows: AIS, and ISS, Conclusion: It appears there is no clinically important difference in predicting outcomes for ED trauma activations when the total GCS score is used compared with the GCS-M. The GCS-M alone displays similar accuracy. United States, approved by the Residency Review Committee of Emergency Medicine, was conducted by attending physicians and residents during a 1-year period. The study survey was administered by 1 of the study investigators, with a physician from each of the residency programs. Data collected included ED demographics, method used to determine sobriety, use of serum ethanol levels, use of breath alcohol measurement, use of calculated ethanol metabolic rates, and use of any other methods. The main study endpoint was the type of method used for sobriety determination. Programs were also analyzed by region, annual patient volume, and type of community served (ie, urban, suburban, rural) .

Results: Responses were obtained from 93 programs. The majority (73%; 95% confidence interval [CI] 64% to 82%) used purely clinical criteria for evaluation of sobriety. Other responses included calculated level (9.7%; 95% CI 3.7% to 15.7%), serial measurements (5.4%; 95% CI 0.8% to 9.9%), breath alcohol measurement (2.2%; 95% CI 0% to 5.2%), various methods (6.5%; 95% CI 1.5% to 11.5%), and other methods (3.2%; 95% CI 0% to 6.8%). There were no significant findings in subgroup analysis.

Conclusion: Most academic EDs in the United States use solely clinical criteria in the evaluation of sobriety of alcohol-intoxicated patients.

Fosnocht DE, Caravati EM/University of Utah, Salt Lake City, UT Study objectives: Acetaminophen is contained in a large number of analgesic medications available over the counter and by prescription alone and in combination with other medications. Patient knowledge of acetaminophen content of common analgesic medications has not been evaluated in a large ambulatory emergency department (ED) population. The objective of this study was to evaluate ED patient knowledge of the acetaminophen content and potential overuse of commonly used analgesic medications.

Methods: This was a prospective survey using a standardized questionnaire administered by trained research associates throughout a 4-month period. The study was conducted in an academic ED with a volume of 32,000 patients. A convenience sample of ED patients presenting with pain was used. Exclusion criteria were patients younger than 18 years, those critically ill or injured, or those unable to complete the survey. Patients were asked whether the following medications contained acetaminophen: Tylenol, Vicoden, Lortab, Percocet, generic non-aspirin-containing pain reliever, Motrin, Advil, and ibuprofen. Patients were also asked the recommended maximum daily dose of acetaminophen, if they had ever taken more than the recommended dose, and the reason.

Results: One thousand five patients were enrolled, and 529 (53%) patients did not know the maximum daily dose of acetaminophen (4,000 mg). The range of responses (n=476) to the maximum dosing for acetaminophen was 100 to 6,000 mg in 24 hours. One hundred eighty-three (18%) patients had taken more than the recommended dose of acetaminophen at some time in the past, with 180 (98%) taking more than the recommended dose for "improvement in pain relief" as the reason (Table) .

Conclusion: Patient knowledge of analgesic medications that contain acetaminophen and the maximum recommended daily dose is limited, which suggests the potential for errors in medication dosing when medications containing acetaminophen are prescribed for ED patients. Non-ASA, Generic "non-aspirin-containing pain reliever." throughout the past few years. Concern in some countries center around possible terrorist events, whereas other countries deal with large numbers of intentional ingestions.

Study objective: We describe the characteristics and treatment of severe organophosphate poisonings in Hungary.

Methods: This was a case series of 742 consecutive cases of organophosphate poisonings presenting to Josa Andras County Hospital, a tertiary care hospital in Nyiregyhaza, Hungary. Information about patient characteristics, treatment, outcome, and complications was recorded in a centralized computer database.

Results: There were a total of 605 men and 137 women treated for organophosphate poisonings. The mean age was 47 years, with a range of 15 to 82 years. Most patients (82%) ingested the insecticide in an attempt to commit suicide. Treatment consisted primarily of atropine, with a mean dose of 531 mg. Patients required treatment for an average of 10.8 days, with a range of 1 to 39 days. Overall mortality was 10.04%, with 29.9% of the patients requiring mechanical ventilation.

Conclusion: Many patients in eastern Hungary attempt suicide by ingesting insecticides. These ingestions require intensive treatment with large doses of anticholinergics. The level of complications after intentional organophosphate poisonings is high, with a significant amount of respiratory failure requiring mechanical ventilation. Clinical studies looking at human organophosphate poisonings are needed to improve therapy.

Johnston JJE, Corrigan N, McKinney LA, Steele J/Altnagelvin Area Hospital, Londonderry, United Kingdom Study objectives: We determine whether there is a greater role for emergency medicine in the management of childhood poisoning and determine recent trends in childhood poisoning locally and compare them with national trends. We also determine the efficacy of child-resistant containers for children of different age groups.

Methods: Information was collected prospectively on all children presenting to the emergency department (ED) with poisoning for a 4-month period (March 1 to July 1, 2001), as well as on telephone calls to the department about advice. Children of different ages were tested in their ability to open a child-resistant container. Inpatient information was collected for the preceding 7 years (January 1, 1994, to January 1, 2001) to determine trends.

Results: Most childhood poisoning admissions to pediatrics of are of low toxicity in the clinically well child, with a short length of stay. Only 23% of children are discharged directly from the ED. Advice given over the telephone was often reassurance (60%), and this service allayed parental anxiety and avoided unnecessary attendances. None of the children younger than 4 years were able to open the child-resistant container, although all those older than 8 years were. There has been a decline in admissions locally and nationally since 1997.

Conclusion: There is a role for emergency medicine in the management of childhood poisoning, which would be in large departments with a substantial pediatric workload, with nursing and medical staff trained in pediatrics. This role would avoid duplication of work, reduce inpatient admissions, further reduce the length of stay for the children, and reduce hospital costs, although it would have training and workload implications for emergency medicine.

Telephone advice effectively reduces workload and parental anxiety. Child-resistant containers are effective for preschool children.

The decreasing trend of admissions can be explained by primary prevention campaigns by the Royal Society for the Prevention of Accidents, the child accident prevention trust, as well as a higher threshold for admission by emergency physicians.

Hochman S, Hung O, Shih RD/Morristown Memorial Hospital, Morristown, NJ Background: Emergency departments (EDs) in the United States use a wide variety of methods to determine when an alcohol-intoxicated patient is clinically sober and ready for discharge. The method used and the time the patient spends in the ED have potential implications for the safety of the patient, as well as the legal responsibilities of the physician.

Study objective: We determine the ED practice pattern in the assessment of sobriety for ethanol-intoxicated patients.

Methods: A survey of all 127 emergency medicine residency programs in the [MSE]=.232, P<.001), a main effect of transport system F(1,660)=136 (MSE=.232, P<.001), and a transport system×test interaction F(1,660)=20 (MSE=.232, P<.001).

Hemolysis rate did not differ between systems (χ 2 =1.74, P>.15). Conclusion: The use of a pneumatic tube delivery system for transporting blood samples from the ED to the laboratory can significantly reduce the turnaround times of results and possibly reduce the rate of hemolysis of samples. Background: Laboratory investigations such as blood and urine analyses are essential to patient care and are conducted on the majority of patients presenting to emergency departments (EDs). Accuracy and timeliness of the investigations are the 2 most important criteria affecting patient care. Standards for turnaround times for reporting the results of these investigations have been published. As part of every ED continuous quality improvement process, periodic assessments of turnaround times should be conducted and compared with the published benchmarks. This study reports the turnaround times at 1 tertiary care ED and demonstrates the application of root cause analysis to identify the causes of delays.

Study objectives: The primary objective of this study was to measure the time intervals in the collection and analysis of 2 of the most commonly performed laboratory tests, measurement of serum hemoglobin and potassium, and to compare these turnaround times with the benchmark times established by the College of American Pathologists' Q-Probes Study. Hemolysis rates are used as a measure of test quality. The secondary objective was to identify root causes for delays in the ED receiving the laboratory test results.

Methods: All consecutive measurements of serum hemoglobin and potassium ordered from the ED during an 8-day period were included, except for those on critically ill patients. Synchronized time stamps were used to record process event times including physician order, sample collection, transport, accessioning, cue, instrument, and reporting. Turnaround times incorporated all of these except the time of physician order.

Analysis: A log transformation (ln[minutes+1]) was performed to normalize the data, and comparison of the transformed values were conducted by using analysis of variance. These data were then compared with published benchmarks by using 1sample t tests.

Results: The turnaround time for hemoglobin was significantly less than that identified in the Q-Probes study (t[140]=-5.69, P<.001), whereas the turnaround time for potassium was significantly greater (t[120]=12.65, P<.001; Table) . The hemolysis rate for potassium was 5.8%, with few samples needing recollection.

Conclusion: The median hemoglobin turnaround time (18 minutes) met the benchmark established by the Q-Probes study, whereas the median potassium turnaround time of 49 minutes did not. Causes of delays identified by root cause analysis were order processing time, high volume of laboratory tests ordered for admitted patients in the ED, and excessive cue and instrument times for potassium. Solutions for reducing these delays are discussed. 

Singh H, Chan CW, Nguygen TL, Diercks DB, Holmes JF/University of California-Davis Medical Center, Sacramento, CA Study objectives: We identify the prevalence of "positive" toxicologic examinations in trauma patients and determine whether the information from these tests alters patient treatment.

Methods: The study was a prospective observational study of injured patients presenting to a Level I trauma center and undergoing toxicologic screening as part of their routine evaluation. Toxicologic screening included a urinary analysis for amphetamine, barbiturates, benzodiazepines, and cocaine. In addition, serum ethanol levels were measured. Upon receiving the results of the toxicologic screening, the physician providing care to the patient was questioned about whether these results altered patient treatment.

Results: Two hundred eighty-four patients with a mean age of 36.4±23.6 years were enrolled. All had ethanol-level tests and 250 had urine toxicologic screening. Results of ethanol level tests were positive in 102 (36%) patients and were more than 100 mg/dL in 83 (29%) patients. Urine toxicologic screening result was positive in 77 (31%) patients for the following substances: amphetamine (30 [12%]), barbiturates (6 [2%]), benzodiazepines (35 [14%]), and cocaine (20 [8%]). The results of the toxicologic screening were stated to influence patient treatment in 42 (15%) cases. Results of the toxicologic screening on patient treatment included the following: explanation of altered level of consciousness or abnormal behavior (29), failed to explain patient's altered mental status (1), additional tests obtained (3), changed patient disposition (3), and other, not specified (6).

Conclusion: Routine toxicologic screening is frequently positive in trauma patients. Although most patients do not have their clinical treatment altered because of the results of the toxicologic screening, a small percentage will have their treatment changed. Future study should identify the population of patients that will best benefit from toxicologic screening. Background: Laboratory investigations such as blood tests are routinely performed on emergency department (ED) patients and are, therefore, an essential part of patient care. ED continuous quality improvement processes include evaluation of the accuracy and timeliness of the results of these investigations to ensure that the former is maximized and the latter meets established benchmarks. Although the accuracy of these tests results is primarily affected by laboratory rather than ED processes, the quality of the sample delivered to the laboratory is the initial factor determining the accuracy of the results. Decreased accuracy can result in delays because the tests are repeated and even errors in patient care if undetected. This study was designed to determine the effect, if any, a pneumatic tube delivery system has on the quality and timeliness of laboratory test results for ED patients.

Study objectives: We compare the hemolysis rates of blood samples and the turnaround times for the results of commonly performed laboratory tests delivered to the laboratory from an ED by a pneumatic tube system with those delivered by human courier.

Methods: We studied all consecutive measurements of serum hemoglobin and potassium ordered from 2 EDs of a multisite tertiary care hospital system, 1 with a pneumatic tube system and the other using human couriers. Synchronized time stamps were used to record turnaround times for the results. Hemolysis was determined by using a standardized laboratory measure. Time data were normalized by log transformation (ln[minutes+1]), and then a comparison of sites was conducted with analysis of variance. Hemolysis rates of the 2 delivery systems were compared by χ 2 analyses.

Results: The Table demonstrates significantly shorter median turnaround times for hemoglobin potassium when transported via a tube system. Its hemolysis rate was 7/121 (5.79%). Analysis of variance performed on the normalized turnaround time data revealed a main effect of test F(1,660)=406 (standard error of the mean

had a positive D-dimer test result, with a threshold of more than 0.35 µg/mL. Of these patients, 76 (11%) had a diagnosis of PE, and 74 of these patients had a positive D-dimer (>0.35) test result. With a D-dimer threshold of more than 0.35, the sensitivity was 0.97 and the specificity was 0.29. The positive predictive value (PPV) was 0.15 and the negative predictive value (NPV) was 0.99. With a D-dimer threshold of more than 0.50 µg/mL, the sensitivity was 0.93 and the specificity was 0.41. The PPV was 0.17, and the NPV was 0.98. Conclusion: Our data are consistent with the findings of other studies that evaluated the sensitivity and specificity of D-dimer testing in the diagnosis of PE. Our ED currently performs D-dimer testing based solely on physicians' clinical judgment. The specificity of our D-dimer testing may be improved by the use of previously described clinical criteria to increase the pretest probability of PE. Raising our D-dimer threshold to more than 0.50 would increase our D-dimer test specificity, with a slight decrease in sensitivity.

Jauch EC, Wagner KR, Dean C, Smulian AG, Carrozzella J/University of Cincinnati Medical Center, Cincinnati, OH; Veterans Administration Medical Center, Cincinnati, OH Study objectives: Thrombolytic therapy in acute ischemic stroke offers the potential for patient improvement and return to neurologic baseline but is not without risk of intracerebral hemorrhage and death. New diagnostic tests to identify patients at increased risk of fibrinolytic-associated hemorrhage would be of value. Currently, we are using proteomics technology to identify plasma proteins that are altered or differentially expressed during the first hours after ischemic stroke. These protein markers could provide information about systemic or brain-specific responses to stroke. Prospectively, they could identify patients at increased risk of intracerebral hemorrhage, should they receive fibrinolytic therapy. We report an initial study that compared protein expression levels in plasma or serum obtained from stroke patients and healthy controls.

Methods: We performed a prospective, unblinded study of pretreatment serum and plasma samples from patients who presented to 1 of 15 regional hospitals within the greater Cincinnati and northern Kentucky area, who were experiencing an acute ischemic stroke, and who would receive fibrinolytic therapy. Patient serum or plasma samples (N=15) were collected before administration of fibrinolytic therapy. Healthy adults served as controls (N=3). Plasma and serum proteins were separated by 2dimensional polyacrylamide gel electrophoresis. Separated proteins were detected by silver staining, and the stained gels were scanned with Phoretix image analysis software. Protein spots were compared between stroke and control gels to determine uniquely or differentially expressed proteins. Target proteins were cut out of the silver-stained gels, in-gel digested, and subjected to matrix-assisted laser desorption and ionization. Peptide mass fingerprint data from matrix-assisted laser desorption and ionization were submitted to standard databases for protein identification.

Results: On average, 275 protein spots were present in 2-dimensional gels from controls and stroke patients. Identification of many spots was suggested from published reports and the SWISS 2-dimensional polyacrylamide gel electrophoresis database. Selected protein spots were conclusively identified by matrix-assisted laser desorption and ionization. Of the differences in the plasma proteomes, there was a noteworthy twofold increase (P<.05) in the spot intensities corresponding to the glycosylation chains of haptoglobin-β in expression patterns between stroke patients and controls.

Conclusion: Expression proteomics can identify unique protein profiles in plasma within the first hours in patients with acute ischemic stroke. Although haptoglobin is generally considered to be an acute phase protein, it is also elevated in chronic inflammation and is associated, along with other inflammation-sensitive plasma proteins, with an increased risk of later stroke. Because the ultra-early time course of expression proteomics has not been examined in stroke, serial blood sampling beginning as soon as possible after stroke could provide evidence for a chronic versus acute inflammatory state in stroke patients. Better understanding this temporal response may assist in the development of future diagnostic tests and therapeutic interventions for acute stroke.

Perez A, Abbet P, Drescher M/Hartford Hospital, Hartford, CT; St. Joseph's Hospital, Bangor, ME Background: Aortic dissection (AD) is the most common acute aortic condition Study objectives: This study was undertaken to determine the effect of infusion rate on the temperature of the infusate reaching the distal end of an infusion tubing with active fluid warming.

Methods: One-liter bags of normal saline solution were warmed to 60°C (140°F) by using a microwave oven and run through standard infusion tubing at 200, 400, 600, 800, and 1,000 mL/h. Temperatures were measured at the distal end of an infusion tubing and every 1 minute for first 10 minutes and then every 10 minutes for 1 hour. Subsequently, by using a fluid warmer, warmed normal saline solution and packed RBCs were infused at 200, 400, 600, 800, and 1,000 mL/h. Temperatures were recorded when the readings had stabilized within a range of ±0.1°C.

Results: Warmed normal saline solution by microwave oven can be delivered through standard infusion tubing more than 37°C (98.6°F) if the normal saline solution is preheated to 60°C (140°F) and infused through long tubing (210 cm) at 800 and 1,000 mL/h. Normal saline solution and packed RBCs warmed to 41°C (105.8°F) with fluid warmer were delivered at or near 37°C (98.6°F) through relatively short tubing (110 cm) at 1,000 mL/h. The fluid composition influenced the infusate temperature (P<.05).

Conclusion: When patients with hypothermia or major trauma are treated by infusing warmed normal saline solution or packed RBCs, we should consider appropriate flow rates of fluid to maintain therapeutic delivery temperature by using practical infusion tubing in the ED setting.

Dula D, DeLo D, Liang L/Geisinger Medical Center, Danville, PA Study objectives: Despite advances in medical technology, the diagnosis of pulmonary embolism (PE) continues to be a difficult clinical problem. In recent years, the use of the D-dimer has been used as a screening test for thromboembolic disease. In previous studies involving the diagnosis of PE, an elevated serum D-dimer had a sensitivity ranging from 90% to 95% and a specificity of 23% to 55.5%. Historically, several different tests for measuring D-dimer have been used. This study evaluates the utility of the Micro-latex Agglutination D-dimer test in the diagnosis of PE in patients presenting to the emergency department (ED).

Methods: This study is a retrospective review of all ED patients who had a Microlatex Agglutination D-dimer assay during 2001 to 2002. Medical records of these patients were reviewed to compare D-dimer levels with their final diagnosis. The final diagnosis was based on the results of the ventilation/perfusion scan, spiral computed tomography, or angiography. The sensitivity and specificity of the D-dimer in the diagnosis of PE was calculated by using several diagnostic thresholds.

Results: A total of 666 medical records were reviewed, and 490 (74%) patients ity, and the presence of blood, protein, and ketones. Associations between urine findings and an abnormal serum bicarbonate level (serum bicarbonate ≤20 mEq/L) were compared by using logistic regression. Results: Two hundred eighteen patients were identified; 41 were excluded because the urine sample was obtained after the intravenous fluids had been administered and the electrolyte panel drawn. Of the remaining 177, 169 had ketones, 158 had protein, and 149 had blood in their urine samples. The mean specific gravity was 1.023 (95% confidence interval [CI] 1.021 to 1.024). The mean serum bicarbonate level was 20.1 mEq/L (95% CI 19.5 to 20.6); 84 patients had a serum bicarbonate level less than 20 mEq/L. Logistic regression demonstrated an association between specific gravity and serum bicarbonate level (P=.01); there was no association between ketones, protein, and blood presence with an abnormal serum bicarbonate level.

Conclusion: There is an association between urine specific gravity and serum bicarbonate level. We did not detect other urinalysis findings associated with an abnormal serum bicarbonate level.

Vinson DR/The Permanente Medical Group, Sacramento, CA Study objective: Acute akathisia is a common adverse reaction of parenteral dopamine receptor D 2 antagonists. Conventional diagnostic criteria used in emergency medicine akathisia research are difficult to remember and cumbersome to use. A simplified scale would facilitate the detection and grading of akathisia in clinical research and daily practice. We sought to develop a short akathisia scale and to evaluate its accuracy in comparison with the conventional long scale.

Methods: The long akathisia scale was prospectively applied to a cohort of 360 adults aged 17 to 65 years and presenting to the emergency department. These patients were receiving 10 mg of intravenous prochlorperazine in the treatment of headache or nausea in a series of 3 studies conducted in a 400-bed, academic, tertiary-care US military hospital that has an annual census of 65,000 emergency patient visits. The long scale consists of 3 subjective questions asked of the patient (each graded from 0 to 3), as well as 2 objective measurements made by physician observation (each graded from 0 to 4). Total scoring can range from 0 to 17 points. The scale was administered before drug infusion and 1 hour thereafter, and the change in score was calculated. Akathisia diagnosis by the long scale required an elevation change of at least 2 points in the subjective element, combined with an elevation change of at least 1 point in the objective element. When present, akathisia was graded as mild, moderate, or major, according the degree of score change. To simplify the long scale, we reduced the number of subjective questions from 3 to 1. The Wilcoxon rank sum test was used to determine the ability of a change in the score of each question to distinguish the presence of akathisia from its absence, as diagnosed by the long scale. The short scale also integrated the 2 objective elements into 1 measure: semipurposeful or purposeless restless motor activity of torso, legs, or feet while sitting. The patient's postdrug report of akathisia severity was taken as the overall akathisia grade. We retrospectively applied the short scale to the cohort. We used sensitivity and specificity with exact confidence intervals (CIs) and receiver operating characteristic (ROC) curves to analyze the impact of instrument form on diagnostic accuracy. We used pair-wise analyses to compare the severity grade results obtained using the short versus long scales.

Results: The z-scores from the Wilcoxon rank sum test for each of 3 patient questions were as follows: (1) -16.71 for "Do you feel a restless urge to move, especially in your legs?" (2) -14.84 for "Are you unable to keep your legs still?" and (3) -16.29 for "Are you unable to remain still, either standing or sitting?" The first and third items were strongly associated with akathisia. We selected the first item for the short scale for this study. The short scale diagnostic criteria require a postinfusion elevation of 1 subjective grade on a 4-grade categoric severity scale, with a corroborating elevation of at least 1 grade in observed akathisic motor activity. The long scale diagnosed akathisia in 118 (32.8%; 95% CI 28.0 to 37.9) patients and the short scale diagnosed akathisia in 120 (33.3%; 95% CI 28.5 to 38.5) patients. When akathisia was characterized categorically as either absent or present, the short criteria agreed with the long criteria in 358 cases (99.4%; 95% CI 98.0 to 99.9). The short scale overdiagnosed akathisia in 2 patients and failed to detect the disorder in no cases. When the long scale was treated as the criterion standard, the short scale had a sensitivity of 100.0% (95% CI 96.9% to 100.0%) and a specificity of 99.2% (95% CI 97.0% to 99.9%). The area under the ROC curve was 0.999 (95% CI 0.997 to 1.000). Regarding the accuracy of grading akathisia into absent, mild, moderate,

A N N A L S O F E M E R G E N C Y M E D I C I N E 4 2 : 4 O C T O B E R 2 0 0 3 requiring urgent surgery. AD, if not diagnosed in the emergency department (ED), is frequently fatal. AD is a difficult antemortem diagnosis. Study objective: We determine whether acute AD is associated with an elevation of fibrin degradation products, D-dimers. In most cases of aortic dissection, there is an intense yet local activation of the coagulation and fibrinolytic systems, leading to a significant elevation of the fibrin degradation products, including D-dimers. We postulate that D-dimers can be useful in the early diagnostic evaluation of aortic dissection.

Methods: This was a retrospective medical record review of patients diagnosed with AD in the ED. Every record was reviewed for D-dimer results obtained in the ED before any therapeutic intervention. The study was conducted in an urban Level I trauma center from October 1996 to September 2000. Exclusion criteria were referred patients with known diagnosis of AD. D-Dimer assay was a semiquantitative latex agglutination assay (Diagnostica Stago, France), with a normal range up to 0.5 µg/mL.

Results: One hundred fifty-six patients were diagnosed with AD in the ED. Seven patients had a D-dimer assay during their evaluation. All 7 had a positive test result.

Conclusion: All 7 patients with an AD who had D-dimer assays performed in the ED had positive results by latex agglutination. In EDs where computed tomography scan is not readily available, a negative D-dimer result combined with a normal chest radiograph result may be sufficient to rule out an AD. The study is limited by its retrospective nature and sample size.

Patients With Quantitative β-Human Chorionic Gonadotropin Values Below the Ultrasonographic Discriminatory Threshold of 2,000 mIU/mL Park CW, Banks SL, Riffenburgh RH, Buckley RG/Naval Medical Center San Diego, San Diego, CA Study objectives: We assess the accuracy of a progesterone (PRG) cutoff value of less than 5 ng/mL for detecting ectopic pregnancy in patients with a β-human chorionic gonadotropin (β-hCG) level less than the ultrasonographic discriminatory threshold of 2,000 mIU/mL.

Methods: A prospective study of progesterone level was performed in all hemodynamically stable, first-trimester patients with abdominal pain or vaginal bleeding presenting to a military teaching hospital emergency department from August 1994 to September 1996. All patients who had a β-hCG level of less than 2,000 mIU/mL were included in the study and were followed up longitudinally until an outcome of intrauterine pregnancy or ectopic pregnancy was established. Progesterone values were measured from the original β-hCG sample. Patients were excluded from the study for any of the following reasons: a previous documentation of intrauterine pregnancy; the estimated gestational age based on the first day of the last normal menstrual cycle and the corresponding uterine size were 13 weeks or more; deemed hemodynamically unstable as defined by an altered mentation, orthostatic syncope, or hypotension. No patients were lost to follow-up.

Results: One hundred eighty-six patients met eligibility criteria. Thirty-three patients were diagnosed with ectopic pregnancy. Of these 33 patients, a PRG level of less than 5.0 ng/mL correctly identified 16 patients (sensitivity 48%; 95% confidence interval [CI] 40% to 55%). Of the remaining 153 patients who did not have an ectopic pregnancy, 95 had a PRG level less than 5.0 ng/mL (specificity 38%; 95% CI 31% to 46%).

Conclusion: A PRG level less than 5.0 ng/mL lacks sensitivity and specificity in detecting ectopic pregnancy in patients with a β-hCG level less than the ultrasonographic discriminatory threshold of 2,000 mIU/mL.

Miner JR, Hick J, Kasak C, Danigelis M/Hennepin County Medical Center, Minneapolis, MN Study objectives: We determine whether urinary protein, blood, or ketone levels or specific gravity test results were predictive of an abnormal serum bicarbonate level in children presenting to the emergency department (ED) with the clinical diagnosis of dehydration.

Methods: This was a retrospective review of children who were younger than 3 years and diagnosed with dehydration and who received intravenous fluids in the ED between June 1, 2001, and June 1, 2002. Data were collected by using the EMSTAT electronic medical record, including diagnosis, time of sample collection (blood and urine), serum bicarbonate level, urinalysis findings of the specific grav-the answer or answered incorrectly about adverse effects of the herbs. Further education on herbal indications and adverse effects is needed.

Can We Predict Those at Risk for Emesis?

Silverman ME, Peters JR, McClister PB/Morristown Memorial Hospital, Morristown, NJ Study objectives: Oral steroids have become part of the regimen used to treat multiple diseases of the pediatric population in the emergency department (ED), among these, asthma and croup. In review of multiple studies for outpatient treatment in asthma and croup in children, little is mentioned about the problems or adverse effects of the oral formulations. In one article comparing nebulized dexamethasone versus oral prednisone, the authors note, "Children will frequently refuse to accept or will vomit orally administered corticosteroids." It appears in their study that 15% had emesis after prednisone treatment. No literature was found that looked at the incidence of emesis with oral Prelone syrup. However, in the multiple studies that address the use of prednisolone, no mention of vomiting is apparent. Our objective was to determine the incidence of emesis and whether patient characteristics can predict emesis.

Methods: The study was conducted in a university-affiliated suburban hospital with an ED volume of 65,000 patients. We performed an observational case series of pediatric patients who were 12 years and younger and received oral steroids in the ED. The incidences of emesis and patient characteristics were recorded. Patient characteristics were then examined by using Student's t test and χ 2 tests when appropriate.

Results: The incidence of emesis in 81 pediatric patients receiving Prelone was 13.6%. Characteristics compared in the 2 groups showed no significant difference in emesis by age (emesis=3.8 years, no emesis=4.2 years), weight (emesis=16 kg, no emesis=18.4), respiratory rate (emesis=32 breaths/min, no emesis=32 breaths/min), presence of coughing (emesis=63.6%, no emesis=75.7%), fever (emesis=37.4°C [99.3°F], no emesis=37.5°C [99.5°F]), hypoxia (pulse oximetry: emesis=97%, no emesis=97%), or disease type (croup: emesis=12.9% versus asthma: emesis=14.6%). There was a statistically significant difference when tested by sex (female patient with emesis= 25.9% versus male patient with emesis=7.4%).

Conclusion: The incidence of vomiting for all patients receiving Prelone was 13.6%, similar to the incidence of vomiting in the previously mentioned study. The only factor that was found to have a significant difference between groups was that female patients had more emesis then male patients; otherwise, no factors were found that help predict emesis in patients receiving oral steroids.

Hirsh EL, Trimarco TW, Ward MF, Lee D, Nozzolillo M, Sama AE/North Shore University Hospital, Manhasset, NY Previous studies have indicated that the incidence of drug-drug interactions (DDI) increases as the number of medications administered to a patient increases.

Study objective: We determine the number of potential DDI according to degree of polypharmacy among emergency department (ED) patients.

Methods: We conducted a retrospective medical record review on a convenience sample of 1,348 ED patient visits from May to November 2001. This study was conducted at an academic tertiary care hospital with an annual ED census of 60,000 visits. All age groups were included, as were discharged and admitted patients. Each patient's outpatient (ie, daily) medications, ED-administered medications, and medications prescribed at ED discharge (when appropriate) were recorded on a standard collection tool. Potential DDI were identified by the Physician's Drug Reference 2002 and by the handheld software program ePocrates. All records were reviewed by 1 of 2 physician reviewers, and both physicians read 50% of records. The Cochran-Armitage trend test was applied to analyze whether the incidence of potential DDI increased as the numbers of total medications increased.

Results: Of 1,348 patients, 923 (68.5%) had 2 or more medications and were susceptible to potential DDI. Four hundred twenty-five (31.5%) patients had a total of fewer than 2 medications and, therefore, were not at risk for potential DDI. Of the patients with 2 or more medications total, 505 had the potential for DDI (54.7%), leaving 418 patients (45.3%) who took at least 2 medications and did not have the potential for DDI. The incidence of potential DDI did increase as the numbers of total medications increased. Of patients taking 2 medications total, 23.5% had potential DDI; so did 45.0% of patients with 3 medications, 47.2% with 4, 70.2% and major, the short scale agreed with the long scale in 325 of 360 cases (90.3%; 95% CI 86.7 to 93.1) and misgraded 35 cases (9.4%; 95% CI 6.9 to 13.3). Of the misgraded cases, 34 were within 1 grade of the long scale and 1 case was 2 grades off. Overall, 28 cases (7.8%; 95% CI 5.2 to 11.0) were upgraded by the short scale and 7 cases (1.9%; 95% CI 0.8 to 4.0) were downgraded.

Conclusion: The simplified short scale is highly accurate in diagnosing acute drug-induced akathisia in comparison with the long scale. The short scale also provides accurate grading of akathisia severity. Study objectives: Before the Food and Drug Administration-imposed black box warning, droperidol was used frequently in emergency departments (EDs) for a variety of indications. This study sought to determine how this warning affected the use of droperidol and why EDs changed their practice.

Methods: A 1-page survey was distributed to 260 attendees at a national emergency medicine meeting 10 months after the warning was issued. Respondents answered questions about their use of droperidol and how the warning changed their practice.

Results: Two hundred seven surveys were returned, a response rate of 68%. The majority of respondents (69%) had more than 5 years' experience; 82% work in EDs with a census of more than 25,000 patients and 42% in sites with more than 50,000 patients. Changes in practice were remarkably unaffected by practice volume and years of experience (P>.05). Almost all (97%) respondents were aware of the black box warning, and 81% changed their use of droperidol. Seventy-three percent had been using droperidol at least once a week, with 53% using it almost every shift. Respondents used droperidol for nausea and vomiting (78%), agitation (57%), headache (55%), and other (10%). As a result of the warning, 71% of respondents never use droperidol for the following reasons: ECG and monitoring requirements (42%), concern about killing patients (11%), and hospital discontinuation (42%). When using droperidol, 12% of respondents obtain a 12-lead ECG; cardiac monitoring is used by 14% for varying periods: 1 hour (4%), 2 hours (8%), and 4 to 6 hours (2%). Of those still using droperidol, 63% did not decrease the dose. Alternative drugs used for nausea and vomiting included metoclopramide(60%), promethazine (87%), and ondansetron (57%). Sedative alternatives included haloperidol (58%), lorazepam (68%), haldol/lorazepam (50%), and ketamine (8%). Only 8% of respondents thought the alternative sedative agent was more effective than droperidol.

Conclusion: The black box warning on droperidol has dramatically decreased its use in the ED, often substituted for agents perceived to be less effective. In many institutions, the agent is no longer available.

Gulla J, Singer AJ, Athode HC/State University of New York, Stony Brook, NY Study objectives: A previous survey indicates the frequent use of herbs by the public. We determined the public's knowledge of indications and adverse effects of 10 commonly used herbs.

Methods: The study was a survey at a university-based suburban emergency department (ED). A convenience sample of patients, family, and visitors in the ED was used. We used a standardized collection of demographics and a questionnaire of multiple choice and true/false questions on the indications and adverse effects of 10 herbs. Data were analyzed with descriptive statistics.

Results: Eight hundred thirty-seven people were surveyed. Mean age was 38 years (±37 years), 60% were women, 62% were white, and 70% had a higher education. Information on herbs was obtained from health care practitioners (13%), family members (26%), friends (19%), the Internet (10.5%), stores (10%), and other sources (11%). Percentages of those who answered the questions on herbal indications correctly were as follows: ginseng, 75%; Ginkgo biloba, 68%; echinacea, 45%; ginger, 37%; ephedra, 37%; kava kava, 25%; valerian, 19%; St. John's wort, 44%; saw palmetto, 21%; and tea tree, 25%. The percentage of those who answered the questions on herbal adverse effects correctly were as follows: ginseng, 31%; Ginkgo biloba, 36%; echinacea, 14%; ginger, 11%; ephedra, 41%; kava kava, 29%; valerian, 24%; St. John's wort, 36%; saw palmetto, 8%; and tea tree, 25%.

Conclusion: In total, 61% of those surveyed did not know the answer or answered the questions incorrectly about herbal indications, and 74% did not know

Martel M, Palmer C/Hennepin County Medical Center, Minneapolis, MN Background: The impact the Food and Drug Administration (FDA) warning in December 2001 of QT prolongation and torsade de pointes has had on emergency department (ED) droperidol use has not been determined.

Study objectives: We sought to identify a change or restriction of the use of droperidol in emergency medicine residency training programs (EMRT) after the FDA warning.

Methods: A survey was distributed to all 126 EMRTs via e-mail. If no response was received, a second e-mail was sent, followed by facsimile, telephone call, or directed e-mail. The survey assessed the ED use of droperidol before and after the FDA warning in December 2001, including restrictions on use, the need for rhythm strip or ECG evaluation of the QTc before administration, continuous cardiac monitoring after administration, indications for use, and an opinion on whether droperidol should be available for emergency physician use. Data were analyzed by using descriptive statistics and χ 2 test.

Results: Sixty-seven (53.2%) of 126 EMRTs responded. Sixty-one (91.0%) EMRTs used droperidol before December 2001, and 40 (59.7%) currently use droperidol (P≤.01). Of the EMRTs currently using droperidol, 16 (40.0%) have restrictions on its use, with 12 (30.0%) requiring a rhythm strip or ECG before droperidol administration. Ten (25.0%) EMRTs require continuous cardiac monitoring after administration. Twenty-three (57.5%) EMRTs currently use droperidol for sedation, 24 (60.0%) for nausea or vomiting, 29 (72.5%) for headache, and 4 (10%) for other pain management. Fifty-eight (86.6%) EMRTs indicated they believed droperidol should be available for emergency physician use, 4 (6.0%) believed it should not be available, and 5 (7.5%) did not respond.

Conclusion: There has been a significant decline in the use of droperidol in EMRTs, with many having restrictions associated with droperidol use. The majority of EMRT emergency physicians believe droperidol should be available for ED use.

Thacker J, Miner JR, Biros M/Hennepin County Medical Center, Minneapolis, MN Study objective: Patients frequently present to the emergency department with the complaint of nausea and vomiting. Droperidol is frequently used as an antiemetic. In this study, we compared the efficacy and adverse effects of droperidol as an antiemetic with that of metoclopramide. A secondary objective was to compare the change in ECG-measured QTc after medication administration.

Methods: This was a prospective, randomized study of adult, nonpregnant patients undergoing treatment for nausea. Patients were randomized to receive either 1.25 mg of droperidol or 10 mg of metoclopramide intravenously. A rhythm strip was collected before and 1 hour after drug administration for QTc calculation. Patients completed a visual analog scale (VAS) for nausea and somnolence at 0 and 60 minutes. Physicians evaluated the patients at 30 minutes and 1 hour for adverse effects. Data were analyzed with descriptive statistics and t tests.

Results: Twenty-eight patients were enrolled (mean age 33.8 years; 95% confidence interval [CI] 29.3 to 38.4; 63% women); 16 received droperidol and 12 metoclopramide. The mean baseline nausea VAS for droperidol was 64.4 mm (95% CI 54.1 to 74.7) versus 66.8 mm (95% CI 48.5 to 85.0) for metoclopramide (P=.80). The mean baseline somnolence VAS in the droperidol group was 52.8 mm (95% CI 36.7 to 68.9) versus 61.0 mm (95% CI 39.8 to 82.2) in the metoclopramide group (P=.50). The mean decrease in the nausea VAS for droperidol was 46.9 mm (95% CI 38.5 to 55.3) versus 45.2 mm (95% CI 27.9 to 62.6) for metoclopramide (P=.89). The mean change in VAS for the degree of somnolence for droperidol was +4.1 mm (95% CI -21.5 to 29.7 mm) versus -6.4 mm for metaclopramide (95% CI -28.9 to 16.2 mm; P=.61). The mean change in QTc for droperidol was 0.0 ms (95% CI -30 to 30 ms) versus 40 ms for metoclopramide (95% CI 0.0 to 80 ms; P=.13). Akathisia was noted in 2 (12.5%) patients who received droperidol and 1 (9.1%) who received metoclopramide.

Conclusion: Both medications were effective antiemetics, with similar degrees of perceived somnolence. There was no difference in the QTc before and after administration of either drug.

A N with 5, 83.5% with 6, 86.3% with 7, 88.1% with 8, 96.7% with 9, 85.7% with 10, and 100% with 11 or more medications. These increases were statistically significant (P≤.0001). We were unable to determine whether these interactions occurred solely among outpatient medications, solely among ED medications, or among a combination of the 2. Conclusion: As the number of total medications increases, the incidence of potential DDI increases. Further study is needed to assess whether these interactions occur solely among outpatient medications, solely in the ED, or as a combination of the two. Background: The association between the initiation of intravenous vasoactive therapy for acute heart failure (AHF) and clinical outcomes has not been adequately studied. If a relationship between intravenous vasoactive therapy initiation and clinical outcomes is established, it could serve as the foundation for identifying the clinical predictors of a favorable response to early intravenous vasoactive therapy and for the development of effective treatment algorithms for patients admitted to the hospital emergency department (ED). This study examines the effect of initiation of vasoactive therapy in the ED versus later inhospital initiation on clinical outcomes, health care use, and mortality in patients with AHF.

Methods: Data were analyzed from the Acute Decompensated Heart Failure National Registry (ADHERE) database. As of April 2003, the database included data from 46,599 patients discharged with a primary diagnosis of AHF from 253 participating US hospitals. Statistical analyses were performed to determine, among patients who were first evaluated in the ED, differences in timing of initiation of intravenous vasoactive therapy in the ED versus inhospital. Additional analyses were then performed to determine the effect of ED versus inhospital initiation of intravenous vasoactive therapy on mortality, hospital length of stay, number of patients requiring invasive procedures, number of patients transferred to the ICU/cardiac care unit (CCU), ICU/CCU length of stay, and the number of patients with prolonged hospitalization (>third quartile). Invasive procedures included coronary artery bypass graft surgery, intra-aortic balloon pump, cardiopulmonary resuscitation, percutaneous coronary intervention, left ventricular assist device, or initiation of dialysis in a patient not receiving long-term dialysis. Patients who were hypotensive (<90 mm Hg) or who had evidence of a recent myocardial infarction (elevated cardiac enzymes) were excluded from the analyses. Intravenous vasoactive therapy included treatment with dopamine, dobutamine, milrinone, nesiritide, nitroglycerin, nitroprusside, or other vasoactive medications. Two-sample Wilcoxon tests were used for continuous or ordinal variables, and Cochrane-Mantel-Haenszel χ 2 tests were used to determine differences in proportions.

Results: The median time of 1.1 hours to initiation of intravenous vasoactive treatment in the ED (n=4,096) was significantly shorter than the 22.2 hours elapsing before initiation of intravenous vasoactive treatment inhospital (n=3,500; P<.0001). Moreover, early initiation of intravenous vasoactive treatment in the ED was associated with significantly lower mortality (unadjusted; 4.3%) compared with that of inhospital initiation (10.9%; P<.0001). Compared with patients who received vasoactive therapy in the hospital, fewer patients given vasoactive therapy in the ED required transfer to the ICU/CCU (4% versus 20%; P<.0001), or required invasive procedures (19% versus 27%; P<.001). Compared with patients treated in the hospital, those treated in the ED had a smaller percentage of patients requiring prolonged hospitalizations (26% versus 49%; P<.0001) and a shorter length of stay in the hospital (median 3.0 versus 7.0 days) and in the ICU/CCU (2.1 versus 4.5 days; P<.0001).

Conclusion: Early initiation of intravenous vasoactive therapy is associated with fewer deaths, fewer invasive procedures, fewer transfers to the ICU/CCU, and shorter stays in the hospital and ICU/CCU. Because the vast majority of AHF patients generally receive their first treatment in the ED, it is crucial to be able to develop methods for the rapid identification of patients likely to benefit from early vasoactive therapy. Reported use for radiologic imaging was plain radiographs (37.3 versus 42.4), computed tomography scan (10.9 versus 37.6; P<.001), and ultrasonography (7.8 versus 8.0).

Conclusion: The NHAMCS is a large data set that can provide useful information. There are several statistically significant differences in comparisons between the 2 data sets. Differences in ED diagnostic categories may reflect population differences, variations in coding methods, or intrinsic differences between retrospective and prospective data collection. Differences in admission rate and radiology use may reflect a difference in the severity of illness seen at large urban academic hospitals or differences in the evaluation and treatment practices at academic centers compared with those in community or rural settings.

Lewis LM, Asaro PV, Blanda M, Gerson LW, Hustey FM, Meldon SW, Shapshak D, Banet GA/Washington University School of Medicine, St. Louis, MO Study objective: We determine accuracy of emergency department (ED) diagnoses and describe disposition and outcomes of elderly patients presenting to EDs with abdominal pain.

Methods: This was an 18-week, multicenter, prospective observational study of elderly patients (age >59 years) with abdominal pain and presenting to participating EDs, excluding those with abdominal trauma or surgery within the previous 30 days. We reviewed medical records for patient demographics, ED diagnosis, ED disposition (admit versus discharge), "final" diagnosis (at hospital discharge or 2-week follow-up), need for emergency surgery or "significant" medical intervention (eg, bowel decompression, blood transfusion, intravenous antibiotics >24 hours), and 2-week mortality.

Results: Of 458 patients screened, 338 (74%) met selection criteria (mean age 73±9 years, 66% female patients, and 53% white patients) and had complete followup data for analysis. Leading ED diagnoses were "nonspecific" abdominal pain (24%), urinary tract infection (8%), bowel obstruction (8%), constipation (6%), gastroenteritis (6%), and gallbladder disease (5%). A cardiothoracic etiology was assigned 5% of the time. ED and final diagnoses were concordant 81% of the time. Of the 193 patients (57%) admitted, 136 (72.5%) required surgery or "significant" medical intervention during their hospitalization. Of those discharged from the ED, 17 patients (13%) had a return ED visit or unscheduled admission within the 2week follow-up period. Two-week mortality rate for the entire group was 4.4%.

Conclusion: Abdominal pain in the elderly is associated with high hospital admission and 2-week mortality rates, as well as significant medical and surgical acuity. More than 25% of those admitted did not require surgery or "significant" medical interventions, and more than 10% of the patients discharged had return visits or admissions. There is an opportunity to refine ED assessment and disposition strategies for elderly patients with abdominal pain. Methods: This was a prospective, observational, multicenter study with a convenience sampling of patients aged 60 years or older presenting to 1 of 4 participating EDs with abdominal pain. Exclusion criteria were recent abdominal trauma, surgery, or invasive procedure and chronic abdominal pain. Abdominal pain history and physical examination findings were obtained prospectively and recorded with a standardized data collection form. Radiographic findings, dispositions, diagnoses, and surgical or medical procedures performed were abstracted from medical record review. A resource use scale was developed to include 4 mutually exclusive Background: The initiation of outpatient enoxaparin therapy from the emergency department (ED) has been shown to be a safe and cost-effective way to treat patients with deep venous thrombosis (DVT). However, many patients with DVT are still admitted despite availability of this therapy.

Objective: The goals of the study were to determine the frequency with which a clinical pathway for the outpatient treatment of DVT is used in eligible patients and to characterize the reasons that patients with DVT are hospitalized.

Methods: This was a single-institution, prospective, observational study of consecutive adult ED patients (age >18 years) who had a diagnosis of DVT confirmed by ultrasonography. The setting was an urban university tertiary care center with approximately 50,000 annual ED visits. To perform this study, a clinical pathway was created to facilitate outpatient therapy with enoxaparin. Resources included a step-by-step instruction packet for the physician, literature and an instructional video for the patient, and an on-call coordinator. We collected pertinent demographic and clinical data, as well as reasons for treatment decisions. We reviewed the ED log to ensure that all patients were enrolled. Data are descriptive and reported as percentages and medians with SDs where appropriate.

Results: A total of 97 (99%) of 98 eligible patients were enrolled. The median age was 61 years (SD±17), and 60% were female patients. Among 97 patients, there were 41 (42%) who were eligible for outpatient therapy and could inject drugs as an outpatient. Of the 56 patients who were ineligible, 18 (32%) had contraindications to anticoagulant therapy, 31 (55%) had other medical indications for admission, and 7 (13%) were unable to reliably inject drugs as outpatients. Of the 41 eligible patients, 29 (71%) successfully began the outpatient protocol, whereas 12 (29%) were admitted to the hospital. For the 12 who were admitted, the reasons for admission were as follows: 2 for a hypercoagulable examination, 7 because of emergency physician preference without justification, and 3 because of primary care provider preference without justification. Of these patients, 6 (50%) of 12 were discharged home and receiving enoxaparin the next day. Among the 68 patients admitted for any reason, 29 (43%) were ultimately discharged home and receiving enoxaparin.

Conclusion: The establishment of an organized DVT pathway for outpatient enoxaparin may facilitate the discharge of eligible patients to home from the ED. Despite an efficient pathway, there remain a number of reasons that make hospital admission unavoidable for some of these patients. Study objective: We compare observational data from the National Health Ambulatory Medical Care Survey (NHAMCS) to that from a multicenter Geriatric Abdominal Pain Prospective Study (GAPPS) about the causes, radiographic evaluation, and admission rate of elderly patients who present to the emergency department (ED) with nontraumatic abdominal pain.

Methods: We queried the 1997 to 2000 NHAMCS database about the diagnosis, evaluation, and disposition (hospital admission) of all patients older than 59 years and treated in an ED for abdominal pain and compared these results with those obtained from the 2001 to 2002 GAPPS database. GAPPS was a 4-center study performed at largely urban academic EDs.

Results: During the NHAMCS period, there was an average of 101,525,553 annual visits to US EDs. Of these, an average of 7,238,772 (7.13%) were by patients 60 years or older with nontraumatic abdominal pain. We analyzed 1,217 records from the NHAMCS weighted to represent 5,293,074 ED visits. We report proportions from the 2 databases, NHAMCS and GAPPS, respectively. Demographically, both data sets reported a majority of female patients (68.5% versus 67.1%) and white patients ( to 21% of all ED visits, and the number of older patients visiting the ED is increasing. The objectives of this investigation are to describe the population of older adults receiving care in our ED and compare the older population who visits the ED with older clinic patients who do not visit the ED.

Methods: This was a retrospective cohort study using an electronic medical record system that routinely collects and stores data for each inpatient, clinic, and ED encounter within our health care system. The study was conducted at Wishard ED, an urban, public, university-affiliated Level I trauma center, and 11 outpatient primary care clinics. Our study population consisted of 2 cohorts. The ED cohort included all patients aged 65 years and older who attended our ED in 2001, and the clinic cohort included all patients aged 65 years and older who were treated in a Wishard primary care clinic in 2001 but were not treated in our ED in 2001.

Results: Forty percent of older ED patients arrived by ambulance. The mean time in the ED for older patients was 435 minutes, or 7.25 hours. Although our general adult admission rate was 9%, 27% of ED visits by older patients resulted in hospital admission. In 2001, our ED had a total of 105,463 visits. The number of ED visits by older patients was 7,562, and the number of clinic visits by older patients who did not attend the ED was 12,935. The number of older ED patients was 3,731, and the number of older clinic patients who were not treated in the ED was 2,925. Each cohort had a mean of 3.5 primary care clinic visits for the year. However, 31% of older ED patients were new to our entire health care system, whereas only 12.8% of older clinic patients who were not treated in the ED were new to our entire health care system. Conclusion: The ED was a principal site of care for older patients. Older ED patients are frequently acutely ill, requiring ambulance transportation and hospital admission. One third of older ED patients do not have regular primary care contact. The delivery of health care to older patients is fragmented and appears to be too dependent on emergency care rather than primary care.

Graziano JD, Lee DC, Tang X, Vernugopal-Patton R, Lukin M, Trimarco T/North Shore University Hospital, Manhasset, NY Study objectives: Visual impairment accompanies increased age, which may adversely affect the daily function of the elderly. There is controversy about whether elderly patients with physical impairments should drive and whether they are at increased risk for accidents. We studied the self-reported degree of visual impairment and driving habits in our elderly emergency department (ED) patients. Our goal was to determine the prevalence of visual impairment in elderly patients who drive.

Methods: We performed a prospective pilot study by using a convenience sample of patients in the ED of a suburban tertiary care hospital (annual census of 70,000). Data were collected between September 15, 2002, and April 30, 2003 . Inclusion criteria were ED patients older than 65 years who were clinically stable, able to accurately complete a verbal questionnaire, and able to give verbal consent. All patients with chief complaints related to vision were excluded from the study. All patients completed a standardized questionnaire and had a brief visual acuity screening examination with the Snellen eye chart. Descriptive statistics were used for data analysis.

Results: Two hundred thirty patients were enrolled in the study. Among them, 61% (145 of 230) patients admitted problems with vision; 59% (86 of 145) admitted to driving; 19% (26 of 140) admitted that they drive and could not read street signs at night; 77% (65 of 84) of patients between aged 65 and 75 years drove; 58% (61 of 106) of patients between aged 76 and 85 years drove; and 35% (13 of 37) of patients older than 86 years drove.

Conclusion: A large percentage of our elderly patients are regularly driving. There is also a substantial percentage of these patients who drive and have significant difficulties with their vision, which suggests further interventions to educate and improve the safety of this vulnerable subsegment of our population.

Eberlein CM, Stead LG, Vukov LF, Decker WW/Mayo Medical School, Rochester, MN Study objective: We describe the frequency of ECG abnormalities in the emergency department (ED) nonagenarian population.

Methods: The records of 1,077 consecutive patients aged 90 years and older presenting to the Saint Mary's Hospital ED between January 1, 2002, and December 31, 2002, were examined. Data on rhythm and QTc, QRS, and PR intervals were collected.

A categories: surgical emergency, medical emergency, surgical urgency, or medical urgency. Patients were assigned to 1 of these 4 categories according to final diagnosis and need for acute medical or surgical intervention. Necessity for acute medical or surgical intervention was determined by record review. Proportions with 95% confidence intervals (CIs) are reported. Results: Three hundred thirty-eight (85.4%) of 396 consenting patients were enrolled. The overall prevalence of use of CT scan in the evaluation of acute abdominal pain was 37% (124 of 338 patients; 95% CI 32% to 42%). Diagnostic sensitivity of CT scan for identifying abdominal pain etiology overall was 61% (76 of 124 patients; 95% CI 53% to 70%). CT scans were obtained in 71 of 155 (46%; 95% CI 38% to 54%) patients with medical or surgical emergencies. Diagnostic sensitivity of CT results in this subgroup for the etiology of abdominal pain was 75% (53 of 71 patients; 95% CI 63% to 84%). The sensitivity of CT scan for detecting surgical emergency was 85% (17 of 20 patients; 95% CI 62% to 97%). Three of 20 (15%; 95% CI 3% to 38%) patients with surgical emergency had abnormal but nondiagnostic CT scan results. The sensitivity of CT scan for detecting medical emergency was 71% (36 of 51 patients; 95% CI 56% to 83%). Likelihood ratio for the presence of surgical emergency with normal CT was 0 (95% CI 0 to 0.30). Likelihood ratio for the presence of either medical or surgical emergency with normal CT scan was 0.25 (95% CI 0.05 to 0.93).

Conclusion: The use of CT scan in the assessment of older ED patients with acute abdominal pain is highly prevalent. CT scan is highly sensitive for the detection of emergency surgical conditions. The presence of emergency surgical condition is unlikely in patients with normal CT scan results.

Grossman SA, Shapiro NI, Wolfe RE, VanEpp S, Arnold RC, Moore RB, Lee L, Lipsitz LA/Beth Israel Deaconess Medical Center/Harvard Medical School, Boston, MA Background: Elderly patients with syncope have been considered to be at high risk for adverse outcomes. The American College of Emergency Physicians (ACEP) recently published guidelines recommending admission for patients with syncope and any of the following criteria: (1) history of congestive heart failure; (2) ventricular arrhythmias; (3) associated chest pain or other symptoms compatible with acute coronary syndromes; (4) valvular heart disease on examination; (5) ECG findings of ischemia, arrhythmia, prolonged QT, or bundle branch block; (6) history of coronary artery disease or congenital heart disease; and (7) older than 60 years.

Study objective: We studied the value of the first 6 criteria (excluding age itself) in predicting adverse outcomes among elderly patients admitted with syncope.

Methods: We performed a retrospective study of 323 consecutive patients older than 65 years and presenting with syncope to an academic emergency department; 230 of these patients were admitted. Clinically significant outcomes included myocardial infarction, death, and cardiac interventions, including cardiac catheterization, electrophysiologic studies, pacemaker or implantable cardioverter defibrillator placement or adjustment, echocardiography, or cardiac medication changes.

Results: Of 230 patients, 45 (19%) had clinically significant cardiac outcomes; 111 (48%) of 230 met 1 or more of the ACEP criteria; 119 patients had none of these criteria. In the group who met no criteria other than age, there were 9 (7.6%) of 119 patients with clinically significant outcomes or interventions compared with 36 (32.4%) of 111 in the group with 1 or more criteria (P>.001). Thus, patients with 1 or more of the ACEP criteria were 4.3 times more likely to have a clinically significant outcome or intervention than those not meeting any criteria. However, the criteria had a sensitivity of 80% (95% confidence interval [CI] 68% to 91%) and specificity of 60% (95% CI 52% to 66%) for detecting patients with adverse outcomes and identifying patients without, respectively.

Conclusion: Although ACEP guidelines identify many high-risk syncope patients even in the elderly population, if age alone is discarded as an admission criterion, a large number of patients would still be sent home without a cardiac intervention, potentially leading to significant morbidity and mortality. years) presenting to an academic, urban emergency department (ED) during a 16week period with UTI suggested by urinalysis results and pertinent symptoms. Thirty-seven demographic and clinical variables were analyzed as potential predictors of outcome. Morbidity was defined as inhospital death, ICU admission, hospital length of stay longer than 2 days, or receipt of hospital intravenous antibiotics longer than 2 days. Factors identified by univariate analysis were combined by using multiple logistic regression to identify independent predictors of morbidity.

Results: Two hundred eighty-four patients met selection criteria. Thirteen (4.6%) died during hospitalization, 27 (9.5%) had ICU admission, 139 (48.9%) had length of stay longer than 2 days, and 75 (26.4%) received intravenous antibiotics longer than 2 days. Multivariate analysis identified the following variables as independent predictors of adverse outcomes: mental status change, frequent UTIs, other nonurinary infections, temperature of at least 37.7°C (99.8°F) or less than 35.0°C (<95°F), pulse rate 100 beats/min or greater, systolic blood pressure less than 90 mm Hg, blood urea nitrogen level 30 mg/dL or greater, glucose level 250 mg/dL or greater, WBC count at least 11.0/mm 3 or less than 5.0/mm 3 , and neutrophils 80% or greater. Regression models for adverse outcomes had sensitivities from 74.8% to 96.2% and specificities from 31.1% to 69.0%.

Conclusion: This study defines high rates of morbidity for geriatric patients with UTIs and describes predictive variables that may help identify low-risk patients. These data may lay the foundation for determining specific guidelines for disposition of this high-risk patient population. Methods: We conducted a prospective cohort study of adults presenting to an urban ED with UAP during 28 consecutive days. Patients underwent scripted interview and focused physical examination at the ED visit. UAP was defined as abdominal pain of at least an hour's duration that was nontraumatic; not previously diagnosed; not primarily in the flank; not associated with pregnancy, vaginal bleeding or discharge, frequency, dysuria or foul-smelling urine; and not associated with upper or lower gastrointestinal tract bleed. Physicians were asked for their clinical impression about whether the patient disposition based solely on history and physical examination should be (1) treat and release; (2) observe positive and negative test results and probably discharge; (3) observe positive and negative test results and probably admit; or (4) admit to hospital. Patients were followed up for up to 3 months to determine the most reasonable final diagnosis. We also tested whether certain combinations of history and physical examination predicted need for diagnosis. Statistics are presented as odds ratios (ORs) with 95% confidence intervals (CIs).

Results: Two hundred twenty-four patients presented during the study period with UAP. Ninety-two (41%) patients had detailed study forms filled. Clinical impression based on history and physical examination had a strong association with final ED disposition (OR 9.6; 95% CI 3.4 to 26.9), as well as for final diagnosis at 3month follow-up (OR 3.5; 95% CI 1.1 to 11.1).

Conclusion: Our data suggest that history and physical examination alone is a strong predictor of final ED disposition and final diagnosis at 3 months. Further studies should focus on determining which subgroups of history and physical examination actually need confirmatory laboratory testing or imaging studies in the ED and which subgroups may require more expeditious admission.

McNaughton T, Steele R, Coba V/Loma Linda University Medical Center, Loma Linda, CA Background: T-wave alternans (TWA) is a noninvasive tool for identifying patients at risk for ventricular arrhythmias. Focal areas of poorly conducting myocardium (such as that produced by ischemia or scarring), as well as emotional stress, significantly lower the threshold heart rate at which TWA appears.

Study objective: We prove TWA is present at rest in emergency department (ED) chest pain patients and may be useful in identifying ED chest pain patients with acute cardiac syndrome (ACS).

A prolonged QTc interval was defined as at least 440 ms. A prolonged QRS interval was defined as at least 100 ms. A prolonged PR interval was defined as at least 200 ms. Data for all consecutive patients in the mentioned period were captured; there were no missing records.

Results: Of the 1,077 total patients, 532 had an ECG performed in the ED. Of these, 243 (46%) patients had a prolonged QTc interval. Two hundred (38%) patients had a prolonged QRS interval. Of the 389 patients who had a measurable PR interval, 132 (34%) had a prolonged one. The most common rhythm was a sinus rhythm (72%). Of those patients in sinus rhythm, 9% had tachycardia and 12% had bradycardia. The second most common rhythm (115 patients, 22%) was an atrial tachycardia (atrial fibrillation, atrial flutter, atrial tachycardia, or multifocal atrial tachycardia). Fully 20% had atrial fibrillation.

Conclusion: ECG intervals are prolonged in more than one third of all patients older than 90 years who present to an ED and receive an ECG. Up to 20% of nonagenarians who have an ECG in the ED have atrial fibrillation.

Caterino JM, Scheatzle MD/Allegheny General Hospital, Pittsburgh, PA Study objectives: The objective of this study was to characterize the usefulness of serum procalcitonin (PCT) in the diagnosis of bacterial infection in an elderly emergency department (ED) population. The primary hypothesis was that PCT would be more sensitive and specific than WBC count in the diagnosis of bacteremia in this patient population. The secondary hypothesis was that PCT would be superior to WBC count in detecting the presence of any bacterial infection.

Methods: A prospective blinded cross-sectional study of all ED patients aged 65 years and older for whom blood cultures were drawn was conducted. The study was undertaken in an urban tertiary care hospital ED. Serum for PCT was obtained at ED visit. PCT was measured by an immunoluminometric technique (LUMItest PCT, BRAHMS Diagnostica, Berlin, Germany). Receiver operating characteristic (ROC) curves and proportions were calculated for PCT and WBC in the diagnosis of bacteremia alone and of the presence of any bacterial infection.

Results: One hundred eight patients were enrolled. Fourteen (13%) patients had bacteremia, 78 (72%) had a bacterial infection, and 30 (28%) had no bacterial infection identified. PCT level was significantly higher in bacteremic than nonbacteremic patients; WBC count was not significantly higher. A serum PCT value higher than 0.2 ng/mL had sensitivity of 92.8% (95% confidence interval [CI] 79.3% to 100%), negative predictive value of 97.3% (95% CI 92.1% to 100%), and specificity of 38.3% (95% CI 28.5% to 48.1%) in the diagnosis of bacteremia. Abnormal WBC count had sensitivity of 64.3% (95% CI 39.2% to 89.4%) and specificity of 53.7% (95% CI 43.6% to 63.8%) for diagnosing bacteremia. Presence of an abnormal WBC count or a left shift had sensitivity of 92.8% (95% CI 73.9% to 100%) but specificity of only 10.6% (95% CI 4.4% to 16.8%) for diagnosing bacteremia. Area under an ROC curve was significantly greater for PCT than for WBC count in the diagnosis of bacteremia. In detecting the presence of any bacterial infection, PCT level more than 0.5 ng/mL had sensitivity of 39.7% (95% CI 28.8% to 50.6%), specificity of 83.3% (95% CI 70% to 96.7%), and positive predictive value of 86.1% (95% CI 74.8% to 97.4%). In the detection of any bacterial infection, abnormal WBC count alone had sensitivity of 50% (95% CI 38.9% to 61.1%) and specificity of 63.3% (95% CI 46% to 80.5%), whereas the presence of either WBC count or left shift had sensitivity of 89.7% (95% CI 83% to 96.4%) and specificity of only 16.7% (95% CI 3.4% to 30%).

Conclusion: Serum PCT level is superior to WBC count for the diagnosis of bacteremia according to ROC curve analysis. A serum PCT cutoff of 0.2 ng/mL is highly sensitive for the diagnosis of bacteremia and may be useful in ruling out the diagnosis. A serum PCT level of more than 0.5 ng/mL indicates a high likelihood that some bacterial infection is present. WBC count alone is insensitive for the diagnosis of bacteremia. Presence of a left shift increases sensitivity but results in worse specificity than serum PCT. In determining that some bacterial infection is present, WBC count with or without left shift is less accurate then serum PCT.

Ginde AA, Rhee SH, Katz ED/ Washington University School of Medicine, St. Louis, MO Study objective: We determine the prognosis of geriatric patients with urinary tract infections (UTIs) and identify clinical factors associated with adverse outcomes.

Methods: This retrospective, cohort study identified elderly patients (aged ≥65

construct an ordinal logistic regression model to predict quintile of troponin I peak and were validated with a bootstrap procedure. These results were compared with a binary logistic regression model constructed to predict abnormal troponin I peak according to the same data and were subjected to the same validation procedure. Results: The ordinal logistic regression model constructed has a global likelihood ratio of 281 with df equal to 18 (P<.0001). Moreover, the model has an outcome c index of .889 (0.5 = random correlation, 1.0 = perfect correlation) with a Somer's Dxy rank correlation (Dxy) of 0.778 (0.0 = random discrimination, 1.0 = perfect discrimination). Validation revealed a corrected Dxy of 0.7390.

The binary logistic regression model constructed has a global likelihood ratio of 258 with df equal to 18 (P<.0001). Moreover, the model has an outcome c index of .908 with a Dxy 0.812. Validation revealed a corrected Dxy of 0.7493.

Conclusion: Initial ECG criteria can be used within an ordinal logistic regression model to predict abnormal troponin I levels in adult patients presenting with chest pain. These findings may assist in rapidly stratifying high-risk adult chest pain patients for aggressive treatment.

Thomas SH, Winsor GR, Pang PS, Driscoll KA, Parry BA/Massachusetts General Hospital, Boston, MA; Harvard Medical School, Boston, MA; Boston MedFlight, Boston, MA Study objectives: The goal of this study was to test accuracy of a novel device that uses continuous partial radial artery compression to assess mean arterial pressure (MAP) and other variables. If it proves reliable, the device, which provides data noninvasively and near-continuously (updates every 10 to 15 heartbeats), has appeal for many emergency department (ED) situations.

Methods: The study was a prospective, nonblinded, convenience-sample trial. The study hospital was a Level I trauma center with ED annual volume of 70,000. Study participants were adult (>18 years) ED patients who had indwelling radial arterial pressure catheters and who had palpable and accessible radial pulse on the contralateral arm. MAPs were assessed simultaneously by 3 modalities: test device (TEST group), oscillometric brachial artery cuff (OSC), and arterial line (ART). MAPs were assessed by using visual (Bland-Altman plot), continuous (analysis of variance and 1-sample t test for means, r 2 to assess the proportion of variance of TEST or OSC MAP data accounted for by ART), and categorical (proportion of TEST and OSC readings within an a priori-selected margin of ART ±5 mm Hg) analyses. Significance for all tests was set at the .05 α level.

Results: The study population was composed of 15 patients (9 men, 6 women) ranging in age from 28 to 89 years (median 62 years; interquartile range 38 to 79 years) and with the following diagnoses: neurologic (n=7; 46.7%); cardiac (n=2; 13.3%); other medical (n=6; 40.0%). There was no difference between the 3 groups' MAP means (analysis of variance P=.98), and paired t test failed to identify a means difference between TEST and ART (P=.73) or between OSC and ART (P=.43). The r 2 values for ART/OSC and ART/TEST were 0.96 and 0.95, respectively. TEST and OSC MAP readings were equally likely to be within 5 mm Hg of the ART standard in the overall set of 307 MAPs ( Conclusion: As assessed by a number of mechanisms, and in a patient population including significantly hypotensive cases, the TEST device MAP assessments showed a strong statistical correlation to ART MAPs. The TEST device performed at least as well as simultaneous oscillometric cuff assessment but offered the unique advantage of providing data noninvasively and near-continuously. These characteristics indicate a high potential for ED applicability and should prompt further investigation of the device for use in other acute care settings.

Pereira E, Richard O, Cantineau JP, Clero J, Moro J, Hazan S, Lambert Y/Versailles Hospital, Le Chesnay, France Study objective: Measurement of end-tidal CO 2 pressure (ETCO 2 ) is usually used to estimate PaCO 2 , but the validity of this estimation remains questionable. We examine the agreement between ETCO 2 and PaCO 2 on scene in mechanically ventilated and sedated patients with severe disease. Methods: A convenience sample of patients (N=100) who had a chest pain evaluation in the ED were monitored for 8 minutes with a Microvolt TWA monitor. Correlation was made between TWA results and final patient outcome, as determined by serial cardiac markers, ECG changes, angiogram, and final diagnosis.

Results: Positive TWA defined as more than 2 µV was seen in 19 patients. Lowamplitude TWA (1 to 2 µV) was seen in 43 patients. Negative (no measurable) TWA was seen in 25 patients. Indeterminate (noisy) tracings occurred in 13 patients. Nineteen ruled in for myocardial infarction (MI) or had ACS, but all had at least 1 Microvolt TWA test. All patients with a negative TWA tracing tested negative for ACS. TWA had 100% sensitivity for MI/ACS (95% confidence interval [CI] 79% to 100%) and a negative predictive value of 100% (95% CI 83% to 100%).

Conclusion: TWA is present in ED chest pain patients at rest. TWA testing has a high sensitivity but poor specificity for ischemia-related ED chest pain patients. Further research is needed to conclusively determine whether TWA will become the "rule-out" test of tomorrow.

Schreck DM, Brotea C, Shah SP, Frank JD/Summit Medical Group, Summit, NJ Background: The ECG may be described as a voltage-time lead-vector array and may be derived from 3 standard leads by using a simplex optimization transformation matrix. Factor analysis is a matrix calculus technique performed on each array to calculate the lead-vector eigenvalues (EV) of the information space in the ECG.

Study objective: We derive the EVs of normal and infarction, measured and derived 12-lead standard ECGs and determine whether the EV magnitudes can distinguish acute ECG pathology.

Methods: Twenty-two standard 12-lead ECGs were acquired and digitized from 5 healthy women, 6 healthy men, and 5 women and 6 men with acute myocardial infarction, yielding 300 voltage-time points for each of the 8 measured ECG leads. Factor analysis was applied to each 300×8 voltage-time array for each patient to derive the 8 EVs, 1 for each contributing lead-vector, from which each EV percentage contribution was calculated. Each of the 22 measured ECGs was also derived from leads I, aVF, and V2 by using a simplex optimization transformation matrix. Analysis of variance was used to test for the statistical significance of the contribution of each EV percentage to the information content in each patient array.

Results: No significant morphologic differences between measured and derived ECGs were noted. Significant differences were detected at EV 3% (P=.0001), EV 4% (P=.017), EV 5% (P=.003), EV 6% (P=.02), and EV 7% (P=.049) for measured ECGs. Significant differences were also detected at EV 3% (P=.0028) for derived ECGs (EV4-EV8 are 0 for the derived ECG). The sum of EV 3% through EV 8% was plotted and demonstrated complete differentiation of normal ECGs from infarction ECGs in all 22 cases for measured and derived ECGs.

Conclusion: This study demonstrated that the EVs in a standard ECG may differentiate normal from acute myocardial infarction pathology. With current computer technology, this calculation may occur instantaneously in real time to enhance patient observation capabilities. This study also demonstrated that the 12-lead ECG can be derived from a simplex optimization transformation matrix by using only 3 measured leads from which the EVs can be calculated. We propose that the ECG eigenvalues may be a new marker for real-time continuous prediction of acute myocardial infarction.

Kurz MC, Brady WJ, Beagle C, Perron AD, Ullman EA, Holstege C, Riviello RJ, Ripley A, Ghaemmaghami CA/University of Chicago Hospital, Chicago, IL; University of Virginia, Charlottesville, VA; Beth Israel Deaconess Medical Center, Boston, MA; Thomas Jefferson University, Philadelphia, PA Study objectives: We determine whether troponin I peak can be predicted according to characteristics of initial ECG by using an ordinal logistic regression model.

Methods: This was a retrospective study of 598 consecutive adult chest pain patients who presented to a university hospital emergency department (ED) and underwent a rule-out myocardial infarction evaluation, including ECG and serially measured troponin I levels at intervals longer than 3 hours. In addition to patient age and sex, the following covariate data were collected from the initial ECG: total ST-segment elevation (STE), the anatomic distribution of STE, number of leads in which STE appeared, amount of STE per lead, morphology of STE, total ST-segment depression, and morphology of ST-segment depression. These data were used to Methods: This is a retrospective case series of 55 patients diagnosed with NF between June 1993 and June 2001. Data collected included demographics, presenting signs and symptoms, risk factors, course of disease, comorbidities, microbiology and pathology reports, and final outcomes. Eligibility criteria were microbiology, pathology reports, and patient records consistent with NF.

Results: Of the 83 patients reviewed, 53 had documentation of NF and were eligible for further inclusion. The mean age of patients with NF was 44 years (range 19 to 78 years), and men accounted for nearly three quarters of the cases. Compared with other sites, patients tended to have disease on the extremities (43 of 53 [81%]). The time to clinical presentation was delayed, the median being 10 days (range 6 to 16 days). Risk factors, including diabetes and intravenous drug user, accounted for two thirds of the cases. The classic presentation of fever, edema, and blisters was present in only 4 cases. Laboratory findings were significant for low serum sodium levels, with a mean sodium level of 133 mEq/L, and leukocytosis. The mortality in this group was 17%, and the median length of hospital stay was 29 days (range 3 to 30 days).

Limitations: Medical record reviews do not accurately predict the true prevalence of NF in the population. Further study should be done to evaluate factors that may help clinicians make this difficult diagnosis.

Conclusion: NF in this cohort occurred more often in male patients who presented after a significant delay with fever, edema, and involvement of an extremity. The group was young, had risk-taking behavior, and remained in the hospital for prolonged periods once diagnosed. The mortality in this group was lower than previously reported.

Trzeciak S, Sharer R, Piper D, Chan T, Kessler C, Pursell KJ/University of Illinois at Chicago, Chicago, IL; University of Medicine and Dentistry of New Jersey, Camden, NJ Study objective: We provide a descriptive analysis of infectious processes in organ transplant patients admitted from a university-based emergency department (ED).

Methods: A database of all adult organ transplant recipients at a university medical center was cross-referenced with a computerized record of all ED visits to the same medical center throughout an 18-month period to identify ED visits by transplant patients. ED records, inpatient records (if applicable), and microbiology data were reviewed retrospectively. Final diagnoses and outcomes were analyzed.

Results: There were 352 ED visits by 158 transplant patients (kidney 66%, kidney/pancreas 15%, liver 13%, lung 3%, heart 3%). The admission rate was 62% (versus 19% for the general adult ED population; P<.001). Infections were the most common indications for admission (77 of 217; 35%). Urinary tract infection (33 of 77; 43%) and pneumonia (18 of 77; 23%) were the most common infections. Nine (11.7%) of 77 patients with documented infections developed severe sepsis (sepsis plus organ dysfunction), requiring ICU admission. Seven of these patients had severe sepsis at ED presentation, and 2 additional patients progressed to severe sepsis shortly after hospital admission. Severe sepsis was the most common reason for ICU use.

Conclusion: Transplant patients presenting to the ED have a high incidence of infections requiring hospital admission and a high incidence of severe sepsis. The emergency physician must have a high index of suspicion for infections when evaluating organ transplant recipients.

DeFlitch C, Greiner R, Fox K/Penn State Hershey Medical Center, Hershey, PA Study objective: Clinical criteria and D-dimer studies have suggested some efficacy in screening for deep venous thrombosis (DVT). Strain gauge plethysmography has shown promise in screening for DVT in Europe. The study seeks to determine which study is most effective in screening in the emergency department (ED).

Methods: This was a prospective, blinded study of adults (>18 years) who presented to an academic ED throughout an 18-month period, with suspected (by the attending emergency physician) DVT. After obtaining written consent, the physician recorded standardized signs and symptoms on a form to determine prospective clinical risk. A D-dimer was performed, and the patient underwent computer-assisted strain gauge plethysmography, performed by the emergency technicians in the ED.

Methods: Eighty-seven patients were included retrospectively. One hundred eight gradient values were calculated. The blood gases were analyzed by fluorimetry and spectrophotometry of reflectance to 3 wavelengths (optiCCA, Roche Diagnostic) after a 20-minute period of stable mechanical ventilation. The ETCO 2 was continuously measured with a mainstream sensor (Propaq). The correlation between ETCO 2 and PaCO 2 was determined by Bland-Altman analysis and linear regression.

Results: The mean age of patients was 50 years. The sex ratio (men:women) was 56:31. Among these 87 patients, 27 had trauma, 28 had neurologic diseases, 17 had poisoning, and 15 had cardiac or respiratory failures. All the patients were intubated after a rapid sequence induction and sedated with midazolam and fentanyl. The mean arterial pressure was 127 mm Hg (SD 38 mm Hg). The mean initial SpO 2 was 97% (SD 3). In the studied population, ETCO 2 underreported PaCO 2 (y=1.2x; correlation coefficient, 0.38; P<.01; bias, 8 mm Hg; precision, 10 mm Hg; limits of agreement, -11 to +27 mm Hg). Gradients were 12 mm Hg (SD 10) for cardiac or respiratory failures, 10 mm Hg for poisoning (SD 7), and 5 mm Hg (SD 9) for neurologic diseases. The gradient was 8 mm Hg (SD 10) for trauma patients. Among them, 24 patients had a severe head injury, with a 10 mm Hg (SD 13) mean gradient. For those patients, 10 kinetics (at least 2 measurements) were drawn up. Final gradients were significantly lower than initial ones (P<.005, Student's t test).

Conclusion: This study confirms that, in early treatment, ETCO 2 is not a reliable tool to estimate PaCO 2 because of this large gradient PaCO 2 -ETCO 2 (8 mm Hg). Furthermore, this gradient is unpredictable because of intraindividual and interindividual variations. Only repeated point-of-care arterial blood gas testing gives the correct PaCO 2 to adapt the ventilation of patients in this setting. These conclusions make particular sense for patients with severe head injury when the strict control of PaCO 2 is fundamental and ventilation goals are well established.

Gerard C, Gerard J, Spielberger CD, Lunsford D/Appalachian State University, Boone, NC; Manatee Memorial Hospital, Brandenton, FL; University of South Florida, Tampa, FL Study objectives: We determined the extent to which emergency physicians and nurses assess levels of anxiety and depression in patients with physical symptoms. We hypothesized that patients' self-reports of anxiety and depression would correlate more highly with diagnostic criteria than either physicians' or nurses' assessments.

Methods: We performed a prospective observational study in a large suburban community emergency department (ED). We administered 2 previously validated self-report measures of anxiety and depression, the State Trait Personality Inventory (STPI) and the Brief Patient Health Questionnaire (PHQ) of the Primary Care Evaluation of Mental Disorders. Physician and nurse ratings of patients' anxiety and depression were obtained according to operational definitions. We calculated scores on STPI and PHQ and the ratings by emergency physicians and nurses. The main outcome measures were the degree of correlation between STPI and PHQ and between PHQ and the ratings assigned by emergency physicians and nurses.

Results: A total of 222 (5.4%) of all patients were approached; 207 agreed to participate, and 193 completed the study. The correlation between the patients' scores on PHQ and STPI for anxiety (0.427; 95% confidence interval [CI] 0.305 to 0.535) and depression (0.685; 95% CI 0.603 to 0.753) was significantly greater than the correlation between the scores on PHQ and either the emergency physicians' ratings of anxiety (0.178; 95% CI 0.043 to 0.313) and depression (0.223; 95% CI 0.085 to 0.353) or the nurses' ratings of anxiety (0.085; 95% CI 0.056 to 0.223) and depression (0.149; 95% CI 0.008 to 0.284). Conclusion: Self-report measures of anxiety and depression correlated more highly with diagnostic criteria than either physicians' or nurses' subjective ratings. These findings suggest that self-report questionnaires could be useful in the assessment and management of emotional states in the ED.

Stone SC, Avelar J, Henderson SO/Keck School of Medicine, University of Southern California, Los Angeles, CA Background: Necrotizing fasciitis (NF) is a rapidly progressive tissue infection. Despite high morbidity and mortality, studies of this disease are limited because of the low incidence of 500 to 1,500 cases per year in the United States.

Study objectives: We describe the clinical presentation, risk factors, and outcomes of patients with NF presenting to our medical center.

Results: Of 145 names selected, 101 were possible to contact, and 9% completed the survey. Of the participants, 11% had been state chapter officers, no one had been a national ACEP officer, 11% had served on committees, 78% had attended state chapter meetings (no one currently or regularly attended state chapter meetings), 89% had attended at least 1 ACEP Scientific Assembly, 11% were able to name the current ACEP president, no one could name their current state chapter president, 78% were board certified, 67% were Fellows of ACEP, 33% had considered leadership roles within ACEP, and the average number of years in practice was 21. In addition, 11% had been members of the American Academy of Emergency Medicine but no one was currently a member, and 33% had not heard of the American Academy of Emergency Medicine. Survey participants also identified reasons why they were members and what the benefits of membership were for them. Lobbying at the national level and legislative work were identified by 67% as the major benefit and reason for belonging to ACEP. Annals of Emergency Medicine and educational offerings from ACEP were identified by 56% as reasons to belong to the society. Eleven percent belonged to ACEP only because they felt some obligation to belong to the largest professional society in the specialty of emergency medicine.

Conclusion: A surprisingly small number of ACEP members are aware of who their leadership is. A surprisingly small number are involved, have been involved, or have considered being involved in ACEP leadership roles. Most members identified national legislative action and lobbying as the main benefits of ACEP. Background: Decisions about the use of prescription drugs lie predominantly in the hands of practicing physicians. Physician prescription decisions offer the potential of large cost savings if less expensive therapeutic alternatives or generic drugs can be chosen. Drug cost knowledge is essential to make a cost-sensitive decision. Little information exists about emergency physician knowledge of actual costs of medications.

Study objective: We assess emergency physician knowledge about the actual patient cost of commonly prescribed antibiotics used in the outpatient setting.

Methods: An institutional review board-approved anonymous written survey of emergency medicine house staff and attending emergency physicians was administered at multiple sites during November and December 2002. Respondents were asked to estimate the actual cost to the patient of commonly prescribed antibiotics for a treatment course with a specified dose and quantity of medication. Actual cost of the medication was determined by a survey of local pharmacies. Responses were determined to be correct if the estimate of cost was within 2 SDs of prices found in the pharmacy survey.

Results: Surveys were returned by 158 physicians (response rate of 60%) and were used in the analysis; respondents were 71% residents, 29% attending emergency physicians. Survey respondents characterized their knowledge of patient prescription costs as very knowledgeable 2%, somewhat knowledgeable 55%, very little knowledge 41%, and no knowledge 2%. Less than 50% of the physician estimates of prescription cost were in range from known prices from local pharmacies. Eightyfour percent of physicians report that knowledge of patient prescription costs would alter the choice of antibiotic in prescribing to a self-pay patient.

Conclusion: This pilot study suggests that emergency physician knowledge about the actual patient cost of commonly prescribed antibiotics is limited.

Bhatia K, Finkel M, Emond JA, Camargo CA/Massachusetts General Hospital, Boston, MA Study objective: We describe the epidemiology of emergency department (ED) visits for ectopic pregnancy and compare any identified risk factors with those from other population-based studies.

Methods: Data were obtained from the 1992 to 2000 National Hospital Ambulatory Medical Care Survey. Cases had International Classification of Diseases, Ninth Revision code 633 and were restricted to ages 15 to 39 years. National estimates were obtained by using assigned patient visit weights. Rates per 1,000 live births were computed by using US population data for annual births. Data analysis used χ 2 and logistic regression.

Results: From 1992 to 2000, there were 590,000 ED visits for ectopic pregnancy, All results were recorded. Patients then underwent a venous duplex within 12 hours, with the sonographer being blinded to the results of all 3 screening tools. Results: Of the 131 patients enrolled, 27 (20.6%) were found to have DVT on venous duplex. Deep venous thrombosis was formed in 8 (24.2%) of 33 patients with high clinical risk by Wells' criteria, 16 (23.5%) of 68 patients at moderate risk, and 3 (10.0%) of 30 low-risk patients, excluding 1 chronic, nonobstructive DVT. D-Dimer had a sensitivity of 88.9%, with a specificity of 19.3%. Strain gauge plethysmography alone showed a sensitivity of 52.6%, with a specificity of 65.2%. Twenty-seven percent of the plethysmography studies were invalid (not determined positive or negative according to the computer-generated curve).

Conclusion: Although negative D-dimer or low-risk clinical criteria can suggest absence of DVT, no individual study can reliably screen for DVT. Computer-assisted strain gauge plethysmography was difficult for the technicians to analyze and effective as an individual screening study for DVT.

Garfield J, Birkhahn R, Tloczkowski J, Gaeta T/New York Methodist Hospital, Brooklyn, NY Background: Patients with a classic history of and physical examination for acute appendicitis are taken for operative exploration without delay, whereas those with an equivocal presentation undergo further imaging.

Study objective: We determine the use of and diagnostic delay associated with abdominal computed tomography [CT] and ultrasonography in the evaluation of acute appendicitis at an urban teaching hospital.

Methods: This was a retrospective medical record review of all patients with the preoperative diagnosis of appendicitis who underwent operative exploration between February 2002 and February 2003. Cases were categorized by the presence or absence of advanced radiographic imaging. Records were reviewed to obtain demographic information and intervals from triage to operative incision (reported as means with 95% confidence intervals [CIs]). Linear regression was used to identify variables associated with operative delay.

Results: During the review period, 124 patients underwent operative exploration for presumed appendicitis (41 with no advanced radiographic evaluation, 67 with CT only, 8 with sonogram only, and 8 with CT and sonogram). When no preoperative imaging was performed, a mean of 6.2 hours (95% CI 5.4 to 7.0 hours) elapsed between triage and operative incision (10% perforated, 5% negative laparotomy). In patients who received a CT only, it took 9.2 hours (95% CI 8.3 to 10.2 hours) to obtain the CT and an additional 6.2 hours (95% CI 5.2 to 7.2 hours) before the patient went to surgery (9% perforated, 6% negative laparotomy). When only a sonogram was performed, 10 hours (95% CI 7.4 to 12.5 hours) elapsed before operative intervention (no perforations, 12% negative laparotomy). The combination of a CT and a sonogram delayed operative time by 15.8 hours (95% CI 13.1 to 18.5 hours; no perforations, 12% negative laparotomy). A linear regression found only a weak correlation (R 2 =0.16) in a model including age and sex as independent predictors of time to surgery.

Conclusions: A third of the patients at our institution with acute appendicitis underwent operative exploration without advanced radiographic imaging; the remaining two thirds had a diagnostic delay between 9 and 15 hours, with no change in the perforation or negative laparotomy rate. Once the decision was made, it took approximately 6 hours to get the patient to surgery, regardless of how the therapeutic decision was made.

Brozen R/Dartmouth-Hitchcock Medical Center, Lebanon, NH Study objectives: It is a common belief that leaders must mentor and train future leaders. The American College of Emergency Physicians (ACEP) as a large organization has many leadership positions, including positions at the national, state chapter, and section level, as well as many committee positions, to fill. Many smaller state chapters and sections have difficulty identifying possible leaders and filling leadership positions. This study attempts to identify which ACEP members have knowledge of who their leaders are and whether or not they might be interested in leadership positions.

Methods: A telephone survey was conducted of randomly selected ACEP members. A standardized form was read to describe the survey and request participation. Participants were asked a list of questions in a standard order and format. Results were tracked on a standard Excel spread sheet. disease estimates, and to assess whether this evidence-based medicine method can be expected to yield consistent posttest estimates of disease.

Methods: This cross-sectional cohort study was conducted at an urban county teaching hospital by using a convenience sample of 61 emergency and internal medicine residents and faculty and ED midlevel practitioners. Participants received 4 clinical vignettes designed to raise consideration for acute myocardial infarction, pulmonary embolism, deep venous thrombosis, and urinary tract infection. Clinicians were asked to estimate the likelihood of the suggested diagnosis according to medical history and physical examination findings alone because no information about laboratory findings or imaging studies was provided.

Results: Mean pretest probability estimates of disease ranged from 42% for pulmonary embolism (95% confidence interval [CI] 36.6% to 47.4%) to 77% for urinary tract infection (95% CI 72.9% to 81.1%). The smallest difference in pretest probability magnitude for a single vignette was 70% in the urinary tract infection case (range 10% to 100%; interquartile range [IQR] 64% to 80%), whereas the largest was 95% in the acute myocardial infarction scenario (range 3% to 98%; IQR 30% to 60%). There were no significant differences between the emergency and internal medicine subgroups on pretest probability estimates of disease.

Conclusion: Wide variability in clinicians' pretest probability estimates of disease may present a possible concern about decisionmaking models based on Bayes' theorem because such variation may ultimately yield inconsistent posttest disease estimates. Perhaps to stabilize the initial step of this evidence-based medicine process and achieve more consistent pretest probability determinations by clinicians, future research might generate clinical decision rules for determining pretest probabilities of different disease processes.

Niegelberg E, Jimenez M, Singer AJ/State University of New York, Stony Brook, NY Study objectives: There is concern that there is a tendency for hospitals to transfer patients with no health insurance to tertiary trauma centers. This analysis was conducted to determine whether there is a relationship between the transfer of emergency department (ED) trauma cases to a tertiary trauma center (TC3) and insurance status.

Methods: This was a retrospective analysis in a suburban/rural county served by 12 hospitals, including 4 secondary trauma centers (TC2s) and 1 TC3. Patients were included if they presented to an ED at a community hospital or TC2 in Suffolk County, NY, and were discharged in 2000 and met the New York State eligibility requirements for inclusion in the State Trauma Registry. Pediatric (<18 years) and burn patients were excluded. Data collected for trauma patients included transfer status to the TC3 (dependent variable), originating hospital, age, Injury Severity Score, and insurance payer. Insurance status categories included no fault, Medicaid, commercial, and self-pay. χ 2 Tests and univariate logistic regression were used to identify associations between the dependent variable and categorical and continuous variables, respectively. Multivariate logistic regression was used to determine the association of insurance status with transfer rates after controlling for potential confounding factors.

Results: There were 705 (98%) of 717 cases with complete data; 109 (15%) patients were transferred to the TC3. Mean age was 55 years, and mean Injury Severity Score was 14. Medicaid patients were more likely to be transferred (45%) and no fault patients were least likely to be transferred (12%), whereas 16% each of commercial and self-pay patients were transferred (P=.001). Community hospitals transferred patients more often than TC2s (31% versus 4%; P<.001). Increasing Injury Severity Score was also associated with being transferred to the TC3 (P<.001). Controlling for other factors did not change the association between insurance status and patient transfer.

Conclusion: Although insurance status is associated with transfer to a TC3, patients with no insurance do not appear to be transferred at a higher rate than other patients. Study objectives: We determine emergency physician adherence to national asthma guidelines; surveys were collected examining self-reported assessment, treat-accounting for 0.3% of all ED visits for women aged 15 to 39 years. Overall, there were 16.5 cases per 1,000 live births, and this rate differed by age: 23.6 for patients aged 15 to 19 years, 15.0 for patients aged 20 to 24 years, 13.9 for patients aged 25 to 29 years, 16.6 for patients aged 30 to 34 years, and 21.6 for patients aged 35 to 39 years. Overall, the ectopic pregnancy rate per 1,000 ED visits was 31.6, with rates lowest for patients aged 35 to 39 years (23.0) and highest for those aged 30 to 34 years (36.0). ED visit rates did not differ by race or ethnicity. Cases did not differ from noncases by geographic region, season, time of presentation, or insurance type (private, public, self-pay, or other). Of patients, 68% (95% confidence interval [CI] 60% to 77%) complained of abdominal pain or discomfort and 43% (95% CI 33% to 53%) reported problems related to pregnancy (such as pain and bleeding). The overall admission rate was 44% (95% CI 34% to 53%). In a multivariate model (controlling for age, race, ethnicity, US region, and insurance status), we did not identify any statistically significant predictors of ED presentation with ectopic pregnancy.

Conclusion: Past studies have identified age, race, and insurance status as risk factors for ectopic pregnancy. These findings are not consistent with current ED presentations and suggest differential presentation to the ED among women at risk of ectopic pregnancy.

Levitt AM, Phelps MA/Alameda County Medical Center-Highland Campus, Oakland, CA Study objective: Likelihood ratios (LR) incorporate the stability of diagnostic tests' sensitivities and specificities; however, there is minimal scientific validation for their application in clinical scenarios. This study examines the impact of a range of positive and negative LRs on clinician confidence about treatment decisions and whether clinicians perceive this evidence-based medicine approach as helpful.

Methods: This cross-sectional cohort study was conducted at an urban county teaching hospital by using a sample of emergency and internal medicine residents and faculty. The authors identified 4 diagnostic tests with a range of known positive and negative LRs (high [32.0, 0.03] , moderate [8.0, 0.08], and low [2.6, 0.27 and 1.3, 0.23]) from the emergency medicine literature and developed clinical vignettes suggestive of conditions evaluated by these tests. After each history-and physical examination-based vignette, participants indicated their pretest estimate of disease and level of confidence to make a treatment decision at that time. A nomogram incorporated this pretest estimate with positive and negative LRs of the appropriate test to yield individualized posttest estimates of disease; participants were asked to again indicate their confidence level and state whether this process helped them. The main outcome measures were the change in clinicians' confidence levels according to the magnitude and direction of a given LR and the perception of clinical usefulness across this LR range.

Results: Positive LRs significantly increased mean confidence scores about treatment decisions in every case because 80% to 95% of participants changed their scores in the more certain direction. On a 10-point Likert scale with 10 as most confident, mean differences after a positive LR ranged from +1.5 (95% confidence interval [CI] 1.1 to 1.9) to +4.9 (95% CI 4.4 to 5.4). Positive LRs at or above 8 were perceived as clinically useful. Negative LRs led to variable confidence levels: 1 vignette elicited significantly more certainty (mean difference +1.3; 95% CI 0.4 to 2.2), 1 less certainty (-1.1; 95% CI -0.4 to -1.8), and 2 no true changes. Across all scenarios, 16% to 53% of respondents became more confident after negative LRs, whereas 30% to 64% became less certain.

Conclusion: Clinicians appear comfortable applying the concept of positive LRs to their medical decisionmaking and perceive a positive LR of 8 as the minimum threshold for clinical utility. Regardless of magnitude, negative LRs are applied more inconsistently, indicating a potentially incomplete comprehension of the topic and highlighting further educational opportunities.

Phelps MA, Levitt MA/Alameda County Medical Center-Highland Campus, Oakland, CA Study objectives: Bayes' theorem is an evidence-based medicine model that incorporates clinicians' pretest estimates of disease into pretest odds, which along with the likelihood ratio of a particular diagnostic test ultimately yields individualized posttest probabilities of disease for each patient. To date, there is minimal scientific validation for the clinical application of this approach. This study is designed to evaluate variability in the initial step of this process, clinicians' pretest probability Conclusion: Recidivism of EDOU patients does not differ according to age, sex, or length of stay but does differ according to observation category, with painful conditions showing the highest recidivism rates.

Veser FH, Agrawal V, Bohnstadt J, McMullan JT/Medical University of South Carolina, Charleston, SC Study objectives: Viewpoints of emergency physicians and nurses toward family presence during emergency department resuscitation efforts (FP) were evaluated.

Methods: An anonymous survey addressing FP was distributed to 25 emergency physicians and 58 emergency nurses at a national-level conference. Questions addressed several issues surrounding FP, including family location during resuscitation, emotional support for the family, perceived effects on the grieving process, and opinion on death notification responsibility.

Results: The majority of respondents (63%) disagree with FP, but half would allow FP if asked. Only 29% of respondents would personally want to be present if a family member were being resuscitated. Seventy-six percent believe the family should be placed out of the way, whereas only 8% would tolerate bedside positioning. Most (62%) respondents believe that FP would interfere with efforts to treat the patient. Only 20% of respondents believe that the grief process would be shortened, and 59% feel that available emotional support is lacking. Fear of traumatizing the witnessing family is a primary concern (75%); fears of the family becoming ill (33%) or violent (26%), potential litigation (25%), and decreased staffing (16%) were also expressed. Most respondents believe that physicians (75%), chaplains (58%), or nurses (50%) should provide death notification; only 14% of respondents believe social workers should give the information. One third (34%) of respondents reported no formal training in death and dying issues. There were no significant differences between physician and nurse responses.

Conclusion: Although the literature shows that most family members desire FP, the medical community remains hesitant. Additional training and resources addressing the dying process appear indicated.

Bailey JR, Bagwell SH, Warren KL, Robinson DJ/University of Texas Medical School, Houston, TX Study objectives: The nation's emergency departments (EDs) have developed the stigma of being plagued by crowding. National government data demonstrated a 14% increase in ED visits between 1997 and 2000, coinciding with a 33% increase in the waiting time for nonurgent ED visits. Literature suggests that a major contributing factor to ED crowding is an excessive number of patients presenting with minor problems, and estimates of total ED visits falling into this category vary from 11% to 82%. The purpose of this study was an attempt to pinpoint where in this large range of percentages the true proportion of inappropriate ED patients lies and examine the factors that lead patients with nonurgent complaints to seek care in the ED setting.

Methods: Before receiving medical care, 2 groups of patients matched by low severity of chief complaint were administered a 30-question survey with standardized questions relating to demographics and access to primary care. The first group of patients was surveyed in a large urban academic ED, and group 2 presented as walk-ins at the same institution's family medicine outpatient (FM) clinic.

Results: Twenty-three patients were surveyed in the ED and 38 in the FM clinic. Of those patients presenting to the ED, 57% had what they considered to be a primary care physician (PCP), and of these, 54% attempted to contact their PCP before presentation: 57% were told by their PCP to go to the ED, 43% were unable to promptly get a PCP visit, and 57% were unable to contact them. Fifty-four percent of ED patients believed they would receive better care in the ED, 31% thought that the care would be comparable to the PCP, 15% reported their PCP would give better care, and 54% of ED patients said going to their PCP was a more convenient source of care. These findings differed from those of the FM clinic patients, of whom only 5.3% did not have a PCP, and 87% perceived the care to be better and more convenient at the FM clinic compared with the ED. Of the 44% of ED patients without a PCP, 70% had some financial medical coverage, whereas 30.0% had no coverage. Of the FM clinic patients, 94% had financial medical coverage, whereas 5% had none. Sixty-two percent believed the ED to involve more out-of-pocket expenses, 27% thought the PCP to be just as expensive as the ED, and 11% believed the PCP to be

ment, and discharge for acute asthma. Medical record reviews of these same facilities indicated that survey responses do not correlate with actual practice, as documented in the patient chart. Our objective was to measure these differences and determine where the largest discrepancies were found, comparing survey responses with record review of acute asthma care in the emergency department (ED). Methods: Surveys were distributed to regional emergency physicians. Survey questions were used to develop a data collection tool to perform record reviews of patients treated in the ED for acute asthma. One to 5 months after survey distribution, the record reviews were performed at these same facilities, exploring assessment, treatment, and discharge practice for patients treated in the ED for acute asthma.

Results: Five facilities participated in the survey and corresponding record reviews. A total of 40 emergency physician surveys were completed, and 238 records were reviewed. The largest differences between survey responses and record reviews were noted in the following areas about care of a patient with acute asthma. "How frequently do you use peak flow as an assessment modality?" Ninety-two percent reported always to often, whereas record review indicated 29% of the time. "How frequently do you use intravenous or oral corticosteroids as treatment for acute exacerbation of asthma?" Intravenous (98%) and oral (100%) were reported always to often: record review showed 30% to 42%, respectively. "How often do you typically prescribe oral steroids when discharging a patient with asthma?" Always to often was reported by 100%, but record review showed 54%. "How often do you typically prescribe inhaled corticosteroids at discharge, if the patient is not currently taking?" Always to often was reported by 82%, but record review showed 9%. Last, "How often do you use written discharge instructions?" Always to often was reported by 100%, but record review indicated 52%.

Conclusion: Our survey shows that most emergency physicians believe that they are practicing in accordance with published guidelines; however, our record review indicates that less than half of these emergency physicians are actually doing so. The reasons for this discrepancy are likely numerous and may include time restrictions, failure of documentation, and poor communication skills. Future explorations will be aimed at more clearly delineating the most important barriers to appropriate asthma care in the ED.

Ross MA, Schwab K, Compton S, Kaminskas E/William Beaumont Hospital, Royal Oak, MI Study objectives: We describe recidivism in an adult emergency department observation unit (EDOU) population and determine whether rates differ according to patient demographics or observation category.

Methods: This is a prospective observational cohort study. Its setting is the EDOU of a high-volume suburban teaching hospital. The study population included all discharged EDOU patients, including those who returned within 14 days of discharge for an unscheduled emergency department (ED) visit or direct inpatient admission to the study hospital or a second proximate hospital, from January 1 to June 30, 2002. EDOU patients were treated by emergency physicians using EDOU protocols. Measurements included age, sex, initial EDOU diagnosis, times, and return visit disposition determined from EDOU and hospital databases and confirmed by medical record review. RVs were classified as possibly related (RRV) or unrelated to the index visit by record review.

Results: During 6 months, there were 55,727 ED visits, 4,348 EDOU patients, of whom 80.7% (3,509) were discharged, with 379 (10.8%) return visits and 278 (7.9%) RRV. Of return visit patients, 86.3% had 1 return visit, 11.6% had 2, and 2.1% had more than 3 return visits. Of return visits, 4.2% occurred at the second hospital. Average time to first return visit was 5.2±4.1 days for all return visit patients and 4.5±3.9 days for RRV patients. On return visit, 37.4% patients were treated and discharged from the ED, 37.1% patients were treated in the ED and admitted, 14.4% patients were treated in the ED then the EDOU and discharged home, 7.9% patients were directly admitted to the hospital, and 3.2% patients were treated in the ED and then the EDOU and admitted. There was no significant difference (P>.05) between return visit and non-return visit patients in terms of age (58.2±21.1 years versus 56.8±19.3 years), percentage of male patients (39.2% versus 40.2%), and EDOU length of stay (15.3±5.8 hours versus 15.0±6.0 hours). Among common conditions, RRV rates were highest for headache (16.1%), back pain (13.8%), and abdominal pain (12.7%) and lowest for chest pain (3.3%). As a group, therapeutic protocols had significantly higher RRV rates than diagnostic protocols (10.3% versus 5.1%; P<.0001).

whereas patients with chest tubes were more likely to have definitive treatment subsequently (pleurodesis, thorocotomy). Half (50%) of aspiration patients required a second procedure. Mean duration of hospitalization was significantly longer for chest-tube insertion (7.40 days) compared with that of other groups (needle aspiration 4.04 days). There was no significant difference in complication rates between treatment groups at 1 year.

Conclusion: Treatment with chest-tube insertion is still the most popular method at the ED. Needle aspiration does result in a high proportion requiring a second procedure; however, complication rates do not seem to differ from chest-tube insertion.

Mansbach JM, Emond JA, Camargo CA/Children's Hospital Boston, Boston, MA; Massachusetts General Hospital, Boston, MA Study objectives: We describe the epidemiology of US emergency department (ED) visits for bronchiolitis, including predictors for admission and variation in ED care.

Methods: Data were obtained from the 1992 to 2000 National Hospital Ambulatory Medical Care Survey. Cases had International Classification of Diseases, Ninth Revision code 466 and were younger than 2 years. National estimates were obtained by using assigned patient visit weights. Ninety-five percent confidence intervals (CIs) were calculated by using the relative standard error of the estimate; analysis used χ 2 and logistic regression.

Results: From 1992 to 2000, bronchiolitis accounted for approximately 1,868,000 ED visits for children younger than 2 years. Among this same age group, the overall rate per 1,000 US population was 26 (95% CI 22 to 31) and 31 (95% CI 26 to 36) per 1,000 ED visits. These rates were stable throughout the 9-year period. Comparing children with bronchiolitis with those presenting with other problems, children with bronchiolitis were more likely male patients (61% versus 53%; P=.01) and Hispanic (27% versus 20%; P=.008). They also were more likely to have Medicaid (54% versus 46%; P=.03) and to present in the southern United States (45% versus 39%; P=.01) during January to March (49% versus 28%; P<.001); the time of presentation was similar (P=.18). Therapeutic interventions varied, with 53% of patients receiving short-acting β-agonists, 13% receiving systemic corticosteroids, and 37% receiving antibiotics. Multivariate predictors for receiving systemic steroids were urgent or emergency status at triage (odds ratio [OR] 4.0; 95% CI 1.9 to 8.4) and possibly male sex (OR 2.0; 95% CI 0.9 to 4.1). Approximately 46% of patients underwent chest radiograph, and 19% were admitted to the hospital. On multivariate testing, the only predictors of admission were urgent or emergency status at triage (OR 3.6; 95% CI 2.0 to 6.7) and possibly urban setting (OR 6.5; 95% CI 1.5 to 27.1).

Conclusion: Bronchiolitis is a common reason for ED visits among children younger than 2 years. ED visit rates for bronchiolitis were stable between 1992 and 2000. Despite the ongoing magnitude of the problem, children receive highly variable treatment, especially with respect to administration of systemic corticosteroids. Boys may be more likely to receive corticosteroids than girls. The observed practice variation may reflect current scientific uncertainty about the optimal treatment of this common condition.

Kelly JJ, Sorondo B/Albert Einstein Medical Center, Philadelphia, PA Study objectives: Community acquired pneumonia (CAP) is a frequent cause of hospitalization and mortality. Administering the appropriate antibiotic in the emergency department (ED) is associated with improved survival and shortened length of stay. An Institutional Practice Guideline (IPG) for CAP was introduced in 2002 and was based on Joint Commission Core Measures/Centers for Disease Control and Prevention guidelines. The objective was to investigate the continued compliance with the IPG in CAP patients admitted through the ED.

Methods: This was an observational, retrospective, consecutive medical records review in an academic urban ED of adult patients with ED admitting diagnosis of CAP, comparing first months of IPG implementation on January 1, 2002, to January 3, 2002, and almost 1 year later (October 2002 to December 2002 .

Results: Results are shown in the Table. Conclusion: Compliance with most guidelines tends to decrease over time. Compliance with IPG in CAP patients was consistently maintained throughout a 1-more costly. Sixty percent of patients in each group did not have a means of contacting a physician for medical advice 24 hours a day. Ninety-two percent of ED patients would have attempted to first contact a physician by telephone had a 24hour service been available to them.

Conclusion: In this survey, patients would more often prefer to go to their established PCP for care of their nonurgent ailments than the ED, provided that prompt service were available. Patients are more likely to go to the ED when they are unable to immediately contact their PCP or if instructed by their PCP to go to the ED, regardless of their insurance coverage or their perception that the ED is more expensive. The survey suggests that facilitated communications between the patient and their PCP may help divert patients from the ED.

Constantinos K, Chathampally Y, Kohilas C/Metropolitan Hospital Center, New York Medical College, New York, NY Background: Previous studies have suggested that non-English-speaking patients are at greater risk for longer emergency department (ED) length of stay and increased use of diagnostic tests. Without medical translators, language barriers may affect many aspects of ED care, including knowledge of discharge diagnosis and instructions.

Study objective: We compare English-speaking and non-English-speaking patients with respect to their knowledge of their ED discharge diagnosis and instructions.

Methods: All discharged adult ED patients were instructed to visit the clerical station to receive their ED visit receipt and clinic follow-up instructions. Bilingual clerical staff were trained to administer a short 4-question discharge interview to ascertain knowledge of the following: (1) English speaking versus non-English speaking; (2) discharge diagnosis; (3) discharge instructions; and (4) use of translator. Responses were compared with ED record discharge section, including medication use and reasons to return to the ED.

Results: A total of 214 patients discharged from the ED were interviewed of a possible 237 consecutively registered nonadmitted patients. Twenty-three patients walked out before evaluation or did not visit the clerical staff before leaving the ED. One hundred forty-two (66%) patients spoke English. Of these, 6 (4%) patients were not aware of their discharge diagnosis or their discharge instructions. Seventytwo (33%) patients were non-English speaking. Of these, 22 (31%) patients were not aware of their discharge diagnosis or their discharge instructions. Twenty of these 22 patients spoke Spanish only. Of the 50 patients who were aware of both items, 47 (94%) had either a Spanish-speaking physician or a Spanish interpreter (member of ED staff or patient's family) compared with only 4 (18%) of the 22 who were unaware. The relative risk of being unaware of discharge diagnosis and instructions was 7.3 (95% confidence interval [CI] 3.1 to 17.0) for non-Englishspeaking patients compared with those who could speak English. The relative risk of the outcome increased to 10.9 (95% CI 4.2 to 28.5) if in addition to being non-English speaking there was also no interpreter used.

Conclusion: Non-English-speaking patients are much less likely to be aware of their diagnosis or discharge instructions than English-speaking patients. The use of an interpreter improves the discharge knowledge but is still inadequate for an important proportion of non-English-speaking patients. Further study on the medical knowledge and language skills of the interpreter is required.

Ong MEH, Chan YH/Singapore General Hospital, Singapore Study objectives: We aim to describe the incidence of patients with spontaneous pneumothorax presenting to the emergency department (ED) at the largest tertiary hospital in Singapore, the treatment given, and outcomes.

Methods: We conducted a retrospective review of patients with spontaneous pneumothorax presenting to the ED of the Singapore General Hospital from January 1, 2000, to December 31, 2001.

Results: There were 159 patients, mean age 36.1 years (SD 19 .3), of whom 94.3% were men. Seventy-five (47.2%) patients were treated at the ED with chest-tube insertion, 28 (17.6%) patients had needle aspiration, and 56 (35.2%) patients were observed. Significantly more patients had complete reexpansion with chest tube (65.3%) compared with needle aspiration (17.9%; P<.001). Patients with needle aspiration or observation were more likely to have a second procedure (P<.001), and number of patients with ultimate ED or hospital diagnosis of PE (by International Classification of Diseases, Ninth Revision code). Analysis of pre-ELISA and post-ELISA findings was performed with odds ratios (ORs) and 95% confidence intervals (CIs).

Results: Compared with the pre-ELISA period, the post-ELISA period was associated with a fivefold increase in number of D-dimer assays ordered (503 versus 100 assays; OR 4.86; 95% CI 3.92 to 6.02), but the number of diagnosed PEs remained the same in pre-ELISA (103) and post-ELISA (101) periods (OR 0.93; 95% CI 0.71 to 1.23). Also similar in the 2 periods were the numbers of advanced radiologic studies ordered (899 in pre-ELISA and 969 in post-ELISA period; OR 1.02; 95% CI 0.93 to 1.12).

Conclusion: Institution of the ELISA D-dimer assay was associated with markedly increased frequency of emergency physician ordering of D-dimer levels, but this increase was not associated with an increase in either PE diagnosis or use of advanced radiologic imaging to evaluate for PE. Thus, it appears that the increased D-dimer use occurred primarily in patients with low disease probability who otherwise may not have warranted any PE evaluation. According to our results, doubt remains as to the cost and benefit of using the rapid ELISA D-dimer assay.

Chumsri S, Kelly JJ, Sorondo B, Sivak S/Albert Einstein Medical Center, Philadelphia, PA Background: Pneumonia was the seventh leading cause of death in the United States in 1999, resulting in approximately 500,000 hospitalizations and 63,730 deaths (2.7% of total deaths), at an average annual cost for treatment of $8.4 billion, $7.5 billion of which was for inpatient care. In the summer of 2002, Jefferson Health System adopted guidelines from the Joint Commission on Accreditation of Healthcare Organization (JCAHO) for the treatment of community acquired pneumonia (CAP). In JCAHO ORYX Core Performance Measures of Community Acquired Pneumonia, blood cultures are routinely required before the first antibiotic administration.

Study objectives: We determine the incidence of positive blood cultures among patients without defined risk factors admitted to Albert Einstein Medical Center with CAP and treated according to the Jefferson Health System and JCAHO guidelines for CAP, we determine cost-effectiveness of routine blood cultures obtained according to the guidelines, we determine the impact of positive blood cultures in the management of CAP, and we identify groups at risk for positive blood culture results.

Methods: This was a retrospective medical record review of patients treated at Albert Einstein Medical Center after implementation of a pathway for CAP between January and December 2002. All patients older than 18 years, presenting with respiratory symptoms and a lobar infiltrate on chest radiograph, and having 2 or more sets of blood cultures obtained within 48 hours of hospital admission were included. Patients with conditions such as HIV-related illness, malignancy, recent chemotherapy, steroid therapy, sickle cell disease, nursing home residency, and hospitalization within the past 14 days were excluded.

Results: One hundred thirty-eight patients were admitted and treated according to the CAP pathway. Of 138 patients, 93 (67.39%) patients met inclusion criteria and 15.05% had a positive blood culture. In 2 patients, antibiotics were changed according to blood culture results and included cefuroxime to penicillin V (Streptococcus pneumoniae) and azithromycin to amoxicillin (S pneumoniae). Total charges for blood cultures were $12,276, assuming only 2 sets of cultures per patient. Savings of $309 were calculated from changing antibiotics.

Conclusion: In this study, only 4% of patients admitted with CAP without defined risk factors had positive blood culture results that were not contaminants. Routine blood culture tests may not be cost-effective despite CAP guideline recommendations in patients without defined risk factors. The finding of multilobar infiltration increases the possibility of positive blood culture results. Limitations of this study include the small number of eligible cases the retrospective analysis of the methodology. Future analysis will include data from other hospitals within the Jefferson Healthcare System.

Richman PB, Wood J, Matthews J, Friese J, Petri RW, Hentz J, Hernandez JL, Field AG/Mayo Clinic Hospital, Scottsdale, AZ Study objectives: We recently described the use of a computed tomography (CT) protocol that uses indirect lower extremity venography (CTV) in addition to chest angiography (CTA) to identify patients with thromboembolic disease. As a prelimi-

year period. Monthly reminders at resident morbidity and mortality conference and attending faculty meetings reinforced the IPG details. This monthly education may serve to make IPG part of the "ingrained culture" for CAP treatment compliance in the ED. 

Cutro RG, Seaberg DC, Stevens G/University of Florida, Gainesville, FL Study objectives: Currently, only 20% to 25% of eligible patients presenting to the emergency department (ED) have received the pneumococcal vaccine. This study examined the cost-effectiveness of an ED-based vaccination strategy aimed at reducing the incidence of Streptococcus pneumoniae bacteremia and pneumonia in the population older than 65 years.

Methods: This was a retrospective cost-effective analysis. The number of patients aged 65 years or older who presented to the ED in 2000 was obtained from the Centers for Disease Control and Prevention and US Census Bureau data. By using the incidence of pneumococcal pneumonia and bacteremia in this age group, potential cost savings were calculated, assuming an effectiveness of 66% for the pneumococcal vaccine. The literature was searched, and current estimates of vaccination and hospitalization costs were obtained. Potential cost savings were then calculated by taking the difference of costs of vaccination of all eligible patients and the costs of treatment of pneumococcal pneumonia and bacteremia.

Results: In 2000, 13,472,792 patients aged 65 and older visited the ED. The incidence of patients hospitalized with pneumococcal pneumonia was 230 per 100,000, the incidence of those treated as outpatients for pneumococcal pneumonia was 536 per 100,000, and the incidence of pneumococcal bacteremia was 72 per 100,000. The cost of the vaccination was estimated at $15.92 per vaccination, which included the adverse event rate of 1 per 100,000 people. A net cost savings of $121,860,291 was found by using an ED-based vaccination strategy .

Conclusion: Immunizing all eligible patients aged 65 years or older in the ED with the pneumococcal vaccine is cost-effective. The cost in terms of quality-adjusted life years needs to be determined.

Kabrhel C, Thomas SH/Massachusetts General Hospital, Harvard Medical School, Boston, MA Study objectives: D-Dimer can be a useful test to exclude pulmonary embolism (PE), depending on the sensitivity of the particular test used. The most sensitive test available to emergency physicians is the rapid enzyme-linked immunosorbent assay (ELISA; VIDAS, BioMerieux), which is 96.4% to 100% sensitive. However, the ELISA method is characterized by unresolved cost and benefit questions. We sought to determine how the introduction of this test would affect the diagnosis of PE and the ordering of related diagnostic studies.

Methods: This study was performed in an urban teaching hospital (annual emergency department [ED] volume 77,000) that converted from using a latex agglutination D-dimer assay to the ELISA method. Study patients were divided into 2 groups: those presenting to the ED in the 120 days before institution of the ELISA test (pre-ELISA) and those presenting in the first 120 days of the ELISA test availability (post-ELISA). Study endpoints included number of D-dimers performed in the ED, number of ED-ordered advanced radiologic studies (defined as combined number of ventilation or perfusion and chest computed tomography scans) ordered for PE diagnosis,

fairly specific marker of myocardial ischemia useful in risk stratification of chest pain patients in the emergency department (ED). By using ED portions of abstracted records of patients entered into a prospective study for chest pain syndrome suggestive of non-ST-elevation acute coronary syndrome and evaluated with and without IMA assay, the clinical utility and economic impact of IMA in making risk stratification decisions at ED presentation were projected. ED cardiac risk assessment (CRA) was subsequently compared with hospital course, including radionuclide myocardial perfusion imaging, which was by protocol performed on every patient. Methods: After receiving limited but standardized training on IMA test performance, 6 academic emergency physicians from varied geographic regions and ED settings in North America reviewed 251 records that included demographics, history and risk profile, 12-lead ECG results, and presentation necrosis markers. A risk level was assigned to each patient, and using published average US cost and reimbursement data, the economic impact of each CRA was calculated. After 2 weeks, without access to previous results, the patients were reevaluated according to the same data plus presentation IMA. Cost, charge, and projected reimbursement data were used to calculate the economic impact of the second CRA.

Results: The study population was at aggregate low risk, with a 10% frequency of ACS. Intrarater reliability of CRAs was very high (κ 0.8 to 0.98), interrater reliability was good (R i >0.67), and paired t test analysis indicated that all 6 evaluators adjusted CRAs concordantly with IMA results. No patients with negative IMA were found to have ACS; negative predictive value was 100%. Initial CRAs assigned by the group accounted for aggregate evaluation costs of US $549,453 and an expected reimbursement of US $571,887 (net positive US $22,434). In patients with negative IMA, an average of 55% were downgraded in the second CRA; 39% remained the same, and 6% were upgraded. Corresponding numbers for positive IMA were 5%, 35%, and 60%. This response indicates a significantly positive impact on resource use and ED throughput time. In addition, economic analysis of the second set of CRAs indicates a net (reimbursement less cost) gain for hospitals of positive US $27,919, an increase of 23% over net without IMA use.

Conclusion: IMA is useful in the presentation risk stratification of chest pain in the ED, is associated with a decreased likelihood of discharging patients with incipient ACS, and results in improved resource use and net financial performance. Study objectives: We determine awareness of risk factors for coronary artery disease (CAD) among patients with or at risk of disease and the frequency with which patients are being tested or treated for homocystinemia compared with hypercholesterolemia.

Methods: This was a prospective survey approved by the institutional review board and conducted in an urban emergency department. Participants were patients who gave verbal consent and were older than 18 years. Patients were asked standardized demographic questions, medical history, awareness of CAD risk factors, and whether they had undergone testing, treatment, or advice for treatment for elevated cholesterol or homocysteine levels. Main outcome measures were factors increasing ability to identify risk and whether patients at risk (more than 2 risk factors or family history) or with a history of CAD had ever been tested for cholesterol or homocysteine levels or advised to be treated to lower cholesterol or homocysteine levels. Appropriate parametric and nonparametric tests were performed.

Results: Three hundred three patients (192 women) with a mean age of 47 years were surveyed. Two hundred four of 303 patients had cholesterol checked, and 34 were undergoing treatment for elevated levels. Two patients had homocysteine checked, whereas 4 were being treated to decrease levels. Patients with increasing numbers of risk factors or a history of CAD were more likely to have had cholesterol checked (P<.001). Of 237 patients with more than 2 risk factors and a history of or family history of CAD, the only predictors of accurately reporting more risks were having a provider who had discussed risk factors and increasing education level (P<.001 and P<.005). Ability to report risks was unaffected by family history of CAD, cardiac history, or number of risk factors.

Conclusion: Most patients, including those at high risk, are undergoing cholesterol but not homocysteine checks and do understand homocysteine level as a CAD risk factor. Primary providers affect patients' awareness of CAD risks. Emergency nary study for the eventual development of a clinical decision rule involving CTA/ CTV, we conducted the current investigation to define clinical characteristics of emergency department (ED) patients who rule in versus rule out for thromboembolic disease by our CT protocol.

Methods: This was a retrospective cohort study. The study was conducted in a suburban ED. ED patients with symptoms suspicious for pulmonary embolism (PE) during the 1-year study period were identified by an electronic database. Physician data abstractors conducted structured reviews of the ED medical record, CT interpretation, and hospital course.

Categorical variables are presented as frequency of occurrence and were analyzed by χ 2 tests. Continuous variables are presented as means and were analyzed by t tests. α Was set at .05. Logistic modeling was used to analyze the overall effects of the variables (risk factors) that were significant in a univariate setting to the final diagnosis of PE. κ Values were calculated to determine interobserver agreement for data collection method.

Results: Three hundred forty patients composed the study group (mean age 61±18 years; 58% female patients). Overall interobserver agreement with data collection for the sample of concomitantly reviewed charts was almost perfect (κ=0.89). A total of 51 (15%) patients were diagnosed with PE. Several variables and risk factors in the logistic regression model were found to have a positive association with the diagnosis of PE, including recent surgery (odds ratio [OR] 3.91; 95% confidence interval [CI] 1.77 to 8.62), cancer (OR 2.38; 95% CI 1.08 to 5.25), and history of deep venous thrombosis or PE (OR 3.38; 95% CI 1.40 to 8.21). Patients with and without PE had a similar prevalence for the complaint of chest pain (63% versus 68%; P=.51), as well as shortness of breath (83% versus 76%; P=.37). Four percent of patients with positive CT studies in our study group had none of the clinical characteristics that were significantly associated with PE.

Conclusion: In our ED cohort of patients with suspected PE who were evaluated by a protocol including CTA and CTV, we found several clinical characteristics that were significantly associated with a diagnosis of thromboembolic disease.

Cameron PA, Rainer TH, Smit P deV/Chinese University of Hong Kong, Shatin, Hong Kong Study objectives: In March 2003, an outbreak of severe acute respiratory syndrome (SARS) occurred at the Prince of Wales Hospital in Hong Kong, which resulted in the first major outbreak outside mainland China. This study aims to describe the epidemiology, clinical aspects, and consequences of this outbreak.

Methods: Prince of Wales Hospital is a 1,400-bed tertiary teaching hospital in Hong Kong, a Special Administrative Region of China, with a population of 7 million. An outbreak resulted from 1 unrecognized index case on a medical ward admitted March 4, 2003. An epidemiologic study of the contact cases (staff, patients, and visitors), including demographics, contact history, clinical course, and followup, was undertaken. The hospital and political response was also studied.

Results: Within 2 weeks of the first staff becoming affected, 138 hospital cases were reported, the hospital was closed, and spread to the community had become apparent. Clinical features of SARS (especially respiratory symptoms) required by the World Health Organization definition were not prominent, with fever, chills, myalgia, abdominal pain, and malaise being the only defining features. Many patients had radiologic pneumonia without symptoms. Blood test results typically showed leukopenia and lymphopenia. The typical course of the illness was throughout 2 weeks, with 20% of patients requiring ICU and 5% of patients dying. The economic and political impact of this disease was immediately apparent.

Conclusion: This is a rapidly evolving epidemic that may have a profound impact on the world beyond the morbidity and mortality associated with the disease itself.

Making the diagnosis is difficult, and at this stage, there is no test that will define patients with SARS. Study objective: We evaluate time saved by using standing orders for chest pain in the emergency department (ED). We hypothesized that implementation of standing orders would improve time to diagnosis, therapeutics, and disposition of chest pain patients in the ED.

Methods: A retrospective medical record review was conducted to identify and compare 2 cohorts of patients with chest pain presenting to the ED: those presenting while standing orders were in effect versus those presenting before standing orders were in effect. Eligible patients were men aged 50 years and older, presenting to the ED, triaged as emergency, and diagnosed with chest pain (International Classification of Diseases, Ninth Revision: 786.50 chest pain NOS; 786.59 chest pain NEC; and 786.52 painful respiration). Data were collected for two 2-month periods: June to July 2002 (standing orders) and June to July 2001 (no standing orders). Outcome measures were time from triage to (1) diagnostic ECG, (2) therapeutic aspirin (ASA) or nitroglycerin (NTG), and (3) disposition.

Results: Ninety-nine eligible patients presented to the ED during the standing orders period; 81 patients presented during the no standing orders period. Time to ECG was 3 minutes faster in the standing orders group (14±13 minutes, 95% confidence interval [CI] 10.9 to 16.1 versus 17±15 minutes, 95% CI 14.0 to 20.4). Time to ASA/NTG was 5 minutes faster in the standing orders group (34±17 minutes, 95% CI 27.3 to 40.7 versus 39±13 minutes, 95% CI 34.4 to 44.6). Time to disposition, however, was 28 minutes longer in the standing orders group (273±175 minutes, 95% CI 237.9 to 308.4 versus 245±188 minutes, 95% CI 203.0 to 286.6). When only those patients who were discharged home were compared, time to disposition remained longer in the standing orders group (262±174 minutes, 95% CI 211.6 to 311.8 versus 254±122 minutes, 95% CI 205.2 to 302.4).

Conclusion: Standing orders reduced time to ECG and time to ASA/NTG by 3 and 5 minutes, respectively. These savings, although potentially clinically important, are not statistically significant. Further, time in the ED increased after implementation of standing orders, albeit not significantly. In the standing orders period, the odds of having a cardiac injury panel drawn is 3.46 times greater (α=.02). Awaiting the results of this panel increases time spent in the ED and, therefore, may explain the nonstatistical increase in length of stay in the standing orders period. We conclude that implementation of standing orders in this ED does not improve time to diagnosis, therapeutics, or disposition of chest pain patients in the ED. Future studies should focus on a cost-effective analysis of standing orders.

Lee CC, Park BK, Crupi RS/Flushing Hospital Medical Center, Flushing, NY Study objectives: Gastrointestinal (GI) bleeding is a common problem seen in the emergency department (ED). Acute upper GI bleeding in adults is more common than lower GI bleeding, with an overall annual incidence of 100 per 100,000 for upper GI bleeding and 20 per 100,000 for lower GI bleeding. In the setting of acute GI bleed, we hypothesized that patients are at increased risk for silent myocardial infarction. We performed this study to investigate the incidence of non-ST elevation myocardial infarction (NSTE-MI) during acute GI bleeding in ED.

Methods: The retrospective medical record review was done to determine whether patients with acute GI bleed and a hemoglobin level of 8 g/dL or less have cardiac enzyme (creatine phosphokinase [CPK] or troponin I) elevation. Patients receiving vasopressor therapy in the ED to maintain hemodynamic stability were excluded from the study. A total of 50 patients were reviewed. Thirty (62%) of 50 patients had preexisting coronary artery disease. The average age of patients was 70.5 years. Male to female patient ratio was 4.5:1.

Results: Six patients (12%) of 50 had an increase of cardiac marker for MI while being treated in the ED. The average hemoglobin level studied was 7.2 g/dL. All 6 patients had history of coronary artery disease and were admitted with a diagnosis of an upper GI bleed. None of our patients had an acute ST elevation in their ECG. Only 1 patient (17%) of 6 had a complaint of chest pain or equivalent.

Conclusion: A significant number of patients had a silent NSTE-MI associated with acute GI bleed. According to our preliminary study, we recommend that CPK

A Study objectives: We determine whether patients with complaints of potential cardiac origin provide accurate histories and whether accuracy is predicated on question type (typical versus simple or descriptive).

Methods: This was a prospective clinical trial approved by the institutional review board and conducted in an urban emergency department. Participants were patients who gave verbal consent, were older than 40 years, and had chest pain, syncope, weakness, shortness of breath, or palpitations during June to November 2002.

Patients were asked open-ended questions about their history, followed by a structured closed-ended format terminating in a simple descriptive question. Responses, when given (open versus closed-ended phase), and change in response upon question simplification were recorded. Main outcome measures were (1) frequency with which patients with atypical chest pain added or altered history to suggest pain of possible cardiac origin; and (2) switched from no to yes on descriptive questioning about chest pain, history of cardiac disease, or previous cardiac evaluation.

Results: Seventy-nine patients underwent detailed structured interviews. Of 23 patients denying chest pain, 7 (30%) stated they had chest pain with further questioning. Thirteen of 78 (17%) patients modified initial history to include descriptive chest pain of possible cardiac origin with further questioning. Of 49 patients stating they had no chest pain or were unsure, 23 (47%) stated they had exertional chest pain with descriptive questioning. Ten of 42 (24%) patients initially denying significant cardiac history stated they had a cardiac history, whereas 6 patients who denied any cardiac history revealed an abnormal evaluation on further questioning. There was no significant difference in age, sex, insurance status, or provider type as to whether patients changed histories with more descriptive questioning.

Conclusion: A significant number of patients modify histories when descriptive simple questioning is used. Patients who answer no to a question may not understand or identify that they do not understand it. A simple descriptive questioning format should be included when patient history accuracy is paramount.

Tijunelis MA, Herbert, ME/Keck School of Medicine, University of Southern California, Los Angeles, CA Background: Patients with atrial fibrillation with a rapid ventricular response are common in the emergency department. Approximately 0.1% to 0.5% of these patients have an underlying diagnosis of Wolff-Parkinson-White syndrome (WPW-AF). The most feared complication of atrial fibrillation in WPW-AF is deterioration to ventricular fibrillation and death. The American Heart Association has provided guidelines for the treatment of a stable patient with WPW-AF. The current recommendations suggest first-line therapy as 1 of the following antiarrhythmics: amiodarone, procainamide, sotalol, propafenone, or flecainide.

Study objectives: The purpose of this study is to review and analyze the evidence supporting the American Heart Association recommendations for the use of intravenous amiodarone in WPW-AF.

Methods: The OVID database was used to search the English literature during 1969 to 2003. The search terms were "Wolff-Parkinson-White," "WPW," "amiodarone," and "ventricular fibrillation." The search included randomized trials, case series, and case reports. Studies of oral amiodarone were excluded because of a different mechanism of action.

Results: We identified a total of 15 documented cases of WPW-AF in which patients received amiodarone intravenously. Of these cases, there were 8 case reports of complications. All 8 patients initially were in atrial fibrillation that degenerated into ventricular tachyarrhythmias.

Conclusion: Contrary to advanced cardiac life support (ACLS) guidelines, a stable patient with WPW-AF should not receive intravenous amiodarone as first-line treatment. There is a paucity of data validating the safety of this approach. Further research should focus on the relative safety and efficacy of the recommended ACLS drugs versus early cardioversion.

S 3 9 tion). Only non-ACS patients aged 18 to 65 years were finally included. Patients were asked whether they had a known family history of any cardiovascular disease, personal history of any cardiovascular disease, hypertension (HTN), diabetes (DM), hypercholesterolemia (C), or hypertriglyceridemia (TG). They were also asked whether they had any CVRF, whether or not it was treated (Tx), and if treated, whether or not CVRF was under control (Ct). Age, sex, smoking habit, and body mass index (<25 normal; 25 to 30 overweight; >30 obesity) were also recorded. Results between sex were compared by using Student's t and χ 2 tests. Conclusion: Even though men have a significantly higher prevalence of previously known cardiovascular disease, smoking habit, hypercholesterolemia, and being overweight compared with women, all patients discharged from the CPU have a high prevalence of cardiovascular risk factors. The percentage of patients treated and treated and controlled is surprisingly low. The question of whether the CPU should also play an educational and preventive role is raised because otherwise young and active people make up this group. Further studies are needed to determine, in a cost-effective fashion, whether an intervention could prevent these patients from having any further cardiovascular events. Background: Patient likelihood of having an acute coronary syndrome (ACS) according to complete evaluation of epidemiologic characteristics, clinical and ECG findings, and laboratory testing has been published elsewhere. However, these rules cannot be applied to identify chest pain patients at risk of having an ACS and waiting to be treated in a busy emergency department or chest pain unit. The aim was to develop 2 predictive models to be useful, within the triage process, for quickly and accurately identifying ACS patients and helping to make the decision of whom to admit.

Methods: This was a prospective observational study. The first 900 chest pain patients (July to September 2002) were consecutively included. From all patients, we recorded usual triage variables: epidemiologic (ie, age, sex, referred or not by a physician, cardiovascular risk factors, any known vascular disease, current treatment), clinical (ie, pain characteristics, pain within 48 previous hours, ongoing pain), and physical (ie, blood pressure, pulse rate, respiratory rate). Accordingly to our protocol, the attending physician prioritized patients. The waiting time to be treated was recorded. Then patients were treated and, once initial clinical evaluation and first ECG were carried out, categorized into non-ACS or possible or definitive ACS, including ST-elevation myocardial infarction (STEMI), other than STEMI ACS, and those in whom, after continuous ECG monitoring for 8 hours and serial cardiac markers, an ECG exercise test (EET) was performed. When all patients had been evaluated, they were then classified into either discharged (including all patients from group 1 and those with a negative EET result) or admitted (including patients with STEMI, other than STEMI ACS, and with a positive EET result). The relation among variables was assessed by multiple logistic regression analysis. The 2 models (for being possible or definitive ACS and for being admitted) were calculated according to goodness-of-fit and discriminated by receiver operating characteristic (ROC) curve methodology.

Results: We obtained data from 822 (91.3%) patients who were first classified into 380 (46.2%) non-ACS patients and 442 (53.8%) possible or definitive ACS or troponin treatment be ordered for acute GI bleed patients to detect the presence of NSTE-MI in the ED.

Schrock JW, Collins SP, Lindsell CJ, Storrow AB/University of Cincinnati, Cincinnati, OH Study objective: We compare emergency physicians' medical treatment for acute myocardial infarction (AMI) and non-ST elevation myocardial infarction/unstable angina (NSTEMI/UA) with adherence to the American College of Cardiology/ American Heart Association (ACC/AHA) guidelines for these diseases published in 1999 and 2002, respectively.

Methods: We used a cross-sectional design with a convenience sample of emergency medicine residencies representing a broad geographic distribution and all program formats. From October 2002 through January 2003, surveys were provided, unannounced, to emergency department residents and faculty through a contact person in their own program. The surveys asked the participant to provide pharmacologic treatment to a patient with AMI and a second patient with NSTEMI/UA. Respondents were asked to provide all medications they would order for each patient. Treatment adherence for AMI required that a patient receive nitroglycerin, aspirin, β-blocker, heparin, and thrombolytics according to guidelines published in 1999.

Adherence for NSTEMI/UA required that a patient receive aspirin, β-blockers, heparin, clopidogrel, and a glycoprotein IIb/IIIa inhibitor (GP IIb/IIIa) according to guidelines published in 2002. Faculty were used as the reference group, and data were reported as frequencies and percentages. Confidence intervals (CIs) of 95% were computed for proportions (score-method, no continuity correction), and logistic regression was used to determine influence of training year.

Results: A total of 348 respondents returned the survey, 60 faculty and 288 residents, for a resident response rate of 50.5% (288 of 570 possible residents). There were 81 first-year respondents (23.3%), 95 second-year respondents (27.3%), 84 third-year respondents (24.1%), and 28 fourth-year respondents (8%). Overall adherence was 63.8% (95% CI 58.5% to 68.8%) and 4.9% (95% CI 3.0% to 7.9%) for AMI and NSTEMI/UA, respectively. First-year residents were 4 times less likely to be adherent than faculty in the treatment of AMI (odds ratio 0.267; 95% CI 0.128 to 0.554; P<.001). With second-, third-, and fourth-year residents, adherence increased with advancing year of education (60.0%, 73.8%, and 78.6%, respectively), but the adherence did not differ significantly with faculty (75.0%). There was no change in participant adherence with advanced training for treatment of NSTEMI/UA. The most infrequent medications offered to a patient with NSTEMI/UA were clopidogrel (10.9%) and GP IIb/IIIa (36.4% overall). Adherence increased to 81.9% if these 2 medications were removed from the guidelines.

Conclusion: Emergency department resident and faculty adherence with ACC/ AHA recommendations for a simulated patient presenting with AMI approaches that of previous studies but remains suboptimal. There is evidence of increasing adherence with advancing education. Adherence with ACC/AHA recommendations for NSTEMI/UA is poor for all levels of training, including faculty. The lack of adherence is due to underuse of clopidogrel and GP IIb/IIIa. Educational initiatives for residents and faculty will be needed to improve adherence with published guidelines for acute coronary syndrome. Background: Two main aims of chest pain units (CPUs) are accurate evaluation of patients with possible acute coronary syndrome (ACS) and early treatment of patients with acute myocardial infarction. However, less attention is paid to the "other patients," who usually are quickly discharged once ACS is ruled out. The aim of the present study was to know the prevalence of cardiovascular risk factors (CVRFs) among active-aged patients discharged from our newly created CPU, as well as the percentage of such patients treated and treated and controlled.

Methods: This was a prospective observational study. The first 900 chest pain patients (July to September 2002) were consecutively selected. Patients were classified into definitive ACS (patients admitted to cardiology service from the CPU when an ACS was confirmed) and non-ACS (patients discharged when ACS was ruled out either after initial clinical assessment and first ECG or after CPU protocol comple-Conclusion: Although the percentage of patients returning to the CPU because of medical mistakes was relatively low, one third should have been diagnosed at least with intermediate-risk unstable angina according to current guidelines. The epidemiologic and clinical profile of such patients may help physicians recognize and properly admit and treat these patients. Because the population is getting older, a higher prevalence of coronary disease, along with patients' lower capability to complete EET, is expected. Therefore, in the near future, new approaches will have to be established for continued treatment of these patients in a cost-effective fashion.

Trial Hunt GS, Spencer MT/University of Rochester Medical Center, Rochester, NY Study objectives: We evaluated whether there was a difference in the time of sedation and time to patient disposition in patients undergoing procedural sedation with etomidate and midazolam.

Methods: This study is a prospective, randomized, double-blind trial comparing etomidate with midazolam for consecutive emergency department patients older than 10 years and requiring procedural sedation for reduction of joint dislocations or long bone fractures. Patients were randomized to receive either 0.10 mg/kg of etomidate or 0.035 mg/kg of midazolam. The emergency physician was allowed to administer up to 3 doses of the study drug. Data were collected for the primary outcomes of length of sedation (as measured by a modified Aldrete score) and time to disposition (admission or discharge), as well as the secondary outcomes of number of doses required, quality of sedation (scale of 1 to 10), complications, procedure success, and patient recall.

Results: Thirty-nine patients were enrolled (20 etomidate, 19 midazolam). Patients receiving etomidate and midazolam had no statistical difference in mean age (respectively) (42 years, 55 years), percentage of male patients (68%, 83%), or analgesia (fentanyl) dosing (88 µg, 56 µg). Mean length of sedation for etomidate was 12.84 minutes (SD 9.39) and for midazolam was 32.00 minutes (SD 15.63; P<.001). Mean time to disposition for etomidate was 119.53 minutes (SD 71.00) and for midazolam was 122.50 minutes (SD 99.59; P=.920). The mean number of doses for etomidate was 2.00 and for midazolam was 2.26 (P=.130). Mean quality of sedation for etomidate was 7.84 (SD 1.54), whereas that for midazolam was 7.26 (SD 3.23; P=.560). Successful reduction occurred in 36 (92%) of 39 patients. Midazolam was the sedation medication used in each unsuccessful reduction. Complications (eg, myoclonus, desaturation, apnea, hypotension) were seen in 26% of etomidate patients and 21% of midazolam patients (P=.703). Recollection of the reduction procedure was expressed by 21% of etomidate patients and no midazolam patients, P=.045.

Conclusion: Our data indicate that the use of etomidate compared with midazolam for procedural sedation provides a significant reduction in recovery time, without a reduction in time to patient disposition, while providing equal sedation quality without an increase in necessary dosage or adverse effects. Etomidate may also have a lower failure rate and fewer amnestic properties than midazolam, but larger numbers of study patients are necessary to confirm these findings.

Pines JM, Gilbert O, Kim C, Shurtz E, Rothenberger S, Perron AD/University of Virginia Health Sciences Center, Charlottesville, VA Study objectives: The chief complaint of acute extremity injury is often encountered in the emergency department (ED), and radiographs are frequently obtained to determine the diagnosis of fracture versus soft-tissue injury. As a result of these injuries, patients present in variable amounts of subjective pain from minimal to excruciating. Physicians, particularly emergency physicians, have been shown to consistently undertreat pain in the ED. The hypothesis we sought to examine was whether patients with radiographic evidence of fracture would receive more aggressive pain control in the ED and at discharge compared with patients with only softtissue injury, regardless of initial or subsequent pain rating on an analog scale.

Methods: A prospective observational series of patients presenting with extremity injury to a suburban academic ED were surveyed on arrival and discharge. Patients with extremity injuries were asked to rate their pain on a visual analog scale (0 to 10) at ED arrival and on ED discharge. ED patients aged 18 years and older and with a single extremity injury for which a radiograph was ordered were eligible for inclusion in the study. Physicians were unaware of patient participation in this study. Patient demographics, radiograph results, ED admission and discharge pain score, Background: Revisit rate is calculated as the percentage of discharged patients returning to the emergency department within 3 or 7 days. However, for acute coronary syndromes (ACS), any intervention is not usually assessed before 1 month. From this point of view, any previously discharged patient record should be reviewed to rule out a mistake if they return within a month because of symptom recurrence. The aim was to define the clinical and epidemiologic profile of discharged patients returning to a newly created chest pain unit (CPU) within 1 month, as well as the reasons for such a return.

Methods: This was a prospective observational study. Because only discharged patients from the CPU could become a revisit, from the first 900 chest pain patients (July to September 2002), only those finally discharged were included. According to our protocol, patients could be discharged at 2 points: (1) by the emergency department attending physician once initial clinical evaluation and first ECG was carried out (they were rated as a non-ACS); and (2) by the cardiologist after the evaluation (ECG monitoring, cardiac markers, and ECG exercise testing [EET]) was completed (they were rated as a ruled-out ACS). Epidemiologic and clinical data from discharged patients were recorded at the first visit. CPU register system detected all returning patients within 1 month. Their records were analyzed to determine the return causes, which were categorized into attending mistake (definitive ACS patient who was rated as a non-ACS at the first visit), cardiologist mistake (definitive ACS patient who was rated as a ruled-out ACS at the first visit), patient mistake (definitive non-ACS patient who was rated as either non-ACS or ruled-out ACS at the first visit), and no mistake (if rated as a ruled-out ACS at both visits). The results were compared using Student's t and χ 2 tests.

Results: We obtained data from 822 (91.3%) patients, of whom 622 (75.7%) were finally discharged: 380 (61.1%) patients as non-ACS and 242 (38.9%) as ruledout ACS. From these data, we detected 21 (3.4%) revisits. Compared with no-revisit patients, revisiting patients significantly were older than 65 years (76.2% versus 32%; odds ratio [OR] 6.8; 95% confidence interval [CI] 2.5 to 18.9), had hypertension (60% versus 31.4%; OR 3.3; 95% CI 1.3 to 8.2), and known coronary disease (50% versus 20.1%; OR 4; 95% CI 1.6 to 9.7), had had another chest pain episode within the 2 previous days (68.4% versus 38.2%; OR 3.5; 95% CI 1.3 to 9.4), had received out-of-hospital treatment for ACS (55.6% versus 27.5%; OR 3.3; 95% CI 1.3 to 8.5), and had a higher Thrombolysis in Myocardial Infarction (TIMI) risk score (2.5±1.3 versus 1.4±1.3, mean difference 1.1, 95% CI 0.2 to 2; OR 1.7, 95% CI 1.1 to 2.7). At the first visit, the 21 revisit patients had previously been discharged as 9 (42.9%) non-ACS and 12 (57.1%) ruled-out ACS. As revisited, return patients were finally classified into 2 (9.5%) attending mistake (0.5% of all non-ACS), 5 (23.9%) cardiologist mistake (2.1% of all ruled-out ACS), 7 (33.3%) patient mistake (1.1% of all discharged patients), and 7 (33.3%) no mistake (2.9% of all ruled-out ACS). The 7 revisiting patients who had been discharged by either the attending physician or the cardiologist had at least 5 of 6 revisiting patient characteristics listed above. For all 5 patients discharged by the cardiologist, EET was not performed because of physical patient limitations.

Methods: This was a before-and-after trial in a university-based ED with 75,000 annual visits. Participants were consecutive patients with a painful chief complaint. Intervention was an educational effort, including didactic lectures and use of posters and pocket cards, about proper use of analgesics in the ED setting. Demographic and clinical data were extracted from the medical records. Outcomes measured were percentage of patients receiving analgesics while in the ED and time from triage to analgesic administration. χ 2 And t tests were used as appropriate. A sample size of 360 patients had 90% power to detect a 10% absolute difference in analgesic administration (2-tailed α=.05).

Results: One hundred ninety-seven patient encounters before educational intervention were compared with 181 encounters after intervention. Groups were similar in regard to age and sex. Analgesia administration before and after intervention was 56% and 54%, respectively (P=.80). Time until analgesic administration before and after intervention was 90 and 82 minutes, respectively (P=.45). The likelihood of receiving analgesics increased with higher pain scores (χ 2 linear-by-linear test, P<.001). Analgesics were given less often in the urgent care area (35% versus 71%, P<.01). Analgesic use was unrelated to patient sex or age.

Conclusion: In our institution, introduction of an educational intervention about analgesia use in the ED did not reduce the rates of oligoanalgesia.

Study objectives: Elevations in pulse rate and systolic blood pressure are sometimes used to detect untreated pain. If pulse rate and systolic blood pressure are not correlated with a patient's report of pain, it is possible that some patients with untreated pain may go undetected. Furthermore, it is possible that patients with abnormal vital signs may have them inaccurately attributed to the pain they are experiencing. In this study, we sought to determine whether there was a correlation between the patients' reported pain and heart rate or systolic blood pressure.

Methods: This was a prospective observational study. Data were collected by trained research associates. Patients were asked to complete a 100-mm visual analog pain scale before treatment. Pulse rate and blood pressure were measured by triage personnel. Data were analyzed by using descriptive statistics and regression analysis.

Results: Two hundred fifteen subjects participated in the study (mean age 39.9 years, 95% confidence interval [CI] 38.2 to 41.6 years). There was a correlation (r=0.10, 95% CI 0.08 to 0.29) between the initial reported visual analog scale (VAS) score (mean 73.2, 95% CI 70.0 to 76.5) and the initial pulse rate (mean 87.6, 95% CI 85.3 to 90.0). There was not a correlation (r=0.07, 95% CI -0.11 to 0.18) between the VAS score and systolic blood pressure (mean 135.1, 95% CI 132.0 to 138.2).

Conclusion: There is a weak relationship between a patient's perceived level of pain and elevations in pulse rate but not in systolic blood pressure. Health care providers should avoid using elevated blood pressure or pulse rate to detect untreated pain or as the sole explanation of abnormal vital signs.

Fosnocht DE, Swanson ER, Chapman CR, Donaldson GW/University of Utah, Salt Lake City, UT Study objectives: The visual analog scale (VAS) has been validated and is commonly used to determine pain intensity. However, change in VAS has not been validated as a measure of pain relief in the emergency department (ED). The objective of this study is to evaluate the correlation between change in VAS and a patientreported verbal descriptor scale (VDS) of change in pain. We seek to determine whether change in VAS score is an appropriate method for tracking change in pain for individual patients in the ED.

Methods: The study design was prospective and observational. The setting was a university ED with an annual volume of 32,000 patients. Participants were a convenience sample of ED patients enrolled from January 2000 to April 2001. Inclusion criteria were patients presenting to the ED with pain. Patients excluded were younger than 18 years or too ill to participate. We measured pain with a 100-mm VAS at presentation, after medication, and at discharge, as well as VDS of change in pain (much less, little less, about the same, little more, much more) after medication and at discharge. Outcome variables were change in VAS for each verbal category and correlation of change in VAS with the VDS of change in pain.

Results: One thousand four hundred ninety patients yielded 1,999 comparisons and medication provided in the ED and at discharge were recorded by enrollers. t Tests and Fisher's exact test were used to compare the data. This study was determined exempt from informed consent by the institutional review board. Results: One hundred three patients were enrolled throughout a 6-month period. Of those patients, 30.1% (31) had fractures. There were no significant differences in age, sex, or body area (shoulder, wrist, etc.) of the fracture group versus the nonfracture group. Initial ED pain score between fracture and no-fracture groups was similar. For the fracture group, initial pain was 5.97 versus 6.21 for no fracture (P= 0.68), whereas discharge pain scores were 4.23 for the fracture group and 4.54 for no fracture (P=.60). The pain differential was also similar between the groups, 1.81 for the fracture group and 1.64 for no fracture (P=.80). Of patients with fractures, 45.2% received narcotics in the ED and 51.6% received narcotics at discharge. A smaller percentage with no fracture received narcotic pain medicine in the ED (36.4%) and at discharge (38.9%); however, the results were not significant (P=.51 and P=.20, respectively). While in the ED, patients with fractures received pain medications (narcotics, nonsteroidal anti-inflammatory drugs, or other) more frequently than those without fractures (77.4% versus 58.3%, P=.08). Similarly, on discharge, patients with fractures received pain medications more frequently than those without fractures (80.6% versus 73.3%, P=.32).

Conclusion: These data suggest that, despite having similar pain scores on ED arrival, patients with a documented fracture are more likely to receive narcotic analgesia in the ED and at discharge than those without evidence of an acute fracture. These data support also the finding that pain is undertreated in the ED, with patients from both groups being discharged with continued significant pain. A surprisingly high percentage of patients received no analgesia either in the ED or at discharge. Undertreatment of pain tended to occur more frequently in patients without fractures.

Fosnocht DE, Swanson ER, Linscott MS/University of Utah, Salt Lake City, UT Study objectives: The verbally administered 11-point (0 to 10) numeric rating scale (NRS) has been validated and is commonly used in the emergency department (ED). The 4-point (none, mild, moderate, severe) verbal descriptive scale (VDS) is simple, easily understood, and often used clinically to measure pain intensity. However, there are few data about validity of the VDS in the ED setting. This study objective was to evaluate the correlation between pain intensity measured using the VDS and the NRS.

Methods: Study design was prospective and observational. The setting was an academic ED with an annual volume of 32,000 patients. Participants were a convenience sample of ED patients presenting with pain from January 2002 to April 2002. Patients excluded were younger than 18 years or too ill to participate. The NRS is an 11-point (0 to 10) scale described with 0 equals no pain and 10 equals worst possible pain. The VDS is a 4-point scale with the words "none," "mild," "moderate," and "severe" used to quantify pain intensity. We measured pain with the NRS and VDS at patient presentation to the ED. Outcome variables were pain intensity measured by VDS and NRS. Correlation of the NRS and VDS was performed with Spearman's ρ.

Results: One thousand twenty-five patients were enrolled; 33 patients were excluded because of missing data, yielding a total of 992 comparisons of pain intensity. Mean age of patients was 38 years (95% confidence interval [CI] 37 to 39 years). There were 567 (57%) female patients. Mean pain intensity measured by the NRS at presentation was 6.8 (95% CI 6.6 to 7.0). Mean pain intensity measured by the NRS for each VDS category was: mild (n=166), 3.8 (95% CI 3.5 to 4.1); moderate (n=395), 6.2 (95% CI 6.1 to 6.4); and severe (n=431), 8.7 (95% CI 8.6 to 8.8).

Overall correlation of the NRS and VDS was ρ equal to 0.82 (95% CI 0.81 to 0.84).

Conclusion: Pain intensity measured with the VDS strongly correlates with pain intensity measured with the NRS. The mean pain intensity measured with the NRS increases in stepwise fashion (3.8 to 6.2 to 8.7) for patients describing mild, moderate, and severe pain with the VDS. The 4-point VDS appears to be a valid measure of pain intensity for ED patients. Limitations include convenience sampling and broad inclusion criteria. emergency department (ED) presentation of sickle cell patients. Tools such as the visual analog scale (VAS) are available for the assessment of pain. Although clinically significant changes in the VAS have been determined for a number of conditions, it has not been specifically examined in VOC. Our purpose was to determine the minimum clinically meaningful changes in the VAS during the ED treatment of VOC.

Methods: Patients older than 18 years and with uncomplicated VOC were eligible. Patients were treated with standard therapy consisting of parenteral narcotics administered every 2 hours for up to a maximum of 3 doses, according to ED protocols. A 100-mm VAS was administered at baseline and just before each analgesic injection. At each assessment, patients were asked to describe their pain as being "much better," "a little better," or "the same or worse." Correspondence between self-reported improvement and ∆VAS (contrast scores) was assessed. The minimally significant change was defined as the difference between treatment measures in patients reporting the pain as being "a little better." Data were analyzed with analysis of variance on ranks with Dunn's multiple pairwise comparison.

Results: Thirty-five patients reported a total of 69 VAS contrast scores. The median ∆VAS in the "much better" group (n=5) was 51 mm (interquartile range [IQR] 36.25 to 65 mm) and was 16 mm (IQR 5.75 to 26.25) in the "a little better" group (n=33). These scores were significantly different than the median ∆VAS of 2.0 mm (IQR -1 to 6.0; P<.001) in the "same or worse" group.

Conclusion: Improvement in the 100-mm VAS of 16 mm is the minimal clinically significant change during ED treatment of VOC. This finding may assist the clinician in the assessment of pain improvement for adult sickle cell patients with VOC.

Shoukry K, Bruno R/University Hospital Pitie-Salpetriere, Paris, France Study objectives: Patients presenting with onychocryptosis (OCP) in the emergency department (ED) experienced long periods of pain, difficulty in walking, and limitation of daily function. They had difficulties in obtaining a specialized consultation. Moreover, the actual methods of treatment included a high rate of infection, recurrence, and required adapted equipment. This work evaluated a method that surgically treated OCP by onysectomy, matricectomy, and nail plasty (bed and fold) in the ED.

Methods: This open prospective study was conducted during 12 months among consecutive patients presenting in the ED with OCP. Preoperatively, we noted the stage of the OCP, duration, pain, difficulty in walking, and primary recurrence. Surgery consisted of an incision of the dorsal fold, in a backward continuum of the lateral fold, to liberate the embedded lateral nail plate, followed by excision and avulsion of the lateral nail plate and corresponding underlying nail bed (partial onysectomy). A plasty of the lateral nail fold beyond the angle of the dorsal fold was performed, combined with excision of the granuloma. Finally, a lateral matricectomy was done using a number 15 scalpel blade. We used a braided polyester thread on 17-mm tapered needle to apply simple sutures on the dorsal incision, the lateral nail fold through nail holes, and the tip of the toe approximating the anterior border of the lateral nail fold to the nail bed. At follow-up, we recorded morbidity (pain, scar discomfort, and wound exudates), infection, growth of nail spikes, recurrence, and time to complete recovery (wearing shoes, performing normal activities, work). Data were collected and processed in a database, programmed on MS Access. Patients were examined postoperatively at day 4, day 7, and day 14. Sutures were removed at day 7.

Results: Thirty-six nail margins were operated on for 29 patients aged 33±10 years. Male to female patient ratio was 17:12. OCP lasted 6±4 weeks. Painful toe and difficulty in walking were present for 5±2 days. Preoperative recurrence was reported among 4 patients (13.8%). Stage I OCP was recorded 3 times, stage II 6 times, stage III 21 times, and stage IV 6 times. Infection was present 6 times, associated with bilateral OCP among stages III and IV. Big toe to small toes ratio was 32:4. Postoperative morbidity was noted among stages III and IV on 6 nail margins (16.7%; pain for 24 hours in 3 margins, discomfort among 4, and exudates among 6). Operated stages I and II healed in 8±2 days, whereas stages III and IV needed 2±1 week. Antibiotics were prescribed to 4 patients: 1 patient presented diffuse toe infection and systemic symptoms, whereas 3 patients (at the beginning of the study) presented local mild infection and needed frequent dressings only. One nail spicule was noted. OCP recurred among 2 treated margins (5.6%, in 2 weeks), after primary bilateral matricectomy of stage III or IV OCP. When reoperated with the same technique, margins healed without complications.

Conclusion: Patients with OCP could be treated with standard material available at the ED. This technique (onysectomy, matricectomy, and plasty) resulted in the (Table) . Conclusion: Change in VAS is moderately correlated with a VDS of change in pain. However, wide variability in change in VAS compared with a VDS of change in pain suggests that change in VAS is not a valid indicator of change in pain for individual patients. Limitations include convenience sampling of patients. Study objectives: There have been few investigations of propofol's amnestic effects at the low doses used during emergency department procedural sedation. Our objective was to assess patients' level of awareness by bispectral index monitoring (BIS) and their ability to repeat and subsequently recall words given during procedural sedation with propofol.

Methods: This was a prospective single intervention study of consenting adult patients undergoing procedural sedation with propofol between August 1, 2002, and January 4, 2003. BIS monitoring was performed starting 4 minutes before the procedure until the patient had regained baseline mental status. Every minute, a word from a standardized list was read to and immediately repeated by the patient. BIS scores at the time the word was read and the patient's ability to repeat the word were recorded. After the procedure, patients were asked to state all of the words they could recall, the last word they could recall from before the start of the procedure, and the first word they could recall after the procedure was completed.

Results: Twenty-four patients were enrolled; the mean time of data collection was 16.4 minutes (95% confidence interval [CI] 13.1 to 19.6). The mean initial BIS was 96.9 (95% CI 13.1 to 19.6). The mean lowest BIS corresponding to words repeated was 80.5 (95% CI 77.1 to 83.8). The mean highest BIS corresponding to words the patient was unable to repeat was 75.2 (95% CI 68.8 to 81.7). There were no patients who could not recall the first word read. No patients recalled words they had not repeated. The mean BIS corresponding to the last word recalled read before the procedure was 95.8 (95% CI 94.2 to 97.4, mean time 1.2 minutes). The mean BIS corresponding to the first word recalled read after the procedure was 87.9 (95% CI 79.1 to 95.7, mean time 15.9 minutes). The mean lowest BIS measured for a word the patient could later recall was 89.0 (95% CI 87.1 to 91.9). The mean BIS low recording during the procedures was 65.4 (95% CI 59.8 to 71.1).

Conclusion: There is a range of BIS scores during which patients were able to repeat words (mean BIS low of 80.5) that they were unable to recall after the procedure (mean BIS low of 89.0). Patients had no recall of words repeated before procedural sedation in BIS ranges associated with recall after procedural sedation (mean BIS of 95.8), suggestive of retrograde amnesia. with a known case of smallpox, personnel do not need fit-tested N-95 masks if they have been vaccinated; and that typically the rash of smallpox begins in the mouth (average 20% correct). Conclusion: Current full-time emergency physicians in Washington, DC, scored an average of 55% on a true/false test about the initial presentation and care of a patient with smallpox. Some facts from the current CDC smallpox Web site are apparently well known by a majority of full-time emergency physicians in Washington, DC. Some questions about the initial presentation and care of smallpox patients were answered incorrectly by a majority of the emergency physicians surveyed. Study objectives: Ethiopia has one of the highest rates of fatalities per auto in Africa, with 180 fatalities per 10,000 cars per year. Although reports indicate that the problem of injury and trauma is growing, we are unaware of any published report detailing the characteristics of injury and trauma patients who reach the hospital in Ethiopia. This study was an initial attempt to understand the characteristics of these patients to allow for future research, training, and systems development.

Methods: Injured patients were prospectively evaluated by using a trauma survey. The survey was conducted on patients registering to be treated in the emergency department (ED) at 8 hospitals in Addis Ababa. Patients were enrolled every other day for a 24-hour period, with follow-up medical record review until discharge or death. Data on trauma were obtained throughout a 6-month period and evaluated with Epi Info 6.

Results: Of the 8 hospitals, 7 returned surveys, with 1 not participating because of administrative and legal considerations within that hospital. Overall, 28% of all ED visits in the 6-month study period were related to injury or trauma, with 353 surveys completed. Two hundred seventy-six (78.2%) patients arriving at the EDs were men, with 214 (60%) in the 21-to 40-year-old age group. Two hundred eightysix (81%) patients arrived by private car, with 4 (1.2 %) receiving out-of-hospital treatment. The majority of injuries were caused by motor vehicle crashes, 315 (89.2%). There was a 16.7% overall fatality rate for all patients reaching the ED.

Conclusion: Injury is a growing problem in Addis Ababa. A large percentage of the injured are young men, with road traffic crashes responsible for the majority of trauma. The poor infrastructure and large number of cars contribute to the high injury rate. Developing countries need interventions aimed at injury prevention, out-of-hospital and hospital care to address this growing need.

Stone SC, Rudis MI, Lee VM, Chitchyan A/University of Southern California Keck School of Medicine, Los Angeles, CA Study objective: Immunization against Streptococcus pneumoniae has been shown to be cost-effective for prevention of invasive S pneumoniae, with more than $10,000/ quality-adjusted life year. Yet this vaccine is widely underused. Previous studies in emergency departments (EDs) suggest that immunization against S pneumoniae is feasible in this setting, provides for the least expensive venue for this intervention, and captures the greatest number of patients. The objectives of this survey were to determine the extent of pneumococcal vaccination among all adult patients presenting to the ED at a county-based, urban, tertiary care medical center and to determine the potential need for pneumococcal vaccine administration in this setting.

Methods: A quality assurance survey was performed in the ED during September 9 to 13, 2002. A survey tool was developed with the eligibility criteria (inclusion/ exclusion criteria) established by the Centers for Disease Control and Prevention. The tool was also translated into Spanish. All patients in the waiting areas and patient rooms were interviewed and completed the survey, with assistance from a research student in the patients' language of preference. Descriptive statistics were performed to quantify the proportion of patients who were immunized and those who were eligible for vaccination and to quantify the indications for vaccination.

Results: A total of 250 patients completed the survey. Of all patients who completed the survey, only 48 (19%) had a primary care provider; the majority used the ED as their source of primary care. Only 22 (8.8%) patients had received the vaccine. A total of 112 (45%) patients fit the Center for Communicable Diseases and Pre-absence of morbidity and recurrence in stages I and II. Low rate recurrence was noted with primary bilateral matricectomy among stages III and IV.

Hiestand B, Huang A, Dick M, Woodard S/The Ohio State University, Columbus, OH Study objectives: Multiple studies have documented undertreatment of pain in the emergency department (ED). The majority of these studies, however, have focused on specific age, race, and diagnosis groups. We performed data analysis on consecutive undifferentiated ED population to evaluate ED physician practice patterns and factors that influence ED physician behavior.

Methods: A retrospective medical record review was performed on 1,019 consecutive ED patients treated in a university-affiliated urban community hospital. Initial triage pain scores were obtained, as were pain duration and analgesic use before ED visit. ED actions were also abstracted from the record, including diagnostic studies (laboratory or radiologic evaluation), analgesic use (oral versus parenteral, narcotic versus nonnarcotic), the use of other medications in the ED, admission status, discharge analgesics, and the use of physical measures such as splinting and ice. Because of the nonparametric distribution of initial pain scores, the Wilcoxon rank sum test was used to analyze the data. Multivariable logistic regression models were used to find the best subset of variables that predicted the use of ED parenteral narcotics, discharge narcotics, admission status, and the use of ED diagnostic studies.

Results: Patients with higher initial pain scores at triage were significantly more likely to have been taking narcotic analgesics before the ED visit, to receive parenteral narcotics and oral nonnarcotics in the ED, and to be discharged after receiving narcotic analgesics (P<.001). Patients with increased duration of pain were significantly more likely to have been receiving narcotic analgesics before the ED visit, were more likely to receive oral narcotics in the ED, and were more likely to be discharged after receiving oral nonnarcotic analgesics (P<.001). Logistic regression analysis found that pain score, previous narcotic use, and hospital admission were the best combined predictors of ED parenteral narcotic use. High initial pain scores and previous nonnarcotic use were the best combined predictors of the use of discharge narcotics.

Conclusion: Emergency physician behavior appears to be strongly influenced by patient pain scores. In addition, emergency physicians are likely to increase analgesic therapy when the patient has tried other measures without relief. Continued surveillance is warranted; however, results from this study indicate that emergency physicians are more likely to treat patients' pain than previous studies have indicated.

Freed HA, Milzman D, Ervin MG/Howard University Hospital, Georgetown University School of Medicine, Washington, DC Study objectives: The purpose of this study was to assess the current knowledge of Washington, DC's, full-time emergency physicians about the initial presentation of smallpox and the initial care of the smallpox patient. Information from the current Centers for Disease Control and Prevention (CDC) Web site was used as the criterion standard.

Methods: A 20-question true/false test was prepared according to information about smallpox found on the current Web site of the CDC (www.bt.cdc.gov/agent/ smallpox/index.asp). Half the questions were based on correctly making the diagnosis, and half were based on appropriately protecting the staff and patients once the diagnosis is suspected. The questionnaire was administered to full-time emergency physicians practicing emergency medicine in Washington, DC. Physicians from the emergency departments (EDs) at 1 community hospital and 1 academic health sciences center were surveyed in this pilot study.

Results: The test was anonymously completed by 100% of the full-time emergency physicians working at the 2 hospitals (17 of 17 full-time emergency physicians). The average percentage of correct answers ("average score") was 55%; the median score was 55% correct; the mode (single most common) score was 45% correct. The facts most likely to be known by the emergency physicians were that the symptoms of smallpox begin with a 2-to 4-day prodrome of fever and myalgias (before the rash), that a person with smallpox may be contagious before the appearance of a rash, and that no antiviral treatment is of more proven value than the vaccination of contacts (average 90% correct). The facts least likely to be known were that the rash of smallpox begins as erythematous macules (not as papules or vesicles); that when dealing rent, 2% were former, and 69% were nonusers. Only 2% had documented referrals or counseling offered. Documentation of drug use was less likely in teens (41% aged 12 to 18 years; 63% aged 19 to 54 years; and 56% aged at least 55 years; P<.005), if patients were treated only by an attending physician/nurse practitioner (resident + attending physician 68%, medical student + attending physician 80%, attending physician/nurse practitioner 37%; P<.0001), and in white patients (white 51%, black 63%, Hispanic 63%; P<.05). Triage category, insurance status, and disposition did not significantly affect drug use documentation. Of 338 patients with drug use documented, 7% were current, 6% were former, and 87% were nonusers. Twenty-two percent of drug users had documented referrals or counseling offered.

Conclusion: The urban ED is often the only access to health care for many indigent and substance-abusing patients. Even so, documentation of alcohol (62%) and drug (56%) use is limited and referrals even more so. Greater emphasis should be placed on screening, particularly of adolescents, who may be more amenable to lifestyle changes and less likely to have experienced the deleterious effects of EtOH or drug use. The low rate of referral or counseling may be due to lack of documentation or limited services. Both warrant further investigation.

Flaherty JJ, Handler JA, Gillam MT, Wurtz RM/Evanston Northwestern Healthcare, Northwestern University Feinberg School of Medicine, Evanston, IL Study objectives: The purpose of this project was to use commonly available opensource computing resources to build a system for syndromic surveillance involving multiple institutions. The concept was to import our local electronic daily log into a database and then use query protocols to determine baseline frequencies of chief complaints, discharge diagnosis, various syndromes, and quality markers.

Methods: Freely available open-source tools were used, including MySQL, an structured query language (SQL)-compliant database, Perl, and PHP Web scripting languages. A flexible existing network of 3 suburban-based hospitals was chosen for initial ease of data management. These 3 hospitals have a total emergency department (ED) annual census of 90,000 patients, and the existing computerized logs were translated into XML and downloaded over the Internet into a dedicated server by using a SQL protocol. The XML data files were later queried by using specifically designed protocols and graphically presented by using a low-cost charting tool to present graphs in real time and by using importation into statistical analysis programs such as SPSS, SAS, or Excel.

Results: The data were collected throughout a 12-month period, showing typical seasonal spikes in presentations for respiratory illness, gastrointestinal illness, and febrile illness, as well as a distribution of chief complaints. Baseline frequency distributions of these conditions can be determined, and outbreaks can be identified. Additional performance characteristics and information were easily identified and accumulated, showing total ED census per hour, admissions per hospital per hour, and total visit times.

Conclusion: We conclusively demonstrated that syndromic surveillance can be set up and can function well over the Internet for 3 related but geographically distinct locations by using open-source computing languages. This is a lightweight, low-cost, readily usable method of implementing tools for syndromic surveillance in the ED. These data can be collected, stored, and queried to find the baseline presenting frequencies for establishing syndromic surveillance and may be useful in enhancing the rapid detections of outbreaks and epidemiologic analysis of unusual or atypical presentations of disease. Study objectives: We describe characteristics of emergency department (ED) patients who smoke tobacco, shown in previous studies to be linked to willingness to participate in interventions and successful cessation.

Methods: Consecutive ED patients aged 18 years or older were interviewed. Patients who were severely ill or who had altered mental status were excluded. The following criteria were assessed: smoking rate, desire to quit, self-efficacy, presence of a smoking-related illness, interest in an ED-based smoking intervention, willingness

vention inclusion criteria for vaccination and had no exclusion criteria, and 99 (39.6%) were eligible and willing to receive the vaccine during their ED visit. Of patients who were eligible and desired the vaccine, fewer than 8% were older than 65 years. Most patients (92%) fulfilled the criteria for receiving the vaccine because of the presence of comorbid illnesses. The majority of patients cited lack of awareness of the existence of the vaccine or of its potential benefit as reasons for not receiving the vaccine. Conclusion: The outcome of this quality improvement survey indicated that 39.6% of patients in the ED of a county-based urban medical center were eligible for and desired vaccination against S pneumoniae, which is consistent with existing data in the literature. Therefore, at our institution, a need exists for provision of an S pneumoniae immunization program. A program for immunization against S pneumoniae in the ED is being developed, and a cost-effectiveness analysis evaluating this intervention in our ED setting is planned.

Leikin J, Brand R, Flaherty JJ, Gillam MT, Feinglass J/Evanston Northwestern Healthcare, Northwestern University Feinberg School of Medicine, Evanston, IL Study objectives: Previous studies have clearly documented the effects of environmental air pollution, especially ozone and particulate matter, on the admission rates and death rates of patients with chronic lung disease. Ozone is an irritant in air pollution that can cause increased airway reactivity. The purpose of this study is to analyze the influence of regional levels of ozone and atmospheric particulate matter in the type and number of emergency department (ED) visits in the northeastern section of Illinois.

Methods: A passive syndromic surveillance system was set up by using a computerized patient log. Daily downloads of data in XML format from 3 regional hospitals with an annual census of 90,000 patients, linked in an existing network, were accumulated in a dedicated server using a structured query language (SQL) protocol throughout a 12-month period. A retrospective cohort study using SQL queries was run to search for chief complaints and discharge diagnoses of a respiratory etiology. Positive matches were compiled in a graphic and geographic database. Ozone levels measured hourly were averaged over 24 hours. Comparative and descriptive statistical analysis was performed with SPSS.

Results: ED daily visits from the 3 hospitals ranged from 140 to 326. Hourly ozone levels ranged from 0.063 parts per million to 1.468 parts per million. Particulates ranged from 0.08 to 43.3. There was no correlation between ozone levels and ED visits or admissions for adults or children. There was a mild correlation with pediatric admissions and elevated particulate levels (r=0.24, P=0.03). There was no correlation between ED cases of asthma and ozone or particulate levels.

Conclusion: Previous studies have documented increased incidence of airwayrelated conditions with elevated ozone levels higher than 0.8 parts per million. Evaluations of the ozone levels well below this level appear to have little effect on ED visits and admissions for respiratory-related conditions in the area studied. Further study and larger regional ED data may show a positive correlation. The availability of air conditioning and ozone alerts may also have a significant effect of limiting environmental exposure in this area. Methods: Six hundred eight consecutive ED records were reviewed by using a standardized abstraction form for demographic data, documentation of alcohol (EtOH) and drug use (present/former), and documentation of counseling or referral (present/absent).

Results: Documentation of EtOH was less likely in teenagers (41% aged 12 to 18 years; 68% aged 19 to 54 years; and 71% aged at least 55 years; P<.001), patients triaged as routine (emergency 67%, urgent 71%, routine 50%; P<.0001), or if treated only by an attending physician or nurse practitioner (resident + attending physician 76%, medical student + attending physician 83%, attending physician/nurse practitioner 40%; P<.0001). Race, sex, and insurance status were not associated with EtOH documentation. Of 376 patients with EtOH use documented, 29% were cur-

urban teaching ED were administered a health questionnaire by trained research assistants during 5 rotating 8-hour shifts a week. Patients were ineligible if they were alcohol dependent, if they were from an extended care facility, if illness precluded interview, if they were pregnant, or if they presented with an acute psychiatric complaint. Hazardous drinking was defined according to National Institute on Alcohol Abuse and Alcoholism (NIAAA) criteria: more than 14 drinks a week or 4 drinks per occasion for men; more than 7 drinks a week or 3 drinks per occasion for women or men older than 65 years. Harmful drinking was defined as an injury, with a blood alcohol concentration of at least 0. Conclusion: Five percent of ED patients are harmful or hazardous drinkers who are more likely to be young men, to be injured, to smoke, and not to consistently use seatbelts. These at-risk patients may not be identified by use of the standard AUDIT screen for harmful or hazardous drinking.

Hoffman RJ, Morgenstern SS, Hoffman RS, Nelson LS/Beth Israel Medical Center, New York City Department of Health Poison Control Center and Schneider Children's Hospital, New York, NY Study objectives: In observance of the Sabbath and other religious holidays, many Orthodox Jews maintain a burning lamp by using paraffin lamp oil as fuel. Unintentional pediatric exposure to this hydrocarbon, typically by ingestion, carries risk of aspiration with subsequent pneumonitis. Prompted by an apparent increase of paraffin lamp oil exposures during the Jewish Sabbath, from sunset Friday until sunset Saturday, noted by the staff of our regional poison control center, we sought to determine whether there is an increased incidence of paraffin exposure in Orthodox Jewish children.

Methods: In this institutional review board-exempted investigation, we retrospectively reviewed all exposures to paraffin lamp oil occurring in our large urban city in children younger than 18 years and reported to our regional poison control center between January 1, 2000, and February 1, 2003. Reports were investigated to ascertain the frequency of occurrence on the Jewish Sabbath and religious holidays. Caregivers of involved children were surveyed by telephone to determine the child's religion and circumstances of exposure.

Results: During these 25 months, 45 cases met inclusion criteria, and all were ingestions. Orthodox Jews accounted for 32 cases (71%), 4 cases (9%) occurred in children who were not Orthodox Jews, and demographic data were unavailable in 9 cases (20%). Twenty-four cases (53%) occurred within 10 hours before the Jewish Sabbath or religious holidays or during them.

Conclusion: The relative risk of Orthodox Jewish children to ingest paraffin lamp oil, calculated by using census data, is 829 times that of other children. Public health authorities and caregivers of Orthodox Jewish children should be cognizant of this phenomenon. Educational efforts directed toward these preventable poison exposures are warranted.

Patterson LA, Costantino T/Drexel University College of Medicine, Philadelphia, PA Study objectives: We determine the prevalence of disease discovered by emergency ultrasonographic scans performed at an emergency medicine residency pro-to remain an extra 15 minutes after discharge to participate in such a program, and whether a registered nurse or physician had screened for or counseled the patient on smoking.

Results: Of the 2,513 adult patients treated in the ED, 1,543 (61.4%) were successfully interviewed. Six hundred thirteen (39.7%) were current smokers, 399 (25.9%) were former smokers, and 531 (34.4%) had never smoked. Three hundred fifty-three (57.6%) smokers rated their desire to quit as "very strong," and 269 (43.9%) were "very confident" they could do so. Two hundred twenty (35.9%) smokers intended to quit within the next 30 days, and 359 (58.6%) had gone at least 24 hours in the last month without smoking. One hundred five (17.1%) patients stated they believed they had a smoking-related illness. Three hundred seventy (60.4%) patients were at least "somewhat interested" in an ED-initiated intervention, with 287 (46.8%) stating they would stay 15 extra minutes to receive such counseling. Three hundred fifty-two (57.4%) current smokers stated that a registered nurse or physician had asked them about smoking during their ED visit, but only 37 (6.0%) patients were given a referral or information about cessation.

Conclusion: More than 50% of current smokers treated in the ED were strongly interested in quitting and were confident in their ability to succeed. Many smokers appeared receptive to interventions initiated in the ED, with nearly 50% agreeing to stay an extra 15 minutes after discharge to receive help. However, only 6% reported being given any information about cessation during their visit. Findings will be discussed in light of recent mandates to provide routine tobacco screening and referral in the ED.

Bernstein E, Bernstein J, Tassiopoulos K, Heeren T, Levenson S, Hingson R/Boston University School of Medicine, Boston University School of Public Health, Boston, MA; Harvard School of Public Health, Boston, MA Study objectives: Research is needed to identify practical methods to help drugusing patients. We determine whether peer educator-led brief motivational interview and active referral is effective among cocaine and heroin users in the clinical setting.

Methods: A randomized controlled trial was conducted during May 1998 to November 2000 in urban teaching hospital clinics. Of 23,669 patients screened, 1,232 (5%) were eligible (Drug Abuse Severity Score of at least 3), 1,175 enrolled, 590 were assigned to intervention, and 585 were assigned to a control group. Twenty-nine percent of enrollees were women, 62% were non-Hispanic black, 23% were Hispanic, 46% completed high school, 83% were unemployed, and 46% were homeless; mean age was 38 years. Intervention consisted of a 20-minute conversation between peer educators and patients to explore pros and cons of drug use, readiness to change, options, and plans and to facilitate treatment entry. Main outcome measures were 30-to 90-day drug treatment contact rates, 3-to 6-month Addiction Severity Index (ASI) scores, and negative hair test result for cocaine or heroin use.

Results: Two hundred ninety-four of 1,175 (25%) enrollees were found in the state's treatment database at 30 days after enrollment, and 39% were found at 90 days after enrollment. No significant difference between groups was found for contact. Both groups improved over time on all ASI subscales, although the intervention group showed a strong trend on the 3-month drug and medical subscales. Negative cocaine hair test results occurred at 6 months in 84 (22.3%) of 376 patients in the intervention group who were cocaine positive at baseline compared with 58 ( Study objectives: We determine whether bedside ultrasonography in the emergency department (ECUS) can be of value to emergency physicians in the evaluation of possible ascites and accompanying decisions to perform emergency paracentesis.

Methods: During a 7-month period, all patients suspected of having ascites and potentially requiring paracentesis were prospectively entered into a randomized study in an urban public hospital emergency department (>140,000 annual visits). Patients were randomized to receive paracentesis by using traditional landmarks or the bedside ECUS technique. Indications for paracentesis included those with known liver disease and obvious ascites, as well as those with suspected ascites or suspected subacute bacterial peritonitis. Participating physicians had received a minimum of 1 hour of formal didactic ultrasonography training that included gallbladder, renal, vascular, and bladder studies, as well as the focused abdominal sonography in trauma examination for trauma and the detection of ascites. A portable Terason 2000 laptop ECUS machine with a 5-MHz probe was used to scan the patients. Data collected included the extent of the emergency physicians' previous ECUS experience, patient characteristics, estimation of ascitic fluid volume, number of attempts made to obtain fluid, speed of paracentesis, and the operator's overall evaluation of the ECUS technique, if used.

Results: Of 60 enrolled patients, 34 received ECUS, and 13 of these patients did not receive paracentesis because no fluid or insignificant amounts of ascites were visualized. One patient was noted to have a large cystic mass in the left lower quadrant, and another patient had a ventral hernia. Of the 26 patients randomized to traditional techniques, all received paracentesis. Among 7 (27%) of these patients, fluid could not be obtained by using traditional methods, and the ECUS was then used. Of these 7 patients, ascitic fluid was obtained for 5 patients, and the remaining 2 patients did not have enough fluid to sample. Regardless of training and experience, emergency physicians reported an ease in detection of ascites and that the ECUS technique was especially helpful in determining cases of absent or subtle ascites, as well as in the eventual collection of samples.

Conclusion: All patients in the ECUS group who had a needle paracentesis performed had ascitic fluid successfully obtained. ECUS also helped to locate and obtain fluid not retrieved by traditional techniques. Although it is not absolutely known whether the bedside ECUS fails to detect some cases of occult ascites, bedside ECUS appears to avoid many unnecessary paracenteses. Further study will help to confirm the true sensitivity and specificity of the ECUS.

Ko P, Chirocolo J, Theodoro D, Hormodzdi S, Trimarco T, Ward MF/North Shore University Hospital, Manhasset, NY Background: It is commonly accepted that residents develop proficiency in procedures such as emergency ultrasonography with progressive training.

Study objectives: We hypothesized that the sensitivity and specificity of residentperformed emergency ultrasonographic studies improved with advancing postgraduate year (PGY) level.

A Results: Eight thousand nine hundred six total emergency ultrasonography scans were performed. There was an average of 50 residents per year, for an average of 178 scans per resident. There were 3,469 focused assessment by sonography in trauma (FAST) scans, of which 302 were positive for free fluid (prevalence of 8.7%). Of 1,406 right upper quadrant scans, 409 (29%) were positive for stones. Of 1,026 aorta scans, there were 108 (11%) abdominal aortic aneurysmectomy. Of 770 kidney scans, 307 (40%) were positive for hydronephrosis. Of 708 pelvic ultrasonographic scans, 528 (74.6%) were positive for intrauterine pregnancy. Of 286 cardiac scans, 70 (24%) were positive for pericardial effusion, 25 (9%) of which were noted to be large. Of 212 deep venous thrombosis studies, 43 (20%) had deep venous thrombosis. The remaining 1,029 scans were mostly ultrasonography-guided procedures.

Conclusion: The overall prevalence of pathology demonstrated using emergency ultrasonographic scans was approximately 15%, and the average resident performed 2 to 8 scans demonstrating pathology from each category. The categories of scans least frequently performed revealed the highest prevalence of pathology. Although this study did not assess competency, the results highlight the importance of a structured ultrasonography didactic program at each residency to adequately prepare residents to recognize and interpret the relatively few pathologic ultrasonographic scans they will perform.

Milniker LA/New York Methodist Hospital, Brooklyn, NY Study objectives: I compare the cost-effectiveness of evaluation of blunt abdominal trauma with an algorithm including focused abdominal sonography in trauma (FAST) and an algorithm without FAST.

Methods: I used decision tree sensitivity analysis to construct a theoretical model of the cost-effectiveness of the decisionmaking process by using National Trauma Data Bank and other published data to estimate the prevalence of intraabdominal injury (IAI) and the usage rates, accuracies, and charges of computed tomography (CT), diagnostic peritoneal lavage (DPL), and early ultrasonography (EUS).

Results: For the decision analysis, with an IAI prevalence of 10% (all for CT, DPL, and FAST, respectively), usage rates were 50%, 20%, and 0% for the No FAST algorithm and 25%, 2%, 50% for the FAST algorithm; accuracies were 97%, 75%, and 94%; and charges (2002 dollars) of $1,658, $368, and $258 revealed a 30% reduction in charges per patient evaluation, with a significantly lower nontherapeutic laparotomy rate with the FAST algorithm. When a cost-to-charge ratio of 40% is used, the potential savings of the FAST algorithm is $15 million per 100,000 patients evaluated.

Conclusion: Millions of patients are evaluated nationwide each year for blunt abdominal trauma; more than 500,000 patients require operative intervention. Not all patients are stable for transfer to the CT suite, and the DPL leads to an unacceptable rate of nontherapeutic laparotomies. Using FAST examinations as the primary screening modality, with complementary use of CT, leads to rapid and accurate identification of patients in need of operative intervention. A normal FAST examination is associated with a low probability of IAI. The use of FAST can lead to 2-fold reduction in the use of CT and 10-fold reduction in the use of DPL, without sacrificing accuracy and with excellent cost-effectiveness.

Milniker LA/New York Methodist Hospital, Brooklyn, NY Study objectives: The National Trauma Data Bank (NTDB) of the American College of Surgeons was used to compare the charges for care of patients with blunt abdominal trauma evaluated with early ultrasonographic (EUS) examinations with those evaluated without early ultrasonography (No EUS).

Methods: We used International Classification of Diseases, Ninth Revision (ICD-9) codes to identify the patients in the NTDB with blunt abdominal trauma and Current Procedural Terminology-4 codes to further identify the subset of patients documented as having had an ultrasonographic examination less than 3 hours after presentation

Results: No condom leakage was observed. Two specimens were excluded because the positive control plates had equivocal results. No plates of either group had CFU after 48 hours' incubation (N=36; alcohol only 18, alcohol and Transeptic spray 18). One plate of a swab before cleaning and before use had CFUs.

Conclusion: The use of a condom and a simple cleansing method such as an alcohol wipe with or without using Transeptic spray is sufficient to prevent transmission of CFUs.

Sierzenski PR, Blaivas M, Leech SJ, Theodoro D/Christiana Care Health System, Newark, DE; Medical College of Georgia, Augusta, GA; North Shore University Hospital, Manhassett, NY Study objectives: We describe format, goals, and requirements of several emergency medicine ultrasonography fellowship programs (EMUS-FP).

Methods: A descriptive survey including length of training, clinical work hours, academic duties and hours, ultrasonography requirements, and fellowship goals were compiled from EMUS-FP directors at 4 institutions. Fellow clinical, academic, and ultrasonography hours per month were calculated from work logs and EMUS-FP director logs. Average total ultrasonographic examinations per month performed by fellows was calculated from ultrasonography logs. Each fellowship is located at an institution with an Accreditation Council for Graduate Medical Education postgraduate year 1 to 3 emergency medicine residency program. Each EMUS-FP primary hospital is a Level I trauma center with more than 70,000 emergency department (ED) visits. Each EMUS-FP reviewed is directed by a board-certified emergency physician certified as a registered diagnostic medical sonographer (RDMS). Seventy-five percent of EMUS-FP directors surveyed were fellowship trained.

Results: The EMUS-FP reviewed had several similarities. All EMUS-FP met American Institute for Ultrasound in Medicine guidelines for physician training in multiple applications, with fellow credentialing as an RDMS a primary goal. The average minimum number of ultrasonographic examinations to be performed was 600. Average required clinical, academic, and ultrasonography hours were 17.4, 22.4, and 8 per week, respectively. All EMUS-FP offered basic and advanced applications. Seventy-five percent require ultrasonography research and article submission. All EMUS-FP in this review required at least 4 distinct ED-based ultrasonography shifts each month. All EMUS-FP included staff education as an academic duty, as well as quality assurance and review duties for all ED-performed ultrasonographic examinations. The average number of EMUS performed by each fellow is 76 per month.

Conclusion: Current EMUS-FP likely entail a 1-year format, with delineation of clinical, academic, and ultrasonography requirements, including RDMS certification, 600 ultrasonographic scans, and ultrasonography research. The format described in this study may represent a standard for EMUS-FP training to meet recognized national requirements for physician training and credentialing in emergency ultrasonography.

Clero J, Richard O, Boutot F, Pereira E, Lambert Y/Centre Hospitalier de Versailles, Le Chesnay, France Study objectives: Rapid sequence intubation (RSI), according to experts' recommendations, is a common practice in emergency medicine. Nowadays, only vital signs such as consciousness (Ramsay score), systolic blood pressure, pulse rate, and physical findings evaluate depth of anesthesia. The bispectral index (BIS) electroencephalographic device is a dimensionless scale based on electroencephalograph, β activity, and burst suppression. It is considered validated for monitoring depth of anesthesia in the operating room. The aim of this study was to evaluate the level of sedation in patients undergoing emergency intubation by using the BIS monitor.

Methods: This pilot prospective study included adults who were intubated outof-hospital by mobile ICUs. Investigators applied the BIS to the patient before intubation. BIS values and vital signs were recorded as a function of time: before and after induction and then every 10 minutes throughout the next 30 minutes. Data were recorded on a computer, and descriptive statistics were generated. A Wilcoxon signed rank test was used to compare BIS values. A P value less than .05 was considered significant.

Results: Forty adults (28 men/12 women) were included. Mean patient's age was

Methods: This is a retrospective review of an educational tracking database gathered at a large community teaching hospital with an annual census of approximately 60,000 during a period of 22 months. At our institution, residents are required to record their emergency ultrasonography results on tape and on a data form for later correlation with a follow-up study and tape review by the emergency medicine ultrasonography program director and fellows. For the purposes of this study, we compared resident PGY level, ultrasonographic findings, and correlation with the follow-up study.

Results: A total of 1,256 examinations were performed. Of these, 252 examinations were performed by the emergency medicine residents alone. One hundred eighty-three examinations were eligible for the study analysis. PGY-1 residents performed a total of 11 scans, with a sensitivity of 83% (95% confidence interval [CI] 36% to 99.5%) and specificity of 100% (95% CI 55% to 1.0). PGY-2 performed 52 studies, with a sensitivity of 89% 95% CI 67% to 97%) and specificity of 88% (95% CI 72% to 97%). PGY-3 performed 120 studies, with a sensitivity of 93% (95% CI 82% to 98%) and a specificity of 86% (95% CI 76% to 94%).

Conclusion: Contrary to our hypothesis, emergency medicine residents gain proficiency in emergency ultrasonography for screening but not for diagnostic purposes with increased PGY level of training and number of emergency ultrasonographic examinations. Further evaluation of emergency medicine ultrasonography training is warranted to explore the reasons for these results and may imply that further training is necessary for diagnostic proficiency.

Chang J, Heendeniya K, Peralta M, Shiblee T, Hsu CK/The Brooklyn Hospital Center, Weill College of Medicine, Cornell University, New York, NY Study objectives: Previous investigations have demonstrated the use of ultrasonography to confirm the diagnosis of depressed skull fracture (DSF). We determine whether residents using a model for DSF can confirm the diagnosis with ultrasonography.

Methods: Porcine heads were harvested from an Institutional Animal Care and Use Committee-approved laboratory. Skin was reflected from the skull by using a flap incision. "Fracture" site was prepared by drilling a hole 1.5 cm in diameter, pressing the fragment to a depth of 1 cm, and filling the cavity formed with gel. A "control" site was prepared in a similar fashion without drilling. The skin flaps in both fractures and controls were sutured into place. Residents trained on known fracture and control sites by using ultrasonography with a 10-MHz linear probe. During the study, 15 fracture sites and 15 control sites were presented randomly to blinded residents. The probe was placed by a nonblinded assistant. Resident's manipulation of probe was limited to twist or tilt the probe.

Results: The presence of a DSF was confirmed within 30 seconds (N=18; sensitivity 98.9%; specificity 98.2%; positive predictive value 98.2%; negative predictive value 98.9%).

Conclusion: Using this model for DSF, residents can be trained to use ultrasonography to rapidly evaluate and confirm the diagnosis of a depressed skull fracture. Limitations include a nonliving animal model and small sample size. The potential differential effect of fresh and clotted blood in the DSF cavity versus ultrasonographic conductive gel was not studied. Study objectives: We determine whether VUS probes act as a vector for crosscontamination if manufacturer-recommended cleaning procedures are followed.

Methods: We determined whether VUS probes can transfer colony-forming units (CFU) from patients to agar plates after 2 simple cleaning methods. Before use, each probe was cleaned with an alcohol wipe and Transeptic spray and then swabbed to inoculate 1 of 4 blood agar plates. After use and removal of the condom, the probe was swabbed to inoculate a second plate. All probes were cleaned by either alcohol wipe only or Transeptic spray and an alcohol wipe. After drying, each probe was swabbed to inoculate a third plate. The fourth blood agar plate was inoculated by using a swab from the investigator's mouth as a positive control. All used condoms were inspected for leakage. All plates were examined for growth by a laboratory blinded to the cleaning method.

residence. The fact that many opiate overdoses occur at home lends itself to such preventive interventions as prescribing naloxone for home use. It was difficult to assess the adequacy of EMS medical responses for opiate overdoses because of missing data. According to the database, naloxone usage did not appear to follow standard clinical criteria, such as respiratory rate of 12 breaths/min and miotic pupils. Updating the Rhode Island EMS protocols on naloxone to include these clinical criteria would promote more appropriate and consistent administration of this medication. Study objectives: Although the question of whether etomidate is superior to midazolam for facilitated intubation (without paralytics) has been a source of considerable controversy, the 2 agents have never been compared in a prospective, controlled, double-blind, randomized trial, to the authors' knowledge. This is a preliminary report on an ongoing out-of-hospital study designed to answer this question.

Methods: All patients aged 18 years or older, transported by the 2 participating advanced life support ground out-of-hospital systems, who require out-of-hospital endotracheal intubation with a sedative agent are eligible. Paramedics determine the need for intubation according to usual practice: low pulse oximetry with supplemental oxygen or an inability to protect the airway. Criteria for intubation with a sedative agent are the need for intubation and the presence of 1 or more of the following: agitated behavior, clenched teeth, or gagging. Exclusion criteria are pregnancy and age younger than 18 years. A systolic blood pressure less than 90 mm Hg requires medical command before study entry. Each ambulance is stocked with identical-appearing, randomized, blinded, numbered syringes equally filled with either 7 mg of midazolam or 20 mg of etomidate. No paralytics are used. If intubation conditions are not met with the study drug, the medic contacts medical command for consideration of additional sedation with midazolam or use of the standard rescue airway.

Results: Twenty-five patients have been enrolled in the study to date; 12 received midazolam and 13 received etomidate. There were 2 failures in the etomidate group and 1 for midazolam. There has been no statistically significant difference in change in blood pressure before and after intubation between the 2 study groups. However, in the 3 hypotensive (blood pressure ≤100 mm Hg) study patients, all of whom received midazolam, there was a clinically significant decline in blood pressure. Fifty percent of midazolam cases were described as "most" or "fairly" difficult versus 31% of the etomidate group. When asked which study drug they used, medics were correct 15 (60%) of 25 times.

Conclusion: This preliminary analysis suggests that for out-of-hospital intubation, a 7-mg dose of midazolam in hemodynamically stable patients may be similar in terms of safety and efficacy to 20 mg of etomidate.

Lim SH, Teo WS, Anantharaman V, Chan H/Singapore General Hospital, Singapore; National Heart Centre, Singapore; Ministry of Health, Singapore Study objectives: Singapore is an island with an area of 648 km 2 , with a population of 3.67 million. In 1994, all the ambulance officers of emergency medical services (EMS) were trained to perform out-of-hospital defibrillation with semiautomated external defibrillation.

Methods: All nontraumatic cardiac arrest in patients older than 10 years was included. Contraindications included patients who were obviously dead and children younger than 10 years or weighing less than 36 kg. The data from the Heartsave project were collected prospectively by ambulance officers in a standard report according to Utstein guidelines.

Results: From February 1, 1994, to January 31, 1999, there was a total of 968 patients with cardiac arrest of nontraumatic origin on whom resuscitation was attempted. In 15% of the cases, the etiology of the arrests was not of cardiac origin. The overall survival rate (cardiac etiology group) is 39 of 819 patients (4.77%; 95% confidence interval [CI] 3.4% to 6.5%) and for all patients was 40 of 968 (4.1%; 95% CI 2.9% to 5.6%). From the 968 cardiac arrests, 22 of 136 (16.2%; 95% CI 10.4% to 23.5%), 18 of 622 (2.9%; 95% CI 1.7% to 4.5%), and 0 of 210 (0%; 95% CI 0% to 1.7%) survived in the EMS-witnessed, bystander-witnessed, and unwitnessed groups, respectively (P<.001). Collapses witnessed by EMS had a significantly higher

51±22 years. BIS monitoring was feasible and easily performed on scene and during transport. All patients underwent RSI (severe trauma, cardiac insufficiency, neurologic coma, intoxications) and were sedated with midazolam and fentanyl. BIS values were 68±21.6 before RSI, decreased to 32±16 just after induction, increased to 55±24.6 10 minutes after induction, and increased to 62±24 10 minutes later (P<.05). Sedation was adjusted for 20% of the patients. There was no change in systolic blood pressure or pulse rate correlated with BIS changes. Conclusion: BIS monitoring has a high sensitivity to assess consciousness during RSI and sedation from general anesthesia. This new tool is not reliably correlated with clinical status of the patient (Ramsay, systolic blood pressure, pulse rate). Recommended values for adequate sedation are reached as expected and allow an objective control of induction. During steady-state anesthesia, BIS should be considered to detect early waking and to adjust anesthetic drug infusion. The use of that characteristic should help avoid excessive sedation.

Merchant RC, Schwartzapfel BL, Wolf FA, Rich JD/Rhode Island Hospital, The Miriam Hospital, Brown University School of Medicine, Providence, RI.

Study objectives: Morbidity and mortality from opiate overdose is a significant public health issue, but given the illicit nature of recreational opiate use, it is difficult to estimate the scope of the problem. Because patients who overdose often use the emergency medical system (EMS), the number of EMS ambulance runs may be a proxy for the frequency of opiate overdose events. In this study, we aimed to estimate the frequency of opiate overdoses occurring in Rhode Island between 1997 and 2002 and to assess the adequacy of EMS interventions.

Methods: For each ambulance run in the state, EMS crew members must complete a standardized form that summarizes the event, the patient demographic and clinical information, and EMS interventions. The Rhode Island Department of Health maintains a database composed of information from these forms. We requested a query of the database of all ambulance runs from 1997 through 2002 in which EMS crew members recorded "poison/OD" (ie, overdose) as a primary or secondary suspected illness. Because opiate overdose is not a discrete data entry point in the database, we identified suspected opiate overdose cases by EMS use of naloxone. Cases in which naloxone was not administered were designated as "other overdoses." We conducted our analyses with STATA version 7.

Results: From 1997 to 2002, there were 732,004 total ambulance runs in Rhode Island, for a mean of 122,000 runs per year. Of the total, 8,763 (1.2%) were for overdose. Of the 8,763 total runs for overdose, 1,629 (18.6%) were for suspected opiate overdoses, and the remainder (81.4%) were for other overdoses. The mean number of suspected opiate overdoses per year was 272. Suspected opiate overdoses were more likely to occur in a private residence, either the patient's (59%) or another residence (12%), than any other place (P<.001). Suspected overdoses happened more frequently on Fridays (15.9%) and Saturdays (16.0%), were more likely to occur on the first of the month than on any other day (4.7%; P<.002), and from 8 to 9 PM than any other hour (P<.001). Ninety-five percent of the suspected opiate overdoses were in patients younger than 54 years. For the patients' clinical status and EMS interventions, there was no difference between the suspected opiate overdose (48%) and other overdose groups (52%) in the percentage of patients with respiratory rates of 12 breaths/min (P<.20). However, in only 50% of suspected opiate overdoses and 71% of other overdoses, EMS crew members recorded a normal respiratory rate (>12 and equal to 20 breaths/min; P<.001). Of the patients for whom pupillary size was recorded, suspected opiate overdose patients were equally likely to have normal, dilated, or constricted pupils (P<.20). EMS crew members did not record respiratory rates for 22% of patients or pupillary size for 40% of patients. Ninety-three percent of patients were transported to an emergency department, 2% refused transport, 0.9% were treated on scene and not transported, and 0.25% were dead on arrival.

Conclusion: To our knowledge, this is the first study in the United States that uses EMS data to estimate the frequency of opiate overdose occurrences. We found that opiate overdoses in Rhode Island happen more often than previously estimated by using other methods. However, this study probably underestimates their frequency because it is well documented that heroin users often do not call for EMS assistance. Ambulance runs for suspected opiate overdoses occurred primarily at the first of the month, on Fridays and Saturdays, and between 8 and 9 PM. These temporal associations indicate specific times that might be targeted by prevention programs. EMS evaluated at least 71% of the suspected opiate overdose patients at a private

LBB or placement of FSS around patient), log roll, and lifting. During each phase, the amount of sagittal flexion, lateral flexion, and axial rotation were computed. Comfort and perceived security were also assessed on a visual analog scale. Repeated measures analysis of variance was used to test the hypothesis for each phase. Results: Degrees of lateral and sagittal movement during the application of the LBB were significantly more than the FSS. (10.6±0.7 degrees versus 2.9±0.2 degrees; P<.001). No difference was found during a secured log-roll maneuver. The FSS resulted in more sagittal flexion during the lift than the LBB (4.8±0.3 degrees versus 3.2±0.2 degrees). The FSS demonstrated superior comfort and perceived security.

Conclusion: The FSS had less movement on application and increased comfort levels. Decreased movement with the FSS may reduce the risk of further spinal cord injury.

Fullagar CJ, Prasad NH, Cherry RA, Brown LH/SUNY Upstate Medical University, Syracuse, NY Study objectives: There is no national regulation of field emergency medical services (EMS) providers. The requirements are left to and vary among the 50 states. Physicians sometimes choose to work alongside their EMS provider counterparts as an extra person or as an observer, but the requirements of physicians wishing to work as a primary EMS provider (ie, not an extra person) are also left to and differ among the states. The purpose of this study is to describe the various requirements of physicians wishing to function as primary field EMS providers.

Methods: A simple 1-page questionnaire was developed and distributed to all 50 state EMS directors in the United States. The survey requested information about each state's regulations about physicians riding as a primary EMS crew member, whether such physicians are allowed to provide physician-level care when functioning in that capacity, and the training and testing required of such physicians. The survey asked about basic life support (BLS) and advanced life support (ALS) ambulances. The state directors (or their delegate) were requested to complete the survey and return it by mail, e-mail, or fax to a central data receiving point. Nonrespondents were recontacted twice. The primary investigator also contacted respondents to clarify vague or missing responses. Data are reported by using simple descriptive statistics, including proportions and 95% confidence intervals (CIs).

Results: Forty-four (88%) of the states responded. In 32 of the states (73%; 95% CI 63% to 89%), physicians can work as a primary member of either a BLS or ALS ambulance crew without any specific training. In 30 of the states (68%; 95% CI 54% to 82%), physicians can work as a primary member of either a BLS or ALS ambulance crew without any specific out-of-hospital certification. All 44 responding states (100%) will allow a physician to become certified as an emergency medical technician (EMT) or paramedic and then work as a primary member of a BLS or ALS ambulance crew. Thirty-four states (79%; 95% CI 67% to 91%) allow the provision of physician-level care on BLS ambulances, and 35 (81%; 95% CI 70% to 93%) allow the provision of physician-level care on ALS ambulances. There was no meaningful difference between the training requirements for becoming a certified BLS versus an ALS provider: full course 54% versus 54%, abbreviated course 22% versus 20%, and no specific training 24% versus 27%. States were significantly less likely to require a skills examination of physicians wishing to become certified as a BLS provider (9%; 95% CI 1% to 18%) compared with those wishing to become certified as an ALS provider (82%; 95% CI 70% to 93%).

Conclusion: Although most states allow physicians to become certified out-ofhospital care providers, few states require physicians wishing to work as a primary EMS provider to do so or even to undergo any specific EMS training. Further, the requirements for physicians who opt to undertake EMS training and certification vary among the states, particularly in terms of the skills testing required for BLS and ALS certification. Better standardization of the preparatory requirements of physicians wishing to provide in-field EMS care might better enable physicians to practice safe and effective out-of-hospital care. Background: Out-of-hospital patients with shortness of breath (SOB) are often difficult to assess diagnostically. Treatment for these patients depends on the spe-survival rate (P<.001) compared with that of the other 2 groups. Similarly, the survival rate of the bystander-witnessed group was also significantly higher than that of the unwitnessed group (P=.010). Within the EMS-witnessed group, the ventricular fibrillation/ventricular tachycardia (VF/VT) patients had a significantly higher survival rate (30.6%) compared with those without VF/VT (4.1%; P<.001, odds ratio [OR] 10.3; 95% CI 2.9% to 36.9%). Similarly, for patient collapses witnessed by bystanders, the survival rate of VF/VT patients (4.5%) was significantly higher than the non-VF/VT patients (1.0%; P=.011; OR 4.4; 95% CI 1.3% to 15.4%). For patients with bystander-witnessed VF/VT arrest who received bystander cardiopulmonary resuscitation (CPR), the survival rate was 9.4% compared with 1.0% for patients whose collapse was not witnessed (P=.037; OR 4.4; 95% CI 1.01% to 20.1%). For the VF/VT group, significant survival rates were obtained at least up to 20 minutes from collapse to shock.

Conclusion: The out-of-hospital semiautomated external defibrillation program improved survival rate of EMS personnel-witnessed arrest. Our survival rate of bystander-witnessed VF/VT arrest is comparable to that of large cities in the United States. Increased bystander CPR rate and quality of bystander CPR may improve outcome of bystander-witnessed cardiac arrest. This study will serve as a baseline to assess the incremental benefits of implementation of public-access defibrillation and advanced life support by EMS in the future.

Lo B, Quinn S, Hostler D, Callaway C/University of Pittsburgh School of Medicine, Pittsburgh, PA Study objectives: Studies suggest that performing cardiopulmonary resuscitation (CPR) and administering drugs before rescue shocks (RSs) can improve resuscitation from ventricular fibrillation (VF). However, it is important not to delay RSs to patients for whom they are effective. This study tested whether clinical features of patients with out-of-hospital cardiac arrest (OOHCA) could predict success of the initial 3 shocks (RS1-3).

Methods: All resuscitation attempts for OOHCA by a municipal paramedic service were collected for 52 months (1998 to 2002) . Clinical data and outcomes of RS1-3 were tabulated. Successful RS resulted in return of organized electrical activity (ROEA) or spontaneous circulation (ROSC). Unsuccessful RS resulted in persistent VF or asystole. Relationships between clinical data and RS success were evaluated with a χ 2 test, and variables with significant univariate associations were entered into a logistic regression model. A t test was used to compare continuous data.

Results: Of 1,596 cases of OOHCA, 711 patients received RS. Success rates (ROEA or ROSC) were similar after RS1 (21%; n=597 with valid data), RS2 (17%; n=399), and RS3 (26%; n= 298), as was the rate of ROSC after RS1 (8.9%), RS2 (6.3%), and RS3 (8.1%). Successful outcome after RS1-3 occurred more often after witnessed collapse (43%) than unwitnessed collapse (21%; P<.001). ROSC followed RS1-3 more often after witnessed collapse (18%) than unwitnessed collapse (9%; P=.004). ROSC after RS1-3 occurred more often with bystander CPR (20%) than without (10%; P=.001). Odds ratio for bystander CPR was 1.38 (95% confidence interval [CI] 1.07 to 1.77) and for witnessed collapse was 1.45 (95% CI 1.07 to 1.97) as predictors of ROSC after RS1-3. Sex (63% men), age (65±16 years), and initial rhythm (61% VF) did not differ between subjects according to RS success or ROSC.

Conclusion: Success of RS1-3 and ROSC is more likely after witnessed collapse in patients who receive bystander CPR. These clinical characteristics should be considered when RS is delayed.

Krell JM, McCoy MS, Sparto PJ, Fisher GL, Stoy WA, Hostler DP/University of Pittsburgh, Pittsburgh, PA Study objectives: Spinal immobilization is essential in reducing risk of further spinal injuries in trauma patients. The traditional long backboard (LBB) is compared with the Ferno Scoop stretcher (FSS; model 65-EXL), which has a more rigid design than the original model. We hypothesize that there is no difference in the amount of movement during immobilization between the new FSS and the LBB.

Methods: Thirty-one volunteer subjects had electromagnetic angular position sensors secured over the forehead, C3, and T12. Subjects were placed in a rigid cervical collar and tested on the FSS and the LBB (test order randomly assigned). For each device, there were 4 measurement phases: baseline, application (log roll onto Study objectives: We describe the characteristics of individuals presenting to an urban ED with an acute cerebrovascular accident (CVA), specifically with regard to delay and mode of transport.

Methods: This was a retrospective review of a convenience sample of patients diagnosed with a CVA (International Classification of Diseases, between May 1995 and June 2001. Data collected included demographics, mode of transport, comorbidities, time of symptom onset, risk factors, hospital course, and final disposition. For inclusion, patients had to be presenting to the emergency department (ED) for the first time with signs and symptoms consistent with an acute CVA. Individuals who are transferred from other facilities, direct admissions to the hospital, and individuals who experienced their onset of symptoms after admission were excluded. Delay to presentation (defined as presentation longer than 3 hours after symptom onset) was capped at 7 days to avoid excessive skewing of the data.

Results: Three hundred forty-four medical records were reviewed, and 210 met inclusion criteria, but 5 patients had delayed longer than 7 days and were therefore removed from further analysis. For the 205 remaining patients, mean age was 60.5 years (range 20.3 to 91.6 years). Time from symptom onset to presentation to the ED averaged 37.65 hours for Latino patients, 39.5 hours for black patients, 40.3 hours for Asian patients, and 36 hours for white patients. When the definition of delay as more than 3 hours after symptom onset was used, delay was protective by using death as an endpoint (odds ratio 0.688; 95% confidence interval 0.487 to 0.973). There was a highly significant relationship (P=.002) in the distribution of type of transportation and ethnicity, with only 33% of Latino patients accessing the ED by emergency medical service (EMS) compared with 67% of black patients and 56% of white patients. Modeling the data against final disposition, age, tobacco use, transient ischemic attack history, sex, and delay in presentation did not affect the final disposition of the patient. Only mode of transport affected the final disposition, with ambulance transport associated with increased morbidity and mortality (P<.001).

Conclusion: Latino and Asian patients arrive at the ED after experiencing a CVA primarily by car, whereas black and white patients use EMS. Patients who presented early or by EMS had a higher mortality rate than those who presented later or by automobile. Background: We recently found that many patients who presented to the emergency department (ED) with "migraine" headache did not meet International Headache Society Criteria (IHSC) for the diagnosis of acute migraine.

Study objectives: We determine whether ED patients with migraine more frequently meet Canadian Headache Society Criteria (CHSC) than IHSC.

Methods: This was a prospective, observational study in a community ED. Consecutive patients who presented to the study authors with chief complaint of headache enrolled. Historical and clinical data were collected on a standardized form. Statistical tests included 95% confidence intervals (CIs) and Fisher's exact test, as appropriate.

Results: One hundred eighty-nine patients enrolled, with a mean age of 38±13.7 years; 69% were women, 37% had previous ED visits for headaches, 44% had previous diagnostic imaging for headache, 35% had a positive migraine family history, and 43% had a previous diagnosis of migraine. Eighteen percent of all patients met the CHSC compared with 15% meeting the IHSC. Forty-one percent of patients had a discharge diagnosis of migraine. Of these patients, 28% (22 of 78) met IHSC and 33% (26 of 78) met CHSC (P=.30). For patients with a discharge diagnosis of migraine, 26% (17 of 64) of female patients and 36% (5 of 14) of male patients met IHSC (P=.34), whereas 33% of female patients (21 of 64) and 36% of male patients (5 of 14) met CHSC (P=.53). For patients with a previous diagnosis of migraine, 26% (21 of 81) met IHSC and 32% (26 of 81) met CHSC (P=.24). Of patients with a previous diagnosis of migraine or a discharge diagnosis of migraine, 25% met IHSC (24 of 95, 95% CI 16% to 34%) versus 31% (29 of 95, 95% CI 22% to 40%) for the CHSC (P=.26). Three patients without a discharge or previous "migraine" diagnosis met IHSC; 4 met CHSC.

Conclusion: In our population, only a minority of headache patients with prior diagnosis or ED diagnosis of migraine met CHSC. The utility of CHSC and IHSC to standardize ED patients for headache research may be limited. cific diagnosis. This study evaluates patient characteristics that may be useful in predicting congestive heart failure (CHF) in out-of-hospital patients with SOB. Methods: This is an institutional review board-approved retrospective study. The inclusion criteria were out-of-hospital patients with the chief complaint of SOB throughout a 1-year period. Data collected included vital signs, skin examination, mucous membrane examination, capillary refill, neck vein distention, peripheral edema, lung examination, cardiac rhythm, and past medical history. All hospital records were located, and the final clinical diagnosis was determined. Logistic stepwise regression was performed to identify independent predictors for patients with chronic obstructive pulmonary disease (COPD).

Results: Two hundred thirty-six patients were enrolled. Methods: The study was a 3-year multicenter retrospective medical record review in a tertiary care center (60,000 visits yearly) and a community hospital (40,000 visits yearly). Patients older than 3 years and presenting with fever, headache, neck pain, or photophobia and who had lumbar puncture were included. Exclusion criteria were altered mental status, HIV positive, individuals taking antibiotics, or negative (0 to 5 WBCs) lumbar puncture results. Final diagnosis was determined by a composite endpoint of final culture results and discharge or expiration diagnosis. Univariate and multivariate regression analyses were used, with appropriate tests for continuous and categorical data. Data are presented as percentage or means with 95% confidence intervals (CI Conclusion: It appears possible to rule out BM by using basic findings; however, emergency physicians continue to admit and treat a majority of VM cases. Establishing clinical guidelines might significantly decrease admission rates and antibiotic administration. VM admissions appear to be driven by total number of CSF WBCs. emergency department (ED) may be an ideal location for focused patient stroke education because waiting in less critically ill patients is ubiquitous and evitable. This population to be treated may be particularly receptive to receiving medical information. We propose using the television and videocassette recorder already set up in most waiting rooms for playing educational videotapes. The purpose of this study is to explore whether passive watching of a videotape is an effective method for stroke education.

Methods: The setting was an urban, inner-city teaching hospital. The design was a prospective, randomized, pretest-posttest design with 2 arms: patients watching a 12-minute educational video on stroke developed by the American Stroke Association (VID) and those only taking the pretest (CON). Patients enrolled were contacted 1 month later to determine knowledge retention. The test administered was previously validated on a cross-section of patients. All patients provided informed consent. Contingency analysis (χ 2 ) was performed on categorical data, and nonparametric tests (Wilcoxon, etc.) were calculated for continuous data.

Results: One hundred ninety-eight patients were enrolled into the study, 100 in the control group and 98 in the video group. Seventy-seven (39%) patients responded to follow-up at 1 month: 36 CON and 41 VID. There was no statistically significant difference between groups by age, race, or sex. There were no differences in pretest scores between groups, and the VID group demonstrated knowledge retention at 1 month longer than pretest (Table) .

Conclusion: An educational videotape may be a valuable and cost-effective tool for focused education for ED patients in the waiting room. Patients retained stroke knowledge up to 1 month after intervention. What remains to be seen is whether the information will be retained for a longer period and whether the increase in knowledge will translate into change in behavior. Study objectives: We simulate atlanto-occipital (AO) injury that occurred to 3 children involved in a frontal motor vehicle crash by using mathematical dynamic models (MADYMO) computer simulation. Cervical spine injury to young children is fairly uncommon, but when it does occur, the occipital-C2 region is most commonly injured. These injuries are ligamentous or other soft tissue disruptions rather than fractures. These injuries, however, demand prompt diagnosis for optimal treatment. AO injury in children is associated with a disruption of the tectorial membrane and surrounding ligamentous structures. It has been shown that damage to this membrane is related to the classification of an AO injury as stable or unstable. Most AO injuries are fatal injuries, and the medical and automotive communities have attempted to reproduce this type of injury by using anthropomorphic test devices (ATD). The industry has attempted to assess the loading of the child's neck by reproducing the crash environment and evaluating the load characteristics of the neck of an ATD. These simulations are based on AO injuries that have exceeded the injury threshold enough to cause death. Unique to this case is the fact that all 3 children survived and had different degrees of AO injury.

Methods: This was an analysis of a frontal crash to a 1994 Ford Explorer in which Methods: Institutional review board approval was granted for this medical record review. This study was conducted at a large urban ED. All ED patients aged 16 to 65 years and administered PHE or FOS from January 2001 to June 2001 were identified by using a computerized system. No patients were excluded. Patients awake, alert, and oriented and able to walk and talk were considered to be candidates for oral medication. Confidence intervals were calculated with S-PLUS software.

Results: One hundred twenty-seven patients were included. All patients received either 1 g of PHE or 1 g of PHE equivalents of FOS intravenously or intramuscularly. None received PHE intravenously or intramuscularly or a combination of PHE and FOS. One hundred (79%) patients were candidates for oral medication. Only 30% (30 of 100) of these patients received PHE. Seventy percent (95% confidence interval [CI] 46 to 63) were administered FOS intravenously or intramuscularly when they were candidates for PHE. In a subgroup of patients with a Glasgow Coma Scale score of more than 13 (n=112), only 37% (42 of 112; 95% CI 28 to 47) were administered oral medication when indicated. If oral medication had been administered as indicated, $7,350 in medication cost alone could have been saved in 6 months.

Conclusion: This study demonstrates the overuse of FOS in an ED. Correct identification of patients who can receive oral PHE can result in significant savings to health care facilities and patients. Study objectives: Intracranial hemorrhage (ICH) accounts for approximately 10% of all strokes and carries a high mortality rate. Although hypertension (HTN) remains a frequent cause of ICH, blood pressure management in ICH remains highly controversial. The study objective was to determine the effects on mortality of antihypertensive treatment in primary ICH patients with severe, moderate, or nonelevated blood pressure.

Methods: A retrospective medical record review was performed on adult patients with primary ICH. Patients were stratified into 3 categories according to initial blood pressure: severe HTN (systolic blood pressure [SBP] 200 mm Hg or higher), moderate HTN (SBP between 140 and 200 mm Hg), and nonelevated SBP (SBP 140 mm Hg or lower). Comparisons were made between patients treated with antihypertensive medications and those who were not. Mortality was the outcome measure investigated.

Results: Of 66 patients studied, 68.2% had moderate HTN, 21.2% had severe HTN, and 10.6% had nonelevated blood pressure. The overall mortality was 30.3%. Patients with moderately elevated blood pressure were 76% less likely to die if initially treated with an antihypertensive (relative risk 0.24; 95% confidence interval [CI] 0.07 to 0.79; P=.0211) and 4.18 times more likely to die if treated with mannitol (relative risk 4.18; 95% CI 1.00 to 17.56; P=.0351). Dexamethasone had no significant effect on outcome.

Conclusion: In the largest group of patients with ICH, those with moderate HTN, there is significant decrease in mortality with antihypertensive treatment. The use of mannitol in the moderate HTN group was associated with a deleterious effect on patient outcome.

Chan YF, Lavery R, Pullicino P, Livinston D, Kwon R, Fox N, Massone R, Zinzuwadia S, Low R/University of Medicine and Dentistry of New Jersey, New Jersey Medical School, Newark, NJ Study objectives: Stroke is the third leading cause of death in the United States, all the more so in urban, medically underserved areas. Numerous studies have shown the need for community education programs for stroke because the delayed recognition of its often vague and varied symptoms results in delayed medical care. The Singer AJ, Gulla J, Thode HC/State University of New York, Stony Brook, NY Study objectives: The American Academy of Pediatrics (AAP) has published simple guidelines for the first aid treatment of ill or injured children. We determined knowledge of these first aid practices in adults.

Methods: The design was a descriptive survey. Participants were a convenience sample of adults presenting to an ED. Standardized collection of demographics was performed, and adults were given a multiple choice questionnaire about the proper first aid treatment of stings and bites, burns and scalds, eye injuries, fractures and sprains, fever, head injuries, fainting, poisoning, dental injuries, nosebleeds, seizures, and skin wounds. Outcome measured was the percentage of adults responding to questions appropriately based on AAP guidelines. Data analysis was done with descriptive statistics and χ 2 tests.

Results: Six hundred fifty-four adults were surveyed. Mean age was 38.5 years (SD 13.8), 56% were female patients, 69% were white patients, and 56% of patients had at least high school education. None of those surveyed knew all guidelines, with roughly half being familiar with 60% of the guidelines. Knowledge of specific guidelines ranged from 21% to 92%. Subjects especially lacked knowledge about the need to rapidly remove all bee stingers (only 36% aware), keep wounds moist and covered (79% felt that drying wounds was beneficial), cover victims of large burns (only 43% aware), and seek medical care after tick bites (only 47% aware). Knowledge was unaffected by age, sex, and education.

Conclusion: Many adults are unfamiliar with AAP first aid guidelines, and most believe that wounds should be allowed to dry. Further education is required to improve knowledge of proper first aid practices. Results: By using the 3-and 6-year-old ATD, the MADYMO simulation shows the IARV to have been exceeded in tensile loading. The model also shows the occupant kinematics of these restrained rear-seated passengers. The degree of injury to the AO area is shown to differ in severity between the 3-and 7-year-old victims clinically and through evaluation of the computed tomographic scan images.

Conclusion: There is no Federal Motor Vehicle Safety Standard criterion to address children's neck injuries in a dynamic environment such as a motor vehicle crash. The current neck injury criteria are based on scaling of adults and the testing of pigs. The federal requirements also address only the situation of airbag deployment on children's neck injuries. We have a case in which the environment can be recreated with tools accepted by the crash injury community that show the neck IARVs to have been exceeded without fatal injury. We believe that, according to the children's injuries to the AO region, we have seen the tolerance limit for an AO injury and the tectorial membrane. There are currently no data for the variation of tectorial membrane failure stress as a function of age for humans. It is rare that children of various ages survive an AO injury and are close enough in age and weight to be modeled by using MADYMO.

Junkins EP, Mallory MD, Guenther EA, Hutchings C, Rassmussen M, McDonough Q, Mann NC/University of Utah, Primary Children's Medical Center, Salt Lake City, UT Background: Children with serious injuries account for only a small percentage of trauma patients treated in hospital emergency departments (EDs).

Study objectives: We sought to determine whether the amount and type of clinical training in pediatric trauma treatment by state emergency physicians results in improved comfort and fewer referrals to a regional pediatric trauma center.

Methods: A self-administered survey was mailed to a comprehensive list of emergency physicians staffing hospital-based EDs in Utah. Respondents answered questions about clinical training, practice type, participation and certification in advanced trauma life support (ATLS)/pediatric advanced life support (PALS), amount of pediatric trauma exposure, and comfort level with these patients (aged <16 years), assessed with a Likert scale. Regression analysis was used to assess the relationship between training and experience and comfort with pediatric trauma.

Results: Of the 277 surveys mailed, 193 (70%) were returned. Eighty-five of the respondents were men, with a mean age of 43.7 years (range 31 to 68 years). The majority of physicians were located in an urban (51%) setting, followed by suburban (24%) and rural (25%) settings. Forty-five percent of the respondents were emergency medicine-trained physicians. Fifty-two percent of respondents reported having PALS certification, whereas 46% reported taking little to no trauma-specific continuing medical education. Exposure to PALS was significantly associated with whether the respondent was confident in his or her ability to care for severe pediatric trauma patients (odds ratio [OR] 0.59; 95% confidence interval [CI] 0.36 to 0.97). Location of practice was significantly associated with whether the respondent cares for most complicated pediatric trauma patients without referral (OR 0.67; 95% CI 0.49 to 0.93). Moreover, 31% considered their own experience and expertise to be a significant limitation in their ability to provide care.

Conclusion: Wide variation in clinical training, experience, and comfort exists in hospital emergency physicians about care for severely injured children.

pediatric procedural sedation or analgesia (PSA). This study examines the continued need for PSA in the era of fast helical computed tomography (CT) scanners.

Methods: The study was conducted at a university-affiliated community hospital. Children younger than 72 months and undergoing CT scans were identified from the hospital UB-92 billing database for the period November 1, 2001, through October 31, 2002. All identified medical records underwent a redundant review searching for either a physician order for PSA in association with a CT study or the presence of the procedural sedation monitoring form required for all patients undergoing PSA. Statistical analysis was through analysis of variance, χ 2 , and regression modeling.

Results: A total of 104 patients underwent 122 CT studies. Requests originated from 3 sources: emergency department (ED; N=73), neonatal ICU (N=47), and pediatric ward (N=2). Studies performed included head and facial (97), abdomen or pelvis (19), and other (6). The mean age of all study patients was 23.4 months (95% confidence interval [CI] 18.6 to 28.1), with a median of 14 months. The mean age of the ED patients was 35.4 months (95% CI 29.5 to 41.4), with a median of 34 months. Only 9 (8.6%) patients underwent any PSA, 8 from the ED, 1 from the pediatric ward. The mean age of the PSA patients was 18.8 months (95% CI 12.5 to 25.1), with a median of 21 months. With the exception of one 4-month-old patient, all other sedated children were between 12 and 30 months of age. PSA use occurred in 36.4% of patients in this age group, which was much more likely to require sedation than any other age group (odds ratio 56.5; P<.001). PSA agents used included diphenhydramine (4), pentobarbital (1), propofol (1), ketamine (1), midazolam (1), and diphenhydramine/pentobarbital (1). No adverse events were associated with PSA use.

Conclusion: PSA is rarely required for infants and young children undergoing helical CT studies. Children aged 12 to 36 months are most likely to require PSA agents.

Haltiwanger KA, Pines JM, Williamson AL, Ross PA, Yost MM, Alladin KP, Olamigoke OC, Martin ML/University of Virginia Health Sciences Center, Charlottesville, VA; University of Virginia College of Arts and Sciences, Charlottesville, VA Study objectives: Pediatric emergency department (ED) patients often present with complaints that would best be handled by a primary care provider (PCP) during clinic hours. Continuity of care is often compromised by unnecessary ED visits. We attempted to correlate the perceived degree of urgency of the visit with several variables, identify deficiencies in the primary care of our pediatric patients, and determine ways to improve health care service to this population.

Methods: A prospective survey was completed by parents of children (aged ≤17 years) and by patients (aged 18 to 21 years) presenting to a suburban academic pediatric ED that treats approximately 20,000 patients per year. A convenience sample of participants was enrolled between 8 AM and midnight 7 days a week. Patients or parents who did not want or were too sick to participate, pregnant women, prisoners, those with altered mental status, and ED transfers were excluded from the study. Data were analyzed by using Microsoft Excel. This study was found to be exempt from informed consent by the institutional review board.

Results: Three hundred five of 334 surveys (91% response rate) were filled out throughout a 3-month period. Twenty-four percent of the chief complaints were perceived by patients surveyed as emergency or possibly life-threatening, 23% were thought to be urgent, and 52% were deemed somewhat urgent or minor. Twentyfive percent of patients with somewhat urgent or minor complaints arrived by ambulance. Twenty-seven percent of the respondents were unaware of their PCP's office hours. Of the 17-to 21-year age group, 58% found the pediatric ED more convenient than the clinic. Weekend visits and minority race (black or Hispanic) correlated with a lower degree of urgency. A distance from the pediatric ED of more than 30 miles correlated with a higher degree of urgency. Overall, 79% of patients surveyed identified a primary care provider for themselves or their child. Of those patients, 46% had attempted to contact them before coming to the pediatric ED. Three percent of patients could not contact their PCP, and 22% were told to come to the pediatric ED. Twenty-four percent of all patients who responded said they would have preferred to see the appropriate PCP. More than 50% of patients who presented with fever, cold, and cough would have rather been treated by their PCP.

Conclusion: The pediatric ED often becomes overburdened with patients being treated for nonemergency problems that could safely be dealt with in a clinic. Access to primary care can be difficult for patients with limited resources, restrictive work schedules, or no designated PCP. Identifying and removing some of these barriers may encourage patients to seek appropriate care for nonemergency health issues. a prospective, randomized, double-blind clinical trial. Secretory diarrheal illnesses are associated with increased production of prostaglandin E and prostaglandin F. Prostaglandins and cyclic adenosine monophosphate (cAMP) are known to stimulate intestinal secretions. Enteric pathogens are also known to stimulate intestinal secretions via the same mechanism. Nonsteroidal antiinflammatory drugs such as aspirin and ibuprofen decrease cAMP and prostaglandin production. We therefore hypothesize that the use of ibuprofen in febrile diarrhea should reduce the volume of fluid loss when compared to acetaminophen, which does not reduce the intestinal secretion of either cAMP or prostaglandin.

Methods: All patients who presented to our pediatric emergency department (ED) with febrile gastroenteritis and were between the ages of 4 months and 6 years were evaluated by clinical examination and urine specific gravity for dehydration, and only parents of patients with mild dehydration were asked for informed consent. Patients who were either receiving antibiotics or had a known nonsteroidal antiinflammatory drug allergy were excluded from the study. All enrollees were randomly divided into 2 groups, A and B, and treated for 24 hours. They therefore received a double-blinded suspension of either acetaminophen or ibuprofen during this period. The patients were recalled after 24 hours to the ED, where they were weighed and examined for a second time

Results: A total of 66 patients were enrolled in the study from January 2002 to January 2003. Both drug categories had 33 (50%) enrollees each. Eighteen (27%) patients were lost to follow-up, 10 (55%) from group A and 8 (45%) from group B. The average weight difference between the first and second weight measurements were -0.10 kg (0.50 to 1.5 kg) for group A and -0.06 kg (0.80 to -0.80 kg) for group B. The 2-sample t test on the data sets for groups A and B resulted in a P value equal to .825.

Conclusion: This prospective, randomized, double-blind, clinical trial did not show any significant difference between the use of acetaminophen and ibuprofen on the amount of fluid loss in febrile diarrhea as hypothesized.

Ryan JG, Barata I, Ward MF, Hormozdi S, Shenkman A, Sama AE, Roit Z/New York Hospital Queens, Flushing, NY; North Shore University Hospital, Manhasset, NY Study objectives: The purpose of this study is to determine whether the WBC count is predictive of a positive urine culture in children aged 90 days to 36 months who present to the emergency department (ED) with a temperature of 103°F (39.4°C) or higher.

Methods: This prospective study was conducted at an academic ED that treats approximately 16,000 pediatric patients per year. Consecutive enrollment of patients was performed by the principal investigator when the principal investigator was present. Patients were eligible for enrollment if they were aged 3 to 36 months, had a rectal temperature of 103°F (39.4°C ) or higher, were immunocompetent, did not receive antibiotics before evaluation, and did not have otitis media or pneumonia on physical examination. Eligible patients were evaluated by physical examination, temperature, CBC count, blood culture, catheter-obtained urine analysis and culture, and chest radiography. Sensitivity, specificity, predictive values, and likelihood ratios (LRs) were calculated for WBC count, with a positive result at 10,000, 15,000, 20,000, and 25,000 cells/mm 3 .

Results: Throughout a 17-month period, 125 patients were enrolled in this study. Of these patients, 8 (6.4%) had positive urine culture results. When the patients were divided according to their WBC count, 82 patients had a WBC count more than 10,000 (sensitivity 88%, specificity 36%, LR 1.37); 38 had a WBC count more than 15,000 (sensitivity 38%, specificity 70%, LR 1.25); 15 had a WBC count more than 20,000 (sensitivity 0%, specificity 87%, LR NA); and 10 had a WBC count more than 25,000 (sensitivity 0%, specificity 91%, LR NA). Of the 8 patients with a positive urine culture, 5 (63%) had a WBC count less than 15,000, and all had a WBC less than 20,000.

Conclusion: The WBC count is not a sensitive or specific indicator of urinary tract infection in febrile children. In this study, most of the patients with positive urine culture results had a WBC count less than 15,000, and all patients with a positive culture had a WBC count less than 20,000. Study objectives: Family presence during resuscitation of adult patients infrequently occurs in the United Kingdom. In contrast, parental presence during pediatric resuscitation is often encouraged, depending on circumstances. Research into the reasons behind this practice has been inadequate. The research that has been conducted suggests that physicians are uncomfortable with family presence in cases of adult resuscitation. This study sought to identify the factors influencing physician opinion on this matter.

Methods: An anonymous questionnaire was sent to physicians in 2 district general hospitals in the county of Cheshire in the United Kingdom. All acute medical and surgical specialties were included, in addition to anesthesia and emergency medicine. All grades of physicians were asked whether they approved of family presence during adult resuscitation. Data collected included grade and specialty of physician, as well as the number of advanced life support courses attended. Factors that may influence opinion were inquired about, including fears of family members witnessing errors, increased litigation, and more prolonged resuscitation. Statistical analysis was carried out with NCSS (Number Crunching Statistical Systems, 2001 edition). Proportions were compared by using the χ 2 test and multivariate analysis by forward stepwise logistic regression.

Results: Responses were received from 147 of 256 physicians (response rate 57.4%). By specialty, responses were received from 40 physicians in emergency medicine, 30 in anesthesia, 38 in the acute medical specialties, and 39 in the acute surgical specialties. Physicians in emergency medicine were significantly more likely to approve of witnessed resuscitation compared with physicians in the other specialties (P<.001). There was a trend toward approval among the higher grades of physicians, but it was not significant. Attendance at life support courses was associated with approval (P<.001), which increased with the number of courses attended (P<.001). Fears of more prolonged resuscitation with family presence and errors noticed by family members did not influence approval. However, fear of litigation was associated with disapproval (P<.001), particularly among physicians not in emergency medicine (P<.05). Physicians of lower grade had greater concerns about errors being noticed and litigation (P<.001 and P<.05, respectively). By forward stepwise logistic regression analysis, specialty, grade, fear of litigation, and the number of life support courses attended were found to significantly influence approval of family presence during resuscitation (R 2 =0.26, P<.001). Other concerns expressed by physicians included emotional trauma to relatives, distraction of resuscitation team, lack of space in the resuscitation room, and lack of staff to look after relatives (47, 30, 13, and 9 responses, respectively).

Conclusion: Approval of family presence during resuscitation of adults depends on how comfortable physicians are in these situations. In the United Kingdom, physicians in emergency medicine deal with the majority of out-of-hospital medical and trauma resuscitations and therefore appear to be more comfortable with family presence. Advanced life support courses give physicians a structured approach to resuscitation, and thus they are less likely to disapprove of family presence. Although no evidence exists to suggest that family presence would result in greater litigation, it is an important factor influencing opinion.

Takeyama Y, Mori K, Nara S, Itoh Y, Hase M, Imaizumi H, Kaneko M, Asai Y/Sapporo Medical University, Sapporo, Japan Study objectives: Cardiac arrest patients who do not respond to advanced cardiac life support (ACLS) have a poor neurologic outcome. However, some patients obtain good neurologic recovery with a cardiac support device. We sought to determine the usefulness of brain hypothermic therapy (BHT) after cardiopulmonary bypass (CPB) in cardiac arrest patients.

Methods: We performed a retrospective medical record review of cardiac arrest patients with CPB who could not respond to ACLS between 1999 and 2003 in the emergency department. We have performed BHT (34°C, 2 to 3 days) after CPB in patients who conformed to the following inclusion criteria: their cardiac arrest was witnessed, they failed to respond to ACLS, a successful intervention for original cause of cardiac arrest was performed, and circulation was stabilized after intervention. We evaluated their ECG on admission (ventricular fibrillation [VF] or non-

A Background: Informed consent requires a physician to provide adequate information on the nature and purpose of the proposed treatment or intervention, anticipated risks and benefits, possible alternatives, probability of success, and implications for not consenting. Information should be comprehensible so that a competent individual could make an informed voluntary and autonomous decision. The informed consent process may be suboptimal in a minority population because of educational, language, and cultural factors.

Study objectives: We study the informed consent process in children undergoing spinal taps in a predominantly minority population. The study was conducted in a pediatric emergency department (census 18,000 annual visits) of a 600-bed innercity community hospital.

Methods: A convenience sample of parents or guardians of children needing spinal taps were the study participants. All individuals who agreed to participate were interviewed within 48 hours of the procedure by using a standard pretested questionnaire by 1 of the study authors (ND) who was not involved with the consent process. Basic demographic data, knowledge of risks and benefits, and knowledge of alternatives to spinal tap were collected.

Results: Parents or guardians of 55 children were interviewed between July 2002 and March 2003. The majority of the parents (51%) were black; the rest were Hispanic (29%), white (7%), Asian (4%), and mixed (5%). The mothers consented for spinal tap for 78% of children. The majority of mothers had high school education (51%), 20% did not complete high school, and 29% had college education. Most parents (n=49, 89%) responded that they had clearly understood the informed consent process, 7% indicated that they did not understand, and 4% were not sure. Of the parents who understood, 63% did not recall discussions about risks of the procedure, 46% did not know about right to refuse, and 77% did not recall being told of possible alternatives.

Conclusion: We have used a common invasive procedure to study the informed consent process in our patient population. Although the majority of our participants recalled the necessity or advantages of having spinal taps, not many recalled any discussion of the anticipated and unanticipated risks, right to refuse, or any alternatives to the procedure. There is a need to improve our informed consent process.

Ong MEH, Chan YH, Anantharaman V, Lau ST, Lim SH, Seldrup J/Singapore General Hospital, Singapore; Ministry of Health, Singapore Study objectives: We describe the epidemiology of out-of-hospital cardiac arrest (OHCA) in Singapore and the emergency medical services (EMS) response and identify possible areas for improvement.

Methods: This prospective observational study constitutes phase I of the Cardiac Arrest and Resuscitation Epidemiology (CARE) project. Included were all patients who had nontraumatic OHCA and were conveyed by the national emergency ambulance service. Patient characteristics, cardiac arrest circumstances, EMS response, and outcomes were recorded according to the Utstein style.

Results: From October 1, 2001, to April 30, 2002, 544 patients were enrolled in the study. Mean age was 62.3 (SD 17.7) years, with male predominance (65.5%). Of the 60.1% of collapses that occurred at home, 35.4% of arrests were not witnessed, and bystander cardiopulmonary resuscitation was initiated for 20.4%. Mean time from collapse to call received by EMS was 10.5 (SD 12.9) minutes. Mean EMS response time was 10.5 (SD 4.4) minutes. Mean time from call to defibrillation was 16.7 (SD 17.2) minutes. Mean on-scene time was 9.8 (SD 4.5) minutes. First presenting rhythm was asystole in 54.5%, pulseless electrical activity 23.2%, ventricular fibrillation 19.5%, and ventricular tachycardia 0.4%. Of all cardiac arrests, 359 had resuscitation attempted and were of cardiac origin, 18.1% had return of spontaneous circulation, 8.4% survived to hospital admission, and 2.2% survived to discharge.

Conclusion: CARE I establishes the baseline for the evaluation of incremental introduction of out-of-hospital advanced cardiac life support interventions planned for future phases. Continuing efforts should be made to strengthen the chain of survival. To the authors' knowledge, this represents the most comprehensive OHCA study yet conducted in Singapore. with cervical collars in place and in 17 of 18 (94%; 95% CI 73% to 99%) patients with CPR in progress. Conclusion: According to our data gathered from a retrospective medical record review, the supraclavicular pocket approach to subclavian vein cannulation may be a useful and safe method of central venous access.

Mori K, Takeyama Y, Nara S, Itou Y, Hase M, Imaizumi H, Kanieko M, Asai Y/Sapporo Medical University, Sapporo, Japan Study objectives: Brain hypothermic therapy (BHT) improves neurologic recovery after resuscitation from cardiac arrest, although it has not yet been determined which patients are most likely to benefit from this therapy. In 2000, we presented indications for BHT in cardiac arrest patients by a multivariate analysis of prognostic factors. In the present investigation, we sought to verify the indications for BHT in cardiac arrest patients used in the prospective study.

Method: In the 2000 study, we determined the indications for BHT in cardiac arrest patients as arrest time of less than 20 minutes and Glasgow Coma Scale score greater than 5. Patients were rapidly cooled by an ice-mounted blanket decreased to 34°C (93.2°F) within 2 hours, BHT (34°C [93.2°F]) was continued for 3 days after resuscitation, and the patient was then rewarmed 1°C per day to 36°C (96.8°F). We evaluated the patients in the emergency department who conformed to the aforementioned criteria, in terms of neurologic outcome (ie, dead, vegetative state, severe disabilities, moderate disabilities, good recovery), as measured by 1-month Glasgow Outcome Scale.

Results: BHT was performed in the 15 patients (age 56.4±13.7 years; 10 men, 5 women) after resuscitation from cardiac arrest. The average duration of time onset of cardiac arrest to return of heartbeat was 29.3±12.6 minutes. Eleven patients obtained good neurologic outcome (moderate disabilities and good recovery; 73.3%), no patient died during BHT, and 1 patient died on the 12th hospital day. This outcome rate (73.3%) was statistically the same as the previous rate (70.6%) in the prospective study between 1996 and 2000.

Conclusion: The results of this study confirmed that the hypothesized indications for BHT in cardiac arrest patients that were used in the previous prospective study (1996 to 2000) were adequate. Study objectives: For the critically ill emergency department (ED) patient, research shows that intensive goal-directed global resuscitation improves mortality. Less research has focused on optimizing cerebral resuscitation, specifically cerebral oxygen delivery and consumption. Invasive methods of evaluating cerebral oxygenation, such as jugular bulb venous saturation, predict neurologic outcome. Noninvasive real-time regional cerebral oximetry has compared favorably with these invasive methods. Cerebral oximetry in the ED is predictive of the adequacy of resuscitation and neurologic outcome.

Methods: A prospective observational study was performed on a convenience sample of adult critically ill ED patients with altered mental status. The real-time regional cerebral oximetry by the In-Vivo Optical Spectroscopy (INVOS) cerebral oximeter (Somanetics Corporation) was compared with the adequacy of resuscitation and neurologic status at hospital discharge. Outcome measures for adequacy of resuscitation were defined by base deficit, lactate, pH, HCO 3 , and vital signs. Neurologic outcome was defined by the Glasgow Outcome Scale. Data were analyzed with descriptive statistics.

Results: Thirty-six patients were enrolled. The predictability of regional cerebral oximetry for adequacy of resuscitation showed a sensitivity of 75%, negative predictive value of 78%, positive predictive value of 67%, and specificity of 70%. The predictability of regional cerebral oximetry for poor neurologic outcome was sensitivity of 60%, negative predictive value of 75%, positive predictive value of 60%, and specificity of 75%. However, false positives were noted in patients with computed tomography (CT) scans displaying acute signs of cerebrovascular accident or intracranial bleeding. If patients with positive CT findings are excluded, the predictability of regional cerebral oximetry for neurologic outcome is sensitivity of 80%, negative predictive value of 82%, positive predictive value of 67%, and specificity of 82%. VF) and neurologic outcome (dead, vegetative state, severe disabilities, moderate disabilities, good recovery) as measured by 1-month Glasgow Outcome Scale.

Results: Resuscitation with CPB was attempted in 30 patients (mean age 52 years; 26 men, 4 women) and successfully achieved in 17 (57%) patients. Eight (53%) of 15 patients who received BHT after CPB obtained a good neurologic outcome (moderate disabilities and good recovery), and 2 (13%) died during BHT. Nine (38%) of 24 patients who were VF on admission obtained a good neurologic outcome (moderate disabilities and good recovery). Six patients were non-VF, and none of them had a good neurologic outcome.

Conclusion: We concluded that BHT after CPB is useful for cardiac arrest patients with VF who could not respond to ACLS.

Vozenilek J, Kharasch M, Flaherty J/Evanston Northwestern Healthcare, Evanston, IL Study objectives: High-fidelity simulation is a powerful, experiential, educational tool that allows the accurate and repetitive reenactment of a variety of clinical scenarios. It is our objective to use this tool in the education of emergency medicine residents in the care of the critically ill. We have devised an educational model that allows residents to review and experience the presentation of critically ill patients as they actually present to our emergency department (ED) and our "code teams."

Methods: At our institution, emergency resuscitation equipment and medications are stored in so-called crash carts. Use of a crash cart either in the ED or on the inpatient wards stimulates a standard institutional response: after the intervention, the equipment is replaced, medications are restocked, and the crash cart is restored to its former position. Now, in addition to the standard institutional response described, our simulator center staff is also notified that a crash cart was used. In this way, data on all resuscitations that occur at our 3 community-based hospitals can be collected and reviewed.

These data are used to create scenarios that attempt to reproduce the conditions directly before, during, and after the critical interventions. These scenarios are programmed into high-fidelity simulators that can simulate human physiology and pharmacokinetics. These scenarios are then brought to life for the residency in general in our simulation center.

Our simulation center is equipped like a small emergency ward containing up to 3 patients simultaneously. It is this level of complexity that we believe creates a more exact reproduction of the conditions the intervening physicians experienced in the peri-intervention period. Data are collected for analysis on physician interventions, timing of interventions, and confounders to interventions. The residency class reviews these data live and by video. The entire residency may, if desired, run through the same simulation for educational purposes.

Conclusion: We believe that, in this manner, the relatively few truly critical cases may be shared experientially by the residency class as a whole. We believe this to be a powerful novel educational model.

Pentheroudakis E, Everett W, Singh R, Gorchynski J/University of California-Irvine Medical Center, Orange, CA; University of Pennsylvania, Philadelphia, PA Study objectives: We examine the feasibility of central venous catheter placement by using the supraclavicular landmarks for subclavian vein access.

Methods: In the first portion of this study, central venous catheter cannulation was performed in 28 cadavers by using a specified set of landmarks through the technique that we call the pocket approach. The second portion of the study was a retrospective medical record review of 75 emergency department patients who had subclavian vein catheter placement attempted by using supraclavicular landmarks by the pocket approach. The success of subclavian line placement, the incidence of pneumothorax, and the use of supraclavicular subclavian access in the setting of trauma or cardiopulmonary resuscitation (CPR) or with the presence of a cervical collar (C-collar) were assessed.

Results: The success rate for placement in 28 cadavers was 100% (34 of 34; 95% confidence interval [CI] 90% to 100%) by using the pocket approach landmarks. There were no complications. Medical record review of 75 patients revealed a success rate for supraclavicular central line placement in this study of 88% (66 of 75; 95% CI 78% to 94%). No pneumothorax occurred (0 of 75; 95% CI 0% to 5%). This technique was used successfully in 12 of 12 (100%; 95% CI 74% to 100%) patients bulge, 3 with soft tissue swelling, 3 with edema, 2 with increased intraspinous signal, and 3 with multiple abnormalities. Eighteen patients were discharged with a cervical orthotic and nursing follow-up (P<.0001). There is still a difference, assuming that only half the patients with normal cervical spine radiograph results would have received orthosis ( Conclusion: Cervical spine MRI of blunt trauma patients with normal radiographs and suspicion for ligamentous injury changes ED treatment. A significant majority of these patients can be discharged from the ED without orthosis and neurosurgery follow-up. Also, MRI may reveal otherwise unsuspected injuries that require hospital admission. Although patients with neurologic symptoms and unrestrained motor vehicle crash patients tended to have a greater likelihood of an abnormal cervical spine MRI result, our study lacks the power to detect a difference. Future studies may investigate the utility of this diagnostic modality prospectively and evaluate which subsets of patients would benefit the most.

Allen TL, Cummins BF, Bonk RT, Harker CP, Stevens MH/LDS Hospital, Salt Lake City, UT Background: Computed tomography (CT) with oral contrast material has traditionally been advocated for the detection of bowel and diaphragmatic injuries in blunt abdominal trauma. The necessity of oral contrast for these injuries has recently been called into question.

Study objectives: We determine the sensitivity and specificity of CT without oral contrast for diaphragm injuries in blunt abdominal trauma.

Methods: We prospectively enrolled 500 consecutive "trauma-1" patients who received CT imaging and interpretation (CT-Read1) of the abdomen within 45 minutes of their arrival from July 2000 to December 2001. All patients were imaged without oral contrast but with intravenous contrast. CT images were reviewed within 24 hours of admission by research radiologists (CT-Read2) blinded to CT-Read1. True injuries were determined hierarchically by laparotomy, autopsy, discharge diagnoses, CT-Read2, and 3-month telephone follow-up.

Results: There were 9 patients with laparotomy-or autopsy-proven blunt diaphragm injury. Six were female patients, and 8 involved the left hemidiaphragm. For patients with diaphragm injury, the average age was 47.4 years (compared with 34.8 years for those without). The average presenting Glasgow Coma Scale score was 10.7 and 11.9 for patients with and without diaphragm injury, whereas the Injury Severity Score for those with and without diaphragm injury was 34.4 and 18.5, respectively. CT-Read1 correctly detected only 6 of 9 blunt diaphragmatic injuries. CT-Read1 missed 3 diaphragm injuries. One of these involved the right hemidiaphragm, whereas the other 2 were left sided. There were no false-positive findings with CT-Read1 for blunt diaphragm injury. The sensitivity and specificity of CT imaging with respect to blunt bowel and mesenteric injuries was 66.7% and 100%, respectively.

Conclusion: The low incidence of blunt diaphragm injuries in this study limits it applicability, but CT without oral contrast appears to have an unsatisfactorily low sensitivity to be used reliably for this injury.

Gerhardt RT, Kernan L, Nelson BK, Lane M, Keenan S, MacKersie A/Brooke Army Medical Center, San Antonio Uniformed Services Health Education Consortium, San Antonio, TX; Texas Tech University Health Sciences Center, San Antonio, TX Study objectives: We compare decision rules with imaging for accuracy of the prediction of need for acute intervention in patients with abdominal pain. The study was conducted in an urban military emergency department (ED) with 62,000 yearly visits.

Methods: An institutional review board approved the prospective case-control study. A convenience sample of adult patients presenting to the ED with a chief complaint of abdominal pain had a standardized database of 82 clinical predictors (CPs) consisting of demographic, historical, physical examination, and laboratory variables collected. All patients had an acute abdominal radiograph series (AAS) and

A Conclusion: Noninvasive cerebral oximetry is predictive of adequacy of resuscitation and neurologic outcome.

Bittner DQ, Director TD, Nanavati K, Mehta S, Rathlev NK/Boston University Medical Center, Boston, MA Study objectives: We determine the incidence and types of complications requiring intervention after tube thoracostomy (TT) performed by emergency medicine residents.

Methods: Between December 1, 2002, and February 1, 2003, we performed a prospective observational study of patients aged at least 15 years and undergoing primary TT in an academic, urban, Level I trauma center. The indications for the procedure and complications were documented on a TT data collection sheet by the resident supervising the procedure. Immediate and delayed complications were documented with use of a data abstract form through review of the emergency department (ED) record, TT data collection sheet, hospital discharge summary, and official radiograph interpretation. Complications that required intervention were documented.

Results: Twenty-seven patients aged at least 15 years had TT performed in the ED during the study period and were eligible for analysis. The mean age of patients was 36.7 years, and 74% were men. The indications for TT were pneumothorax (56%), tension pneumothorax (7%), hemothorax (22%), and suspected or documented hemopneumothorax (15%) . The mechanisms of injury were spontaneous pneumothorax (18%), fall (4%), gunshot wound (15%), motor vehicle crash (30%), and stab wound (33%). Minor complications including repositioning or replacement of the TT were required in 6 (22%; 95% confidence interval [CI] 6.5% to 37.9%) cases because of tube kinked (7%), tube advanced too far (11%), and tube too far out (4%). In 1 other case, an additional TT had to be performed on the same side because of an unresolved pneumothorax. No adverse outcomes were noted in these 7 cases. Major complications including empyema and iatrogenic injuries requiring operative intervention occurred in 1 (3.7%; 95% CI 0% to 10.8%) patient. This individual sustained an intercostal artery laceration that required operative repair after emergency TT for a stab wound to the chest. Long-term complications did not develop after repair of this injury.

Conclusion: Major complications of TT are rare when the procedure is performed by emergency medicine residents and were documented in only 1 (3.7%) patient in this series. Minor complications requiring repositioning or replacement of a TT occurred in 6 (22%) patients.

Sattari M, Nufer KE/University of New Mexico, Albuquerque, NM Study objectives: Evaluation of cervical spine injury is a significant and common clinical challenge. The role, if any, of magnetic resonance imaging (MRI) to evaluate emergency department (ED) patients with suspected ligamentous injury despite normal cervical spine radiograph results has yet to be defined. Our study question is, "Does cervical MRI change treatment of ED patients with normal cervical spine radiograph results and clinical suspicion for ligamentous injury?"

Methods: This was a retrospective medical record review in an urban Level I trauma center. Inclusion criteria were blunt trauma victims with normal cervical spine radiograph results who received cervical spine MRI. Data were analyzed with χ 2 . Our primary outcome measure was number of patients discharged without cervical spine orthosis. We compared this measure to the assumption that all 66 patients would have received an orthotic and neurosurgery follow-up. Given that the availability of cervical spine MRI may have led some clinicians to obtain cervical spine MRI in patients whom they may not have discharged with a cervical orthotic, we reexamined the data, assuming that only half the patients with normal cervical spine MRI results would be discharged with an orthotic.

Results: Sixty-six patients met the inclusion criteria. Median age was 31 years (range 11 to 77 years). Forty-seven (71.2%) patients had normal cervical spine MRI results, 18 (27.3%) had abnormal MRI results, and 1 MRI result was missing. Eighteen patients received cervical spine orthosis and neurosurgery follow-up; 1 was admitted to neurosurgery for an MRI finding of disk bulge with spinal cord edema; none received surgery in 24 hours for cervical spine injuries. Of the 18 patients with abnormal MRI results, there were 4 patients with ligamentous injury, 3 with disk and female patients 14%). The negative operative rate for 110 patients evaluated with CECT was 5.5% (male patients 1.5% and female patients 16.6%). Conclusion: Use of CECT improves the accuracy of diagnosis of AAP compared with clinical assessment alone (NOPR 5.5% versus 11%). When male patients were compared with female patients, we found no difference in the NOPR for female patients who underwent CECT compared with female patients who underwent surgery according to clinical evaluation alone (NOPR 14% versus 16.6%). CECT performed on male patients considered to have equivocal examinations by a surgical consultant decreased the NOPR from 10% to 1.5%. Our results support the added value of CECT scanning in male patients with equivocal examinations and do not support the added value of CT scanning in female patients with equivocal examinations. Results: Fifty percent (2,725) of the patients in the registry were diagnosed with DVT as outpatients, including in the ED. Mean duration of symptoms before diagnosis was 7.4±19.9 days. Recurrent DVT was reported in more than 50% of patients, and a family history of venous thromboembolism was reported in more than two thirds of patients. Ninety-eight percent of patients experienced symptoms before ultrasonographic confirmation. Most commonly reported symptoms included edema (82%), extremity discomfort (70%), erythema (17%), dyspnea (13%), and difficulty in walking (13%). The most frequent comorbidities were family history of venous thromboembolism (69%), hormone replacement therapy (65%), minor trauma (58%), previous DVT (57%), previous pulmonary embolism (56%), and current smoker (55%). Within 30 days before DVT diagnosis, 21% of patients received some prophylaxis and only 9% received treatment. The most common treatment in these patients was "bridge" therapy to warfarin by low-molecular-weight heparin (46%) or intravenous unfractionated heparin (39%).

Conclusion: Almost all patients diagnosed with DVT as outpatients experienced recognized high-risk clinical signs or symptoms of the disease. For appropriate treatment of these at-risk patients, it is essential to recognize them as candidates for the initiation of DVT prophylaxis on arrival to the ED. Initiation of DVT prophylaxis in at-risk patients being admitted for inpatient care may also be in the purview of the emergency physician.

Paluska T, Sise M, Egan M, Biondi M/Naval Medical Center San Diego, San Diego, CA; Scripps Mercy Hospital, San Diego, CA Background: Helical computed tomography (CT) has revolutionized the evaluation of trauma patients because of its speed and high sensitivity and specificity for identifying injury. Helical CT is now widely available and often used in the evaluation of the head, face, neck, chest, abdomen, pelvis, and spine. As a result, many unrelated or "incidental" findings are detected. The spectrum of incidental findings is broad and can pose extra, unexpected disposition issues for the trauma team.

Study objectives: We identify the incidental findings in selected Level I trauma population and stratify into follow-up priority.

Methods: A retrospective medical record review was conducted of 1,015 consecutive admissions to an urban Level 1 trauma center from January 2002 to November 2002. Twenty-one records were either unavailable or incomplete. The percentage of patients receiving CT scans was recorded. All CT studies of the head, neck, chest, abdomen, pelvis, and spine were reviewed for incidental findings. The data were then further stratified by age, sex, and study. Mucus retention cysts, sinusitis (except mastoiditis), degenerative joint disease, evidence of previous surgery, and agerelated cerebral atrophy were excluded. All other incidental findings were reviewed and placed into 1 of 3 categories according to follow-up priority: (1) finding requires attention before discharge from trauma admission; (2) finding requires fol-an unenhanced computed tomographic scan of the abdomen and pelvis (UHCT). AAS and UHCT results were classified as suggesting need or no need for acute intervention (AASAI and UHCTAI). A criterion standard of need for acute intervention (NAI) was defined at 6-month follow-up by hospital admission for observation, operation, or endoscopic intervention within 24 hours of ED presentation. All variables were individually correlated with NAI. Those with significant correlations were entered as candidate variables into classification and regression tree models. Three models were generated. Model 1 used the candidate variables without imaging, model 2 used the candidate variables with AASAI, and model 3 used the candidate variables with UHCTAI.

Results: One hundred eighty-four patients were entered in the study and 19 were excluded, leaving 165 for analysis. Thirty-one CPs correlated with NAI. AASAI and CT scan predicting need for acute intervention (CATAI) were strongly correlated with NAI. Model 1 generated a tree with 10 nodes. Predictors were elevated amylase, alkaline phosphatase, prothrombin time, and WBC count more than 13,400 cells/mm 3 . Model 2 generated a tree with 6 nodes. Predictors were WBC count more than 13,400 cells/mm 3 and AASAI. Model 3 generate a tree with only 2 nodes. The only predictor was UHCTAI. Sensitivity, specificity, positive likelihood ratio, and negative likelihood ratio, respectively, were for AASAI 0.29, 0.93, 4 

Daniel EJ, Miller AH, Delaney KA/University of Texas Southwestern Medical Center, Dallas, TX Study objectives: We evaluate the added contribution of contrast enhanced computed tomography (CECT) of the abdomen to the clinical diagnosis of patients with suspected acute appendicitis (AAP) in male and female patients.

Methods: We conducted a retrospective study of patients with suspected AAP in an urban public teaching hospital emergency department (ED) that treats more than 140,000 patients a year. Because of chronically limited computed tomography (CT) scanning resources, patients with suspected appendicitis are first examined by a surgical consultant who orders the CT scan if the clinical evaluation seems equivocal. Patients with examinations considered to be unequivocal routinely undergo appendectomy without diagnostic CT scanning. Patients with suspected AAP were first examined by an emergency physician and subsequently by a surgical consultant. The study identified 449 patients admitted to a surgical service between December 1999 and January 2003 with a final ED diagnosis of suspected AAP. Patients were excluded if the medical records did not include a final diagnosis or surgical pathology report. In cases in which there was a discrepancy between the results of the pathology report and the surgeons' impression of the pathology, the pathology report was used as the final arbiter. The pathologic diagnosis of appendicitis was based on standard published criteria. Outcome measures included CECT sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy in the diagnosis of AAP, as well as the impact on the negative operative rate (NOPR). The finding of surgical disease not specifically caused by appendicitis was not counted as a negative operation. Further subgroup analysis was performed on the male and female patients.

Results: A group of 448 patients, 354 male patients (79%) and 95 female patients (21%) aged 14 to 69 years were admitted by the department of general surgery for suspected AAP throughout a 4-year period. One hundred ten (25%) of these patients were imaged with CECT. Nineteen percent (69) of the male patients and 43% (41) of the female patients were imaged with CECT. Of the 110 CECT scans performed, there were 82 true positives, 7 false positives, 10 true negatives, 5 false negatives, and 6 considered nondiagnostic, which resulted in an overall sensitivity of 94%, specificity of 59%, PPV of 92%, NPV of 67%, and accuracy of 88%. In the subgroup analysis of 69 male patients, the sensitivity of CECT was 95%, specificity was 75%, PPV was 98%, NPV was 50%, and overall accuracy was 94%. In the subgroup analysis of 41 female patients, the sensitivity of CECT was 92%, specificity was 54%, PPV was 79%, NPV was 78%, and overall accuracy was 78%. The negative operative rate for 338 patients who were not evaluated with CECT was 11% (male patients 10% personnel. Of EDs with ED security personnel, 93% had 24-hour coverage (7% of EDs were uncovered at times), all security personnel wore uniforms, 40% were armed, 66% were stationed exclusively in the ED, and 60% had the power to arrest and were called 2.5 times a day on average. Ninety-three percent of EDs have limited visitor access, 61% have locked doors, 23% have badge card entry, 21% have bulletproof glass, 21% have metal detectors, 64% have a direct phone line to security, 76% have a "panic button," 42% have a radio in the ED to communicate with security, and 86% have closed-circuit television monitors to view hidden areas. Seventy-one percent of EDs have a training program on handling violent incidents (offered to physicians 64%, nursing 64%, residents 57%, and medical students 21%), 85% have a written policy on handling violent incidents, 100% have a written policy on use of restraints, 86% have a psychiatric seclusion room, and 100% have used physical restraints. On a Likert scale with a range of 1 to 5 (1=safe to 5=unsafe), ED staff rated safety as 2.24 (mean).

Conclusion: EDs are at risk for violent threats and incidents. Compared with that of previous reports, the risk appears to have increased in recent years. Measures are being taken to ensure the safety and security of ED staff, patients, and visitors, especially in terms of the presence of security personnel in the ED and the ED physical design, which may account for many people feeling "somewhat" safe in their ED. However, improvements can be made, especially with regard to training. Background: PLS seek to eliminate MCO policies that hamper patients' timely access to emergency medical care. In 1996, New York enacted a PLS that defined emergency conditions and required that MCO enrollees be notified (1) that emergency services are not subject to previous authorization, (2) that reimbursement for such services is not to be denied on retrospective review, provided that such services were medically necessary to stabilize or treat an emergency condition; and (3) that procedures for obtaining emergency services be described, including a definition of emergency services. Each MCO was also required to provide informational materials to actual and potential enrollees on request. To our knowledge, no studies have assessed MCO compliance with PLS.

Methods: In November and December 2002, informational materials were requested from each MCO registered with the New York State Department of Insurance. Using an explicit list of criteria and questions, 2 of the principal investigators independently performed a text content analysis of each MCO member handbook chapter addressing coverage for emergency care. These investigators then reviewed the analyses for discrepancies. Consensus was achieved by joint review of the primary materials.

Results: Twenty (80%) of 25 MCOs provided member handbooks. No MCOs informed enrollees of their legal right to emergency care under the PLS, and 2 (10%) failed to define emergency services. Eight (40%) gave diagnoses rather than symptoms as examples of emergencies. One (5%) MCO required and 9 (45%) MCOs encouraged primary care provider (PCP) contact before seeking care. An additional 4 (20%) MCOs left the issue of PCP preauthorization unspecified. Fifteen (75%) MCOs required PCP contact after emergency care, 3 of which would deny payment without such contact. The remaining 12 MCOs left this issue unspecified. Eight (40%) MCOs placed conditions on reimbursement: 1 covered only the first visit for a clinical condition and the other 7 conducted retrospective reviews based at least in part on final diagnosis. Enrollees in 3 (15%) MCOs were encouraged to go to an affiliated, rather than the closest, emergency department (ED).

Conclusion: Seven years after the passage of New York State's PLS, MCOs still fail to fully comply with the law. Violations include warnings of retrospective denial of coverage, suggestions to attend in-system EDs, encouragement of PCP preauthorization, and post hoc PCP notification. Increased enforcement is necessary to ensure that MCOs do not hamper patients' timely access to emergency medical care.

Welch SJ, Allen TL/LDS Hospital, Salt Lake City, UT Background: Although many community hospitals have embraced the concept of total quality management (TQM), teaching and tertiary care hospitals have been

low-up with primary physician within 1 week; and (3) finding requires no specific follow-up. Results: Nine hundred ninety-four (680 men, 314 women) patients of 1,015 were included in the study. Eight hundred thirty-one patients received at least 1 CT scan. A total of 343 incidental findings were discovered, 194 of which were determined to be category 1 or 2. Thirty-three were determined to be category 1. One hundred thirty-nine incidentals occurred in women (32% of patients) compared with 204 in men (68% of patients). Two hundred forty-four (71%) of incidental findings occurred in patients older than 40 years. Scans of the abdomen or pelvis yielded 108 (55.7%) category 1 or 2 incidental findings, followed by the chest (43 [ Conclusion: The use of helical CT in the evaluation of trauma patients produces many unexpected findings but in this retrospective study affected management of approximately 3% of trauma hospital stays.

Worster A, Gilboy N, Fernandes C, Eitel D, Eva K, Geisler R, Tanabe P/Hamilton Health Sciences, McMaster University, Hamilton, Ontario, Canada; Brigham and Women's Hospital, Harvard University, Boston, MA Background: The Emergency Severity Index (ESI version 2) has been developed as a triage acuity scale for use in the emergency department (ED). This 5-level tool rates patients from level 1 (most serious) to level 5 (least serious). Unlike other triage acuity scales, such as the Canadian Acuity Triage Scale (CTAS), the ESI combines the level of acuity and the patient's clinical resource needs into a single measure. It has been proven to be reliable and valid in several US ED settings but has yet to be evaluated in a Canadian ED setting or compared with CTAS.

Study objectives: We compare the interrater reliability of CTAS and ESI version 2. Methods: Ten triage nurses from 4 urban EDs were randomly assigned to 1 of 2 groups for 3 hours of CTAS refresher or ESI introductory triage training. After their training, the nurses in each group independently and in a blinded fashion assigned triage scores to 200 prospectively collected, local ED cases.

Results: There were no significant differences found between the 2 groups with respect to age (P=.053), years in the ED (P=.13), hours of previous CTAS training (P=.57), or years of experience in triage (P=.61). An intraclass correlation to measure the interrater reliability of the CTAS group was 0.9079 (95% confidence interval [CI] 0.897 to 0.987) and not significantly different from that of the ESI 0.8926 (95% CI 0.882 to 0.985). An intertest G-study performed on the variance components derived from an analysis of variance revealed G(5)=0.896 (95% CI 0.817 to 0.992).

Conclusion: After just 3 hours of introductory training, experienced triage nurses were able to perform triage assessments using ESI version 2 with the same interrater reliability as those with experience and refresher training in CTAS. This comparison study also shows that the triage scores assigned by using CTAS generalize to those assigned by using ESI version 2.

Mace SE, Clauser S/Cleveland Clinic Foundation, Ohio State University School of Medicine, Cleveland, OH; University of Michigan, Ann Arbor, MI Background: Violence in the United States is epidemic. Homicide rates in the United States are the highest of any industrial country. Death rates from firearms are 8 times higher in the United States compared with those in other industrialized countries. Emergency department (ED) and hospital personnel have been assaulted, injured, and even killed while at work.

Study objectives: We obtain data about violence in EDs throughout the United States to determine what measures EDs are taking to improve and ensure the safety of ED staff, patients, and visitors.

Methods: A telephone survey of emergency medicine residency programs was done in August 2002. The survey assessed prevalence and types of violent incidents in the ED, ED security, ED design, and ED training about handling and preventing violence.

Results: Preliminary results (90.6% response rate, 115/127 programs) were as follows: 76% of EDs had incidents in which a weapon was confiscated or displayed (72% knives, 27% guns); 92% of ED staff admitted receiving verbal threats, 84% received physical threats, 62% received verbal threats daily, and 42% received physical threats at least once a month; 42% of EDs had violence resulting in an injury or death; 71% of all violent patients were intoxicated; and 53% of all violent incidents occur during evenings and weekends. Ninety-three percent of EDs had ED security Study objectives: We sought to explore the relations between patient acuity, throughput times, and emergency department (ED) patient satisfaction. We hypothesized that emergency patients would be the most satisfied with their throughput times and that this satisfaction would translate into greater overall satisfaction with their ED visit. We also hypothesized that overall ED satisfaction would be more strongly associated with perceived throughput times compared with actual throughput times, regardless of acuity.

Methods: We prospectively surveyed 1,865 patients treated in the ED of a large, inner-city hospital during a 1-month period. Data were collected on patient demographics, acuity of patient illness, actual waiting time for evaluation by a physician, and actual overall length of stay. Patient satisfaction with various throughput times and overall ED visit was assessed by using a 7-point scale (1=poor, 7=excellent). We conducted analysis of variance (ANOVA), analysis of covariance (ANCOVA), and correlations to explore the hypotheses.

Results: Emergency patients were more satisfied with their throughput times and overall ED visit compared with urgent and routine patients (all P<.01). Once the effects of throughput time satisfaction were controlled for by using an ANCOVA, acuity no longer predicted overall ED satisfaction. Correlations showed that overall ED satisfaction was more closely linked to throughput time satisfaction than to actual throughput times (average r=.62 versus -.12).

Conclusion: Emergency patients are more satisfied than urgent and routine patients with their ED visit. This appears to be, in large part, because emergency patients are more satisfied with their throughput times, especially the wait for physician evaluation. Changing perceptions of throughput times may yield larger improvements in satisfaction than improving actual throughput times, regardless of acuity.

Tijunelis MA/Keck School of Medicine, University of Southern California, Los Angeles, CA Background: The impact of noise pollution on patient and patient care provider has been extensively studied in the neonatal intensive care unit and in other critical care units. Noise pollution makes errors more probable and is one of the risk factors for provider burnout and negative outcomes for patients. Regulations under the Occupational Safety and Health Administration (OSHA) state that workers are at risk when noise levels exceed an 8-hour average of 90 dB (equal to the noise produced by a pneumatic drill). There are no known published studies of the typical noise level or the effects of noise in the emergency department (ED).

Study objectives: The purpose of this pilot study was to record and analyze noise in a large urban Level I trauma center. The hypothesis is that there are excessive noise levels in the ED that contribute to the already stressful work environment.

Methods: A 3-channel dosimeter Quest Q300 was placed as a stand-alone unit at a 4.5-m elevation, on the wall of the resuscitation booth in the ED. Sound was sampled 16 times per second for 12 hours and recorded as peaks and averages for each minute.

Results: Given that the maximum time-weighted average (TWA) allowed by OSHA is 90 dB in 8 hours (100% dose) and given that the percentage dose doubles for every 5-dB change, the TWA (43 dB) for this study represents a dose of only 0.14% for the 12 hours studied. The average sound levels peaked approximately 25 times throughout 12 hours. Individually measured peak levels of 90.1 (pneumatic drill) to 127.2 dB (equal to a boom box in a car) occurred every minute.

Conclusion: In this study, the TWA was well within OSHA noise standards; however, significant intermittent peaks occurred with regular frequency. Further research is needed to determine how these intermittent peak levels might contribute to stress in the ED.

White AA, Blanco R, Lemonds B, Sisco J, Bledsoe S, Irwin C, Isenhour J, Wright SW, Pichert JW/Vanderbilt University School of Medicine, Nashville, TN Study objectives: Medical errors occur in all settings, including the emergency department (ED). The causes and risk factors for medical errors are poorly understood despite their frequency. Claim files opened by medical malpractice carriers slower to implement these programs. We describe how a TQM program was designed and implemented in the emergency department (ED) of a tertiary care environment and demonstrate improvement in markers of patient care.

Study objective: We improve the efficiency and quality of the care as demonstrated by specific TQM markers in a tertiary care facility.

Methods: A TQM program was designed and implemented in our ED in 1998. This program involved monthly data collection and analysis and staff education in the core concepts of patient satisfaction. Components of the program included census data, physician profiling, and focused clinical audits. Baseline indices collected at the beginning of the program and at the 4-year mark included patient complaint rates, turnaround times, rates of left against medical advice and left without being seen, and patient satisfaction surveys performed by an outside consulting firm contracted by hospital administration.

Results: During the 4 years since its implementation, the program demonstrated improvement in all measured areas. Despite an increase in patient volume of 32% to 37,000 visits a year and only minimal staffing adjustments, the rate of complaints dropped by 56.1% (2.1 per 1,000 patients to 0.92), the turnaround times decreased from 183 minutes to 161 minutes (12%), and patients leaving against medical advice and left without being seen decreased 64.0% from 2.5% to 0.9%. Similarly, 44.8% of ED patients rated their care as "excellent" (compared with the national benchmark level of 30% for teaching institutions). The program led to a number of focused initiatives that resulted in specific improvements in an array of clinical problems, including endotracheal intubation, pain management, and laceration care.

Conclusion: Heretofore, tertiary care centers have not been thought to be environments conducive to strong TQM programs nor to a culture committed to patient satisfaction. We demonstrate how such a program was designed and tracked and how that process demonstrated improvement in specific TQM characteristics. Methods: This prospective, observational, institutional review board-approved study, conducted in a busy large urban ED, used audiotaping of medical interviews between medical care providers and Spanish-speaking patients, as well as a 4-question patient satisfaction survey. A 2-question satisfaction survey was also given to the health care providers. Questions on the patients' survey were as follows: "Did the doctor understand your problem?," "Was the interview respectful to you?," "Did you feel that the doctor was genuinely interested in you?," and "Did the doctor understand your problem?" Questions on the physicians' survey were as follows: "Do you think that the patient understood you?" and "Do you believe that the patient understood you?" The physician providers were divided into 2 groups: Group I was made up of physician providers who are not native speakers but have achieved a basic medical Spanish-speaking aptitude and have been certified by the institution as competent translators, officially called "ñ" providers. Group II was made up of the non-Spanish speaking physicians who used an official translator provided by the hospital's professional translation department. The audio recordings were then evaluated for errors (which could affect the medical treatment plan) by a blinded, native bilingual physician not involved in the study and who was unfamiliar with the voices on the tape.

Results: Among 26 medical interviews conducted by Group I (the ñ group), 14 errors were assessed. However, 7 of the 26 interviews were later disqualified because the ñ was, in fact, an Argentinean who made no errors in his 7 interviews. Among 23 patients interviewed by Group II, 46 errors were assessed. Nevertheless, all of questions on the patient and provider satisfaction surveys were uniformly positive in every case.

Conclusion: More errors in patient history-taking occurred in situations in which translators were used as an interface between the patients and the physicians versus situations in which the physicians had learned basic medical language skills and conducted the interviews themselves. Emergency physicians are encouraged to develop bilingual (Spanish-speaking) skills to diminish medical-history errors. legally defensible by the Sullivan Group standard. The reasons for the unexpected results of this study were not assessed.

Cox RD, Galli RL/University of Mississippi Medical Center, Jackson, MS Study objectives: This state began experiencing a medical malpractice crisis in 2001 that resulted in a number of specialists leaving the state and closure of some small emergency departments (EDs). The purpose of this study was to examine the impact of a state malpractice crisis on requests for transfers to the only tertiary care, university-based teaching hospital in the state.

Methods: The ED manages all requests for outpatient transfers. These requests were recorded on standardized sheets and then entered into a database. A retrospective review was performed by using the database, examining request volume and service requested throughout the previous 4 years. Data from 1999 and 2000 were compared with those from 2001 and 2002.

Results: Between 1999 and 2001, there was a 4.9% yearly increase in all transfer requests. In 2002, there was a 17.5% increase in transfer requests. The overall transfer volume increased from 3,944 in 1999 to 5,042 in 2002. All requested services showed an increase throughout this period. The requested medical services showing the greatest increase were psychiatry 142%, neurosurgery 114%, otolaryngology 95%, surgery 83%, and ophthalmology 87%. In contrast, the percentage increase for the 2 most requested services, orthopedics and trauma surgery, increased only 44% throughout this period.

Conclusion: During the 2-year period in which the state was undergoing a malpractice crisis, there was a significant increase in transfers to the state's only tertiary care, university-based teaching hospital. The increase was greatest for psychiatric services and neurosurgery. A number of neurosurgeons left the state during this period. Several local psychiatric hospitals closed during this period, but these closures were believed to be due more to financial reasons than to the malpractice situation. The university hospital did not increase its capacity for these services during this period. Methods: We conducted a retrospective review of patient visits to emergency departments (EDs) throughout a 1-year period in the greater Indianapolis area by using the Indiana Network for Patient Care. We examined data from 9 hospitals that represented more than 92% of the emergency care delivered in the metropolitan area according to self-reported data. These hospitals represent 5 independent and competing health care systems, which provide the majority of acute medical care in Indianapolis. The Indiana Network for Patient Care uses a "global patient index" to match patients across unrelated hospital systems. Data were exported to a Microsoft Access Database (Microsoft Corporation, Redmond, WA). Structured query language (SQL) queries were performed to determine the annual ED visit census from the selected institutions, the number of unique patients accounting for the total number of visits, and the average number of visits per treating hospital system. Results: The 9 hospitals included in this analysis provided 474,712 episodes of emergency medical care from September 1, 2001, to August 31, 2002. We excluded 3,072 visits (0.6%) without patient identifiers that could not be matched to the global patient index file, leaving 471,640 ED visits by 288,696 patients. On average, each patient had 1.6 (range 1 to 140) ED visits during the study period. Thirty-one percent (89,043) of the patients had more than 1 visit, accounting for 58% (275,059) of the total number of visits. Of patients with more than 1 visit, 25% (22,605) went to more than 1 hospital system and accounted for 33% (91,729) of all repeat visits and 19% (89,611) of the total number of visits. As patients presented to additional hospital systems, the average number of visits per patient increased dramatically. More than half (266,689) of patients remained within a single hospital system, with an average of 1.4 visits per patient, whereas 97 patients visited all 5 hospital systems, with an average of 21.2 visits per patient.

Conclusion: Our data underscore the need for improving the communication between health care systems. Because these systems are separate and distinct hospital networks, patients traveled throughout the system irrespective of geographic and financial boundaries. A small number of patients accounts for a considerable number of visits. Access to pertinent medical record information across hospital systems could potentially result in a reduction of diagnostic testing, reduced time and cost savings for our patients and physicians, and improved health care outcomes.

offer a unique method of detecting patterns and causes of medical errors. The purpose of this study was to examine claim files to assess the presentations, causes, and specific factors of potential medical errors. Methods: We reviewed 74 ED claim files opened by the medical malpractice carrier of a University Hospital. The hospital is a Level I trauma center and regional pediatric center. An ED nurse reviewer trained in cause-and-effect analysis examined each claim file and generated a cause-and-effect diagram that portrayed the various contributors to the adverse event. Next, the reviewer assigned 1 of 120 malpractice claims description codes to each potential contributing cause. The reviewer presented each claim to a similarly trained team for further review and professional consensus about underlying causes for each adverse outcome. Descriptive statistics were calculated to identify common trends and patterns across ED cases.

Results: Injuries, abdominal pain, infections, and psychiatric emergencies accounted for 70% of the cases. Chest pain (4%) and headache (4%) were uncommon reasons for opening a claim file. Common reasons for the alleged adverse outcome included issues relating to diagnosis (45%), resident supervision (31%), communication (30%), patient behavior (26%), administrative (24%), documentation (20%), and treatment (19%). Specific factors within these groups were identified, eg, diagnostic errors were caused in part by initial problematic radiograph reading in 18 cases (24%) and were the sole factor in 13 cases. Similarly, administrative issues included failures to follow policies or protocols in 14% of cases. Review of 6 (8%) case files identified no underlying, preventable cause of the alleged adverse outcome.

Conclusion: Institutional assessment of adverse outcomes is often done case by case. Analyzing and aggregating claim files may help identify clusters of errors that can drive quality improvements in EDs on the individual and organizational scale. This study has demonstrated that the methodology is feasible for identifying highrisk areas for ED-related medical errors. These clusters of common causes of adverse outcomes suggest meaningful opportunities for quality improvement.

Chathampally Y, Moore S, Riesz L, Hunt H, Almond G/Metropolitan Hospital Center, New York Medical College, New York, NY Background: Template-based charting has been shown to increase physician satisfaction and increase billing, but to our knowledge, no study has assessed the medicolegal defensibility of a templated emergency department (ED) chart versus a handwritten chart. Many EDs have instituted protocols for evaluation of patients presenting with a chief complaint of atraumatic chest pain. Our quality improvement data show 100% compliance with our protocol when medical records of patients who are admitted with chest pain are reviewed.

Study objectives: We compare the mean number of missed elements required to maximize medicolegal defensibility in templated ED chart versus handwritten ED chart for patients discharged after ED evaluation for a chief complaint of atraumatic chest pain. We quantify the overall medicolegal incompleteness of medical records for discharged patients who presented with a chief complaint of chest pain.

Methods: This was a retrospective study of 191 consecutive charts overlapping the transition from handwritten ED records (99) to template-based ED records (92). Data were extracted by using 12 elements listed by the Sullivan Group (emergency medicine malpractice education company) as necessary for medicolegal record completeness (ie, location, aggravating factors, radiation, history). Data were collected on the number of missing elements for each record type and how frequently each data element was missed. The templated chart lists all 12 elements on its chest pain template. Because the purpose of the study was to evaluate the completeness of documentation of the medical record, we did not confirm whether the missing elements had in fact been assessed clinically. Completeness using other areas of the chart were included only if the physician referred to that area explicitly.

Results: Three of the 92 handwritten records were complete. Five of the 99 template charts were complete. Only 8 (4%) of all the discharged patients' charts were complete as defined by the Sullivan Group. The mean number of missing elements for the templated records was 4, ranging from 1 to 6, and the mean number of missing elements for the handwritten records was 5, ranging from 1 to 6 (P>.05). The most frequently missed element was documentation of the reading of the ECG within 15 minutes of arrival for both groups. There were no statistically significant differences.

Conclusion: According to the recommendations of the Sullivan Group, templated ED records did not improve the recording of key data elements compared with the handwritten chart. Only a small number of all charts evaluated would be deemed

Results: When the bolus and infusion protocols were compared by Failure Modes and Effects Analysis, there was a decrease in the risk profile number from 1,437 (bolus) to 1,143 (RPD). Forty patients who received the treatment were evaluated. No adverse effects occurred in the preparation, administration, or patient response to therapy. Also, hypotension, hypokalemia, and renal insufficiency did not occur in any of the patients. Nine patients were excluded from the LOS portion of the study: 4 patients were treated for pneumonia, 2 patients were treated for cancer, 1 patient was treated for infarction, 1 patient was treated for valve disorder, and 1 patient was treated for cellulitis. The average LOS for 31 RDP patients was 5.4 days (95% confidence interval [CI] 4.3 to 6.5). An increase in urine output documentation was observed after educational efforts (95% CI 61% to 90%).

Conclusion: Our RDP for CHF patients was found to be safe, efficacious, and easily implemented in the ED.

Study objectives: I establish conformity with the advanced cardiac life support (ACLS) protocols among the ACLS-certified and non-ACLS-certified physicians.

Methods: This was a retrospective analysis of successive cardiac arrests from January 1, 2000, to December 1, 2000, including assessment of the resuscitation leaders' ACLS certification. The Aga Khan University Hospital is a main urban 550bed teaching University hospital in Karachi, Pakistan, with approximately 100 daily admissions. All patients experienced inhospital cardiac arrests.

Results: There were 403 arrests studied, with a total of 1,007 cardiac rhythms. There were 179 (44.4%) resuscitations with return of spontaneous circulation. A total of 4,514 interventions were recorded for all rhythms, with 2,618 (58%) acquiescent with ACLS protocols compared with 1,896 (42%) deviations. Cardioversion, atropine sulphate, calcium carbonate, and sodium bicarbonate were used, with significantly higher nonconformity. Ventricular fibrillation and unstable ventricular tachycardia had significantly higher mean rhythm deviation scores, whereas scores were significantly lower for sinus rhythm and stable paroxysmal supraventricular tachycardia (P<.001). The mean number of deviations per resuscitation attempt and mean number of treatments per arrest were compared for the resuscitations, and there were significantly more deviations and treatments in resuscitations led by non-ACLS-certified physicians (P<.005). The return of spontaneous circulation and survival to discharge rates were higher in resuscitation led by ACLS-certified physicians (P<.001).

Conclusion: Nonconformity with ACLS protocols was found in 42% of treatments. ACLS certification and ACLS protocol conformity are positively correlated. It is prudent for all residents to take regular ACLS courses and receive certification for better patient outcome.

Nagurney JT, Brown DF, Chae C, Chang Y, Chung WG, Cranmer H, Dan L, Fisher J, Grossman SA, Tedrow U, Lewandrowski K, Jang IK/Massachusetts General Hospital and Harvard Medical School, Boston, MA Study objectives: We compare a strategy by using cardiac markers in a subpopulation of patients with potential cardiac symptoms that began more than 6 hours before their emergency department (ED) blood draws (6HR) to a strategy using them in patients with symptom onset at 0 to 24 hours (ALL) before ED blood draws.

Methods: This was a prospective descriptive study with cardiac markers obtained approximately 0, 2, and 8 hours after patients presented to the ED. Time to blood draw was calculated by combining symptom onset to ED delay and ED arrival to blood-draw times. The study included all patients aged at least 30 years and presenting to a university ED with symptoms consistent with an acute coronary syndrome.

Results: Three hundred seventy-five patients were enrolled, with time of symptom onset available in 365 patients. The median age was 65 years (interquartile range 51 to 75); 59% of patients were men, and 14% experienced an acute myocardial infarction. At the initial (time 0) blood draw, 56% of patients had symptom onset less than 6 hours earlier, 32% of patients had symptom onset 6 to 24 hours earlier, and 12% had symptom onset more than 24 hours earlier (excluded from analysis; Table) .

Although a formal statistical comparison of the sensitivities in the 2 groups (ALL and 6HR) violates the assumption of independence, it appears that a strategy using a 6-hour rule greatly enhances sensitivity at the time 0 blood draw for all 3 markers but becomes much less valuable at the 2-hour blood draw. By the 8-hour blood draw, the 2 populations become equivalent.

Conclusion: Using a strategy with a 6-hour rule from onset of cardiac symptoms

Morris D, Jerrard D/University of Maryland, Baltimore VA Hospital, Baltimore, MD Background: Institutional and Joint Commission on Accreditation of Healthcare Organizations (JCAHO) policy requires that all physicians have annual purified protein derivative (PPD) testing. Particularly in urban facilities, emergency physicians continue to be at high risk for exposure to patients with tuberculosis.

Study objectives: We determine the compliance with annual mandatory PPD testing among faculty and residents in a university-based emergency medicine residency during a JCAHO visit year.

Methods: A retrospective cohort analysis of employee health data pertaining to emergency medicine faculty and house staff was performed by looking at compliance with PPD testing, as well as conversion rates before and during employment at the institution for the involved physicians. Investigators also tallied health care nurse and administrative assistant time spent in ensuring compliance. The study took place between July 2001 and June 2002 at a training program staffing an urban academic emergency department treating 80,000 patients yearly. Institutional review board approval was granted. Data are reported as proportions.

Results: A total of 50 emergency medicine residents and 29 emergency medicine faculty health care records as maintained by the employee health nurses were examined for the study period. Ten percent of the residents (n=5) were found to have tested positive for PPD (+PPD) before enrolling in the training program. Two of the remaining 45 residents (4.4%) converted while in the residency, resulting in 14% (n=7) of the emergency medicine housestaff being +PPD. Compliance for emergency medicine house staff in obtaining skin testing for the year (or answering a questionnaire if before +PPD results) was 98% (49/50). Administrative assistant time spent in ensuring resident compliance was 45 minutes for each of the 12 months of the study period. Two (6.9%) of the 29 attending physicians were known to be +PPD from previous screens and were known to have converted while employed at the institution. Only 12 (41%) of 29 attending physicians were compliant (2 who were +PPD answered a health risk questionnaire that fulfilled compliance criteria as determined by hospital policy). Employee health nurse time spent ensuring faculty and resident compliance included attending Grand Rounds and performing skin testing while there (9 hours total for the 12 months) and memorandums to noncompliant physicians (3 hours total for the 12-month period). The health care nurse was never able to perform skin testing on the remaining 17 (59%) attendings during the year because these physicians did not avail themselves of the opportunity.

Conclusion: Although significant ancillary staff time is spent cajoling emergency medicine house staff and faculty to comply with hospital and JCAHO policy about PPD testing, attending physician cooperation remains poor despite intensive efforts to bring all personnel into compliance in preparation for a review by JCAHO the same year. The real possibility of conversion dictates that emergency physicians take the initiative and pursue testing of their own volition. This study should remind residents of the need for continued compliance even once training is finished.

Pharm RA, Silka PA, Brooks K/Cedars Sinai Medical Center, Los Angeles, CA Study objectives: We describe the process of implementing a rapid diuresis protocol (RDP) in a private, urban emergency department (ED) for patients with congestive heart failure (CHF).

Methods: All steps leading to the implementation of an RDP are described. A failure mode effect analysis of the RDP by using a furosemide infusion was performed and compared with the traditional furosemide bolus treatment for patients with CHF. ED in-service training was conducted for nursing and physician staff after a furosemide RDP was agreed on by the ED chairman and the director of the Heart Failure program. Educational efforts were carried out by means of newsletters, emails, and group meetings to ensure safe implementation of the aggressive diuretic treatment. In-service meetings outlined the following: literature-based support for the treatment, patient eligibility, and key aspects of monitoring adequate diuresis. The furosemide RDP was made available to clinical personnel on the hospital's Intranet and called for an initial dose of 10 mg/hour intravenously. A cohort of CHF patients was treated with the RDP, and outcomes were analyzed by retrospective medical record review for electrolyte levels, blood urea nitrogen, creatinine, urine output, and length of stay (LOS). Patients with discharge codes other than CHF, heart failure, or hypertension with heart failure were excluded from the LOS portion of the study. chest pain unit (CPU) are subjected to cardiac catheterization (CATH) and then receive no subsequent significant cardiac interventions.

Methods: This institutional review board-approved retrospective study is a descriptive analysis of patients admitted from the CPU at a tertiary referral center from November 2000 to October 2001. Patients presenting to the emergency department (ED) with complaints of cardiac ischemia, who had negative initial ED evaluation results, and who were admitted to the CPU for serial enzyme level tests, ECGs, and functional testing were included. Patients who had myocardial ischemia, had a positive functional test, or who were believed to require further cardiac evaluation were admitted to cardiology. Medical records were identified through a computerized database and reviewed by using an explicit data abstraction form to identify patients who underwent CATH and significant interventions (defined as angioplasty/stent placement, coronary artery bypass grafting, aggressive anti-anginal medication, or pacemaker placement).

Results: One thousand forty-eight patients were admitted to the CPU. Two hundred eighty-two (27%) patients were admitted to the hospital, of whom 123 (44%) underwent CATH. Seventy-six (62%) of the post-CATH patients received significant cardiac interventions and 47 (38%) did not.

Conclusion: Although a majority of patients subjected to the high-risk procedure of CATH received significant cardiac interventions, a substantial minority did not. Future studies are needed to determine whether the benefits of this evaluation paradigm outweigh the risks of CATH in patients receiving no further cardiac interventions and whether those likely to benefit from CATH can be identified.

Hess JJ, Johnson GG, Lohse CM, Goyal DG/Mayo Clinic, Rochester, MN Study objectives: We examine the incidence of significant cardiac interventions performed after chest pain unit (CPU) evaluation and admission to cardiology.

Methods: This institutional review board-approved retrospective study is a descriptive analysis of patients admitted from a CPU at a tertiary referral center from November 2000 to October 2001. The patients included presented to the emergency department (ED) with complaints of cardiac ischemia and were admitted to the CPU after negative initial ECG and troponin T test results. Patients underwent serial troponin T, ECG, and functional cardiac testing in the CPU. If they ruled in for myocardial ischemia, had a positive functional test, or were believed to require further cardiac evaluation, they were admitted to cardiology. Medical records were identified by using a computerized database and reviewed using an explicit abstraction form to identify those who underwent significant interventions (defined as angioplasty/stenting [PCI], initiation of aggressive anti-anginal medications, coronary artery bypass grafting [CABG], or pacemaker placement).

Results: We identified 1,048 patients who were observed in the CPU, of whom 282 (27%) were admitted to the hospital. One hundred ninety-nine (71%) patients received no significant interventions and 83 (29%) received at least 1 during their admission: 50 patients had PCI, 21 patients received aggressive anti-anginal medications, 15 patients underwent CABG, and 1 patient underwent pacemaker placement.

Conclusion: Although many patients were discharged from the CPU, the majority of those admitted received no significant interventions. Further studies are required to determine whether the CPU paradigm efficiently risk stratifies patients with suspected cardiac ischemia and whether those requiring interventions can better be identified.

Study objectives: A delay in the arrival at the hospital is a factor that offers negative prognostic diagnosis to patients affected with an acute myocardial infarction (AMI). The politics of medical education attempts to teach the general population the symptoms of AMI so that on arrival the patients could be treated immediately. In theory, the patient with a history of ischemic heart disease knows the clinic and the circuits and is able to receive treatment at once if he is affected by an AMI. The aim of this study is to find out the effect of the delay from the beginning of the symptoms until arrival at the hospital in patients with previous ischemic heart disease before those who did not have this history.

Methods: The delay was analyzed in minutes from the beginning of the pain until the arrival at the hospital of 1,358 patients affected with a coronary syndrome with ST elevation between March 1, 2001, and March 1, 2003, in 27 Spanish hospitals

to time of blood draw appears to markedly enhance the sensitivity of cardiac markers drawn at arrival (time 0) to an ED but becomes of much less valuable at the 2hour blood draw. Study objectives: We compare the risk strata established by emergency department (ED) house staff when predicting acute coronary syndrome (ACS) versus acute myocardial infarction (AMI) and compare their ability to correctly categorize patients within these 2 groups.

Methods: This was a prospective descriptive study. While evaluating patients with symptoms consistent with ACS (either AMI or unstable angina pectoris [UAP]), house staff were asked to predict the probability of ACS and of AMI. They were reminded that one (AMI) represented a subset of the other. The probability scale was defined as low/very low (0% to 15%), low/moderate (16% to 40%), moderate (41% to 60%), moderate/high (61% to 85%), and high/very high (86% to 100%). House staff were postgraduate year 2 to 3 residents in emergency medicine or internal medicine; patients were all aged at least 30 years and being evaluated in a university hospital ED for symptoms consistent with ACS.

Results: Three hundred seventy-five patients were enrolled, with estimates available on 370. The median patient age was 65 years (interquartile range 51 to 75); 59% were men, and 25% had ACS (53 AMI; 39 unstable angina pectoris). The number of patients predicted to be in each risk stratum and the percentage with AMI and ACS are shown in the Table. Almost 3 times as many patients (44%/15%) were estimated to be at moderate to very high risk for ACS than for AMI. Among patients who did have ACS, only 7% (95% confidence interval [CI] 2% to 12%) were incorrectly placed in the very low/low risk group compared with 23% (95% CI 12% to 34%) of the MI patients. Similarly, only 20% (95% CI 12% to 28%) of the ACS patients were incorrectly placed in the very low or low/moderate risk groups compared with 38% (95% CI 38% to 51%) of those with AMI.

Conclusion: House staff attach a higher probability to the likelihood of ACS than to AMI in ED patients presenting with possible ACS symptoms. Fewer ACS than AMI patients are incorrectly predicted to be in very low through low/moderate risk categories for these respective outcomes. 

Hess JJ, Johnson GG, Lohse CM, Goyal DG/Mayo Clinic, Rochester, MN Study objectives: We assess the incidence with which patients admitted from a

ing complication (95% CI 1.4 to 13.5; P=.01) compared with patients with a Ho clinical score less than or equal to 2. Conclusion: The simplified Ho clinical score risk performs as well as the more complicated Morise and Framingham scores in the risk stratification of ED chest pain patients. Future studies are needed to determine whether this information can be used to influence initial triage and management decisions in the evaluation of chest pain patients with suspected ACS in the ED setting.

Edwards J, Sloan EP/University of Illinois at Chicago, Chicago, IL Study objectives: The objective of this study was to examine emergency physicians' treatment of patients with atrial fibrillation or atrial flutter.

Methods: Data analyzed in this study were from the National Hospital Ambulatory Medical Care Survey. Patient selection was based on International Classification of Diseases, Ninth Revision, Clinical Modification codes.

Results: An estimated 1.8 million patients diagnosed with atrial fibrillation or atrial flutter in US emergency departments (EDs) accounted for 0.5% of US ED visits during 1997 to 2000. In this group, 55% were women, 85% were white, and the mean age was 72±14 years. Many patients (44%) with atrial fibrillation or flutter complained of chest pain or shortness of breath. The most frequently administered individual medications were diltiazem (22%), digoxin (17%), furosemide (16%), warfarin (9%), and heparin (8%). Patients were most often treated with calcium channel blockers (35%), cardiac glycosides (26%), and pain medications (22%). Overall, an antidysrrhythmic agent was given to 77% of patients; 4% received adenosine. Two thirds of patients were hospitalized, with 15% being admitted to the ICU or cardiac care unit. Patients receiving a βblocker were less likely to be admitted (47% versus 69%; odds ratio 0.4; P=.02).

Conclusion: Most patients who presented to the ED with atrial fibrillation or atrial flutter were older and complained of chest pain or shortness of breath. Most patients received an antidysrrhythmic, and a few received adenosine despite its not being indicated for use in atrial fibrillation or atrial flutter. Patients who received a β-blocker were less likely to be admitted. Study objectives: Recent studies analyzed cardiac troponin I (cTnI) as a marker in heart failure. To the authors' knowledge, no one has studied the relation between cTnI and several clinical variables in heart failure. We examined the power of cTnI as diagnostic and prognostic marker in a large population of severe heart failure patients. These are preliminary data to a study on cTnI use in heart failure care unit.

Methods: We performed a retrospective analysis of heart failure patient records discharged from the emergency department (ED) from March to December 2002 by heart failure codes according to International Classification of Diseases, Ninth Revision. We included 98 patients, 53 with cTnI negative (<0.05 ng/mL) and 45 with cTnI positive (>0.05 ng/mL). The heart failure severity was founded on clinical variables (New York Heart Association [NYHA] class). One hundred eight patients were excluded because of incomplete data, renal failure, recent infections, or acute coronary syndrome. A 3month follow-up was performed for death, heart failure, and acute coronary syndrome endpoints. We used a second-generation immunoassay test for cTnI (Dade Behring). The lowest concentration detected was 0.04 ng/mL. Comparison between 2 groups was performed by t and χ 2 test, among more groups by analysis of variance test.

Results: The 2 groups of patients with positive cTnI results (cTnI+) and negative cTnI results (cTnI-) were homogeneous for clinical characteristics (P>.05). They differ for cTnI serum concentration (P<.01), for oxygen saturation (P<.01), and for NYHA class (P<.05). We divided our population into 5 groups by using cTnI levels, and we found significant differences between the average of NYHA class and oxygen saturation in these groups (P<.01). Heart failure patients with cTnI+ had had more relapses of heart failure in the past than heart failure patients with cTnI-(P<.01). There were significant differences in days of hospitalization between these groups (P=.05). After follow-up, heart failure patients with cTnI+ had a higher incidence of death than heart failure patients with cTnI-(15.2% versus 3.8%; P<.05), with a higher relative risk for death (relative risk 4; 95% confidence interval 3.53 to 4.39).

Conclusion: Our data show that patients with heart failure and elevated cTnI had a worse clinical presentation and short-term outcomes than those with heart failure and cTnI negative. These results could be explained by minor myocardial damage in heart failure; causes of damage are fibrosis, hypoxia, and apoptosis. cTnI for its high that participate in the RESIM group (Registry of Acute Myocardial Infarction in the Emergency Services of Spain).

Results: Results are shown in the Tables. Conclusion: The patient with previous ischemic cardiopathy does not arrive faster at the hospital when he or she is affected by an AMI than the patient with previous coronary disease. 

Fesmire FM, Wali GO, Adams DE/University of Tennessee College of Medicine, Chattanooga, TN Study objectives: We determine predictive value of the simplified Ho clinical score, which uses only 5 clinical variables, compared with the more complex Morise and Framingham clinical scores for predicting 30-day adverse outcome (AO) in emergency department (ED) chest pain patients with suspected acute coronary syndromes (ACS) and absence of preexisting ischemic heart disease.

Methods: This was a retrospective analysis of a prospectively acquired database of 1,378 consecutive ED chest pain patients with no history of acute myocardial infarction (AMI), percutaneous coronary intervention (PCI), or coronary artery bypass surgery (CABG) and with absence of injury on initial ECG who underwent a standardized chest pain evaluation protocol. All patients were followed up for 30day AO that was defined as AMI, PCI/CABG, life-threatening complication, or cardiac death within 30 days of ED presentation. The Ho clinical score was calculated as follows: 1 point for each decade of patient's age older than 40 years, 1 point for male sex, 1 point for previous myocardial infarction (excluded from present study), 1 point for typical chest pain, 1 point for non-insulin-dependent diabetes, and 2 points for insulin-dependent diabetes. The Morise clinical scores and the Framingham clinical scores were calculated according to previously defined scoring systems by using a combination of typical risk factors and select clinical findings.

Results: Thirty-day AO occurred in 148 (10.7%) patients. Area under the receiver operating characteristic curve for 30-day AO for the Ho score, the Morise score, and Framingham score was 0.73±0.02, 0.72±0.03, and 0.71±0.02, respectively. There were no significant differences in areas between the Ho score compared with the Morise score (95% confidence interval [CI] -0.03 to 0.05), and Ho clinical score compared with the Framingham score (95% CI -0.02 to 0.06). Increasing clinical score points was associated with increasing risk of 30-day AO for all 3 scoring systems. Using the most accurate cutoff value of the Ho clinical score risk stratified patients for 30-day AO. Patients with a Ho clinical score more than 2 had 6.4 times increased odds of AMI (95% CI 3.7 to 10.8; P<.0001), 4.8 times increased odds of 30-day PCI (95% CI 2.6 to 8.6; P<.0001), 7.7 times increased odds of 30-day CABG (95% CI 3.2 to 18.6; P<.0001), and 4.3 times increased odds of 30-day life-threaten-

Levin WJ, Levitan RM/New York Medical College, New York, NY; Hospital of the University of Pennsylvania, Philadelphia, PA Background: Thirty-centimeter rhinolaryngoscopes cannot be used for fiberoptic guided oral intubation because the device is not long enough to serve as a stent for 30-cm-long standard tracheal tubes. Even if the tracheal tube is shortened, it is difficult to manipulate the rhinolaryngoscope into proper position and adequately place the tracheal tube through the vocal cords. The objective of this study was to develop and test a new technique for oral intubation using a rhinolaryngoscope.

Methods: Two operators tested the technique on 12 nonembalmed, recently deceased cadavers. In addition to a 30-cm-long rhinolaryngoscope (Olympus ENF-GP; Olympus North America, Melville, NY), the technique used a newly developed plastic endoscopic sheath with a working channel (Vision Sciences, Boston, MA), a retrograde intubation wire, and a stiffener (both from Cook Critical Care, Bloomington, IN). To assist oral intubation with the fiberscope, either a Williams or Ovassapian airway was inserted into the mouth. The fiberscope and endoscopic sheath were then guided through the vocal cords and into the trachea under direct vision. The wire from a retrograde intubation kit was then inserted anterograde through the working channel of the endoscopic sheath. After removal of the scope and the sheath, the stiffener from a retrograde intubation kit was then placed over the wire, down the mouth, and into the trachea. A standard 7.0-mm internal diameter tracheal tube was subsequently slid down the stiffener, achieving oral tracheal intubation. All cadavers had surgical incisions permitting direct visualization of the tracheal tube and confirming intratracheal placement.

Results: Intubation was successful by operator 1 in 10 of 12 cadavers and by operator 2 in 11 of 12 cadavers. Of 24 total attempts in the 12 cadavers, there were 3 failures. One of the failures for each operator occurred in the same cadaver, and operator 1 had an additional failure. Time from placement of the oral airway to successful intubation in the 11 of 12 cadavers that were able to be intubated averaged 107 seconds, with a range from 70 to 129 seconds (95% confidence interval [CI] 93 to 121 seconds). Operator 1 had a mean time of 117 seconds (95% CI 105 to 129 seconds); operator 2 had a mean time of 100 seconds (95% CI 81 to 109 seconds).

Conclusion: This preliminary study proves the feasibility of fiberoptic intubation using a short rhinolaryngoscope with the use of an endoscopic sheath and a retrograde wire and stiffener. Additional experimentation with the technique is necessary to shorten the intubation time and solve visualization difficulties that prevented intubation in 3 of the 24 attempts.

Levin WJ, Levitan RM/New York Medical College, New York, NY; Hospital of the University of Pennsylvania, Philadelphia, PA Study objectives: The purpose of this preliminary study was to assess the feasibility of using the Shikani Optical Stylet (SOS) in place of a standard stylet during routine laryngoscopy. The SOS was used for added visualization of the anatomical landmarks to facilitate placement of the endotracheal tube into the trachea.

Methods: A prospective study of physician intubation time and success rate was performed on nonembalmed, recently deceased human cadavers. Laryngoscopy and intubation were performed in the routine manner. However, in place of the standard stylet, the tracheal tube was preloaded on the SOS. Time to successful placement of the tracheal tube into the trachea was recorded with a stopwatch by independent observers. Verification of placement in the trachea was performed by using direct laryngoscopy or fiberoptic assessment through the tracheal tube. Statistical analysis of laryngoscopy performance (time in seconds) was done using Microsoft Excel Office 98 (Microsoft Corporation, Redmond, WA).

Results: The technique using the SOS for intubation was successful in 30 of 32 attempts. Two failures (1 unrecognized misplacement into the esophagus and 1 inability to place the tracheal tube) occurred on the first cadaver. All laryngoscopies were successful on the second cadaver. The mean time to intubation using the SOS as a laryngoscopy adjunct on the first cadaver was 29 seconds, with a range from 13 seconds to 93 seconds (95% confidence interval [CI] 18 to 40 seconds). On the second cadaver, the average was 21 seconds, with a range from 12 to 42 seconds (95% CI 17 to 25 seconds) and no failures.

Conclusion: The initial data suggest that the technique using the SOS in place of the standard stylet can be easily learned. Physician intubators with no previous

sensitivity and specificity for myocardial injury can be a good marker in heart failure. Limitations to our findings are the use of an International Classification of Diseases codes database, other causes of myocardial cell damage (myocarditis and transient silent myocardial ischemia), and difficult-to-compare studies on cTnI made by different standardization immunoassay methods. However, we believe our data results reliable because of their statistical power. We suggest further research: large studies on the power of cTnI as a marker in heart failure.

Jimenez C, Larkin G, Delaney K/University of Texas Southwestern Medical School, Dallas, TX Study objectives: The objective of this study was to evaluate the impact on admission rates, inpatient days, and use of diagnostic radiographs for patients with severe congestive heart failure (CHF) who receive primary care in the emergency department (ED).

Methods: This was a retrospective study conducted in a high-volume public teaching hospital ED. The medical records of all patients with a diagnosis of CHF who visited the ED more than 1 time between December 1999 and October 2002, and every third patient with 1 visit, were reviewed for inclusion in the study. Included patients received all of their care in our institution and had a diagnosis of severe CHF (ejection fraction <30%) or end-stage CHF (ejection fraction <20%). Primary reasons for exclusion were care in clinics not in our database, isolated right-sided heart failure, frequent visits with asthma or chronic obstructive pulmonary disease, end-stage renal disease, and terminal illness other than CHF. Data were collected on CHF severity, number of ED and clinic visits for CHF, comorbidities, admissions per year for decompensated CHF, inpatient days per year for decompensated CHF, and number of chest radiographs per year. Two cohorts were defined according to the number of times the patient visited a primary care clinic during a year. "ED primary care physician (PCP)" patient visits declined 25% below the median number of clinic visits per year: 6 times per year for "severe" CHF and 8 times per year for "end stage" CHF. "Clinic PCP" patients visited the clinics more often than these minimum defined frequencies.

Results: The median number of clinic visits per year was 2.0 for the ED PCP group and 11.7 for the Clinic PCP group. ED visits per year were 7.1 for the ED PCP group (81% of all outpatient visits) and 3.3 for the Clinic PCP group (23% of all outpatients visits). The groups were comparable in the extent of comorbidities. Results are tabulated in the Table. Conclusion: Patients who receive primary care for severe CHF in the ED rather than in the clinics are admitted to the hospital significantly more often than patients who are treated in the clinics, spend significantly more days in the hospital, and have more frequent diagnostic chest radiographs. Efforts to move patients into primary care settings are likely to be cost effective. were 35 (2%) intubations that were ultimately determined to be nontracheal by receiving physicians. Among these, 20 (57%) patients were intubated for trauma indications. Seven (20%) patients were children (<10 years). Fifteen (43%) patients did not have a pulse before intubation attempts. Overall, 21 (60%) had multiple confirmatory techniques used by paramedics. The most commonly documented was "equal lung sounds" (91%), followed by "visualized cords" (37%). Colorimetric end-tidal CO 2 was used selectively among patients with pulses, 9 (45%) of 20. Seventeen (49%) of the nontracheal intubations were potentially recognizable according to paramedic documentation. Conclusion: An unrecognized, nontracheal intubation is a potentially devastating consequence of failed airway management. We report a small but not inconsequential experience with failed out-of-hospital airway management. In this emergency medical system, more frequent use of multiple confirmatory techniques (including end-tidal CO 2 detection) may help to reduce the incidence of this potentially life-threatening scenario.

Panagos PD, Luber S, Lindsell C, Carleton S/Spectrum Health-Butterworth Campus, University of Cincinnati College of Medicine, Cincinnati, OH Study objectives: We describe the methods, success rates, and complications of airway management in the out-of-hospital, aeromedical setting performed by emergency medicine residents.

Methods: Patients were enrolled in a prospective observational study throughout a 26-month period from January 2000 to March 2002. Entry criteria into the study required advanced airway management performed by a member of aeromedical flight crew. Location, indication for intubation, method of intubation, and number of attempts were recorded. Outcomes measures included success rates and complications. Institutional review board exemption was granted for this study.

Results: One thousand six hundred seventy-nine helicopter transport missions on 1,692 patients were performed. One thousand sixty-six were interhospital, and 614 were scene flights. One hundred twenty-six patients (mean age 33.2±19.6 years, 88 men [75.9%], 100 [82%] trauma patients, 86 [72.3%] patients with cervical immobilization, and 90 [78.9%] patients with field intubations) required definitive airway management techniques. Self-reported perceived need for definitive airway included airway protection (75%), oxygenation (43%), ventilation (44%), and patient combativeness (23%). Methods used were rapid sequence intubation in 79 (62.6%) patients, modified rapid sequence intubation in 12 (9.5%) patients, and no medications in 30 (23.8%) patients. Techniques used were curved blade in 116 (92.1%) patients, straight blade in 11 (8.7%) patients, nasal in 5 (4.0%) patients, cricothyrotomy in 3 (2.4%) patients, intubating laryngeal mask airway in 3 (2.4%) patients, Endotrol in 2 (1.6%) patients, and Combitube (esophageal-tracheal twinlumen airway device) in 1 (0.8%) patient. The overall success on first attempt was 73.8%, with the following training level breakdown: postgraduate year (PGY)-1 40%, PGY-2 79%, PGY-3 88%, and PGY-4 75%. There were 25 (19.8% overall rate) complications; 9 (or 39% of the total) were a recognized mainstem bronchial intubation. All 126 patients ultimately received a definitive out-of-hospital airway, 94.4% via the endotracheal or nasotracheal route.

Conclusion: This study indicates that emergency medicine residents, at all training levels, can successfully perform advanced out-of-hospital airway techniques without direct attending supervision. Although the complication rate was almost 20%, most complications were minor, immediately recognized, and ultimately corrected. Residency programs may consider resident participation in aeromedical programs as a safe and effective way to obtain airway management experience while gaining confidence to practice independently in the future.

Tiah L, Wong E, Chien MFJ, Sadarangani SP/Singapore General Hospital, Singapore Study objectives: We evaluate the use of alternative devices by junior officers to secure an airway by comparing the ease of successful insertion and skills retention of the laryngeal mask airway (LMA), the esophageal-tracheal combitube (ETC), and the endotracheal tube (ETT).

Methods: This was a prospective cohort study of 93 third-year medical students who were taught the use of LMA, ETC, and ETT, with skills tested at 0 and 6 months.

Results: There was significant difference (P<.005) between the mean times for experience using this technique had a high success rate and rapid intubation times in fresh cadavers. This technique with the ability to see the laryngeal structures through the SOS may aid in the placement of the endotracheal tube when visualization of the anatomy with standard laryngoscopy is suboptimal and tube placement difficult.

Antro C, Urbino R, Merico F, Suman F, Ferrera P, Gai V/San Giovanni Battista Hospital, Turin, Italy

Study objectives: We describe our emergency department's (ED) experience with noninvasive positive pressure ventilation (NPPV) for patients with hypercapnic acute respiratory failure (ARF).

Methods: We conducted an analysis of the NPPV register data (prospectively collected) and of the medical records (retrospective data abstraction) of a nonselected population of patients with hypercapnic ARF, who were treated during a period of 10 months at our ED (setting consisting of ED and a linked intermediate NPPV unit managed by ED staff). We evaluated the indications for NPPV, the NPPV effect on physiologic variables, and outcomes. Treatment success was defined as no need for endotracheal intubation (ETI), that is the absence of any of the following criteria within 14 days of admission: pH less than 7.20, pH 7.20 to 7.25 on 2 occasions 1 hour apart, hypercapnic coma, PaO 2 less than 45 mm Hg despite maximal tolerated FIO 2 , and cardiorespiratory arrest, or the achievement of previously established clinical goals (such as improvement of dyspnea) for the patients who refused or were not candidates for ETI.

Results: Ninety-one patients (47 women) with a mean (±SE) age of 75.8±10.4 years and Acute Physiology and Chronic Health Evaluation (APACHE) II score 3 of 19.6±5.7 were treated with facial mask bilevel positive airway pressure or continuous positive airway pressure, depending on the physician on-duty decision. Indications for NPPV were acute exacerbation of chronic obstructive pulmonary disease (COPD; 26), acute cardiogenic pulmonary edema (ACPE; 36), COPD with ACPE (7), pneumonia (13), or other forms of ARF (neuromuscular, obesity, postoperative atelectasis; 21). In the overall cohort, NPPV was successful in 56 (61.5%) of 91 trials. The overall average actual and predicted by APACHE II score mortality were 30% and 34%, respectively. Five (8.9%) of 56 patients with NPPV successes died (2 had an advanced lung cancer), whereas 23 (65.7%) of 35 patients with NPPV failures died, and 7 underwent ETI. Intermediate ICU and hospital lengths of stay of NPPV successes were 7.9±5.9 and 19.1±14.7 days, respectively, versus 7.8±6.3 and 15.8±13.9 days of NPPV failures (P=NS). NPPV was successful in 13 (50%) of 26 COPD patients, 29 (80.5%) of 36 cardiogenic pulmonary edema patients, 3 (75%) of 4 mixed COPD-ACPE, 6 (75%) of 8 pneumonia patients, and 11 (52.3%) of 21 of patients with other forms of ARF. After exclusion of the 4 patients with advanced disease, overall mortality was 27.9%. At entrance, mean PaCO 2 (79.1 versus 66.9 mm Hg; P=NS), pH (7.22 versus 7.22) and APACHE II score (21.5 versus 18.4; P=NS) were not better in NPPV responders than in nonresponders. Arterial blood gases response (pH and PaCO 2 ) within 2 to 6 hours of NPPV was not predictive of success.

Conclusion: Our results confirm the global efficacy of NPPV, but show lower survival and success rate of NPPV in "real life" in comparison with those in randomized controlled trials (except for ACPE), apparently because of a nonselected population of treated patients and a minor intensity of monitoring in a non-ICU setting. In addition, we found no significant predictors of treatment success, likely because of a high rate of "late failures" (28.5%).

Smith D, Bair AE, Lichty L/University of California-Davis School of Medicine, Sacramento, CA Study objectives: We studied out-of-hospital intubations performed by paramedics, which were later determined to be nontracheal on arrival to an urban, academic emergency department (ED). Our aim was to characterize the various confirmation techniques used among these unrecognized nontracheal intubations.

Methods: We conducted a retrospective review of the emergency medical services quality assurance database throughout a period of 65 months. Paramedic patient care reports and hospital medical records were compared with regard to techniques used for airway evaluation. Simple descriptive statistics are used to summarize the data.

Results: During this study period, paramedics intubated 1,643 patients. There ies have evaluated the best method to perform this procedure. We compare the accuracy and time to placement of a cricothyrotomy tube in the trachea by using the traditional cricothyrotomy technique versus a single incision approach, using a deer neck model. Methods: This study was performed at an academic emergency department (ED) with a 3-year emergency medicine residency and 24 residents. Postgraduate year (PGY)-1 and PGY-2 emergency medicine residents who had completed advanced trauma life support were eligible. A half-hour lecture describing both techniques was followed by a 1.5-hour procedure laboratory, during which the residents were able to practice these procedures on deer necks obtained from hunting refuse. The traditional cricothyrotomy involved making a vertical skin incision followed by dissection and a horizontal incision in the cricothyroid membrane. This single-incision method involved using a #11 blade scalpel to puncture the cricothyroid membrane after palpation of landmarks and then use the opposite dull end of the scalpel to dilate the incision for subsequent airway placement. After the procedure laboratory, each subject was asked to perform each method of cricothyrotomy on a deer neck. The time from the initial skin incision to placement of the tube in the trachea was measured. Correct placement of the tube was confirmed by cut-down on the trachea by an attending physician.

Results: Eight residents participated in this study. The median time for the residents to perform a single incision cricothyrotomy was significantly shorter than the time required to perform a traditional cricothyrotomy (16.5 versus 32.5 seconds; P=.0234). All tubes were placed correctly for each method.

Conclusion: Emergency medicine residents, using a deer neck model, perform a single-incision cricothyrotomy method significantly quicker and without any change in accuracy compared with a traditional cut-down cricothyrotomy.

Swanson ER, Fosnocht DE, Jensen SC/University of Utah Health Sciences Center, Salt Lake City, UT; Stanford University/Kaiser Permanente, Stanford, CA Study objectives: Etomidate and midazolam are commonly used sedatives for rapid sequence intubation (RSI) in the air medical setting. Midazolam use may be associated with a dose-related incidence of hypotension. We compare etomidate and midazolam as sedative agents for RSI in the air medical setting.

Methods: We conducted a retrospective review of a consecutive series of intubations at a university-based air medical transport program from January 1995 to December 2000. The records of all patients requiring intubation were reviewed. Exclusion criteria were patients not undergoing RSI, age younger than 18 years, and incomplete medical record data. Outcome measures included intubation success rate, incidence of hypotension, and percentage change in heart rate (HR) and systolic blood pressure (SBP).

Results: A total of 372 intubations were performed. RSI with etomidate or midazolam was used in 209 (56%) of 372 patients. Intubation success rate was 110 (98%) of 112 patients for etomidate and 96 (99%) of 97 patients for midazolam. Complete hemodynamic data were present for 74 (77%) of 96 of the adult patients receiving etomidate and 56 (73%) of 77 receiving midazolam. There was no difference in mean age, sex, initial SBP, or HR between the 2 groups. For the etomidate group, the mean dose was 24 mg, the mean percentage change in HR was -1% (95% confidence interval [CI] -6 to 4), and the mean percentage change in SBP was 2% (95% CI -3 to 7). In the midazolam group, the mean dose was 3.4 mg, the mean percentage change in HR was 1% (95% CI -5 to 7), and the mean percentage change in SBP was 3% (95% CI -3 to 9). There was no significant difference in the number of hypotensive episodes with etomidate (7/74) compared with midazolam (3/56) (Fisher exact P=.51).

Conclusion: Intubation success rate was high with etomidate (98%) and midazolam (99%). There was no significant mean percentage change in SBP or HR with either agent. We found a low incidence of hypotension with both agents, although the mean dose of midazolam used was considerably less than typically recommended for induction. Limitations include retrospective data collection and lack of complete hemodynamic data for 43 patients. 

insertion of the LMA (32.1 seconds) compared with the ETC (55.0 seconds) and the ETT (71.5 seconds). The differences in mean times between 0 and 6 months were 13.6 seconds (LMA), 29.6 seconds (ETC) and 31.6 seconds (ETT; P<.005). More students needed more than 1 attempt inserting the ETC (31 versus 8) and ETT (32 versus 17) but not the LMA (6 versus 4) at 6 months. There were significantly more esophageal intubations (20 versus 5) and more failures (6 versus 2) with the ETT at 6 months than the LMA (1 failure) and ETC (no failures) insertions. Conclusion: There is a significant deterioration in the speed and skills in securing the airway over time. Junior physicians should be taught the LMA as a rescue airway device in the event of failed endotracheal intubation.

Jones JS, Harris W, McNinch DE, Alexander C/Spectrum Health-Butterworth Campus, Michigan State University, Grand Rapids, MI Study objectives: The intubating laryngeal mask airway (ILMA) was introduced in 1997 to provide an improved conduit for blind endotracheal intubation, requiring little to no training. The purpose of this study was to compare 3 methods of tracheal intubation through the ILMA by inexperienced emergency physicians.

Methods: A prospective, cohort assessment of emergency medicine house staff and faculty performance was conducted throughout a 6-month study period by using an adult intubation mannequin. Tracheal intubations by each participant were randomly performed by using 3 insertion techniques: blind intubation through the ILMA, light-guided intubation using the ILMA and light wand, and intubation by using the ILMA with a flexible bougie or tube exchanger. The order of the 3 techniques for each participant was chosen by a computer-generated random numbers table. More than 3 attempts at intubation was regarded as failure. χ 2 and analysis of variance tests were used to compare intubation success rates, number of adjusting maneuvers, and the time required for successful intubation and ventilation.

Results: Forty emergency medicine house staff (postgraduate year 1 to 3) and attending physicians were tested; none of the participants had any previous clinical experience intubating through the ILMA. The overall intubation success rates were similar: blind intubation 88%, light wand 90%, and tube exchanger 93% (P=.76). However, the time to successful intubation varied significantly: blind intubation 66 seconds; light wand 74 seconds; and tube exchanger 54 seconds (P<.001). After completion of the study, the majority of participants (85%) preferred the tube exchanger as the best method of intubating through the ILMA.

Conclusion: These results suggest that the inexperienced physician should consider using a flexible tube exchanger for endotracheal intubation through the ILMA.

Kwon MK, Kim HK/The Catholic University of Korea, Seoul, Korea Study objectives: The study was conducted to evaluate the usefulness of a syringe aspiration technique (SAT) to correctly differentiate between esophageal and tracheal intubation in the emergency department (ED).

Methods: This study comprising a prospective case series was conducted in the ED of a university hospital. A 50-mL catheter-tip syringe was attached to the end of the tracheal tube after intubation. The syringe plunger was then withdrawn or aspirated with 50 mL of air. The ability to withdraw air into the syringe without resistance and without syringe plunger rebound was considered evidence of tracheal intubation. If resistance was met as the syringe was aspirated or if the syringe plunger rebounded after aspiration, esophageal intubation was assumed.

Results: Seventy patients were enrolled in the study. Tracheal intubation was performed in 46 patients, and esophageal intubation was performed in 24 patients. Of the 46 tracheal intubations, the SAT correctly indicated tracheal placement in 45 patients and was indeterminate in 1 case because of blockage of the endotracheal tube by gastric material resulting from aspiration. All 24 esophageal intubations were correctly identified.

Conclusion: This study would suggest that the SAT is a safe and effective device for determining endotracheal tube position in patients with a variety of clinical conditions requiring urgent airway control in the ED. DRM are perceived to be effective and practical, whether providers are satisfied with current DRM practice, and whether they would consider changing DRM practice. Methods: We used an anonymous survey of all registered nurses, physician assistants, nurse practitioners (NPs), and resident and attending physicians who work in an academic ED. Questions concerned demographics and DRM for NGTI. Descriptive statistics were calculated.

Results: Eighty-three (49%) of 171 participants returned the survey. Respondents comprised emergency medicine resident physicians (34% of participants), registered nurses (34%), emergency medicine attending physicians (22%), nurse practitioners (6%), and physician assistants (5%). Sixteen percent worked mostly with pediatric patients, 53% with adults, and 31% with both. Ninety-eight percent of respondents thought that awake patients find NGTI uncomfortable. Eighty-four percent used DRM; 26% believed that these measures were adequate. Forty-seven percent were dissatisfied with their current practice. Ninety-one percent would consider changing the way they perform NGTI. The most practical DRM were believed to be lidocaine jelly (30%), topical anesthetic spray (28%), and nebulized lidocaine (16%; multiple responses were allowed). The most effective DRM were felt to be anxiolytics (25%), nebulized lidocaine (22%), and topical anesthetic spray (18%). The DRM most commonly used in actual practice were lubricant jelly (76%), topical anesthetic spray (63%), lidocaine jelly (52%), and child life/distraction (23%).

Conclusion: Almost all ED staff agreed that patients who are awake find NGTI to be uncomfortable; 84% reported use of DRM. However, only 26% of participants believed that DRM were adequate; 47% were dissatisfied with current practice. The measures used most commonly were not necessarily those rated as most effective or practical. Ninety-one percent of participants were willing to change their practice. Improvement in DRM for NGTI is needed and desired by ED staff.

Study objectives: Many patients in the emergency department (ED) are perceived by their physician to be exhibiting drug-seeking behavior. Our goal was to determine whether patients who are perceived as drug seeking receive less pain relief while in the ED and if this perception may be related to ethnicity.

Methods: This was a prospective observational study of patients undergoing treatment in the ED for painful conditions. Before treatment and immediately before discharge, patients were asked to complete a 100-mm visual analog scale (VAS) representing their perceived pain. Physicians were asked to complete a 100-mm VAS representing their perception of the likelihood the patient was drug seeking. Data were analyzed by using descriptive statistics, analysis of variance, and multiple regression.

Results: Two hundred seventeen patients were enrolled (67% men; aged 39.2 years; 95% confidence interval [CI] 37.3 to 41.1). The ethnicity was 48.6% black, 38.3% white, 5.5% Hispanic, and 6.9% Native American. Mean drug-seeking VAS scores were 22.2 (95% CI 16.7 to 27.8) for black patients, 18.3 (95% CI 14.1 to 22.5) for white patients, 7.5 (95% CI 1.8 to 13.2) for Hispanic patients, and 30.4 (95% CI 13.2 to 47.6) for Native American patients (P=.06). Multiple regression revealed a strong (coefficient 26.8; P=.00) correlation between Native American ethnicity and the drug-seeking VAS score. The mean change in pain VAS for all patients was 17.4 (95% CI 14.3 to 20.5), 14.9 (95% CI 10.2 to 19.6) for black patients, 19.5 (95% CI 14.6 to 24.2) for white patients, 29.3 (95% CI 17.3 to 41.4) for Hispanic patients, and 8.5 (95% CI -2.4 to 19.5) for Native American patients (P=.07). Fiftythree (24.4%) of 217 patients had at least a 50% reduction in their VAS pain score. Patients with a 50% reduction had a mean drug-seeking VAS of 14.1 (95% CI 9.4 to 18.9); those without had a mean of 21.9 (95% CI 18.1 to 25.9; P=.03).

Conclusion: Physician perception of drug-seeking behavior was influenced by the ethnicity of the patient and was higher in Native American patients. This perception was higher in patients who did not receive relief from their pain. Differences in the changes in the VAS scores among ethnic groups were noted as well. It is possible that physician perception of drug-seeking behavior may be one of the causes of differences in pain treatment among ethnic groups.

Anesthesia Newton K, Copeland K, Ng M, Mabee J/Keck School of Medicine, University of Southern California, Los Angeles, CA Study objectives: We determine whether buffering lidocaine before subcutaneous infiltration affects the duration of action of lidocaine. experience using propofol for deep procedural sedation by emergency physicians in an urban emergency department (ED).

Methods: During the investigation of the first 100 patients to receive propofol for deep procedural sedation in our ED, a complication incidence of 6% was noted. The current study investigates the incidence of complications in the next 100 patients. Our theory was that the incidence of complications would decrease as familiarity with the drug increased. Data were collected prospectively on patients 101 to 200, who had undergone deep procedural sedation with propofol in our urban ED. Patients were given propofol in a titrated dose at the discretion of the emergency physician until the desired level of sedation was obtained. The drug was given without regard to time of last oral intake. Vital signs and pulse oximetry were recorded during the procedure and until recovery. The nurses recorded complications. Case data and complications were recorded into a database. Hypotension of systolic blood pressure less than 90 mm Hg, hypoxemia of pulse oximetry less than 90%, respiratory depression requiring assisted ventilation or intubation, aspiration, airway obstruction, or anesthesia consultations were considered complications.

Results: Throughout the study period, patients received propofol for orthopedic reduction (30%), lumbar puncture (36%), incision and drainage of abscess (22%), and other indications (12%). Mean total dose of propofol was 150 mg (range 20 to 600 mg). Patient ages ranged from 2 to 92 years. Transient hypotension (systolic blood pressure <90 mm Hg) occurred in 1 patient. The hypotension responded quickly to fluid bolus and did not recur. No patients developed transient hypoxia, required assisted ventilation, required intubation, aspirated, experienced airway obstruction, or required anesthesia consultation.

Conclusion: In the first 100 patients, 3% developed hypotension and 3% developed transient hypoxia. In the next 100 patients, there was only 1 complication, which was transient hypotension that responded quickly to fluid resuscitation, revealing a decrease in the complication rate from 6% to 1% and indicating that, as familiarity with propofol increased, the complication rate greatly improved. The most dreaded complications of aspiration or intubation have not occurred in the 200 patients studied to date.

Kelly KM, Dang CT, Phan V, Gorchynski J/University of California-Irvine Medical Center, Orange, CA Study objectives: There is ongoing controversy about the proper use of narcotic analgesia for patients presenting to the emergency department (ED) with subjective acute exacerbations of chronic pain. Are we treating pain or enabling addiction? We determine whether the presence of specific factors could be used to identify adults complaining of acute exacerbations of pain for suspected drug addiction, to estimate the percentage of drug addiction, to assess the physicians' ability to detect drug addiction, and to evaluate interrater reliability.

Methods: A drug abuse screening test (DAST-20) was administered to 76 ED patients who presented with acute exacerbations of chronic pain, with multiple ED visits for similar pain complaints, specific narcotic requests, or "allergies" to nonnarcotics and to 74 age-matched controls. Treating emergency physicians rated their suspicion for drug addiction with a visual analog scale (VAS).

Results: The overall estimation of drug addiction was 17.3% (26/150). Of the 76 analgesia patients, 21% (16/76) scored positive for drug addiction, and of the 74 control subjects, 13.5% (10/74) scored positive for drug addiction. Positive DAST-20 scores for drug addition was 43.8% (7/16) in patients with multiple visits, 43.8% (7/16) in patients requesting specific narcotics, and 6.3% (1/16) in patients with "allergies" to nonnarcotics. There is no correlation between VAS scores and the DAST-20 scores. There was a significant correlation between resident and attending VAS scores for their suspicion of drug addiction.

Conclusion: There exists a significant drug addiction problem among low-acuity patients presenting to the ED. Suspecting unproven drug addiction is a poor excuse for olioanalgesia.

Juhl GA, Conners GP/University of Rochester Medical Center, Rochester, NY Study objectives: Nasogastric tube insertion (NGTI) is a painful emergency department (ED) procedure. We sought to determine whether NGTI is believed to be uncomfortable by ED staff, whether discomfort-reducing measures (DRM) are used, whether Methods: A retrospective medical record review was performed at a university tertiary referral center ED of 68 ED visits of adult sickle cell patients treated throughout a 4-month period according to an ED-specific treatment guideline for pain related to sickle cell disease. Baseline data were obtained for 1 month before initiation of the guideline. The institutionally developed guideline stressed early pain assessment and analgesic administration with a fixed dose range of opioid analgesic (dilaudid or morphine) in opioid-tolerant or opioid-naive sickle cell patients. In addition, ketorolac was available for adjunct pain relief at the discretion of the emergency physician. Reassessment and treatment of pain occurred at scheduled intervals after each intervention. Patients with a temperature higher than 38.6°C (101.5°F), a SaO 2 less than 90% on room air, or hemoglobin level less than 7 gm/dL were excluded from guideline use and the review. Data abstracted from the charts included time to treatment, length of stay (LOS), documentation of pain assessment (visual analog scale [VAS]) at 3 time intervals (at triage and 2 reassessments), and opioid and adjunct analgesic usage.

Results: Of patients who received ketorolac (n=22), 66% were discharged to home compared with patients who did not receive ketorolac (n=48), of whom 35% were discharged to home (P=.025). Difference between means at baseline compared with study months for time to treatment (81 minutes versus 61, 102, 82, and 57 minutes) and LOS (270 minutes versus 212, 267, 255, and 232 minutes) demonstrated no statistical significance (Wilcoxon rank sum test). Cochran-Armitage trend test for documentation of VAS at 3 time intervals was not statistically significant.

Conclusion: After implementation of a sickle cell guideline, documentation of pain assessment (VAS), time to treatment, and LOS were not statistically improved. The addition of the nonsteroidal anti-inflammatory medication ketorolac to an opioid analgesic, however, had a positive effect in increasing the number of patients discharged to home. Limitations of this study include the inability to consistently differentiate opioid-tolerant and opioid-naive patients and the effect the delay in treatment may have had on the efficacy of pain management.

Bazuro RK, Frei SP, Bond WF, Sierzega GM, Richardson DM, Koenig KE, Belmont BJ/Lehigh Valley Hospital and Health Network, Allentown, PA Study objectives: The objective of this study was to describe the frequency of analgesia use in emergency department (ED) patients with acute appendicitis. The study examined frequency of pain medicine ordering by emergency physicians and private physicians who were treating patients in the ED. The study also compared how often pain medicine was ordered for patients who had received computed tomography (CT) scans versus those who did not.

Methods: This was a structured retrospective medical record review of patients with acute appendicitis confirmed by pathology from January 1999 to December 2001. The setting was a community teaching health network consisting of 3 EDs. Each record was reviewed for demographic data, type of treating physician, CT use, and the type and amount of analgesia that the patient received. Analgesia included primarily intravenous ketorolac and opiates. Patients were excluded if they did not have a confirmed diagnosis of appendicitis by pathology, were not treated in the ED, left against medical advice, or had an incomplete record. All data were collected and submitted for statistical review with χ 2 analysis. The analysis compared analgesia use by the type of treating physician and frequency of analgesia use in relationship to CT use.

Results: Six hundred twenty-six patients had confirmed appendicitis by pathology from 1999 to 2001. Of these patients, 253 (40%) received analgesia in some form while in the ED and 274 underwent CT (44%). We were able to detect a difference in the use of pain medicine between private physicians and emergency physicians. Of the privately treated patients, 29 (29%) of 99 received analgesia, and 224 (42%) of 527 patients treated by emergency physicians received analgesia (χ 2 =6.041; P=.014). Of the 274 patients who had a CT scan, 47% (n=128) received analgesia, whereas of the 352 patients who did not have a CT scan, 36% (n=125) received

Methods: A prospective, randomized, double-blind, crossover clinical trial was conducted by using 25 healthy adult volunteers. Two solutions were used for comparison, unbuffered 1% lidocaine and 1% lidocaine combined in a 9:1 mixture with sodium bicarbonate (44.5 mEq/L). The participants were injected with 0.5 mL of each solution in their forearms in random order by the same investigator blinded to the identity of the solutions. Each forearm was tested for sensation at 5-minute intervals until normal sensation returned.

Results: Three independent measures were compared: (1) total period of absolute anesthesia; (2) period of dysesthesia; and (3) return of normal sensation (absolute anesthesia plus dysesthesia). The use of buffered lidocaine was associated with a clinically significant increase in the duration of absolute anesthesia (53 minutes [95% confidence interval (CI) 40.6 to 65.4]) compared with nonbuffered lidocaine (41 minutes [95% CI 32.1 to 49.9]). When the time from injection to return of normal sensation (period of absolute anesthesia plus dysesthesia) was compared, there was a small but statistically insignificant difference between the 2 solutions (buffered lidocaine 86.4 minutes; 95% CI 74.5 to 98.3) compared with nonbuffered lidocaine (78.6 minutes; 95% CI 66.8 to 90.4).

Conclusion: The addition of sodium bicarbonate to lidocaine appears to increase the duration of absolute anesthesia but does not appear to affect duration of overall anesthesia.

Chang DJ, Desjardins PJ, Black P, Bird SR, Chen E, Petruschke RA, Mockoviak SH, Geba GP/Merck & Co., Inc., West Point, PA Study objectives: Opioids are commonly used in the treatment of acute pain. This study compared the analgesic efficacy of rofecoxib (ROF) with that of multidose oxycodone/acetaminophen (OXY/ACET) throughout 24 hours by using a validated model of acute pain.

Methods: Two hundred seventy-one patients experiencing moderate to severe pain after surgical extraction of 2 or more third molars were randomized to ROF 50 mg or OXY/ACET 10/650 mg, followed by 5/325 mg every 6 hours as required, or placebo during 24 hours. Patients provided assessments of pain at 18 prespecified times during the 24-hour period. The key endpoints were overall analgesic effects during 6 hours and 24 hours, measured by total pain relief (TOPAR) and sum of pain intensity difference (SPID), analyzed with analysis of variance with factors for treatment group, center, and baseline pain intensity. Patient global assessment, time to onset, peak pain relief, and opioid rescue medication use were also measured. Incidence of prespecified clinical adverse events (AEs) was evaluated with Fisher's exact test.

Results: The results are shown in the Table. Significantly fewer (P<.001) patients receiving ROF (47.9%) versus OXY/ACET (75.8%) experienced AEs. The most common AEs were nausea (19.0% versus 42.5%; P<.001), vomiting (9.9% versus 24.2%; P<.001), and dizziness (7.4% versus 31.7%; P<.001), respectively.

Conclusion: In this study, a single dose of ROF provided an analgesic effect at least as effective as a dose of OXY/ACET 10/650 mg during 6 hours and also multiple doses of OXY/ACET during 24 hours while resulting in fewer AEs. 2002, with a primary or secondary diagnosis of CHF. Patients who had a diagnosis of CHF and had a primary care provider in a clinic affiliated with our institution were included in the study. This analysis excluded patients who used the ED as their primary care provider. Major exclusions were patients with isolated right-sided heart failure, frequent visits with asthma or chronic obstructive pulmonary disease, and end-stage renal disease. The medical records of those patients who visited the ED more than 1 time during the study period and every third patient with 1 visit were reviewed for inclusion in the study. Data were collected about severity of CHF, number of admissions per year, number of inpatient days per year, and number of chest radiographs per year. In addition, we analyzed the chief complaints recorded in our ED database to identify patients whose ED visit was precipitated by running out of their medication. Two cohorts of patients were identified; one group had 1 or more visits to the ED because they had run out of medication (OOM), and the second group had no visits to the ED because of running out of medication. Results: Results are presented in the Table. Patients were comparable in the severity of their CHF.

Conclusion: For patients with CHF, barriers to access to medication, whether bureaucratic, financial, or related to transportation, increase the rate of ED use and the frequency of hospitalization. 

Flaherty JJ, Gillam TM/Evanston Northwestern Healthcare, Northwestern University Feinberg School of Medicine, Evanston, IL Study objectives: West Nile virus infection appeared diffusely in Illinois in 2002. Most cases were mild, but there were a total of more than 800 cases and 63 deaths. This number of confirmed cases was the highest in the nation and more than triple the number of deaths than any other state. The purpose of this study was to use a passive syndromic surveillance model to analyze emergency department (ED) patient chief complaints of fever and headache, flu symptoms, and viral syndrome and correlate these with known West Nile virus cases and the epidemic curve of confirmed cases in northern Illinois.

Methods: A passive syndromic surveillance system was set up by using a computerized patient log. Daily downloads of data in extensible markup language (XML) format from 3 regional hospitals with an annual census of 90,000 patients, linked in an existing network, were accumulated in a dedicated server using a structured query language (SQL) protocol throughout a 12-month period. A retrospective cohort study using SQL queries was run to search for chief complaints of fever and headache, flu symptoms, or viral syndrome. Positive matches were compiled in a graphic and geographic database. Comparative and descriptive statistical analysis was performed by using available data from public health databases.

Results: The 12-month output of the SQL queries revealed a biphasic distribution, with the first peak in February corresponding to the influenza A cases, and a second unexpected peak occurring in late August and September. A total of 362 cases were identified. Geocoding analysis and case numbers compiled in a frequency distribution matched the positive serology compiled by the state. Case analysis revealed that most of these patients were discharged, and no deaths occurred. IgM serology was positive in 5% of cases. Regression analysis and analysis of variance using SPSS showed no significant differences in distribution and a coefficient of determination of 0.67.

Conclusion: Passive syndromic surveillance systems can retrospectively detect analgesia (χ 2 =8.031; P=.005). Of patients who were treated by an emergency physician (n=530), 113 (49%) received analgesia of 232 patients who had a CT scan versus 111 (37%) receiving analgesia of 298 patients without a CT scan (χ 2 =7.019; P=.008). In the subgroup of 85 privately treated patients, 15 (33%) received analgesia of 45 who had a CT scan versus 14 (32%) patients receiving analgesia of 44 without a CT scan (χ 2 =0.0650; P=.420).

Conclusion: The percentage of appendicitis patients receiving analgesia remains relatively low overall, regardless of treating physician. Emergency physicians were more likely to use pain medicine than nonemergency physicians. Patients who underwent a CT scan were more likely to receive analgesia compared with those who did not. In particular, patients who were treated by emergency physicians and had a CT scan were more likely to have received pain medicine. In patients who were treated by private physicians and had a CT scan, no statistical difference was observed; however, there was a small sample size.

Nick JB, Lynch JM, Gelotte CK, Bowen DL, VanderBurg KC/McNeil Consumer and Specialty Pharmaceuticals, Fort Washington, PA Study objectives: In the emergency department (ED), acetaminophen (APAP) is the most frequently provided or recommended generic medication. We sought to establish the efficacy of APAP 500 mg and APAP 1,000 mg versus placebo in treating moderate to severe post-oral surgery pain and to compare the relative efficacy of APAP 1,000 mg with APAP 500 mg.

Methods: In this single-dose, double-blind, placebo-controlled study, 453 patients with moderate-to-severe pain (score ≥2 on a 0 to 4 categorical pain scale and a rating of ≥50 mm on a 100-mm visual analog scale [VAS]) after the surgical extraction of a minimum of 3 molars (with ≥1 partial or complete bony mandibular impaction) were randomized to receive a single dose of APAP 500 mg (n=200), APAP 1,000 mg (n=202), or placebo (n=51). After the recording of baseline pain intensity and the administration of study medication, patients evaluated pain intensity and pain relief relative to baseline by using 100-mm VAS at 30, 60, and 90 minutes and at 2, 3, 4, 5, and 6 hours after taking medication, with the sum of weighted pain relief scores over 4 hours (ie, total pain relief-4) as the primary efficacy evaluation. At the end of 6 hours or when rescue medication (including use of ice packs) was taken, the patient reported his or her impression of therapy.

Results: Eighty-six percent of patients had 4 or more teeth extracted, and 88.5% reported moderately severe or severe pain at baseline. Patients treated with APAP 1,000 mg had significantly higher total pain relief-4 scores compared with those treated with APAP 500 mg or placebo: 198, 140, and 70, respectively (all pairwise comparisons were significant; P<.01). In addition, 41.4% of patients taking APAP 1,000 mg rated treatment as better than APAP 500 mg over 4 hours, and 48% of the APAP 1,000 mg-treated patients rated treatment as very good/excellent versus 28% of patients in the APAP 500 mg-treated group and 15% in the placebo-treated group. During the 6-hour period after taking medication, 67% of placebo-treated patients required rescue medication or measures versus 48% of APAP 500 mg-treated and 33% of APAP 1,000 mg-treated subjects (all pairwise comparisons were significant; P<.01).

Conclusion: APAP 1,000 mg was a clinically and statistically superior analgesic compared with APAP 500 mg in this setting of moderate-to-severe pain after multiple molar extractions. Both active drugs were superior to placebo in providing analgesia.

27.2 white patients less than 48.2 black patients. From 1992 to 2000, NTCP rates increased from 27.4 to 31.4 per 1,000 US population (P=.009). Rates per 1,000 ED visits did not differ by sex or race: 85.9 men equal to 82.1 women; 85.0 white patients equal to 80.3 black patients. Overall admission for NTCP was 40%, stable over time (P=.10), but higher among men (43% versus 37%) and white patients (43% versus 28%).

Conclusion: ED visits for ACS and NTCP comprise a significant percentage of total ED visits. Although ACS rates were stable between 1992 and 2000, NTCP rates increased. Hospitalization rates for ACS were the same for women and men, but for chest pain, men were more likely to be hospitalized. Among black patients, hospitalization rates for ACS and chest pain were lower. Given the higher mortality rate in black patients from ACS, these findings have significant clinical and public health implications.

Clark S, Kim S, Hohrmann JL, Camargo CA/Massachusetts General Hospital, Boston, MA Study objectives: Obesity is a health problem affecting millions of Americans. Estimates for 2001 show that the prevalence of obesity in Massachusetts is 16.1%. Our objective was to determine the prevalence of obesity among adults presenting to 4 Massachusetts emergency departments (EDs).

Methods: For two 24-hour periods in July 2002, we interviewed patients aged at least 18 years and presenting to 4 EDs: 2 large urban, 1 small urban, and 1 small suburban. Exclusion criteria included severe illness and emotional disturbance (n=224). Body mass index (BMI) was calculated by using self-reported weight in kilograms divided by height in meters squared. Normal or underweight was defined as BMI less than 25.0 kg/m 2 ; overweight was defined as 25.0 to 29.9 kg/m 2 ; and obese was defined as at least 30.0 kg/m 2 . Data analysis used χ 2 and t test.

Results: Of 754 eligible patients, 530 (70%) were interviewed; 502 patients had BMI data and were included in this analysis. The median BMI was 26.1 kg/m 2 (interquartile range 23.0 to 30.0), or "overweight." Two hundred nine (42%) patients were underweight or normal weight, 172 (34%) were overweight, and 121 (24%) were obese. The proportion of obese patients did not differ across sites (P>.1). ED patients were significantly more obese than the Massachusetts general population (24% versus 16%; P<.001). Obese patients did not differ from nonobese patients according to most sociodemographic factors but were more likely to be black (22% versus 11%; P=.008) and to lack a high school education (64% versus 48%; P=.002). Obese patients were more likely to smoke (34% versus 24%; P=.02) and to report a previous diagnosis of hypertension (43% versus 28%; P=.001), diabetes (17% versus 11%; P=.08), and asthma (26% versus 15%; P=.01).

Conclusion: The prevalence of obesity in this ED sample is significantly higher than the obesity prevalence in the general Massachusetts population. ED patients with obesity experience conditions previously linked with obesity. Because many Americans, particularly in urban settings, use the ED as their main source of primary care, we encourage the development of ED-based strategies for obesity screening and referral to help address this national epidemic.

Richards ME, Fuity P/University of New Mexico School of Medicine, Albuquerque, NM; New Mexico Department of Health HIV Prevention, Santa Fe, NM Study objectives: In response to a rising heroin overdose death rate, the New Mexico Legislature passed the Opioid Antagonist Act in 2001. This act and its subsequent regulations, which were modeled after the public access defibrillation program, are designed to improve access to naloxone. The act provides immunity and authority for a person who is not a licensed health care professional to administer an opioid antagonist in a suspected opioid drug overdose. New Mexico is the only state with this type of legislation. As enabled by this law, the New Mexico Department of Health (NMDOH) sponsored a program to train heroin drug users in harm reduction, overdose recognition, use of 9-1-1, rescue breathing, and intramuscular naloxone administration. After completion of the training, participants are given a prescription for naloxone and supplied with an "overdose kit" that contains 3 prefilled syringes of 0.4 mg of naloxone, instructions, and personal protective equipment. This report describes the preliminary results of this unique program.

Methods: We used the NMDOH databases for program enrollment and followup. Enrollment data are supplied to the NMDOH by the health care practitioner

and identify cases compatible with West Nile virus infection. The system was able to detect an increase in syndromic presentations to the ED during a confirmed outbreak of West Nile virus. Further study is needed to quantify this effect, as well as encompass a larger regional area. Rapid and reliable serologic confirmation will also help in validation.

Pallin DJ, Kim S, Emond JA, Camargo CA/Brigham and Women's Hospital, Boston, MA; Massachusetts General Hospital, Boston, MA Study objectives: Streptococcus pneumoniae kills more than 40,000 people in the United States per year, more than all other vaccine-preventable diseases combined. Influenza is another major pathogen. Studies have found vaccination in the emergency department (ED) to be feasible and cost-effective. We measured pneumococcal vaccination (PV) and influenza vaccination (IV) of US adult ED patients throughout a 9-year period.

Methods: Data were obtained from the 1992 to 2000 National Hospital Ambulatory Medical Care Survey, a probability sample of ED visits. We identified all visits in which adults (≥18 years) were given PV or IV, with tetanus for comparison, and calculated national estimates based on known characteristics of the sample database. Ninety-five percent confidence intervals (CIs) were calculated using the relative SE of the estimate; data analysis used χ 2 .

Results: From 1992 to 2000, 37,000 PV and 247,000 IV were given during 266,000 ED visits (0.04% of all ED visits). Of these visits, 82% were nonurgent versus 51% of other visits (P<.001). For these visits, the mean age was 52 years (95% CI 47 to 56 years) versus 45 years (95% CI 44 to 45 years). In 77% of these visits, the chief complaint was request for vaccination, leaving only 13% of these visits in which patients with unrelated complaints were vaccinated. Vaccination with PV or IV was not associated with sex, race, or urban hospital but was higher in the Midwest compared with other regions (0.11% versus 0.02%; P<.001). In a multivariate model, independent predictors of PV or IV were older age and Midwest region. Tetanus toxoid was given at 25,912,000 visits (4.1% of all ED visits) and accounted for 93% of vaccinations.

Conclusion: S pneumoniae kills more people in the United States than all other vaccine-preventable diseases combined, but ED patients are rarely vaccinated against it or influenza, another major pathogen. Tetanus toxoid is given 700 times as frequently as the pneumococcal vaccine, despite the rarity of tetanus (<100 cases/year). Emergency physicians could save lives and money by vaccinating against pneumococcus and influenza.

Graham GN, Brown DF, Jones DML, Kim S, Emond JA, Camargo CA/Massachusetts General Hospital, Boston, MA Study objectives: Previous data have indicated that women and minorities receive less aggressive care for cardiac disease. In this study, we examine recent trends in the presentation and treatment of suspected acute coronary syndromes (ACS) and nontraumatic chest pain (NTCP) in US emergency departments (EDs) and subsequent rates of hospitalization.

Methods: Data were obtained from the 1992 to 2000 National Hospital Ambulatory Medical Care Survey (NHAMCS). ACS cases were defined by International Classification of Diseases, Ninth Revision (ICD-9) codes 410 to 411, whereas NTCP visits had NHAMCS visit codes 1050.0 to 1050.3 (without ICD-9 codes 800 to 999 or NHAMCS injury code). National estimates were obtained by using assigned patient visit weights, and 95% confidence intervals were calculated by using the relative standard analysis.

Results: From 1992 to 2000, there were 10,975,000 visits for ACS and 44,162,000 visits for NTCP (2.1% and 8.4% of all ED visits, respectively). The overall ACS rate per 1,000 US population was 7.1 and varied by sex but not race: 8.1 men more than 6.2 women; 7.4 white patients, 7.2 black patients. Rates were lower in the South and West (both 5.6) than Northeast and Midwest (8.8 and 9.2). From 1992 to 2000, overall rates were stable (P=.25). Rates per 1,000 ED visits varied by sex and race: 24.8 men more than 17.5 women; 23.1 white patients more than 12.0 black patients. The overall admission rate was 87%, stable over time (P=.12), with no difference by sex; however, white patients were more often admitted than black patients (86% versus 79%; P=.002). With respect to NTCP, there were 28.6 visits per 1,000 US population, and race pattern differed from ACS: 28.0 men equal to 29.1 women;

Methods: This was an epidemiologic analysis of 1991 to 2000 Fatal Analysis Reporting System (FARS) data of crashes involving child passenger deaths. This is a case series report. Study authors used SAS software for statistical analysis. Main outcome measures are driver, passenger, and crash characteristics (eg, age, driving history, restraint use, driver blood alcohol concentration, single versus multiple vehicle collisions).

Results: In 1991 to 2000, there were 17,381 child passenger deaths, with 4,519 (26%) involving a drinking driver (95% confidence interval [CI] 24% to 28%). Among drinking drivers involved with a child passenger death, more than 20% had a previously revoked or suspended license during the preceding 3 years, and 9% had at least 1 previous conviction for driving while intoxicated. Among drinking drivers transporting a child who died, one third (95% CI 31% to 35%) had an inactive or invalid license at the fatal crash.

Conclusion: Drinking drivers contribute to 1 in 4 child passenger deaths in the United States. Drivers with previous alcohol-related driving convictions who transport children represent a high-risk group and warrant aggressive public health interventions.

Methods: This institutional review board-approved, retrospective study included all adult (>18 years) patients with end-stage renal disease who presented to a public urban emergency department (ED) in a 36-month period between 1998 and 2002. Data were gathered from an electronic database, and a cost analysis was performed by using existing standards for ED care at this institution.

Results: The total number of patients identified by the search criteria was 735 patients who were responsible for a total of 4,496 visits. The number of patients with 20 or more visits was 54, and they accounted for 2,620 visits. The number of patients with 10 to 19 visits during the study period was 22. The number of patients with 4 to 9 visits was 78, and the number of patients with 3 visits, 2 visits, and 1 visit was 71, 109, and 401 people, respectively. The estimated minimum cost of a visit to the ED is approximately $150 at this institution; thus, the total minimum ED cost for the entire cohort during the 36-month study period was $674,400. The 54 patients with the most visits collectively accounted for ED costs of $442,050. These costs do not reflect costs for additional procedures, laboratory or radiographic investigations, and costs of dialysis sessions (which at our institution are performed in a dialysis suite with physical plant and personnel separate from the ED).

Conclusion: End-stage renal disease patients and patients with no funding commonly present to the ED for possible emergency dialysis therapy, which places an unnecessary drain on resources compared with scheduled (funded) dialysis. When this dilemma occurs, the identification of the subset of patients accounting for the excessive use of the ED must be made. After their identification, this small group of patients (54/735) who disproportionately use the ED in the hope of receiving compassionate dialysis would benefit from institutionally supported enrollment into a program of scheduled dialysis at a specialized dialysis center. This measure could be a cost-effective intervention by improving use of human resources (ie, nurses and physicians for the institution), as well as contributing to improved quality of life for these patients.

Miller AH, Jimenez CH, Pepe PE, Walters K, Thomas J/University of Texas Southwestern Medical Center, Dallas, TX Study objectives: Disproportionately increasing numbers of people with chronic diseases are seeking their maintenance health care in an emergency department (ED). This study analyzes factors that may begin to explain this evolving observation, particularly factors such as income level, health insurance status, and the designation of primary care providers (PCPs).

Methods: This was a prospective, consecutive enrollment survey of all eligible patients (>18 years) who had congestive heart failure, diabetes, or hypertension and presented to a large urban public hospital ED (>140,000 visits per year) throughout a 6-week period. Patients with severe acute decompensation (incoherent, unable to who writes the prescription for the naloxone. Participants who return to their health care practitioner for refills are interviewed by using a standardized form, and the data are supplied to the NMDOH.

Results: In the first 18 months of the program, a total of 205 participants were enrolled, with a distribution of 556 prefilled syringes of 0.4 mg of naloxone. Participants in the program were aged 18 to 64 years (mean 38.2; SD 10.5), and 72% were men. They were 24.4% non-Hispanic white, 60% Hispanic, 7.3% Native American, and 2.9% black. Most participants were referred to the program through participation in syringe exchange (65.9%) or by friends in the program (8.8%). There were 18 (8.7%) reported uses of naloxone to treat opioid overdoses with administration of 1 to 6 intramuscular doses (mean 1.61; SD 1.29) per overdose. In these 18 reported overdose cases, 5 (28%) people called 9-1-1, 13 (72.2%) provided rescue breathing, and 5 (27.7%) received medical care at an emergency department. In 10 (55.6%) of these cases, the naloxone was administered to a stranger, and in only 1 (5.6%) case was the naloxone administered to the person prescribed the medication. There were 4 additional cases in which rescue breathing was provided without administration of naloxone. There were no reported deaths or adverse outcomes.

Conclusion: Although infrequently used, the administration of naloxone by non-health care practitioners to victims of opioid overdose appears to be efficacious. In addition, rescue breathing was commonly used and used in cases in which naloxone was not administered. However, naloxone was rarely administered to the individual for whom it was prescribed, and notification of 9-1-1 was surprisingly low. A larger-scale trial with closer participant follow-up is needed to fully evaluate the impact of this program. Diseases, diagnosis codes and patient's chief complaint have been advocated for syndromic surveillance. Previously, we found only fair agreement between 2 existing algorithms for respiratory syndrome. The objective of this study was to determine whether the level of agreement could be improved by equalizing and expanding the syndromic definitions.

Methods: This was a retrospective analysis of a computerized database of emergency department (ED) visits in 15 New Jersey EDs. Participants were consecutive nontrauma patients treated by emergency physicians from July to November 2002. The New York State Department of Health extended the chief complaint algorithm originally developed by the New York City Department of Health. The ESSENCE (Efficacy and Safety of Subcutaneous Enoxaparin in Non-Q-Wave Coronary Events) project supplied the ICD-9 algorithm. We examined agreement between the 2 methods in 3 stages: (1) the initial chief complaint and ICD-9 algorithms; (2) after modifying the chief complaint algorithm to match the ICD-9 algorithm; and (3) after expanding both algorithms to include fever. We used the κ statistic to measure agreement. We also calculated sensitivity and specificity of chief complaint by using ICD-9 as the criterion standard.

Results: There were 174,520 patient visits in the database. For stages 1, 2, and 3, respectively, the κ statistics were 0.28, 0.42, and 0.56. The sensitivities were 0.31, 0.53, and 0.71, and the specificities were 0.94, 0.91, and 0.90.

Conclusion: Starting with 2 existing methods for biosurveillance of respiratory illness, agreement was improved by modifying the methods so that the syndromic definitions matched. Agreement was further improved by expanding the syndromic definitions to include fever.

Ramos PC, Crandall C/University of New Mexico, Albuquerque, NM Study objectives: Motor vehicle collisions are a leading cause of mortality for children older than 1 year in the United States. Drinking drivers contribute to 1 in 4 child passenger deaths; however, the role of drivers with previous alcohol-related convictions has not been well studied. mean±SD waiting time per incident was 32.7±21.2 minutes (range 15 to 405). There was a statistically significant trend toward increased number of hours ambulances were out of service during the study period (P=.0357), with the highest levels during the winter (January through March).

Conclusion: ED crowding has resulted in delays for paramedics waiting to transfer patients. This decrease in ambulance availability may have a significant impact on emergency medical services systems' abilities to provide timely response.

Perron AD, Custalow CB, Brady WJ, Johnson DM, Williamson AL, Fields MAH/Maine Medical Center, Portland, ME; University of Virginia, Charlottesville, VA Study objectives: Providing care at mass gathering events can be a daunting task that emergency physicians are frequently asked to perform. In 1999, our academic department of emergency medicine was asked to provide such care at a newly rebuilt football stadium with an enlarged crowd capacity of 61,500. During the past 4 football seasons, our experience has been that the number of patients we will care for during a game correlates closely with heat and humidity (heat index) at game time. The purpose of this study was to determine whether heat index can be used as a reliable predictor of the number of patients who will require care at a mass gathering event.

Methods: This was a retrospective review of all patient care provided at the stadium from 1999 to 2003 (4 years). The Division I college football stadium has a seating capacity of 61,500 patrons and is located in the southeastern United States. The football season runs from August to December. All patrons treated in 1 of 2 emergency care centers were included.

Results: A total of 20 football games occurred during the study period. The heat index ranged from 33 to 92. The number of patients cared for during a game ranged from 15 to 74, and stadium attendance ranged from 53,371 to 61,625. For each game, heat index was compared with the number of patients treated. To control for stadium attendance as a contributing factor to number of patients treated, the Pearson product moment correlation was calculated for each game. This statistical tool is used to determine whether there is a positive correlation between the heat index and the ratio of the number of patients cared for per 10,000 patrons at the game. The Pearson product moment correlation was calculated as 0.607, which indicates a strong positive correlation between heat index and patient volume (P<.005). Linear modeling predicts that for every 10°F increase in heat index, 3 more patients per 10,000 patrons will require care.

Conclusion: In this retrospective study, heat index could be used to predict the volume of patients who would be treated at a mass gathering event.

Milzman D, Ervin M, Wadwa A, Daniels F, Johnson E/Howard University, Washington, DC Study objective: We determine whether there is a difference in presentation for nursing home patients experiencing out-of-hospital cardiac arrest compared with all others and examine the ability of emergency medical services to use existing criteria to forgo cardiac resuscitation on nursing home patients.

Methods: A retrospective study of all nontraumatic cardiac arrests in Washington, DC, for a 4-year period was performed. Arrest characteristics included initial cardiac rhythm, location of arrest, and demographic data to determine whether nursing home patients present differently than other cardiac arrest patients. Direct comparison with t test and χ 2 were used to determine significance, with P less than .05.

Results: Throughout a 4-year period from 1999 to 2003, there were a total of 1,948 cardiac arrest cases initially called cardiac arrest by the first responder. On advanced life support (ALS) arrival, 4.5% of patients were discovered to have a pulse and never were in cardiac standstill. The mean patient age was 75.0 years; 79.6% of calls were for patients in a private residence, 11% in a public setting, and 10.4% in a nursing home facility. Of those cardiac arrest victims in nursing homes, there was an increased number of asystolic arrests, fewer patients pronounced dead on arrival, and reduced survivors (P<.05). Of the 234 nursing home patients in the study, the initial cardiac rhythm in 91% of arrest victims was asystole, whereas 26% of nonnursing home patients experiencing cardiac arrest in their home were not asystole (P<.05). Patients experiencing cardiac arrest in public were found with ventricular fibrillation 47% of the time versus 3.5% in nursing home patients (P<.02).

Conclusion: In this city-wide study, there is a definite misuse of ALS resources

speak, or triaged to receive immediate critical care) were excluded. Survey questions (n=39) were asked by 1 of 4 assigned medical students. Questions addressed demographics, income level, insurance status, point of access locations, how fast patients were able to be treated by PCPs (where applicable), and the frequency of health care access and hospitalizations. Results: Most of the 1,168 total patients studied sought health care providers frequently (nearly 40% doing so monthly), and one third of those with PCPs could get same-day appointments (more than half within a week). Overall, 42% of patients used the ED as their main source of care. Stratified by insurance status, this preferred use of the ED included 33% of the 784 insured and 63% of the 377 uninsured patients (P=<.0001; odds ratio [OR] 0.29; 95% confidence interval [CI] 0.22 to 0.38). Of 564 patients without designated PCPs, 82% preferably used the ED versus 18% of 604 patients with PCPs (P<.0001; OR 4.5; 95% CI 3.8 to 5.4). The number of hospitalizations was not affected by designated PCP or insurance status or by predisposition to use PCPs or the ED for maintenance health care.

Conclusion: Many patients with chronic illnesses are often seeking maintenance care in an ED, even those with insurance coverage and access to PCPs. Still, a lack of insurance or designated PCPs is significantly associated with this behavior. Nevertheless, rates of hospitalizations are not affected by the preferred use of the ED versus PCPs. Study objectives: There are no out-of-hospital predictors of neurologic outcome. This deficiency in monitoring the cerebral function of our cardiac arrest patients forces us to misuse precious time and limited resources on patients with no chance of meaningful survival. Our goal is to determine the utility of cerebral oximetry as a real-time noninvasive prognosticator for neurologic outcome in out-of-hospital cardiac arrest.

Methods: We proposed an observational prospective cohort study to assess the ability of cerebral oximetry to predict neurologic outcome in out-of-hospital cardiac arrest. Throughout a 13-month period, we enrolled patients from 1 of 3 local emergency medical services (EMS) stations. All participating EMS providers were retrained in the new advanced cardiac life support guidelines and use of the cerebral oximeter. Data were collected according to the Utstein criteria for out-of-hospital cardiac arrest, specifically focusing on correlates of neurologic outcome. Neurologic outcomes were categorized by the Glasgow Outcome Scale (GOS).

Results: Thirty-five patients (n=35) were enrolled into the study; of the 5 possible outcomes as identified by the GOS, only 2 outcomes were noted in our patient population: 34 deaths and one neurologically normal survivor. When a regional oxygen saturation level of 15% was used, the cerebral oximeter was able to predict the final neurologic outcome independent of return of spontaneous circulation. The positive predictive value and specificity of the cerebral oximeter in predicting death (GOS=1) was 100%.

Conclusion: Cerebral oximetry has the potential to be the first out-of-hospital independent predictor of neurologic outcome.

Eckstein M, Chan LS/Keck School of Medicine, University of Southern California, Los Angeles, CA; Los Angeles Fire Department, Los Angeles, CA Study objectives: We determine the impact of emergency department (ED) crowding on paramedic ambulance availability. This was a prospective longitudinal study from April 2001 through March 2002 in Los Angeles, CA. All incidents in which Los Angeles Fire Department ambulances were out of service for more than 15 minutes while waiting to transfer a patient because of the lack of open ED beds were captured and analyzed. Data included the total time each ambulance was out of service and the hospital where paramedics were waiting for an open gurney. Analysis was performed to determine weekly and seasonal variations and preponderance at various hospitals.

Results: There were a total of 21,240 incidents in which ambulances were out of service while waiting to transfer their patients to an open ED gurney, which accounted for 1 of every 8 transports. Of these, 8.4% were in excess of 1 hour. The Results: The response rate was 100% (56/56); unavailable data (NA) reflect EMSreported database corruption or inconsistent collection. Of counties with 1 ED (n=13), 46.2% (6) recorded diversion time as an outcome measure in 1995 and 2000 (all as zero hours for both periods). Among counties with 2 to 5 EDs (n=26), diversion status was tracked by 30.8% (8) in 1995 and 80.8% (17) in 2000 (7 and 6 as zero hours, respectively); mean nonzero diversion hours were undeterminable in 1995 (1 NA) and 247 hours (range 42 to 839 hours; 7 NA) in 2000. Counties with 6 to 10 EDs (n=10) had a diversion time recording rate of 50.0% (5) in 1995 (mean 3,103 hours; range 268 to 6,264 hours; 1 NA) and 80.0% (8) in 2000 (mean 8,668 hours; range 46 to 23,883 hours; 2 NA). The proportion of counties with more than 10 EDs (n=7) that tracked diversion time grew from 57.1% (4) in 1995 (mean 6,284 hours; range 876 to 11,691 hours; 2 NA) to 100% (7) in 2000 (mean 12,981 hours; range 2086 to 21,914 hours; 1 NA).

Conclusion: The proportion of California counties tracking ambulance diversion as an outcome variable increased 27% between 1995 and 2000, from 41% to 68%. Counties with 2 to 5 EDs accounted for most of this increase because at least half of counties with more than 6 EDs were already tracking ambulance diversion in 1995. Diversion time more than doubled between 1995 and 2000 in counties with 6 or more EDs.

Cooley C, Borton J, Ellis D/Erie County Medical Center, State University of New York at Buffalo, Buffalo, NY Study objectives: Current management of patients with acute coronary syndromes (ACS) presenting as chest pain, syncope, or shortness of breath requires timely ECG acquisition for appropriate treatment strategies. National guidelines have stressed obtaining ECGs within 10 minutes for possible ACS, which may be difficult in busy emergency departments (EDs). ECGs acquired by emergency medical technicians (EMT) in the field or by ED nurses at triage have demonstrated utility in detecting acute myocardial infarction and unstable angina. This study examines EMTs' attitudes about performing on-the-stretcher ECGs in the ED triage area on arrival to the hospital for patients with possible ACS. This is the first step in attempting to reduce the time required to treat patients with cardiac ischemia.

Methods: A 23-question voluntary anonymous survey was given to EMTs bringing patients to a tertiary care academic ED. EMTs were asked to use the following scale: 5=strongly agree, 4=agree, 3=neutral, 2=disagree, 1=strongly disagree. Statements assessed an EMT's belief about the need for ECG acquisition within 10 minutes of ED arrival in patients with possible ACS. Perceptions of waiting times to initial ECG in the ED were obtained. EMTs were questioned about their current ECG equipment and level of training in 12-lead ECGs. Willingness to learn about and perform 12-lead ECGs at triage was also assessed.

Results: One hundred forty-three EMTs responded to the survey. Age, sex, EMT level, and years of service were gathered. Eighty-six percent believed patients with possible ACS should receive an ECG within 10 minutes of arrival. Only 17% stated this actually happened. Fifty-two percent thought patients receive ECGs within 30 minutes of arrival. Sixty-two percent have no access to 12-lead ECG equipment at this time. Fifty-percent do not think they are capable of performing 12-lead ECGs, and only 41% have received training. Eighty-three percent state willingness to learn how to use out-of-hospital 12-lead ECGs. Seventy percent would be willing to perform ECGs at ED triage. Seventy-three percent could see benefit in being able to discuss the patient's ECG interpretation with an emergency physician. and out-of-hospital effort in near-futile resuscitation efforts on nursing home patients presenting in nonresuscitable rhythm. The primary focus of advanced directive implementation should be focused on these patients.

Goyal M, Spencer S, Adkins EJ, Sweeney TA, University of Maryland, Baltimore, MD; Christiana Care Health System, Newark, DE Study objective: Out-of-hospital endotracheal intubation (ETI) in cardiac arrest and comatose patients is well established as an advanced airway intervention in the out-of-hospital setting. A recent report indicates that up to 24% of patients intubated in the field have improperly located endotracheal tubes (ETTs) in a system with poor medical oversight. The objective of our study is to determine the rate of unrecognized improperly located endotracheal tubes after out-of-hospital ETI. We hypothesize that this rate is minimal in a system with active medical oversight.

Methods: This is a prospective observational study. The study group consists of patients undergoing out-of-hospital ETI who are then transported to 1 of 2 community-academic emergency departments (EDs). Interhospital transfers were excluded. Successful ETI was defined as the ETT placed into the trachea beyond the vocal cords. Each patient was evaluated by an emergency medicine resident or attending physician for proper ETT placement at arrival by direct laryngoscopy, end-tidal CO 2 , chest or abdomen auscultation, and chest radiograph. Statistical analysis was performed with 95% confidence intervals (CIs).

Results: Fifty-six patients who underwent out-of-hospital ETI were enrolled. Fifty-four (96%) of these patients were older than 12 years. The study group consisted of 34 (61%) male patients and 22 (39%) female patients. Fifty-four (96%; 95% CI 89% to 99%) patients were determined to have proper ETT placement, with 98% proper placement in patients older than 12 years. Of the 2 (4%) patients who had improperly located ETTs, 1 was esophageal and 1 was hypopharyngeal. The esophageal intubation was in a 4-year-old child. One case was excluded because independent review indicated that the ETT was possibly dislodged in the ED.

Conclusion: This study demonstrates a low rate of unrecognized improperly located ETTs. The data suggest that the aggressive use of secondary confirmation methods and active medical oversight may decrease the rate of unrecognized improperly located ETTs. However, the rate was not zero, which indicates a need for further effort to ensure proper ETT position.

Compton S, Dunne R, Wu J, Swor RA, Welch RD, Zalenski RJ/Wayne State University, Detroit, MI; William Beaumont Hospital, Detroit, MI Study objectives: Many large urban emergency medical services (EMS) systems have noted low out-of-hospital cardiac arrest (OOHCA) survival rates. These rates have been suggested to be low because of inadequate EMS performance; however, patient or community characteristics may limit potential survivability of OOHCA victims. We assess the potential survivability of OOHCA victims within a large, urban, all advanced life support (ALS) EMS system.

Methods: This was a preliminary analysis of a convenience sample of nontraumatic OOHCA patients transported by the Detroit Fire Department Division of EMS between January 2002 and September 2002. Cardiac arrest variables were captured as part of the ongoing Detroit Cardiac Arrest Registry. Key predictors of cardiac arrest survival, including witnessed arrest, initial cardiac rhythm, and bystander cardiopulmonary resuscitation, were collected. Descriptive statistics are provided.

Results: To date, there have been 371 OOHCA cases reviewed as part of the Detroit Cardiac Arrest Registry. Median age of patients was 65 years (interquartile range [IQR] 26 years), with 40.1% older than 70 years. Approximately half (49.0%) were men. Overall, 34% of arrests were witnessed, and bystander cardiopulmonary resuscitation was noted on EMS arrival in 23.7% of patients. Median time from ambulance call receipt to scene arrival was 8.6 minutes (±5.6 minutes). First recorded cardiac rhythms were asystole 71.2%, ventricular fibrillation/ventricular tachycardia 8.1%, other 12.1%, and none recorded 8.6%.

Conclusion: The potential impact of EMS interventions on survival rates from OOHCA in this large urban setting may be inherently limited because of patient characteristics. different according to patients' ethnicity, age, sex, or insurance status. Investigations of other patient and ED characteristics are warranted to further elucidate the presence of health care disparities in spinal immobilization.

Bracken M, Medzon R, Rathlev N, Hoffman J, Mower W/Boston Medical Center, Boston, MA; University of California at Los Angeles Medical Center, Los Angeles, CA Study objectives: We determine whether blunt trauma victims with evidence of intoxication are at increased risk of significant intracranial injury (ICI). The analysis is based on data from the National Emergency X-Radiography Utilization Study (NEXUS) II derivation study, which developed decision rule indications for head computed tomography (CT) scans in patients with blunt trauma.

Methods: This is a multicenter observational study involving 18 hospitals. Institutional review board approval was obtained. Prospective data were collected on all blunt trauma patients who had head CT. Intoxication was defined as a history of intoxication, positive toxicologic screen, or physical evidence suggesting intoxication. Comparisons between intoxicated and nonintoxicated patients were calculated using χ 2 analysis.

Results: Of the 13,728 patients enrolled in the study, 1,020 patients had ICI. Of injured patients, 22.6% (231/1,020) were intoxicated. The prevalence of ICI among intoxicated patients was 6.9% (231/3,356; 95% confidence interval [CI] 6.0% to 7.8%) versus 9.0% (789/8,752; 95% CI 8.4% to 9.6%) in nonintoxicated patients (χ 2 5.0; P=.02). However, intoxication was frequently associated with an altered level of alertness (ALOA). The prevalence of ALOA among intoxicated patients with ICI was 65% (151/227; 95% CI 60.0% to 72.6%), whereas ALOA among nonintoxicated patients with ICI was 53.4% (412/771; 95% CI 49.8% to 57.0%). Intoxication was also higher in patients with ALOA and no ICI than in ones with ALOA and ICI, 64.2% (1,235/1,925; 95% CI 62.0% to 66.3%) versus 41.0% (231/563; 95% CI 36.9% to 45.2%).

Conclusion: Intoxication was associated with a decrease in prevalence of ICI among imaged blunt trauma patients. Intoxicated patients are more likely than nonintoxicated patients to exhibit ALOA and may therefore be more difficult to assess clinically. The prevalence of ICI is lower in intoxicated than in nonintoxicated patients, probably because providers are more likely to obtain imaging in the former group. These findings require confirmation in a validation study.

Edwards J, Sloan EP/University of Illinois at Chicago, Chicago, IL Study objectives: The objective of this study was to examine emergency physicians' treatment of adult emergency department (ED) seizure patients.

Methods: Data analyzed in this study were from the National Hospital Ambulatory Medical Care Survey. Patient selection was based on International Classification of Diseases, Ninth Revision, Clinical Modification codes.

Results: Patients diagnosed with seizures accounted for 0.7% of US ED visits during 1997 to 2000, or an estimated 2.9 million visits. Patients presenting to the ED with seizures were most often white (66%) male (56%) patients, and the mean age was 44±17 years. Most patients received an antiepileptic drug (78%), and the most frequently administered antiepileptic drugs were phenytoin (38%), lorazepam (10%), carbamazepine (9%), phenobarbital (9%), diazepam (5%), fosphenytoin (4%), and gabapentin (2%). Overall, hydantoins were administered 2.5 times more often than were benzodiazepines (42% versus 16%). Hospitalization was required in 22% of seizure patients, with 3% being admitted to the ICU setting. Patients older than 64 years were 1.4 times more likely to be admitted (44% versus 18%; odds ratio 3.6; P=.001), patients receiving benzodiazepines in the ED were nearly 2 times more likely to be admitted (37% versus 19%; odds ratio 2.4; P=.001), and there was a trend suggesting that patients who received a hydantoin were more often able to avoid hospitalization (19% versus 24%; odds ratio 0.7; P=.09).

Conclusion: More than three quarters of ED seizure patients received an antiepileptic drug while in the ED, most commonly phenytoin. Patients older than 65 years and those who received a benzodiazepine were more likely to be admitted, and there was a trend for patients receiving phenytoin or fosphenytoin to avoid admission.

Jacobson EN, Stead LG, Weaver AL, Decker WW/Mayo Medical School, Rochester, MN Study objectives: We study the association of a prolonged corrected QT interval (QTc) in acute ischemic stroke.

Conclusion: EMTs are willing, with proper training, to perform ECGs at triage. By implementing new triage protocols, further research will look for time reductions that could lead to faster diagnosis and treatment of ACS.

Roberts I, Schierhout G, Wakai A/London School of Hygiene and Tropical Medicine, University of London, London, England; St. James's Hospital, Dublin, Ireland Methods: We assess the effects of different mannitol therapy regimens and of mannitol compared with other intracranial pressure (ICP)-lowering agents and quantify the effectiveness of mannitol administration given at other stages after acute traumatic brain injury. The review drew on the search strategy for the injuries group as a whole. We checked reference lists of trials and review articles and contacted authors of trials. Randomized trials of mannitol in patients with acute traumatic brain injury of any severity were selected. The comparison group could be placebo-controlled, no drug, different dose, or different drug. Trials in which the intervention was started more than 8 weeks after injury and crossover trials were excluded. The reviewers independently rated quality of allocation concealment and extracted the data. Relative risks (RRs) and 95% confidence intervals (CIs) were calculated for each trial on an intention-to-treat basis.

Results: Overall, there were few eligible trials. In the preoperative treatment of patients with acute intracranial hemorrhage, the administration of high-dose mannitol resulted in reduced mortality (RR 0.55; 95% CI 0.36 to 0.84) and reduced death and severe disability (RR 0.58; 95% CI 0.45 to 0.74) compared with conventionaldose mannitol. One trial compared ICP-directed therapy to "standard care" (RR for death 0.83; 95% CI 0.47 to 1.46). One trial compared mannitol with pentobarbital (RR for death 0.85; 95% CI 0.52 to 1.38). No trials compared mannitol with other ICP-lowering agents. One trial tested the effectiveness of out-of-hospital administration of mannitol against placebo (RR for death 1.75; 95% CI 0.48 to 6.38).

Conclusion: High-dose mannitol appears to be preferable to conventional-dose mannitol in the preoperative treatment of patients with acute intracranial hematomas. However, there is little evidence about the use of mannitol as a continuous infusion in patients who have raised ICP and do not have an operable intracranial hematoma. Mannitol therapy for raised ICP may have a beneficial effect on mortality compared with pentobarbital treatment. ICP-directed treatment shows a small beneficial effect compared with treatment directed by neurologic signs and physiologic indicators. There are insufficient data on the effectiveness of out-of-hospital administration of mannitol to preclude either a harmful or a beneficial effect on mortality.

Hoerneman BD, Crandall CS, Sklar DP/University of New Mexico Health Sciences Center, Albuquerque, NM Study objectives: Socioeconomic and ethnic/racial health disparities have been noted in the emergency medicine and general medical literature. Spinal immobilization is an extremely common out-of-hospital practice that places patients in a vulnerable position and subjects them to possible iatrogenic pain and injury. We sought to determine the presence of disparities in the duration of spinal immobilization in the emergency department (ED) according to patient characteristics.

Methods: At a Level I trauma center and teaching hospital, we conducted a prospective cohort study on a convenience sample of adult patients who were brought to the ED in a cervical collar on a rigid backboard. Unstable patients taken to the ED resuscitation rooms before removal of the backboard were excluded from the study. Physician assessment before removal and physician presence during removal from the backboard were not required. The main outcome measure was time spent on the backboard after arrival in the ED. Observational data including patient ethnicity, age, sex, and insurance status were collected.

Results: One hundred nine patients were enrolled in the study. Patient duration of time on the backboard in the ED ranged from 3 to 60 minutes, with mean time of 20.7 (SD 13.6) minutes and median time of 16 minutes. With Student's t test, no statistically significant difference in duration of immobilization in the ED was detected between white and nonwhite patients (P=.57), male and female patients (P=.93), and patients with and without insurance (P=.85). Additionally, by using simple linear regression, there was no detectable association between duration of time on a backboard and patient age (P=.72). The study had adequate power to detect a difference of 7.4 minutes or more in each group.

Conclusion Study objectives: In the "age of information," the current generation of emergency medicine applicants uses the World Wide Web extensively during their everyday endeavors. It is clear that today's medical students search for residency program information online and that a program's Web site has become a necessary part of residency recruitment. The objective of this survey was to explore the importance of emergency medicine residency program Web sites to applicants and to analyze which aspects of a program's Web site are most crucial.

Methods: This is a cross-sectional study of prospective residency applicants, generally fourth-year medical students, interviewing for a position in emergency medicine. One hundred forty-one applicants participated in this institutional review board-approved study by voluntarily completing a survey at 1 of 3 geographically separate emergency medicine residency programs in New York, North Carolina, and Michigan. The surveys were conducted anonymously, separate from the application process.

Results: The survey demonstrated that 97% of applicants researched programs on the Web by using information from the Fellowship and Residency Electronic Database Access (FREIDA) or the Society for Academic Emergency Medicine Web sites. When the first and second choices were tallied overall, 71% identified geography as the overriding factor that led to applying to a specific program. Three areas tied for the single most influential reason to make an application decision: program Web-site information, other online information, and mentor advice. However, each of these was ranked as the first choice by less than 50% of the survey respondents. The single most important aspect of a program's Web site according to the applicants was ease of navigation (36% of responders), followed by rotation schedule (29%), community information (7%), and a listing of current residents (6%). The applicants were also asked to identify the aspects of a Web site that were very important, moderately important, slightly important, and not important. Ease of navigation was clearly an expectation of the applicants, with 52% saying that the ease of navigation was very important (88% said that it was moderately or very important). A listed rotation schedule was very important to 49% and moderately important to 42%. Other aspects of the Web site that were considered very important or moderately important include a description of didactics (29% and 34%, respectively), a listing of elective opportunities (28% and 55%, respectively), benefit information other than salary (23% and 50%, respectively), emergency medical services opportunities (22% and 43%, respectively) and a listing of current residents (22% and 42%, respectively). Interestingly, the survey participants considered other items less important. A listed salary on the Web site was very important to 17% but unimportant to 10%. Program facility information (eg, the hospital, emergency department) was very important to only 16%, whereas 53% said it was moderately important. Graphics, faculty information (biographies), a listing of faculty and resident's publications, and research opportunities were factors considered less important. The quality of a program's Web site in general was moderately or very important in the application decision of 49% of students, although 13% said that a program's Web site was unimportant. The programs identified as having a best Web site, excluding the programs involved in the survey, were Denver (4%), Ohio State University (3%), Region's Hospital St. Paul, MN (2%), and Kalamazoo-Michigan State University (2%).

Conclusion: The Internet is an important source of information to the majority of applicants in emergency medicine. Geography ranks highest overall as the factor that determines where an applicant may choose to interview for an emergency medicine residency program. Online information, including the program Web sites, is a more important resource for guidance of the students than mentor or peer advice. The applicants identified ease of navigation and a rotation schedule as the most important aspects of a program's Web site. Almost all the survey participants used the emergency medicine residency program's Web site in their application process. However, only half the applicants believed that the quality of a program's Web site had a significant impact on their decision to apply to a particular program. An easily navigated, complete Web site may improve the recruitment of candidates to an emergency medicine residency program. 15, 2001, and June 30, 2002 , were examined. The QTc interval was obtained by computer analysis of the ED ECG, and the location of the stroke was categorized by vascular territory on head computed tomography or magnetic resonance imaging. A prolonged QTc was defined as more than 430 ms for men and more than 450 ms for women. Medical record review was standardized by a priori definition of the variables to be studied on a standardized spreadsheet for data abstraction. All consecutive patients in the mentioned period were captured; there were no missing records. The intervals on the ED ECG were compared with those on the previous (baseline) ECG. Factors that could contribute to QTc prolongation were also collected, including current medications and acute electrolyte abnormalities.

Results: Of the 206 consecutive patients presenting to the ED with signs and symptoms consistent with acute ischemic stroke, 196 had an ECG in the ED. Of these, 155 had previous (baseline) ECGs. Among these 155 patients, 48 (31%) had prolonged QTc at baseline. Of the remaining 107 patients, 25 patients had a "new" prolonged QTc in the ED. Of these 25, 6 were receiving medications or had electrolyte abnormalities known to cause QT prolongation, yielding a total of 19 of 107 (18%; 95% confidence interval [CI] 10.5% to 25.0%) patients with QTc prolongation attributed to acute brain ischemia. Although higher than the rate of 12% (95% CI 6.8% to 17.2%) reported in the literature for a control group with similar age and sex composition, the difference was not statistically significant (χ 2 test, P=.20).

Conclusion: There appears to be a definite association between prolonged QTc interval in patients with acute ischemic stroke, independent of previously existing ECG changes or medication regimen. Prolonged QTc appears to be 3 times more frequent in the acute stroke population at baseline and one and a half times more frequent during the acute event, although statistical significance could not be shown in this study to date. However, it does seem reasonable to suggest that the presence of a prolonged QTc may be used as a clue in the diagnostic armamentarium of the ED patient whose presentation of acute ischemic stroke may be clinically vague.

Stead LG, Jacobson EN, Weaver AL, Decker WW/Mayo Clinic, Rochester, MN Study objectives: We determine whether low diastolic blood pressure in the emergency department (ED) correlates with mortality in acute ischemic stroke.

Methods: The records of 206 consecutive patients presenting to the Saint Mary's Hospital ED for acute ischemic stroke between December 15, 2001, and June 30, 2002, were examined. Data on ED vital signs and dates of death and last follow-up were abstracted. All consecutive patients in the mentioned period were captured; there were no missing medical records. Patient survival was estimated with the Kaplan-Meier method, and survival curves were compared by using the log-rank test.

Results: Of the 206 patients studied, 40 died within a mean of 37 days after ED presentation for acute ischemic stroke. Patients with diastolic blood pressure less than 60 mm Hg were 3 times (95% confidence interval for relative risk, 1.4 to 6.5) more likely to die than patients with diastolic blood pressure more than 60 mm Hg (log-rank test, P=.004). In particular, at 6 months, the overall survival was 54.9% (SD 11.9) versus 81.6% (SD 3.1) for those with diastolic blood pressure less than 60 mm Hg versus at least 60 mm Hg, respectively (Table) .

Conclusion: Low ED diastolic blood pressure in the setting of acute ischemic stroke is correlated with an increased risk of mortality. 3 120 84.7 (2.8) 10 61.0 (11.6) 6 96 81.6 (3.1) 8 54.9 (11.9) 9 83 80.7 (3.2) 7 54.9 (11.9) 12 39 80.7 (3.2) 5 54.9 (11.9) * Four patients were missing a diastolic blood pressure measurement.

morbidity because of misinterpretation. We propose the establishment of an education program that includes cervical spine radiography, which has a high error rate for interpretation of simple radiography.

Sparano DM, Sites FD, Shofer FS, Hollander JE/University of Pennsylvania, Philadelphia, PA Study objectives: The Academic Associate (AA) program uses undergraduate students to collect data for clinical studies in the emergency department (ED). Students enrolled in a formal course assist with data collection by providing double coverage 16 hours a day, 7 days a week. Previous studies have shown that this system captures more than 90% of eligible patients. Methods to obtain closer to 100% enrollment are desired. We compared patient enrollment in 6 clinical studies when using shared coverage (24 ED rooms; 2 students share enrollment responsibility) with enrollment using split coverage (12 rooms each per student). This was a before-and-after trial at an urban university hospital.

Methods: During consecutive 15-day periods with the same 38 AAs, the daily ED census, hours of AA coverage, and enrollment in each of 6 studies were evaluated by a log completed prospectively in the ED. Data are presented with means and SD and compared with t tests.

Results: There was no difference between the shared and split enrollment periods with respect to hours of AA coverage (30.3 versus 30.7 h/d; P=.7) or average daily ED census (133.7 versus 141.8; P=.15). Patient enrollment in the 6 studies (2 National Institutes of Health; 1 foundation, 1 industry, 1 institutional-funded, and 1 nonfunded) were similar (P>.3 for all): chest pain study, 5.7 versus 6.6 patients enrolled a day; miscarriage study, 8.7 versus 8.2 patients referred a day; domestic violence survey, 44.3 versus 45.9 patients a day; violence intervention project for youths, 0.2 versus 0.2 patients a day; acupressure study, 0.27 versus 0.27 patients a day; depression and minor trauma, 1.2 versus 1.3 patients a day. After the 30-day trial, AAs were surveyed. Eighteen (75%) of 24 found the split strategy to be "more helpful in enrolling subjects." Twenty-two (92%) of 24 found that the split strategy helped them "keep better track" of patients. Eighteen (75%) of 24 recognized that they did not enroll more patients with the split strategy.

Conclusion: Study subject enrollment was not affected by the use of either shared or split responsibility strategy for recruitment. Students generally preferred the split strategy because it was more helpful and easier to monitor. Therefore, this option may be the best for similar student-oriented data collection programs.

Sparano DM, Shofer FS, Sites FD, Hollander JE/University of Pennsylvania, Philadelphia, PA Study objectives: Several emergency medicine programs are using premedical students as research assistants for clinical research in the emergency department (ED). These programs have been shown to enhance faculty research productivity, but the effects of these programs on the students' goals has not been assessed. We evaluated the effect of the Academic Associate (AA) program on medical school acceptance rates.

Methods: From 1997 to 2002, premedical students who attended informational sessions about the AA program completed a 28-item survey including demographic information, educational experiences, career goals, grade point average (GPA), and previous research participation. Premedical students who had a career goal of becoming a physician were included in the study if they have been accepted to or rejected from medical schools by April 2003. Determination of medical school acceptance was done by contacting students or their parents by telephone and email. Data for students enrolled in the AA program were compared with Association of American Medical Colleges (AAMC) data to determine whether they were more or less likely to be accepted into medical school; 95% confidence intervals (CIs) were used.

Results: During the study period, 149 students were enrolled in the AA program. They were a mean age of 22.5±3.1 years, 48% female students, and 39% nonwhite students. Most students were junior year or later with respect to educational level (93%). Forty-three percent of AAs had previous management experience. Twentyeight percent had previous database experience, and 39% had research experience. Previous research experience included being an author on an abstract (23%) or manuscript (16%) and presenting research at a scientific meeting (14%). The mean

mine whether these descriptors correlate with the number of students beginning training in an emergency medicine residency. Methods: A survey was mailed to the MSRDs at all 127 emergency medicine residency programs accredited by the Accreditation Council for Graduate Medical Education to assess characteristics of their academic position and involvement with medical student education. A repeated mailing and e-mail reminders were sent to programs not responding by 2 months. Data were analyzed by Mann-Whitney test or analysis of variance, as appropriate, to determine predictors of the percentage of students choosing an emergency medicine residency. Data are shown as the mean plus 95% confidence interval. Significance was indicated by P less than .05.

Results: Ninety-eight (77%) MSRDs responded with completed surveys. The average age of MSRDs was 39+1.5 years. Sixty-eight (69.4%) MSRDs were men. MSRDs spent an average of 9.3+1.4 years in emergency medicine practice and 4.3+0.8 years as MSRD. Eighty-nine (90.8%) MSRDs completed a residency in emergency medicine, 10 (10.2%) completed a second residency, and 11 (11.2%) had fellowship training. Sixty-nine (70.4%) MSRDs volunteered for the MSRD position. Eighty-five (86.7%) MSRDs were aware of and 58 (68.2%) were involved in an emergency medicine medical student interest group at their school. Sixty-nine (70.4%) MSRDs served as an advisor to an average of 6.3+1.3 students. Sixty-five (66.3%) MSRDs had protected academic time. The average amount of protected academic time was 25+3.4%. MSRDs also were involved in clinical supervision (99.0%), didactic instruction (88.8%), career counseling (82.7%), and rotation scheduling (67.3%). The percentage of students entering an emergency medicine residency program was 8.8%+1.2% for MSRDs who volunteered for the position versus 6.5%+1.0% for programs with an MSRD assigned to the position (P=.006).

Conclusion: Most MSRDs are residency trained in emergency medicine. Many MSRDs have protected academic time and interact with medical students beyond clinical teaching. The increased percentage of students choosing emergency medicine residencies from programs with volunteer versus assigned MSRDs may be related to a greater enthusiasm for medical student interaction.

Sim MS, Song KJ, Song HG/Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Korea Study objectives: The interpretation of simple radiography is an important training program in emergency residency. The accurate interpretation is important in diagnosis and treatment of emergency patients. We researched the accuracy and clinical significance of an emergency medicine resident's interpretation of simple radiography in order to develop an education program for emergency residents.

Methods: All simple radiography was interpreted for 1 month during February 2003 by 2 emergency medicine residents who were trained for 10 months. The result of interpretation was compared with the result of a formal radiography interpretation report. Clinical significance was determined on the deliberation of radiology staff and emergency medicine staff. Groups and subgroups were assigned according to the following criteria. Group N: The interpretations of the emergency medicine resident are correct. Group M: The interpretations of the emergency medicine resident are incorrect. Subgroup M0: There was no change in treatment or follow-up. Subgroup M1: Discrepancy has no clinical significance. Subgroup M1: Discrepancy is significant, with minor treatment or follow-up. Subgroup M2: Discrepancy is significant, with admission. Subgroup M3: Discrepancy is significant, with emergency treatment or bad outcome. Subgroup M4: Discrepancy is significant, with unnecessary further examination or treatment.

Results: A total of 371 radiographic studies were interpreted from 258 patients. The mean age of the patients was 45.8 years, and 157 were male patients and 101 were female patients. The 5 most frequent radiographs were of the chest, cervical spine, lumbosacral spine, simple abdomen, and the knee. Of 371 radiographic studies eligible for enrollment, group N was 341 studies (91.9%) and group M was 30 studies (8.1%). Subgroup M0 was 20 studies (5.4%), M1 was 6 studies (1.6%), M2 was 1 study (0.5%), M3 was 0 studies (0%), and M4 was 3 studies (0.8%). The error rate of cervical spine was higher than other radiographs, and it was statistically significant. There was no difference in the total error rate and the clinical significance between this result and the previous result of our study conducted in 2001 at the same hospital.

Conclusion: Limitations in our study included the small number of cases and small number of emergency medicine residents. However, there was no mortality or Conclusion: In a survey of fourth-year medical students applying to emergency medicine residency programs, we found that the dissemination of unfavorable information by directors, attending physicians, and residents about other residency programs during emergency medicine interviews is high. We found that, although these statements are consistently made during the interview process, they also occur outside of the formal portion of the interview. We also found that specific personal questions that should not be asked, as outlined by the Council of Residency Directors (CORD), were asked of applicants. The majority of these questions were asked during the formal portion of the interview. The emergency medicine interview plays a significant role in the creation of rank lists. We believe that consistent and fair behavior during this process is crucial. We propose that new national guidelines for regulating and monitoring interview practices in emergency medicine residency programs be developed by CORD.

Williams J, Williams T, Duggan S/Synergy Medical Education Alliance/Michigan State University, Saginaw, MI; St. Mary's Hospital, Michigan State University, Saginaw, MI Study objectives: In the communities where there are sexual assault nurse examiners (SANEs), sexual assault victims are generally taken to see these nurses for a forensic examination and collection of evidence. Because of this procedure, emergency medicine residents are having limited exposure to these types of patients. In a previous study, it was found that the majority of emergency medicine residents are not confident in caring for these patients, and their educational experience was limited. The objective of this study was to determine the degree of interaction, if any, between SANE programs and emergency medicine residency programs.

Methods: An institutional review board-approved questionnaire was sent to 28 SANE/Sexual Assault Response Team (SART) programs identified in the United States that have an emergency medicine residency program in their community. The questionnaire included issues about the number of examinations performed and the interactions the SANE/SART programs have in training emergency medicine residents.

Results: In our study, we had a 78% response rate, with 22 of the 28 SANE/SART programs responding. The majority of examinations (60%) were performed in the emergency department, with an average of 315 examinations performed per year and a range of 60 to 1,000 examinations per year. A total of 6,913 examinations annually were performed at the 22 locations. Only 36% (8 of 22 programs) have a formalized educational program with the emergency medicine residents that ranges from lectures to the residents observing and performing the sexual assault examination. Emergency medicine residents were involved in 5% of the total number of sexual assault examinations performed by the SANEs. All of the programs that are not currently involved in the training of emergency medicine residents indicated that they would be interested in collaborating with the residents in their educational process.

Conclusion: This study shows that a large number of sexual assault victims are being treated by SANEs, with minimal involvement of emergency medicine residents. There is little formalized collaboration between the 2 programs. With their extensive training and experience, SANEs are a significant and valuable resource that should be used in educating emergency medicine residents. The surveyed programs indicated that they would be interested in establishing formalized educational programs. Study objectives: The American College of Emergency Physicians (ACEP) Academic Affairs Committee conducted a survey assessing how ACEP could support the academic mission of emergency medicine and evaluate whether greater collaboration between the major organizations in emergency medicine is needed.

Methods: A 57-question survey was mailed to program directors of the 121 accredited emergency medicine residency programs. Eighty-three (69%) surveys were returned. The survey gathered demographic information and identified respondents' views on topics ranging from graduate medical education to unification of the specialty.

Results: Sixty-eight (82%) respondents were interested in ACEP developing pub-GPA for AAs was 3.41 (95% CI 3.36 to 3.46). Medical school acceptance rates were 83% for students in the AA program (95% CI 77% to 89%), which is higher than expected according to AAMC-published data, with a 49% (95% CI 48.5% to 49.5%) acceptance rate for students with a mean GPA of 3.46. Conclusion: Compared with published AAMC data, data for AA students demonstrate that they have a higher medical school acceptance rate despite a lower GPA, which suggests that participation in the AA program may compensate for a lower GPA.

Weinstein J, Williams E, Bania T, Ortiz C, Sullivan L/St. Luke's-Roosevelt Hospital Center, New York, NY Study objectives: Medical students applying to emergency medicine residencies frequently report that during some emergency medicine interviews, unfavorable information is disseminated about other emergency medicine residency programs. Additionally, discriminatory questions about family planning and marital status have also been asked. Although previous surveys on this subject have been limited to 1 geographic region, this survey seeks to include applicants from medical schools across the country that are interviewing at emergency medicine programs nationwide. We determine the prevalence of unfavorable information that is disseminated about other emergency medicine programs during emergency medicine interviews throughout the United States. We determine the prevalence of discriminatory questions during these interviews. We determine by whom these questions and statements are made and whether they occur during the formal interview.

Methods: We conducted a survey of applicants who interviewed at emergency medicine residency programs across the country. Applicants were invited to visit a Web site posting the questionnaire and to submit responses anonymously over the Internet. Invitations to participate in the survey were sent by e-mail after the conclusion of the interview season and before the deadline for match list submission. Applicants were asked to recall whether they were told unfavorable information about another program or whether they were asked discriminatory questions during any of their emergency medicine interviews during the 2002 to 2003 interview season. In addition, applicants were asked to identify whether these questions and statements were made by a residency director, attending physician, or member of the house staff and to specify whether it occurred during the formal portion of the interview.

Results: In total, 182 applicants participated in the survey. Fifty-three percent (96/182) of applicants were told that postgraduate year (PGY)-2 to PGY-4 and PGY-1 to PGY-4 graduates are better trained. Thirty-two percent (58/182) of applicants were told that another program's residents are unhappy.Twenty-five percent (45/182) of applicants were told during an interview that PGY-1 to PGY-3 graduates have difficulty finding jobs. Thirty-four percent (62/182) of applicants were told that another program was not accredited or was on probation.

Twenty-two percent (40/182) of applicants were told that another program's residents are not well trained. Nine percent (17/182) of applicants were told that another program's attending physicians are unhappy. Twelve percent (22/182) of applicants were told other negative statements about another program. Seven percent (12/182) of applicants were told another program's attending physicians are not well trained. Forty percent (69/182) of applicants were asked their marital status. Nineteen percent (33/182) of applicants were asked their first choice of the match list. Ten percent (17/182) of applicants were asked about family planning issues. A few (0.6%; 1/182) applicants were asked about sexual orientation. Four percent (8/182) of applicants did not submit responses about discriminatory questions. Of all applicants who were told negative statements about another program (352 "yes" responses about negative statements by all interviewers), 39% (137/352) of applicants report statements made by a director, 43% (153/352) report statements made by an attending physician, and 70% (246/352) report statements made by house staff. None of the negative statements by directors, attending physicians, or house staff were made during the formal interview. Of all applicants who were asked discriminatory questions (151 "yes" responses to discriminatory questions asked by all interviewers), 55% (83/151) of applicants report questions asked by a director, 77% (117/151) report questions asked by an attending physician, and 53% (80/151) report questions asked by house staff. Of all discriminatory questions asked by all interviewers, 81% (227/280) were asked during the formal interview.

to code blues to treat critically ill patients and supervise advanced cardiac life support (ACLS) measures. Many hospitals use some combination of resident physicians or hospitalists to direct inhospital cardiopulmonary resuscitation code teams. However, this practice has not been scientifically evaluated. Which physician staffing model is best for inhospital cardiopulmonary resuscitation: emergency medicine residents or staff inpatient hospitalist physicians?

Methods: This was a prospective blinded observational study in a 600-bed community-based teaching hospital in Georgia. All adult inhospital code blue activations that occurred between June 1999 and July 2001 were included. Code blue activations occurring in the emergency department, operating rooms, cardiac catheterization suites, or by emergency medical services activation were excluded. Primary outcomes were patient survival, first physician response to code team activation, total time for primary attending physician to arrive, and Utstein standard intervals from code blue activation until start of compressions, first defibrillation, first ACLS drug, endotracheal intubation, arrival of first physician, and arrival of attending physician. Secondary outcomes were compliance with ACLS guidelines, code team leadership effectiveness, and surrogate markers of a "clinically aggressive style": endotracheal intubation performed by the directing physician, use of highdose epinephrine, duration of code, and use of invasive procedures.

Results: There were 749 adult cardiac arrests during the study period, of which 151 were excluded by protocol as above. The patient's primary attending physician took an average 29.8 minutes to arrive (SD 20.3 minutes; range 0 to 54 minutes). Emergency medicine residents directed 288 codes; hospitalists directed 248 codes; other specialists directed the remaining 62 codes. Baseline patient characteristics between the 2 specialties were similar. There was no difference in immediate survival (emergency medicine residents 61.1% versus hospitalists 58.5%; P=.13). Emergency medicine residents responded quickest for 59.2% of the codes compared with hospitalists arriving first, 28.0% (P<.05). Compliance with ACLS guidelines and code team leadership effectiveness was comparably high in both groups. Residents performed only 19.3% (versus hospitalists 27.4%; P=.18) of all potential endotracheal intubations, delegating the remainder to respiratory therapists. There was no statistical difference among other outcomes.

Conclusion: Because the primary attending physician's arrival may be extensively delayed, hospitals should predesignate physician staffing for code teams. In our study, the resident staffing model produced quicker physician responses to code blue activations than did the hospitalist model. The choice of physician staffing model did not appear to influence mortality or the timeliness of initial ACLS measures. Resident and hospitalist clinical performances were similar in directing code blue teams. Emergency medicine residents gain valuable clinical experience from leadership of code blue teams; they should be encouraged to perform more of their own endotracheal intubations to maximize the potential training benefit.

Sadosty AT, Goyal DG, Stead LG, Weaver AL, Boie ET, Decker WW/Mayo Clinic, Rochester, MN Study objectives: To date, no study in emergency medicine has addressed the educational utility of the residency review committee for emergency medicine's requirement for patient follow-up. The authors examine whether the act of performing patient follow-up improves emergency medicine resident education.

Methods: This is an institutional review board-approved prospective study examining all emergency medicine resident follow-up encounters from September 25, 2001, through September 24, 2002. Emergency medicine residents from a regional tertiary referral emergency department (ED) initiated patient follow-up encounters by entering information about the patient's initial ED presentation into a secure Web-based follow-up system and later, after researching the patient's subsequent inpatient or outpatient course, entered follow-up information. Before followup entry completion, the resident was asked to enter whether he or she judged the specific follow-up encounter to be educational (yes/no). The completed entry was then routed electronically to the faculty member who originally supervised the care of that patient. Blinded to the resident's response about educational utility, the faculty member reviewed the completed follow-up entry and independently evaluated whether he or she thought the follow-up encounter was educational for the resident (yes/no). Data entered into the Web-based follow-up system were then summarized as percentages.

Results: During the 1-year study period, 847 follow-up encounters were completed by 18 emergency medicine residents and 29 emergency medicine faculty. Ninety- lications. Seventy (84%) respondents stated that ACEP should take the lead role in lobbying for graduate medical education funding of emergency medicine residency training on the national and state levels. An overwhelming majority of respondents stated that ACEP should continue to support its current workshops and expand its class offerings at these workshops. Eighty-one (98%) respondents stated that ACEP should continue or expand its role in supporting emergency medicine research.

Responses were mixed about the establishment of a national emergency medicine procedure database to standardize procedure tracking. Seventy-six (92%) respondents thought it important to teach evidence-based medicine to emergency medicine residents. Sixty-two (75%) respondents thought that a bank of evidencebased practice guidelines should be developed. Eighty-two (99%) respondents agreed that the major emergency medicine organizations should work more closely together in solving problems facing the specialty. Conclusion: In conclusion, (1) existing products need to be publicized better;

(2) ACEP should take the lead in political lobbying; (3) ACEP should continue to support research activities and systematic reviews; and (4) ACEP must work with other emergency medicine organizations to achieve these goals.

Boston WS, Hafner Jr JW, Aldag JC/University of Illinois Chicago, College of Medicine, OSF Saint Francis Medical Center, Peoria, IL Study objectives: Chief resident (CR) physicians play an important role in most US emergency medicine residency programs. There is, however, limited published information describing the duties and training of emergency medicine CRs. This abstract describes US emergency medicine CRs' demographics, preparatory training, job responsibilities, departmental influence, benefits, and career goals.

Methods: Each of the 125 emergency medicine residency programs approved by the residency review committee of emergency medicine in the fall of 2002 was mailed a survey questionnaire for the program director (PD) and each CR. The survey used a Likert scale (with 1 being most positive and 5 being most negative), multiple choice, and free-answer questions. Two mailings were used, and a reminder e-mail was sent to the CR by a list-serve e-mail list. All data were recorded in an Excel 97 spreadsheet and analyzed with SPSS 11.0. χ 2 Tests, Student's t tests, and Mann-Whitney U tests were used to compare nominal, ordinal, and interval variables, respectively.

Results: CRs from 69% and PDs from 59% of the total emergency medicine residency programs returned surveys. Of the CRs responding, 42% reported receiving a written job description, and 58% of the PDs reported offering one (P=.02). Of the CRs, 8% reported undergoing local formal job training, whereas 22% of the PDs reported offering local formal job training (P=.004). Sixty-nine percent of the CRs and 76% of the PDs report sending CRs to outside formal training. CRs who received written job descriptions tended to think they were better prepared for their job (mean Likert scale 2.22 versus 2.54; P=.06). CRs (mean Likert scale 2.26 versus 2.73; P=.02) and PDs (Likert scale 2.05 versus 2.59; P=.004) agreed that CRs who received formal outside training were better prepared for the job. PDs and CRs agree on the level of responsibility assigned to CRs (2.64 versus 2.74; P=.18). PDs think CRs are more influential than do CRs about influence in residency program (2.35 versus 1.91; P=.003) and in emergency department (3.05 versus 2.59; P=.001) issues. The most common benefits offered CRs were increased salary and fewer shifts. The majority of CRs would recommend the job to fellow residents. Most CRs state that they plan to practice in a mixed academic and community medicine setting or a community medical setting and believe that the position is beneficial to their career.

Conclusion: PDs and CRs report different levels of training being offered to CRs; however, both agree that more training improves effectiveness of the position. Additionally, CRs do not think they are as influential in emergency department and residency program issues as their PDs believe they are, suggesting that there may be a need for improved communication between the CRs and PDs. In general, most CRs believe that being CR is a beneficial experience and would recommend the position to other residents.

Adams BD/Medical College of Georgia, Augusta, GA * * Dr. Adams is currently affiliated with Brooke Army Medical Center, Fort Sam Houston, TX.

Study objectives: "Code blues," or inhospital cardiopulmonary resuscitations, occur frequently and unexpectedly. Qualified physicians should promptly respond 23 in the right anterior position, 37 in the right posterior position, and 6 in the posterior commissural position. ETH excision revealed a large thrombus 86 times, microthrombi 7 times, and collected blood 4 times. Neither a slipped incision nor late bleeding was noted. We encountered a single minimal persistent bleed at the beginning of the experience with multiple ETH. Three (4.5%) patients with multiple ETH more than 4 cm needed intravenous opiates and were hospitalized, and 1 required further radical surgery. Sixty-five (97%) patients passed stools within 24 hours. Pain disappeared in less than 48 hours among 59 (88.1%) patients. Wounds healed at day 5.

Conclusion: The technique is based on the placement of traction sutures, which pulled out the hidden part of ETH. Suturing the wound after excision en bloc of the recent, edematous, or multiple ETH resulted in the disappearance of pain and in wound healing in 5 days without postoperative bleeding.

Sherman SC, Taggart A, Khatoon A, Snyder E, Galich P/Cook County Hospital, Chicago, IL Study objectives: Rewarming a frostbitten extremity is best if performed in a water bath of 104°F to 107.6°F (40°C to 42°C) and can be destructive to tissue if the temperature of the water bath is 113°F (45°C) or greater. Our objective was to determine the most reliable and efficient way to produce a waterbath of 104°F to 107.6°F (40°C to 42°C) for the immersion of a frostbitten extremity.

Methods: We tested 2 microwaves (Panasonic NN-S732WL, 1,200 W, and Emerson MW8102SS, 1,100 W) to determine the appropriate heating time on the highest setting to warm a 4-L water bath to 104°F to 107.6°F (40°C to 42°C). The reliability of each microwave was tested in 30 subsequent trials and at 2 starting temperatures. Alternatively, the ability of physicians to create a water bath in the desired temperature range by using hot and cold faucets from a standard hospital sink was evaluated with (n=30) and without (n=30) a thermometer.

Results: Using the 1,200-W Panasonic microwave and a 5-minute 30-second heating time was found to produce a temperature closest to the desired range of 104 to 107.6°F (40°C to 42°C). Thirty trials with this microwave on high for 5.5 minutes heated a 4-L water bath to 105.8±1.6°F, with 80% (95% confidence interval [CI] 66% to 94%) of the trials in the desired temperature range. The 1,100-W Emerson microwave (4 L for 5.5 minutes) created a final temperature of 104.3±0.7°F, with 63% (95% CI 46% to 80%) in the desired range. If the starting temperature was less by 2°F (71.5 versus 69.3°F), the water bath heated to only 102.6±1.4°F, with 20% (95% CI 34% to 56%) of the trials in the desired temperature range. No final temperatures were considered harmful to frostbitten tissue (113°F). During the physician trials at a hospital sink, all trials took less than 2 minutes to complete. Without a thermometer, temperatures ranged from 103.6±6.2°F, with only 20% (95% CI 34% to 56%) falling in the range of 104°F to 107.6°F and an additional 10% (95% CI 0% to 21%) that were dangerously high (113°F). With the use of a thermometer, the final temperatures averaged 105.7±1.2°F, with 93% (95% CI 84% to 100%) in the desired temperature range, and no dangerously high temperatures were recorded.

Conclusion: Microwave heating of a water bath for a frostbitten extremity is influenced more by the starting temperature than the type of microwave. In addition, microwave heating is more time consuming than using a conventional hospital sink. The best method is to use a conventional sink with a thermometer to check the final temperature because this process can be accomplished in less than 2 minutes in almost all circumstances. A physician will create a water bath that risks increased tissue injury 10% (95% CI 0% to 21%) of the time if a thermometer is not used, and therefore this method should be avoided.

Hopkins J, Mangolds VB, Cukor J, Darling C, Dickson EW/University of Massachusetts Medical School, Worcester, MA Study objectives: Increasingly, simple wounds are being closed with a variety of suture-free techniques, including tissue glues, staples, and adjustable tissue-closure devices. Although these wound-closure systems offer the clinician an assortment of treatment options, little is known about the relative procedural pain, infection rates, initial patient satisfaction, and long-term patient scar contentment for the particular devices.

Methods: Using a series of 100-mm visual analog scales and questionnaires administered at wound closure, as well as at 1 month and 3 months after closure, we three percent of the follow-up entries were deemed by at least 1 evaluator (resident or faculty) to be educational for the resident. The residents found the act of performing follow-up educational in 81.3% of cases, whereas faculty believed that 80.4% of follow-up cases were educational for the resident. A comparison of resident and faculty assessments revealed that in 68.6% of cases, emergency medicine residents and faculty agreed that the act of performing follow-up was educational.

Conclusion: Although the assessment of educational utility is subjective, our study indicates that emergency medicine residents and faculty find that the act of performing patient follow-up is educational for emergency medicine residents. Methods: We retrospectively analyzed the cohort of emergency department (ED) patients receiving CK and TnI throughout a 4-month period, and correlated assay results with final outcomes as determined by the consulting cardiologists. CK more than 250 U/L with CKI (CKI=CK MB/CK×100) more than 2.5% or TnI more than 1.5 ng/mL were the enzymatic definitions of MI. Odds ratios were calculated to determine the effect of renal insufficiency (creatine >1.5 mg/dL) on TnI. Test utility is presented as sensitivity, specificity, positive predictive value, negative predictive value, and accuracy.

Results: Of 2,289 patients with ED TnI and CK, 2,251 had concordance of TnI and CK/CKI: 2,235 (97%) negative and 16 (0.7%) positive in both assays. Discordant values were noted in 38 patients; 8 (0.4%) patients were CK/CKI positive and TnI negative, and none were diagnosed with MI. Thirty (1.3%) patients were TnI positive and CK/CKI negative, and all were diagnosed with myocardial injury. Compared with presenting TnI, CK/CKI had a sensitivity of 43%, specificity of 99%, positive predictive value of 63%, negative predictive value of 99%, and accuracy of 99%. One thousand five hundred twenty-seven patients whose initial cardiac assay results were negative proceeded to serial TnI and CK determinations. There was concordance in 1,483 patients: 1,477 (97%) negative and 6 (0.4%) positive in both assays. Discordance was present in 44; 3 (0.2%) were CK/CKI positive and TnI negative, and none were diagnosed with MI. Forty-one (2.7%) patients were serial TnI positive and CK/CKI negative; all had myocardial injury. Overall, TnI positivity indicative of myocardial injury was present in 5.1% of the cohort; initial ED TnI was positive in 2.1% of patients. Renal dysfunction did not affect TnI (odds ratio 0.9; P=.7). The elimination of CK/CKI from the ED cardiac evaluation profile would save $933,644 annually in patient charges.

Conclusion: TnI distinguished all cases of myocardial injury; addition of CK/CKI provided no additional diagnostic benefit, and its elimination will result in important cost savings.

Shoukry K, Bruno R/University Hospital Pitie-Salpetriere, Paris, France Study objectives: Simple excision of external thrombosed hemorrhoid (ETH) produces a bleeding perianal zone and delays wound healing. Delimitation of the excision by traction sutures extrudes the ETH and prevents extension of incision to anal mucosa. Suturing the wound reduces bleeding and seems to speed healing.

Methods: During this open prospective study over a 24-month period, we noted the number, site, size, edema, and ulceration of ETH. Polyglactine 910 (rapid Vicryl) 3/0, mounted on a 22-mm half-circle visiblack laser needle, was used for traction and suture. Traction sutures were placed on the perianal skin and on the pectinate line of each ETH. Pulling traction sutures extruded ETH, thus delimiting the zone to excise and easing suturing. Patients were observed for 1 hour postoperatively for bleeding, received recommendations and prescription, and were evaluated at the fifth postoperative day.

Results: Sixty-seven consecutive patients aged 38±13 years were included, 42 men and 25 women. ETH were recent (<72 hours) among 36 (53.7%) patients. We operated on 97 EHT, of which 51 (52.6%) were edematous and 13 (13.4%) ulcerated. Multiple ETH was seen among 24 (35.8%) patients. ETH sizes were 22 (22.7%) 1 cm, 45 (46.4%) 2 cm, 21 (21.6%) 3 cm, and 9 (9.3%) more than 4 cm. ETH was located 27 times in the left lateral position, 4 in the anterior commissural position, that emergency physicians overemphasize the quality of symptoms when attempting to diagnose actual dizzy patients, thereby placing such patients at risk for misdiagnosis.

Methods: After institutional review board approval, we conducted a retrospective medical record review of a sample of patients coming to an urban, tertiary care ED with a triage complaint of dizziness. From our ED triage database, we identified 1,144 patients (aged 18 to 98 years) who had complaints compatible with our operational definition of dizziness throughout a 1-year period (July 1999 to June 2000). Among these patients, there were 632 whose complaint was coded as "dizzy," "dizziness," or "vertigo"; 482 patients coded as "lightheaded," "presyncope," "faint," or "syncope"; and 30 patients coded as "ataxia," "unsteady gait," or "off balance." We reviewed 92 records selected at random to determine whether charted records demonstrated a bias toward recording qualitative aspects of dizziness. A single, unmasked reviewer (neuro-otologist) abstracted 5 elements of history recorded about each patient: date or time of first symptoms, quality of dizziness, associated pain, triggers for dizziness, and episode duration. The specialist reviewer then attempted to arrive at a tentative diagnosis according to the complete charted ED record and compared it with the final ED diagnosis.

Results: Of the 43% (40/92) of records that documented at least 1 of the 5 historical attributes, 90% documented the date or time of first symptoms, 70% documented the quality of the dizziness, 50% documented the presence or absence of pain, 30% documented the presence or absence of triggers, and only 13% documented the episode duration. In 30% of records, no mention was made of pain, triggers, or episode duration. Although there were insufficient data in any record to provide a definitive specialist diagnosis, tentative diagnoses were possible in 48% (44/92) of records according to available information. Among these diagnoses, there was discordance between specialist and attending emergency physician diagnosis in 39% (17/44). Although most of these suspected misdiagnoses were likely of minor clinical significance (eg, a diagnosis of benign positional vertigo in a patient likely to have vestibular migraine), 7% (3/44) of charted records suggested probable missed cerebrovascular events, translating to approximately 1 missed transient ischemic attack or stroke per week in our hospital ED.

Conclusion: Although it is impossible to discern with certainty an emergency physician's diagnostic approach from a retrospective record review, some inferences about the process may be drawn from the information documented (and not documented) in records. The quality of dizziness was frequently charted (70%), whereas the duration of dizziness was infrequently charted (13%). From these data, it appears that the quality of dizziness is given more diagnostic weight by emergency physicians than other attributes of the patient's medical history, most notably, episode duration. However, insights gained through vestibular research throughout the past 30 years have established that episode duration, triggers, and presence or absence of pain usually provide diagnostic information far more important to triage decisions than symptom quality. In conjunction with our other preliminary studies demonstrating a correlation between physician misperceptions and outdated information in ED texts, we believe our results point toward a system-level flaw in the way we train emergency physicians to diagnose dizzy patients, which could be responsible for a substantial number of misdiagnoses of significant clinical impact. Prospective, masked studies examining diagnosis of ED dizzy patients are required to confirm these suspicions.

Pereira E, Richard O, Clero J, Moro J, Richard CD, Caussanel JM, Cantineau JP, Lambert Y/Versailles Hospital, Le Chesnay, France Study objectives: Early hemodynamic resuscitation of severe trauma patients is based on clinical findings such as pulse rate and arterial pressure, which reflect partially the hemodynamic status of the patients. Transesophageal echo Doppler (TED) is currently one of the most promising noninvasive techniques for monitoring cardiac function in the early stage. The purpose of this study was to describe the hemodynamic profiles (HP) obtained and to evaluate the therapeutic impact of this new tool in early treatment.

Methods: This was a preliminary prospective study including 19 patients. The setting time to obtain an echo Doppler signal and its quality and stability were studied. HP were analyzed with mean arterial pressure (MAP), aortic blood flow (ABF) was determined by a HEMOSONIC 100 (Arrow), and total systemic vascular resistances (TSVRa) in aortic circuit were calculated. A Wilcoxon signed rank test was used to compare MAP, ABF, and TSVRa variations in patients for whom the treat-

evaluated patient and clinician satisfaction with either suturing or an adjustable tissue-closure devise (ClozeX) in the setting of extremity laceration repair. Results: Clinicians reported that wound closure took an average of 15+1.5 minutes when sutured and 3+0.5 minutes when an adjustable device was applied, which mimicked the patient's perception of closure time and was directly linked to an improvement in patient satisfaction with the initial closure. Not surprisingly, patients reported less pain when the wound was closed with the adjustable closure device compared with the suture group, 7.3+2.4 versus 31+16, respectively. There was no significant difference in wound satisfaction scores between the 2 groups. There was no significant difference in infection rates between the 2 groups (<2% in both groups).

Conclusion: Overall, patients were satisfied with scar formation in extremity wounds, regardless of the closure technique applied. However, wounds closed with an adjustable tissue restraint device caused less pain and took less time than those closed by suturing.

Richards ME, McLaughlin SA, Hubble MW/University of New Mexico School of Medicine, Albuquerque, NM; Western Carolina University, Cullowhee, NC Study objectives: Pulse oximetry is the primary tool to monitor respiratory status in hypoventilating patients. Two common examples include acute narcotic overdose and overdose during procedural sedation. In these cases, the hypoxia is secondary to decreased respiratory drive. Treatment with supplemental oxygen alone will correct the hypoxia but not the hypercapnia. These patients may have reassuring oxygen saturation levels while experiencing significant hypercapnia secondary to their hypoventilation. The objective was to develop a mathematical model of the hypoventilating patient by using established physiologic formulas to estimate the effect of supplemental oxygen.

Methods: A computer model of the hypoventilating patient was developed with Microsoft Visual Studio 6.0. The model uses well-established mathematical descriptions of the respiratory system, with the alveolar gas equation as the principal driving formula. The model includes the following variables: fraction of inspired oxygen (FIO 2 ), atmospheric pressure, respiratory quotient, and alveolar ventilation. The model calculates values for the following: partial pressure of arterial carbon dioxide (PaCO 2 ), partial pressure of arterial oxygen, bicarbonate level, pH, and oxygen saturation (SaO 2 ). The model assumes a single respiratory acid base abnormality. We estimated the theoretical effect of maintaining a hypoventilating, 70-kg, 170-cm, male patient at a target oxygen saturation of 90% by using various levels of supplemental oxygen.

Results: A patient requiring supplemental oxygen by nasal cannula with a FIO 2 of 24% to maintain an SaO 2 of 90% will have an alveolar ventilation of 1.98 L/minute, a PaCO 2 of 81 mm Hg, and a pH of 7.16. A patient requiring supplemental oxygen by simple mask with a FIO 2 of 50% to maintain an SaO 2 of 90% will have an alveolar ventilation of 0.71 L/minute, a PaCO 2 of 227 mm Hg, and a pH of 6.90. Conclusion: Our model demonstrates that even small amounts of supplemental oxygen in a hypoventilating patient with normal oxygenation capacity can produce clinically reassuring oxygen saturation levels despite significant hypercapnia and respiratory acidosis. Although maintaining appropriate oxygenation is critical, it is also important to recognize that the patient is at continued risk for significant physiologic changes secondary to hypercapnia if alveolar ventilation is not also restored. We conclude that appropriate monitoring of these patients should include a measurement of ventilation (respiratory rate and depth or end-tidal CO 2 ), and interventions for hypoxia should also aim to correct the underlying hypoventilation.

Newman-Toker DE/Johns Hopkins University, Baltimore, MD Study objectives: Dizziness is a common chief complaint in the emergency department (ED) and poses a significant diagnostic challenge because of the myriad illnesses that may cause this symptom. Preliminary research using paper-and-pencil tests suggests that emergency physicians harbor misperceptions about the bedside evaluation of hypothetical dizzy patients. These misperceptions relate principally to an overemphasis on the qualitative, rather than temporal, features of dizziness as a chief complaint and appear to derive from antiquated information presented in emergency medicine textbooks. The purpose of this study was to test the hypothesis

S 8 1 hyperglycemia of 250 mg/L in the emergency department setting. We have compared sensitivity, specificity, and predictive value of ketonuria and ketonemia for the diagnosis of ketoacidosis (blood bicarbonates <20 mmol/L; anion gap >16 mEq/L) in a sample of patients for whom levels of ketone bodies in blood and urine, as well as serum electrolytes, were available. Results: We included 355 hyperglycemic patients. The median time between arrival and dipstick testing was 21 minutes and more than 2 hours in more than 10% of cases. Comparison between ketonuria and ketonemia was performed in 173 patients. Ketonuria equal to or less than 1 cross or a β-OHB value lower than 3 mmol/L enabled ketoacidosis to be excluded (negative predictive value 100%). At 2cross cutoff point for ketonuria and at 3 mmol/L cutoff point for ketonemia, the 2 tests had the same sensitivity (100%), but the specificity of β-OHB (93.9%) was significantly higher (P<.0001) than that of ketonuria (77.3%). The best positive predictive value for ketonemia was obtained at the 5 mmol/L cutoff point (100%) and for ketonuria at the 3-cross cutoff point (25.7%).

Conclusion: The measurement of β-OHB in capillary blood is faster and more effective than the use of dipsticks in the urine to detect ketoacidosis.

Stromski C, Allmon D, Fagerlie S, Kasarda D, Laurora M, Reed J/St. Luke's Hospital, Bethlehem, PA Study objectives: Recent studies have suggested that the urine dipstick can accurately identify patients with renal insufficiency in the emergency department (ED) who present with a hypertensive crisis or are in need of an abdominal computed tomography scan. The purpose of this study was to determine the sensitivity of the urine dipstick in the evaluation of renal insufficiency in an unselected ED population.

Methods: A retrospective chart review was conducted in an academic ED with 55,000 annual visits. Consecutive patients who presented to the ED and had a creatinine level greater than or equal to 1.5 mg/dL were selected. Demographic data were collected on a standardized form. Urine dipsticks were performed with standard technique in the ED by the nursing or ancillary staff and noted on the chart. Urine dipstick results were also used if performed within 2 days of hospital admission. Any amount of proteinuria or hematuria was considered a positive result in the detection of renal insufficiency, which was defined as a creatinine level greater than or equal to 1.5 mg/dL. Overall sensitivity was calculated by using the entire data set and excluding patients with a history of renal insufficiency. α was set at .05.

Results: A total of 859 consecutive patients were included in the study. Urine dipstick results were available for 453 (53%) patients. There were no statistical differences between patients with and without documented urine dipstick results with regard to creatinine level, age, sex, history of renal insufficiency, systolic and diastolic blood pressure, and number of preexisting conditions (0.996 less than P<.051). Overall sensitivity of the urine dipstick in detecting renal insufficiency was 74.6%. Sensitivity remained essentially unchanged at 75.1% when patients with a history of renal insufficiency were removed from the analysis. Sensitivity improved to 87.5% as creatinine level increased to more than or equal to 3.0 mg/dL (P<.019), but the 95% confidence interval would still miss as many as 23.2% of cases.

Conclusion: The urine dipstick is not reliable enough to be a predictor of renal insufficiency in unselected patients presenting to the ED.

Costa M, O'Dunne P, Ong M, Jacoby JL, Reed J/St. Luke's Hospital, Bethlehem, PA Study objectives: There are few data about the presenting signs and symptoms of emergency department (ED) patients ultimately diagnosed with pulmonary embolism (PE). The purpose of this study was to examine the presenting complaint and vital signs of such patients and to determine the frequency with which the diagnosis of PE was made in the ED.

Methods: This was a retrospective medical record review of patients discharged after hospitalization with a primary diagnosis of PE throughout a 1-year period in an active community teaching hospital. A diagnosis of PE was considered valid by using the following diagnostic criteria: positive angiography or computed tomography (CT) scan, high probability ventilation/perfusion ( V /Q) scan, and intermediate probabil-ityV/Q scan with positive Doppler results. We did not attempt to determine the level of clinical suspicion or pretest probability for any study. ment adjustment was guided by the TED. A P value less than .05 was considered significant.

Results: The sex ratio (male:female) was 15:4. The mean age was 37.6 years (+17.8). The mean ISS was 35+23. The median time from injury to HP obtainment was 75 min (interquartile range 60 to 90 minutes). The mean duration of procedure was 6+2 minutes. In all measurements, the signal quality was good and stable during the treatment.

Patients were classified according to initial MAP with a threshold of 90 mm Hg: group 1 MAP more than 90 mm Hg (n=8); ABF more than 2.5 L/min (n=4) (TSVRa >1,400; n=3); ABF less than 2.5 L/min, n=4 (TSVRa <1,400; n=4); and group 2 MAP less than 90 mm Hg (n=11); TSVRa more than 1,400 dyne·s/cm -5 , n=6 (ABF <2.5; n=4); TSVRa less than 1,400 dyne·s/cm -5 , n=5 (ABF <2.5; n=2). For 9 patients (47%; 95 confidence interval 25% to 69%) TED helped to detect specific HP: low ABF (<2.5 L/min) with normal MAP for 5 patients (occult hypoperfusion) and low TSVRa (<1,400 dyn·s/cm -5 )with low MAP for 4 patients (vasoplegic profile). For 11 patients, the treatment (fluid loading or aminergic support) was adjusted with TED data, in real time, during the early treatment. For these patients, the MAP variation (60 to 72 mm Hg) and the ABF variation (2.5 to 3.6 L/min -1 ) were significant (P=.004 and .036, respectively). TSVRa variation was not significant (P=.28).

Conclusion: Early resuscitation in severe trauma patients that is guided by continuous TED monitoring may be useful to minimize tissue ischemia and its consequences.

Lopez BL, Cogen J/Thomas Jefferson University Hospital, Philadelphia, PA Study objectives: Acute, painful vaso-occlusive crisis (VOC) is a condition of localized tissue ischemia and hypoxia and is the most common emergency department (ED) presentation of sickle cell anemia. Pulse oximetry is a noninvasive test commonly used in the ED to assess oxygenation status. Our previous study in 13 adult VOC patients found that pulse oximetry underestimated measured arterial blood gas oxygenation and that a correction factor might be needed. To our knowledge, this is the only ED study examining pulse oximetry in VOC. For this study, we hypothesize that pulse oximetry cannot predict the ED disposition (admission or discharge) of adult VOC subjects.

Methods: We performed a retrospective medical record review of all patients who were older than 18 years, had documented homozygous sickle cell (SS) or heterozygous (SC) disease, presented to the ED with VOC as a primary or secondary diagnosis, and had a pulse oximetry reading obtained in the ED. Historical and physical examination information, laboratory testing, chest radiographs, disposition, and final diagnoses were recorded. Patients were divided into a normal group (pulse oximetry >95%) and a low group (pulse oximetry <95%).

Results: In the normal group (n=48), the mean pulse oximetry value was 97.68%+1.7%; the low group was 92.57%+1.74%. There was no significant difference between normal and low groups with regard to patient disposition (P>.05, χ 2 ); pulse oximetry reading could not predict admission or discharge. In our sample, 11 in the normal and 9 in the low groups had a chest radiograph done, all of which were read as no new infiltrate. There was a trend toward lower hemoglobin levels in the low group (8.55+2.05 g/dL normal versus 6.94+2.08 g/dL low; P=.07).

Conclusion: Pulse oximetry may not be able to predict ED disposition of VOC patients. Low pulse oximetry readings can be found even with no significant chest radiograph findings. Lower pulse oximetry readings may be due to lower hemoglobin content in VOC subjects. Our future study will involve the accrual of larger numbers in this retrospective record review and will ultimately involve a prospective study examining the pulse oximetry reading in the VOC clinical decisionmaking process.

Taboulet P, Haas L, Porcher R, Manamani J, Fontaine JP, Feugeas JP, Gautier JF/ Hopital Saint-Louis, University of Paris, Paris, France Study objectives: We have compared the semiquantitative measurement of acetoacetate by using urinary dipsticks with the bedside quantitative finger-prick measurement of the principal ketone bodies 3-β-hydroxybutyrate (β-OHB), for the diagnosis of ketoacidosis.

Methods: This is a 1-year retrospective study of patients who presented with during office hours (9 AM to 3 PM) throughout a 3-month period. The setting was an urban, academic tertiary hospital ED (75,000 annual visits) in which most patients have a PCP. Patients aged at least 18 years and who were triaged to an intermediateacuity section of the ED were included. Study patients were interviewed by using a survey instrument during their ED visit. The primary analysis was descriptive. Univariate odds ratios (ORs) with 95% confidence intervals (CIs) were calculated, and multivariate logistic regression was used to control for demographic factors and simultaneously assess effects of independent variables. Results: A total of 302 patients were approached; the study set comprised the 268 (88.7%) who had PCPs. Of the 268 patients with PCPs, 128 (48%) called their PCP before coming to the ED. Patients gave several reasons why they did not call their PCP: 81 (58% of 128) thought their condition was too urgent, 17 (12%) stated it did not occur to them, and 10 (7%) failed to call their PCP because they did not think they would be treated soon enough.

The hospital admission rate in ED patients who had failed to call their PCP was 34% (47 of 140). Multivariate analysis was used to simultaneously assess effects of multiple parameters upon the dependent variable "contacted PCP before ED presentation." When factors such as age and primary language spoken were controlled, statistically significant results were that minority patients (compared with white patients; OR 0.52; 95% CI 0.28 to 0.99; P=.046) and men (compared with women; OR 0.61; 95% CI 0.38 to 0.98; P=.040) were less likely to call their PCP before their ED visit.

Of the 128 patients who called their PCP before ED presentation, 55 (43%) were able to reach the PCP. Elderly patients (>55 years) were more likely to reach their PCP (OR 1.95; 95% CI 1.07 to 3.57; P=.028). One hundred fourteen (89% of 128) were told to come to the ED (87 [67%] over the telephone and 27 [21%] after being treated in clinic); 46 (36% of 114) were admitted. Of the 14 (10% of 114) patients who contacted the PCP but were not referred to the ED, 4 (28% of 14) presented to the ED because they did not want to wait for an appointment, and another 7 (50% of 14) came to the ED because they could not get through to anyone; 3 (28%) of these 11 patients were admitted.

Of patients who contacted their PCP, 25 (20% of 128) stated they were able to schedule appointments; 19 (76%) were treated by the PCP on the same day of the ED visit. Of these 19 patients, 8 (42%) were admitted. Of the 6 patients who had contacted their PCP and were assigned a future appointment but instead presented to the ED, 1 (17%) was admitted. Of the 114 patients told to come to the ED (either over the telephone or at their PCP's clinic), 65 (57%) were sent to the ED for hospital admission, 23 (20%) for a radiographic study, 25 (22%) for a specialty consultation, and 3 (3%) because the clinic was too busy. Of the 65 patients referred to the ED for hospital admission, 28 (43%) were ultimately admitted.

Comparison of rates of hospitalization showed that patients were no more likely to be admitted if they called their PCP versus not calling their PCP (OR 0.90; 95% CI 0.55 to 1.49; P=.68) . Patients were also no more likely to be admitted if they were referred to ED versus not referred (OR 1.2; 95% CI 0.77 to 1.88; P=.42).

Conclusion: Most of the patients approached had PCPs; however, less than half of the patients had contacted PCPs before ED visits. The main reason patients failed to call their PCP was the perceived emergency nature of their complaint, rather than not being able to be treated in a timely fashion. Minority and male patients were significantly less likely to attempt to call their PCPs, whereas older patients were more likely to successfully reach their PCPs. The vast majority of patients sampled who called their PCP's office were told to come to the ED. Ultimately, less than half of patients referred for hospital admission were admitted. Overall, interventions to decrease inappropriate ED visits could focus on increased access to PCPs, especially for young and minority patients, facilitation of same-day visits, and education of patients to call their PCP before coming to the ED.

Salazar A, Masip J, Bardes I, Juan A, Riera C, Corbella X/Hospital de Bellvitge, University of Barcelona, Barcelona, Spain Study objectives: We examine the mortality pattern throughout a 3-year period in a university emergency department (ED).

Methods: This was a descriptive, retrospective study. The study took place in the ED of a 1,000-bed teaching hospital in Barcelona, Spain, with a population in the metropolitan area of 1.5 million, which attends about 110,000 emergency visits per year, excluding pediatrics and obstetrics. The ED computer registration database was used to identify all patients presenting to the hospital ED and dying during a 3-

Results: One hundred thirty-six patients were discharged with a primary diagnosis of PE during a 1-year period. Thirty were excluded because they did not meet study design criteria for PE diagnosis, leaving 106 for analysis. PE diagnoses were made with high probabilityV/Q scan (47), CT scan (25), angiography (13), intermediate probabilityV/Q scan with positive Doppler results (8), and a combination of CT and high or intermediate probabilityV/Q (13). Chief complaints were dyspnea (43%), pleuritic chest pain (16%), and chest pain not specified (17%). Neither chest pain nor dyspnea was noted as chief complaint by 23% of patients. The majority of patients presented with normal vital signs (pulse rate 80 to 100 beats/min [64%], respiratory rate less than 25 breaths/min [74%], oxygen saturation 94% to 100% [52%]), and 99% of patients were afebrile. Although 19% of patients were receiving coumadin, 62% were nontherapeutic (international normalized ratio <2). PE was diagnosed in the ED in only 39% of patients.

Conclusion: ED patients who are ultimately found to have PE frequently present with normal vital signs and oxygenation; neither chest pain nor dyspnea is the chief complaint in most cases. The majority of patients are not diagnosed in the ED.

Peacock WF, Summers RL, Emerman CE/The Cleveland Clinic, Cleveland, OH; The University of Mississippi, Jackson, MS; Case Western Reserve University, Cleveland, OH Study objectives: Bioimpedance monitoring is a noninvasive Food and Drug Administration-approved technology that takes only minutes to perform and provides data indicative of the patient's hemodynamic state. We evaluate the frequency of clinical management changes resulting from the addition of BI-derived hemodynamic characteristics to standard clinical evaluation. This was a prospective, blinded, case control study in 2 urban academic emergency departments (EDs). All dyspneic patients who were older than 40 years and presenting to the ED were eligible for entry. Bioimpedance data were collected by a research nurse not involved in patient care. Bioimpedance data were blinded to the attending emergency physician. At the initial ED presentation, after completion of the medical history and physical examination and before any central laboratory or radiograph data were available, the attending emergency physician completed a case report form documenting the anticipated diagnosis, medication orders, and disposition. The bioimpedance data were then immediately unblinded, and the attending emergency physician repeated completion of the same case report form. Prebioimpedance and postbioimpedance data case report forms were compared for differences.

Results: Fifty-five patients were enrolled, with a mean age of 76.1±6.9 years; 28 (50.9%) were white patients and 27 (49.1%) were black patients. The initial respiratory rate was 22 breaths/min, and the mean ejection fraction was 40.5%±7.1%. Heart failure or chronic obstructive pulmonary disease (COPD) were the most common final discharge diagnoses, occurring in 27 (49.1%) patients and 11 (20%) patients, respectively. The mean brain natriuretic peptide was 634.9±585.1 pg/mL. Bioimpedance data changed the diagnosis in 3 (5.4%; 95% confidence interval [CI] 1.3 to 15.4) patients, from heart failure to COPD, or the reverse. In these cases, there were no changes between the postbioimpedance diagnosis and the final ED discharge diagnosis. In 2 cases of pulmonary embolus, bioimpedance data did not alter the diagnosis. Bioimpedance data resulted in 20 medication changes in 13 patients (23.6%; 95% CI 14.2 to 36.5). Stopping or starting diuretics accounted for 5 (25%) of the changes. There were 15 episodes of medications being added and 5 episodes of previously ordered medications being canceled.

Conclusion: Bioimpedance data result in diagnosis changes in 5.3% and medication changes in 23.6% of dyspneic ED patients older than 40 years.

Liu SW, Thomas SH/Massachusetts General Hospital, Harvard Affiliated, Boston, MA Study objectives: Our goal was to describe why patients who have a primary care physician (PCP) seek emergency department (ED) care. We wished to assess the following in a population of nonacute ED patients who had PCPs: (1) What was the frequency with which patients contacted their PCPs before ED presentation? (2) For patients who did not call their PCP, what were the reasons for failure to call? (3) Was there an association between demographic factors and likelihood of calling or reaching the PCP? (4) If patients were sent in by their PCPs for hospital admission, what was their likelihood of actually being admitted?

Methods: A prospective convenience sample of adult ED patients was interviewed recovery strategy. Study recommendations to reduce HCP uneasiness as the main impediment to DCD included implementation of a strong educational program, development of a critical pathway on DCD, and guidelines for determining the likelihood of cessation of heartbeat within the interval demanded by DCD. Methods: Patients at least 18 years old and presenting to an urban ED were prospectively surveyed. Patients with severe conditions or cognitive insufficiency or who refused the interview were excluded. We assessed the following: access to acute and primary health care, 6-month health care use history, and situational and perceptual reasons for choosing the ED for the current visit. χ 2 and logistic regressions were performed.

Results: Eight hundred thirty-nine (61%) of 1,375 patients were interviewed. Black and Hispanic patients (versus white patients) were more likely to seek care exclusively in the ED (black 29%, Hispanic 31%, white 18%; P<.001), to present with a nonurgent complaint (21%, 19%, and 13%, respectively; P<.001), and to have had 2 or more ED visits in the past 6 months (36%, 36%, and 28%, respectively; P<.05). Black and Hispanic patients were less likely to have been sent to the ED by a health care provider (23%, 26%, and 43%, respectively; P<.001), more likely to have transportation problems (15%, 12%, and 8%, respectively; P<.05) , and would have been more likely to seek treatment at a walk-in medical clinic, had one been available (45%, 47%, and 26%, respectively; P<.001). Hispanic patients were also more likely to believe that the care received in an ED is better than that received in an outpatient venue (68%, 83%, and 78%, respectively; P<.001) and to believe that care received in the ED is free (10%, 7%, 3%; P<.05). Logistic regression analyses controlling for socioeconomic confounders accounted for the racial and ethnic differences in routine ED use, frequent ED use, and nonurgent complaints.

Conclusion: The decision to seek treatment in the ED is determined by multiple factors, such as perceptions of severity of illness, a belief that the quality or timeliness of care is better in the ED, and a lack of other convenient alternatives for acute care. Race and ethnic differences in ED use and nonurgent complaints appear to be largely due to socioeconomic confounds. These findings suggest that programs to reduce inappropriate ED use must be sensitive to an array of complex social and psychologic issues and may necessitate substantial changes in the way health care systems provide acute care, such as establishment of 24-hour, nonurgent, walk-in clinics.

Hong R, Baumann BM, Chansky M, Boudreaux ED/Cooper Hospital and University of Medicine and Dentistry of New Jersey, Robert Wood Johnson Medical School, Camden, NJ Study objectives: Emergency department (ED) crowding is partially due to the use of EDs for minor medical problems. We examined patients' choice of health care venue according to hypothetical presenting complaints.

Methods: Patients aged at least 18 years and presenting to an urban ED were prospectively surveyed. Patients with severe conditions or cognitive insufficiency or who refused the interview were excluded. We presented 6 hypothetical medical symptoms that were established by consensus of the investigators as either "ED indicated" (chest pain, unilateral weakness, syncope) or "non-ED indicated" (cough without fever, rash, toothache). We asked patients where they would go for care if they experienced each of the 6 problems (ED, physician's office/clinic, other). We examined associations between several patient characteristics and ED use for non-ED indicated illnesses by using χ 2 analyses.

Results: Of the 1,375 adult patients visiting the ED, 839 (61%) were interviewed. The sample consisted of 49% black patients, 10% Hispanic patients, 36% non-high school graduates, 27% uninsured patients, and 57% low-income status (<$20,000/ year) patients. Most patients chose to use the ED for chest pain (77%), unilateral weakness (75%), and syncope (84%). A minority chose to use the ED for cough without fever (19%), rash (24%), and toothache (11%). Patient characteristics statistically associated (P<.01) with ED use for non-ED indicated problems were no identifiable primary care physician or outpatient physician, problems with transporta-year period from January 1, 1999 , to December 31, 2001 . A cohort composed of 986 patients of the total 331,387 ED visits was identified as making a mortality rate of 0.30%. All medical records for each patient were reviewed to examine demographic characteristics, referral patterns, and ED stay and to determine the cause of death.

Results: Of the 986 deaths, 559 (56.7%) patients were men. Mean age was 72 years (range 16 to 95 years), but only 102 (10.3%) patients were younger than 50 years. Mean age of the female patients (77 years) was 9 years older than that of the male patients (68 years). There were no differences in age and sex along the period: 1999, men 56.4%, mean age 71 years; 2000, men 56.3%, mean age 72 years; and 2001, men 57.6%, mean age 73 years. Mortality rates during the study period were as follows : 1999, 368/108,561, 0.34%; 2000, 302/109,857, 0.27%; and 2001, 316/112,969, 0.28%. Most (65.7%) of the patients presented to the ED on their own with neither a previous visit by the general physician nor a referral from another center, and this annual rate has progressively increased (1999, 60.7%; 2000, 65.2%; 2001, 72%) . Twenty-two percent of the patients were dead on arrival to the ED, and this percentage was not modified throughout the 3-year period. After these patients were excluded, the mean time in the ED was 4 hours 46 minutes. The 3 predominant causes of death in the cohort were respiratory failure, coronary heart disease, and sepsis, but in the subgroup of patients younger than 50 years, mortality in the ED was more frequently associated with diagnosis of polytraumatism, violent death, and tumors.

Conclusion: These data show that there is a low mortality in the ED, with a preponderance of male patients, and that patients' age is a factor that substantially influences the cause of death.

Schears R, Seem D, D'Alessandro A/Mayo Clinic, Rochester, MN; United Network for Organ Sharing, Richmond, VA; University of Wisconsin-Madison, Madison, WI Study objectives: Donation after cardiac death (DCD) is considered a legitimate but underused strategy to recover organs for transplant. Of the potential 1,000 controlled DCD donors annually, only 169 were recovered in 2002, whereas more than 84,000 Americans waited and 6,099 died listed. To better analyze the sources for this unrealized potential, a study exploring current health care professionals' (HCPs) knowledge, beliefs, and ethical concerns was undertaken. Focus groups of HCPs who interface with potential DCD donors were identified for study inclusion by an advisory council sponsored by The United Network for Organ Sharing.

Methods: From August 2001 through May 2002, focus groups were conducted. A standardized data tool was used, and a qualitative researcher led all encounters. Participants were given scripted explanations of controlled and uncontrolled DCD and a case study for each of the scenarios. Responses given during the discussion period were monitored live and recorded for later analysis. HCPs practicing in emergency medicine, critical care, neuroscience, trauma and transplant surgery, anesthesia, and the operating room, and also medical examiners and procurement staff, were invited to participate.

Results: Identified barriers to DCD were categorized with attitudes and beliefs summarized. Of the 206 total participants, 91 physicians, 78 nurses, 10 operating room technicians, and 27 procurement representatives were studied. Universally, participants held the opinion that organ donation was beneficial and supported the current policy of promoting organ donation and transplantation. However, most groups expressed concern that a DCD strategy would likely fail without specific process education and greater practical experience. Broad endorsement was identified for (1) DCD protocol and concept education to include the idea of a "firewall" to separate conversations involving withdrawal of life support between physicians and patients from proximate organ donation discussions with family; (2) greater certainty in predicting the occurrence of heartbeat cessation after life support withdrawal; and (3) the need for medical community consensus on what are and are not adequate, ethical, DCD donor selection and procurement practices. Participant dialogue was split about the ethics of premortem cannulation as defined in uncontrolled DCD donation. Other barriers influencing DCD acceptance were family requests to attend the withdrawal of life support, the possibility of hastening death through the administration of medications for pain or organ preservation, and negative publicity for attempted but failed donation.

Conclusion: There exists an unquestioned benefit of organ donation and nearly reflexive approval of the need to expand procurement to meet ever-increasing demand. This HCP bias is tempered by concerns of public accountability, adequacy of process education and experience, feasibility, and ethicality of using DCD as a The analysis of variance revealed no correlation between emergency physician personality and career satisfaction (F=.477; P=.698) . The reliability analysis for the assertiveness construct of the behavioral instrument revealed α equal to .71, and for the responsiveness construct, α was equal to .73. Cronbach's α on the entire scale revealed α equal to .76. The scores show that the personality instrument did not measure what it was intended to measure.

The personality styles of the respondents included 124 (38%) expressive, 112 (34%) driver, 63 (19.1%) analytical, and 30 (9%) amiable. Driver and expressive personality styles are assertive, fast-paced, quick decisionmakers and "tell oriented." Analytical and amiable styles are responsive, slower-paced, slower decisionmakers and "ask oriented." The career satisfaction of the emergency physician respondents averaged 4.03 on a scale in which 1 is extremely satisfied and 5 is extremely dissatisfied. Each style was equally dissatisfied: expressives averaged 4.07, drivers 4.06, analyticals 3.92, and amiables 4.01.

Conclusion: The physicians studied had low satisfaction with their emergency medicine career and had an assertive personality style. The 5-item questionnaire, validated in an office environment, asked about overall career success and success in meeting career, financial, skills, and advancement goals. Unhappiness in these areas may not translate into career dissatisfaction for emergency physicians. Problems with the reliability of the personality construct made it impossible to determine whether certain personality styles are more satisfied than others. Further study is warranted using different survey instruments.

Marco CA/St. Vincent Mercy Medical Center, Toledo, OH Study objectives: Promotion of prescription drugs represents a growing source of pharmaceutical marketing expenditures. Previous studies have demonstrated the impact of marketing on prescribing practices. This study was undertaken to identify the frequency of items containing pharmaceutical advertising in clinical emergency departments (EDs).

Methods: Emergency physicians quantified a variety of items containing pharmaceutical advertising present at specified representative times and days in clinical EDs.

Results: Measurements were obtained by 57 emergency physicians representing 20 states. Most EDs in this study were community EDs (84% community, 16% university or university affiliate), and most were in urban settings (47% urban, 40% suburban, 13% rural). Observers measured 43 items per ED (mean 43; median 31) containing pharmaceutical advertising in the clinical area. The most commonly observed items included pens (mean 16 per ED, median 10), product brochures (mean 4.8), stethoscope labels (mean 3.8), drug samples (mean 3.5), books (mean 3.4), mugs (mean 2.4), and published literature (3.1). EDs with a policy restricting pharmaceutical representatives in the ED had significantly fewer items containing pharmaceutical advertising (mean 11.25, median 10) compared with EDs without such a policy (mean 45.7, median 35.5; P=.02, . There were no differences in quantities of pharmaceutical advertising for EDs in community compared with university settings (P=.4) or rural compared with urban settings (P=.4).

Conclusion: Numerous items containing pharmaceutical advertising are frequently observed in EDs. Policies restricting pharmaceutical representatives in the ED are associated with reduced pharmaceutical advertising. Study objectives: We test the hypothesis that private patients (PVT) contact their primary care physician (PCP) more often than clinic patients (CLN) and assess patient attitudes toward calling the PCP before coming to the emergency department (ED). To our knowledge, there are no studies that investigate whether patients try to contact their PCPs before coming to the ED.

Methods: From July 2001 through December 2002, we surveyed a convenience sample of patients presenting to 2 suburban EDs in New Jersey with annual volumes of 65,000 and 40,000. Patients with no PCP and patients with traumatic injury within the previous week were excluded. The 30-item survey, available in English and Spanish, collected information on demographics and whether the PCP was

A tion, a belief that ED care is free, no high school diploma, low income, lack of health care insurance, and male sex. Conclusion: Most patients made a distinction between ED versus non-ED indicated symptoms. Nevertheless, a notable percentage of patients stated they would use the ED for problems such as cough without fever, rash, and toothache, which was considerably more likely among patients low in socioeconomic status. Efforts to reduce "inappropriate" ED use without accounting for these trenchant social factors have questionable chances of success.

Irvin CB, Murray T, Vretta C/St. John Hospital and Medical Center and Wayne State University School of Medicine, Detroit, MI Study objectives: Patients who leave without being seen (LWBS) pose a potential Emergency Medical Treatment and Active Labor Act (EMTALA) violation from lack of medical screening examination and have potential for adverse outcomes because of delayed medical care. We determine whether insured or uninsured patients are more likely to LWBS.

Methods: This was a retrospective observational medical record review done at an urban teaching Level I trauma center (80,000 visits/year), where all presenting emergency department patients have a registration sheet (data archived from last medical center visit) printed and information later confirmed through interviewed registration. Data on insurance status and months since last interview registration were collected from a consecutive sample of all LWBS and evaluated patients presenting in November 2002. Because patients may LWBS before interviewed registration, insurance status was assumed reliable only if interviewed registration had occurred in the past year. For data analysis, an assumption of insured status was made for all LWBS with no interviewed registration (NewReg) or with interviewed registration more than 1 year old (OldReg). The LWBS and presenting patients were compared for insured and uninsured status. Study was institutional review board exempt. Data were analyzed with Fisher's exact test.

Results: Of the 6,485 presenting patients (807 [12%] uninsured), 242 LWBS (53 [22%] uninsured) and 6,243 stayed for evaluation. Of all uninsured presenting patients, 53 (6.6%) of 807 LWBS. Of all insured presenting patients, 189 (3.3%) of 5,678 LWBS (P<.001). A limitation to the study was the assumption that all LWBS who were NewReg or had OldReg were insured, which minimized the percentage of uninsured LWBS.

Conclusion: Uninsured patients LWBS more often than insured patients. Efforts to limit the number of uninsured LWBS patients may be necessary to maintain the safety net for uninsured patients.

McEwen JS/American College of Emergency Physicians, Dallas, TX Study objectives: I sought to determine whether there was a correlation between emergency physician personality style and career satisfaction. If a correlation were shown to exist, personality style testing could be used as a tool for use with students considering emergency medicine careers.

Methods: I conducted a random sample Web-based survey of 1,000 members of the American College of Emergency Physicians who have e-mail addresses and have been in practice for at least 3 years. There were 7 demographic questions, a 20-item behavioral style instrument previously validated in the marketing field, and a 5-item career satisfaction instrument previously validated with office employees. The study incorporated prenotification and reminder e-mails. Questionnaires were compiled by using the SurveySaid Web questionnaire software program; data were transferred to an Excel spreadsheet and later imported into SPSS for statistical analysis. Frequencies and percentages were run on the demographic data collected. Scores on the inventories were tallied by using Excel and subjected to 1-way analysis of variance. Cronbach's α was run on the assertiveness and responsiveness dimensions of the scale and on the entire 20-item scale to determine validity.

Results: Of the 1,000 e-mail addresses, 707 (70.7%) were valid. Valid responses were obtained from 329 physicians (46.5%). Demographics included 52.2% between the ages of 36 and 45 years, ranging from 29 to 70 years. More than 92% of respondents were born in the United States, and 86% were white. All but 2 (99.4%) respondents were residency trained in emergency medicine, and 55.9% of respondents have been in practice between 6 and 15 years. The nature of practice is divided among clinical, teaching, research, and administrative duties, with the respondents reporting an average of 69.9% of their time spent treating patients. However, patients transported by ambulance at a later time were included. Patients were given an anonymous short questionnaire asking several questions relating to their crash and reasons for attending the A&E department. The questionnaire was then collected after the patients' treatment was completed.

Results: Complete data from 98 patients were collected throughout the 3-month period. By using the department's computerized records, a total of 120 eligible patients were identified as having attended the department during the study period, giving a response rate of 81%. Demographically, our patients were similar to those in previous studies of motor vehicle crash victims. The mean age was 32 years, and 46% of patients were men. Of the 98 patients, 22 patients reported that they were considering the possibility of compensation after their crash. Of these patients, 9 patients reported the possibility of compensation as being a factor in their reason for attending the hospital that day.

Conclusion: Our results show that approximately one quarter of patients attending A&E departments with minor injuries after a motor vehicle crash are considering compensation relating to the motor vehicle crash. Around 40% of these patients report the possibility of compensation's affecting their decision to attend the hospital. There does not seem to be any particular age or sex group this trend applies to. There is no correlation between time from crash to presentation and likelihood of considering compensation.

Since deregulation of personal injury claims in 1999, the rate of personal injury claims has increased further. There were 744,000 claims in 2000, which is set to top 1 million by 2005. Significant numbers of patients seem to be influenced to attend A&E departments after an accident by the possibility of a personal injury claim. If personal injury claims continue to grow at their present rate, this increase has significant implications for A&E department workloads and staffing levels.

D'Onofrio G, Irwin K, Khoshnood K, Degutis LC/Yale University, New Haven, CT Study objectives: The emergency department (ED) visit provides an opportunity for nonoccupational HIV postexposure prevention (NPEP), including antiretroviral treatment, education, counseling, and referral. There are no federal recommendations for NPEP. We determine the need and feasibility of offering NPEP in 2 New Haven EDs.

Methods: Four focus groups were convened with high-risk populations in New Haven, with cooperation of local AIDS service organizations. Discussion involved receptivity, needs, merits, and limitations of offering ED NPEP. Groups included black women (7) with a recent history of drug abuse, Hispanic women (5) with various risk factors, active injection drug users (11), and men who have sex with men (11). Structured surveys were completed by emergency physicians at both locations about current NPEP practice, knowledge, role responsibility, and attitudes and beliefs. Exposures were categorized as sexual assault, unintentional needlesticks, unsafe sexual practices, and injection drug use.

Results: Common focus group themes included concern about cost, confidentiality, follow-up procedures, mistrust of the ED personnel, and cultural appropriateness of the interventions. Forty-five physicians completed the survey: 93% were supportive of offering NPEP in the ED. Knowledge was low about lack of Centers for Disease Control and Prevention guidelines (24%), time frame for offering drug prophylaxis (15.6%), and assessment of risk (15.6%). Forty-three percent of physicians were not confident in prescribing drug therapy.

Conclusion: High-risk populations have concerns about ED NPEP. Physicians and community members are supportive of delivering NPEP, but collaboration in education, development of protocols, and follow-up strategies is needed to ensure that NPEP initiated in the ED fits into exiting services.

Holsti M, Kadish H, Sill B, Nelson D/University of Utah and Primary Children's Medical Center, Salt Lake City, UT; University of Utah Medical School, Salt Lake City, UT Study objectives: Closed head injury (CHI) is common in childhood and frequently results in hospital admission for observation and treatment. Observation units (OU) have shown significant benefits for emergency department (ED) patients and physicians. At Primary Children's Medical Center, a Level I trauma center, CHI patients are often admitted to an OU for up to 24 hours. We describe characteristics of called before the ED visit. For pediatric visits, adult caretakers filled out the surveys. No attempt was made to count the patients who refused or failed to answer the survey. χ 2 was used for statistical analysis. Institutional review board approval was obtained.

Results: Of the 2,007 patients surveyed, mean age was 47±17 years, 21% were children, 66% were PVT, and 97% had telephones. Both sites had similar demographics. PVT called their PCP more often than CLN (46% versus 21%, respectively; P<.001), and caretakers of pediatric patients called more often than adult patients (49% versus 33%, respectively; P<.03). Participants with limited English proficiency were less likely than those who were proficient to call (24% versus 36%, respectively; P<.001). The most common reasons for not calling for the PVT and CLN groups, respectively, were that the situation was too urgent (21% of PVT) and there was not enough time to call (14% of CLN Study objectives: Needlestick injuries are an occupational hazard for junior physicians, especially in emergency medicine. The emergency department is involved in the treatment of injuries in the hospital setting and in the community. The setting was in an inner-city area with a high incidence of intravenous drug abuse, HIV, and hepatitis B and C. The study was to highlight areas for improvement in treatment.

Methods: This was a retrospective review of all emergency notes coded as needlestick injury for a 12-month period from July 1, 2001, to July 1, 2002. Information recorded included times from incident, arrival at department, to be treated by a physician, and to receive postexposure prophylaxis (PEP) if indicated. Also recorded were the number of tetanus toxoid, hepatitis B immunoglobulin and vaccine, and HIV PEP given, as well as the number indicated. Risk of injury and exposure was assessed and follow-up was checked.

Results: There were 73 needlestick injuries; 35 (48%) presented during normal working hours (9 AM to 5 PM) and 38 (52%) outside these hours. Twenty-six (34%) were health care workers and 51 (66%) in non-health care workers. The average time from incident to arrival was 1.4 hours for health care workers and 22.6 hours for non-health care workers. The median time from arrival in the department to be treated by a physician was 90 minutes. Ten (13.7%) injuries were high risk. Antiretroviral agents were given to 15 (20.1%) patients, and average time from door to HIV PEP was 141 minutes.

Conclusion: Emergency medicine staff should be aware of the risks of bloodborne viral transmission because they have greater exposure than other health care groups. They are at higher risk of percutaneous injury and therefore should adopt universal precautions; shield and sheath devices would also reduce the risk of sharp injury. The HIV PEP is effective if given early, so needlestick injury patients must be assessed urgently, and antiretroviral agents must be given as soon as possible, if indicated. Emergency medicine has had an increasing role in treatment of health care workers' needlestick injuries occurring outside working hours, in out-of-hospital injuries, and in other presentations for HIV PEP. Greater education of emergency staff, other health care workers, and the general public is required for optimal management of needlestick injuries.

Corfield AR, Turner CM/Crosshouse Hospital, Kilmarnock, Scotland, United Kingdom Study objectives: Motor vehicle crashes account for 3% of all attendances to United Kingdom accident and emergency (A&E) departments. We aimed to ascertain whether the prospect of a personal injury claim affected patients' decision to attend an A&E department after involvement in a motor vehicle crash.

Methods: All "walking wounded" patients attending the A&E department of a district general hospital after a motor vehicle crash throughout a 3-month period were included. "Walking wounded" was defined as patients classified as Manchester triage category 4 or 5. Patients involved in a motor vehicle crash as a pedestrian or bicyclist were excluded, as were patients younger than 16 years. Patients transported directly from the scene of a crash by ambulance were excluded.

Delinquent Behavior (F=16.59, df=1, P=0), SAGE Baseline (F=26.63, df=1, P=0), Houston School Cohort (F=15.789, df=1, P=.00), Houston Gun Survey (F=15.789, df=1, P=0) , and the New York City Youth Violence surveys (F=9.649, df=1, P=0) .

Conclusion: This study demonstrated that there is a set of questions that could be used to differentiate youth who are violence victims from those who present to an ED for other reasons. The developed tool could be used to identify youth at high risk for violence-related injuries. Predictive measurement of this tool for victimized youths and enrollment of these youths into an intervention would be a worthwhile endeavor.

Being Seen: Why Don't They Stay and Where Do They Go?

Goldman RD, Macpherson A, Schuh S, Mulligan C, Pirie J/The Hospital for Sick Children, University of Toronto, Toronto, Ontario, Canada Study objectives: Numerous children visiting pediatric emergency departments (EDs) leave before being treated by a physician (LWBS). Of potential concern is the inability to provide them with a timely assessment and treatment. The population of children who LWBS has not been adequately characterized. The objective of this study was to examine the acuity (triage score) of children who LWBS compared with children who have stayed to be treated in the ED. We also examined variables related to LWBS and follow-up health care use.

Methods: We conducted a prospective cohort study during a 3-month period in the ED of a tertiary pediatric hospital in Toronto, Ontario, Canada. All families who LWBS were contacted and asked questions about their child's condition, use of follow-up health services, and sociodemographic variables. Logistic regression analysis was used to compare LWBS children with controls matched for age and sex.

Results: During the study period, 289 (2.6%) families left the ED, of whom 180 (62%) consented to participate in the study. The study and control groups consisted of 158 and 316 children, respectively. Waiting too long and improved symptoms accounted for 58% and 37% of premature leaving. Of the LWBS patients, 14% were triaged as urgent, and two thirds of them sought further medical care; 1 child was admitted. Multivariable analysis showed that patients who left had a lower acuity compared with those who stayed (odds ratio [OR] 4.95; 95% confidence interval [CI] 2.6 to 9.4). Most of them sought further medical attention after leaving (OR 4.63; 95% CI 2.8 to 7.8) and were more likely to register in the ED between midnight and 4 AM (OR 4.86; 95% CI 2.2 to 10.5).

Conclusion: Children who LWBS have a lower acuity level, seek follow-up care elsewhere, and usually leave because they get better or the wait is too long. A small proportion is triaged as urgent but their outcome is favorable. Further studies are recommended to address possible solutions, including improving patient flow in the ED and increasing staff at peak times.

Nibhanipudi KV, Spiller D/Metropolitan Hospital Center, New York Medical College, New York, NY Study objectives: Spastic torticollis is a condition in which the head is laterally flexed and rotated on the neck because of painful spasm of the neck muscles, and the objective is to determine the efficacy and adverse effects of chloroethane spray in the treatment of spastic torticollis. This was a prospective study in an urban innercity hospital pediatric emergency department (ED).

Methods: All children and adolescents (between the ages of 1 and 16 years) presenting to the authors with the complaint of stiff neck were enrolled in the study. The total number of patients enrolled was 132. After complete evaluation to rule out cervical spine injury, a central neurologic cause, patients were enrolled in the study. Sixty-six patients were treated with chloroethane spray to the neck, and the other 66 patients were given normal saline solution as placebo.

Results: Sixty-three of 66 patients treated with chloroethane spray achieved relief within 5 minutes, as demonstrated by painless and free movement of the head and neck. No adverse effects were observed. The patients treated with placebo had no relief within 5 minutes. To test the null hypothesis that the percentages improved in the 2 groups equally, we calculated the Z statistic. The Z statistic of 24 corresponded to a P value less than .0001, whether the test was 1 sided or 2 sided. Chloroethane spray treatment was superior to placebo, with a high statistical significance.

Conclusion: In summary, chloroethane spray is superior to placebo (normal saline solution) for treatment of spastic torticollis. If chloroethane spray does not work, no harm is done. However, when it works, a satisfied patient leaves the ED.

A patients who have CHI and are admitted to a pediatric OU and identify factors associated with the inability to discharge these patients within 24 hours. Methods: This was a retrospective cohort review of all OU admissions for CHIs at Primary Children's Medical Center from August 1999 through July 2001. A standardized data sheet was used for data collection. The database included age, sex, mechanism of injury, presenting symptoms, physical examination, computed tomography (CT) results, diagnosis, and outcome of the injury.

Results: During the study period, 830 patients were treated in the ED for CHI. Two hundred eighty-six (34%) patients were admitted to the OU, 272 (33%) patients were admitted to an inpatient service, and 269 (32%) patients were sent home. One hundred eighty-one medical records were reviewed for abstract submission. The median age of the patients admitted to the OU was 5.2 years (range 2 weeks to 17 years). Sixty-one percent were male patients. The median length of stay in the OU was 12.6 hours (range 1.8 to 24 hours). The most common injuries were the following: 28% of patients fell from 3 to 6 feet, 14% fell from higher than 9 feet, 11% fell from ground level, 7% fell from a bike without a helmet, 7% had an object hit their head, 7% were unrestrained passengers of a motor vehicle crash, and 6% fell from lower than 3 feet. Presenting symptoms in the ED included vomiting (36%), loss of consciousness (20%), amnesia (12%), and seizures (2%). Physical examination included facial abnormalities (49%), altered mental status (43%), scalp abnormalities (34%), and neurologic deficits (9%). All 181 OU admissions had a head CT scan performed. Skull fractures were present in 59 (33%) patients, intracranial pathology (ie, epidural hematoma, subdural hematoma, intraparenchymal contusion) was present in 24 (13.4%) patients, and 3 (1.7%) patients had a basilar skull fracture. Only 4 (2.2%) patients required admission to an inpatient service from the OU for the following reasons: persistent cerebrospinal fluid leakage (N=1), intravenous fluids (1), decreased level of consciousness (1), and pain management (1). No patient deteriorated or required neurosurgery. One patient with basilar skull fracture was readmitted after OU discharge for observation secondary to continuous cerebrospinal fluid leakage.

Conclusion: The vast majority (98%) of pediatric OU patients with small intracranial hematomas, skull fractures, and concussions were discharged safely within 24 hours, with no serious complications. OU admission was an efficient and effective treatment setting for children with stable intracranial pathology, skull fractures, and concussions.

Zun LS, Arakaki D, Patel R, Downey L/Finch University/Chicago Medical School, Mount Sinai Hospital, Chicago, IL Study objectives: The complications of youth violence in the United States are well known. The mechanisms to combat the problem of victimization are elusive, as is the role of the emergency department in reducing the incidence of violence. The first step in the process is to identify youths who are at greatest risk for violence and victimization. The purpose of this project was to determine whether questions from various survey tools could differentiate victimized youths. We hypothesized that the survey responses in victims of violence are the same as those found in the other youths treated in the ED.

Methods: Patients aged 10 to 24 years in an urban Level I trauma ED with 45,000 annual patients were approached between June and August 2002 to participate in the study. One of every 3 non-violence related and all violence-related patients were enrolled. Patients with altered mental status, who were unstable, or who did not speak English were excluded. Patients were asked to complete a 34-question survey that was composed of 5 shorter surveys from the Centers for Disease Control and Prevention Measuring Violence-Related Attitudes, Beliefs, and Behaviors Among Youth. Data were put into an SPSS program (version 10, Chicago, IL), and statistical analyses were performed to compare the groups. This study was institutional review board approved.

Results: One hundred twenty-six patients were enrolled, 61 violence related and 65 non-violence related. Treatment for violence-related injury was associated by using analysis of variance test to age (F=6.877, df=1, P=.01), sex (F=36.41, df=1, P=0), gang related (F=26.46, df=1, P=0), in school (F=12.06, df=2, P=.002), physical fights (F=15.2, df=1, P=0) , detained or arrested (F=33, df=1, P=0), appeared in court (F=33.0, df=1, P=0), alcoholic drinks (F=18.5, df=1, P=0), gang members (F=15.7, df=1, P=0), reason to carry a weapon (F=5.6, df=1, P=.02), and desire to be a gang member (F=9.7, df=1, P=0). All survey scores were different between the groups: Physical df=1, P=.13) , Disciplinary and domized to receive either intravenous morphine or normal saline solution. Clinical data and the emergency physician's confidence with his or her clinical diagnosis (0% to 100%) were systematically collected by using a standardized form. This process was repeated 15 minutes after the administration of the study medication. The surgeon assessed the child within 1 hour and completed a similar data collection sheet. Pain was assessed by using a color analogue scale before and after study medication administration. Each patient was followed up for 2 weeks after enrollment.

Results: One hundred eight children were enrolled; 52 received morphine and 56 were saline solution control patients. There were no differences between groups in demographic variables or degree of pain. There were no differences between groups in the diagnosis of appendicitis or perforated appendicitis or the number of children who were initially observed and then subsequently underwent laparotomy. The mean reduction in the median pain score was significantly greater in the morphine group compared with the placebo group, 2.2 versus 1.2 cm, respectively (P=.026). The emergency physicians' and the surgeons' confidence with their diagnosis was not affected by the administration of morphine.

Conclusion: Our data show that morphine effectively reduces the intensity of pain in children with acute abdominal pain and does not impede the diagnosis of appendicitis.

Zebrack MM, Rentz AC, Kadish HA, Nelson DS/University of Utah School of Medicine, Primary Children's Medical Center, Salt Lake City, UT Study objectives: Infants with elevated total serum bilirubin (TSB) levels in the context of poor feeding or decreased activity usually require inpatient admission. Treatment of these patients in an observation unit (OU) has not previously been described, to the authors' knowledge. OUs provide an alternative to inpatient admission for diagnoses anticipated to need a limited treatment time, usually 24 hours or less. We conducted a study to describe the use of a pediatric OU for treatment of neonatal hyperbilirubinemia not requiring exchange transfusion or intravenous immunoglobulin.

Methods: We identified 127 infants aged 10 days or younger and admitted to the OU throughout a 3-year period with hyperbilirubinemia and coexisting dehydration. Retrospective medical record review was performed.

Results: The median age of patients was 5 days, median gestational age was 38.5 weeks (range 34 to 41 weeks), and median weight loss since birth was 8%. Overall, 70% of patients were exclusively breast fed; 5% were exclusively formula fed. The median TSB for admission was 20 mg/dL (range 12 to 26 mg/dL); 8% of patients had a TSB of 25 mg/dL or greater. ABO blood group incompatibility was present in 16% of patients, more than half of whom had a positive direct antiglobulin test. Rh incompatibility was noted in 9% of patients; all patients were direct antiglobulin test negative. All patients received phototherapy; 73% of patients received intravenous fluid; 68% of patients received intensive lactation consultation; 15% of patients required inpatient admission for continued medical care beyond 24 hours. Hypernatremia at 150 meq/L or greater was associated with need for inpatient admission (odds ratio 1.8; 95% confidence interval 1.0 to 3.2; P<.001). Age, gestational age, TSB, percentage of weight loss, and feeding method were not associated with need for inpatient admission. The median discharge TSB was 15. Patients and TSB were followed up after discharge. There were no adverse outcomes.

Conclusion: The OU is an effective alternative to inpatient admission for the management of neonatal hyperbilirubinemia. With focused therapy, 85% of these patients can be discharged within 24 hours. Jaundiced infants with hypernatremia may not be appropriate for the OU.

Goldman RD, Ong M/The Hospital for Sick Children, Toronto, Ontario, Canada Study objectives: The function of the emergency department (ED) includes assessment and determination of need for inpatient care or outpatient treatment. Patient return within 72 hours contributes to crowding and may indicate failure to give proper assessment, treatment, or follow-up. The objective of this study was to describe return rate and characteristics of patients who return to the ED. Study objectives: This study evaluates the initial diagnostic value of magnetic resonance imaging (MRI) and electroencephalogram (EEG) to find neurologic abnormality and diagnose the possibility of epilepsy of the children who visit the emergency department (ED) with a first afebrile seizure.

Methods: Medical records of children who visited the ED with a first afebrile seizure and received an MRI and EEG were reviewed. The medical records were selected between January 1997 and December 2002.

Results: A total of 46 patients (23 male patients and 23 female patients) were reviewed. Thirty-seven (80%) children showed generalized tonic clonic seizure, and 9 (20%) children showed focal seizure. Twenty-nine (63%) children had 2 or more seizures during a single episode. Eight (17%) children had a seizure episode lasting more than 30 minutes.

MRI abnormalities were found in 20% (9 cases) of the enrolled patients, and all of the abnormalities were significant. The results were as follows: brain atrophy in 2 cases, leukomeningeal enhancement in 2 cases, aneurysm in 1 case, arachnoid cyst in 1 case, ventriculomegaly in 1 case, dysmyelinating disease in 1 case, and hypoxic damage in 1 case. EEG abnormalities were found in 33% (15 cases) of the enrolled cases. Among them were 11 partial seizures and 4 cerebral dysfunctions.

Conclusion: Abnormal findings from MRI and EEG appeared in only a small portions of the cases, but they were unpredictable, serious, and significant problems. We suggest that initial evaluation of MRI and EEG is a valuable diagnostic procedure for finding neurologic abnormalities of children early and treating the afebrile seizures of children who visit the ED.

Coon ME, Chan SB/Resurrection Medical Center, Chicago, IL Study objectives: The relationship between head and cervical spine trauma has not been thoroughly investigated in previous studies. We hypothesized that a negative head computed tomography (CT) scan was insufficient to preclude a thorough cervical spine evaluation.

Methods: This was a retrospective review of the National Pediatric Trauma Registry from April 1994 to March 1998. Inclusion criteria were records that had a sports-related mechanism and included a head CT. The outcome measure that was evaluated was the presence or absence of cervical injury in this population. The risk ratio of a negative head CT and cervical injury was then calculated. The P value was obtained by using Fisher's exact test.

Results: Of the 15,708 patients included in the Pediatric Trauma Registry, 2,368 (15.1%) had a mechanism listed as "sports." One thousand eighty-two patients had a head CT performed and were included in the study group. Six (1.17%) of 512 patients with a positive CT result had a cervical injury, and 5 (0.88%) of 570 patients with a negative CT scan result had a cervical injury. The relative risk for cervical injury with a negative head CT result was not different than when head CT was positive (relative risk 0.75; 95% confidence interval 0.23 to 2.44; Fisher P=.76).

Conclusion: An investigation into the relationship between intracranial injury and cervical injury in pediatric sports trauma reveals that cervical injury can exist despite absence of intracranial injury as identified by a head CT scan. As a result, all pediatric sports-related head trauma should be treated as if a cervical injury also exists. Study objectives: The purpose of this study was to determine whether the administration of morphine to children with acute abdominal pain would impede the diagnosis of appendicitis and to evaluate the efficacy of morphine use in pain relief of children with acute abdominal pain.

Methods: This is a double-blind randomized controlled clinical trial in 5-to 16year-old children presenting at the emergency department (ED) of a children's hospital with acute abdominal pain requiring surgical consultation. Patients were ran-available to read studies 1 morning per week. The hospital has plain radiography and computed tomography suites. A Medison ultrasonography machine with curved linear and flat linear array transducers was available to the emergency physicians. Emergency physicians were asked to fill out a survey about any patient requiring an ultrasonographic examination before the scan. The emergency physician recorded reason for the examination, whether alternative diagnostic imaging was available at the facility, whether transfer would otherwise be necessary, and the prescan differential diagnosis. After the ultrasonographic examination, the emergency physician recorded results, current differential diagnosis, patient disposition, and whether treatment was altered. Descriptive statistics and a paired t test were used to evaluate study results.

Results: A total of 32 ultrasonographic studies were performed. On average, the differential diagnosis decreased from 2.8 to 1.3 after the ultrasonographic examination. The decrease of 1.5 (95% confidence interval 1.1 to 1.8) was statistically significant (P<.0001). In 44% of cases, an alternative imaging test existed at the facility, but in only 1 case would a formal interpretation have been possible. Mean transfer distance was 60 miles. In 13% of cases, the final diagnosis provided by the ultrasonographic examination was not even part of the original differential. Treatment was changed in 75% of cases. In 59% of cases, transfer was avoided. The range of ultrasonographic examinations included pelvic, abdominal aorta, gallbladder, lower extremity duplex, vascular access, trauma ultrasonography, cardiac, testicular, and soft tissue examinations.

Conclusion: Emergency physicians using ultrasonography in a rural ED with limited radiology services frequently change treatment and avoid transferring patients.

Jang TB, Aubin C, Sineff S, Naunheim R/Barnes-Jewish Hospital, St. Louis, MO; Washington University School of Medicine, St. Louis, MO Study objectives: This was a pilot study to assess whether resident confidence correlates with more accurate focused abdominal ultrasonographic examinations (FAUS).

Methods: This was a prospective study of residents with limited experience (>10 and <100 previous FAUS) evaluating patients presenting with abdominal pain by using bedside FAUS before operative or radiology department evaluation. Resident confidence of findings was assessed by self-report on a visual scale from 1 (doubtful) to 5 (certain).

Results: Ninety-nine patients were examined by using bedside FAUS performed by 10 residents with an average previous experience of 36 (95% confidence interval 17 to 55) FAUS. The results are shown by operator confidence level in the Table. Conclusion: There was a trend toward more accurate FAUS examinations when a resident was more confident of his or her findings. Study objectives: We assess the impact of resident-performed ultrasonography on length of stay (LOS) for patients presenting with symptomatic first-trimester pregnancy.

Methods: This was a retrospective medical record review in an urban academic emergency department of patients presenting at less than 20 weeks pregnant with

A Results: Of 97,404 visits recorded for 61,849 patients, 5.4% (5,308) were return visits within 72 hours. Male to female patient ratio was 1.21 and 1.28 for ED population and patients who returned. Thirty-six percent, 40%, and 24% returned in the first, second, and third day, respectively, after the initial visit. Return rate was highest for 1-to 2-year-old patients (7.1%) and during December and January (6.2% and 5.9%, respectively). Forty-eight percent, 30%, and 22% return rate was found in the morning, evening, and overnight shifts, respectively. Triage score was significantly different between children with 1 visit and multiple visits. The more visits, the less likely the child was sent home, and the higher the admission rate. The initial visit length of stay (LOS) for returning patients was 256 minutes and was 324, 373, 349, and 345 minutes in the first, second, third, and fourth return, respectively. Using the average LOS, we calculated that all return visits accounted for a total LOS of 1,210 days, 12.5% of LOS in our department during 2 years (9,673 days).

Conclusion: More than 5% of our visits are return visits within 72 hours, and they occupy 12% of the time in the ED. The reasons for returning should be further investigated, and ways to reduce this number should be found.

Blaivas M, Brannam L, Lyon M/Medical College of Georgia, Augusta, GA Study objectives: Several federally supported organizations have recently mandated ultrasonographic guidance for central line placement, and its use for peripheral access is also becoming more common. According to previous literature, peripheral intravenous line placement by using ultrasonographic guidance is typically accomplished by physicians in the emergency department. We determine whether a short axis (SA) or long axis (LA) ultrasonographic guidance approach is easier for novice nurse ultrasonography users.

Methods: We performed a prospective, randomized, observational study on emergency nurses at a Level I trauma center. After a 30-minute tutorial on ultrasonographyguided vascular access, nurses were randomized to 1 of 2 groups. A Jell-o and Metamucil mixture, providing a realistic ultrasonographic image, was placed inside a synthetic arm skin that was used for training phlebotomists and contained a rubber vein filled with red fluid at a depth of 1.5 cm. Group 1 attempted the SA approach first and then the LA. Group 2 tried LA first. Any nurse not wishing to participate was excused. Time from skin break to cannulation, number of skin breaks and needle redirections, and difficulty on a 10-point Likert scale were recorded. Statistical analysis included paired Student's t test with 95% confidence intervals (CIs).

Results: Seventeen nurses were enrolled. Mean time by nurses to vein cannulation in SA and LA was 0.98 min (95% CI 0.48 to 1.47) and 2.2 min (95% CI 1.32 to 3.8), respectively. The difference of 1.22 min (95% CI 0.3 to 2.13) was statistically significant (P=.01). Mean number of skin breaks for SA and LA was 1.2 (95% CI 0.95 to 1.52) and 2.35 (95% CI 1.24 to 3.46), respectively. The difference of 1.15 skin breaks (95% CI 0.09 to 2.32) was statistically significant (P=.03). Mean number of needle redirections in the SA and LA was 2.35 (95% CI 1.34 to 3.36) and 4.36 (95% CI 3.29 to 5.41), respectively. The difference of 2.01 redirections (95% CI 0.7 to 3.3) was statistically significant (P=.005). Mean difficulty scores for SA and LA were 3.0 (95% CI 1.86 to 4.14) and 5.3 (95% CI 4.3 to 6.3), respectively. The difference of 2.3 (95% CI 0.56 to 4.02) was significant (P=.01).

Conclusion: Novice emergency nurse ultrasonography users obtain vascular access faster by using an SA approach on an inanimate model after a brief tutorial. They also redirect the needle and break the skin fewer times in the SA approach and rate it less difficult on a 10-point Likert scale.

Blaivas M, Lyon M, Brannam L/Medical College of Georgia, Augusta, GA Study objectives: Emergency physicians working in small rural emergency departments (EDs) are often faced with limited resources. Frequently, these limitations include radiologic services. We determine whether introduction of emergency physician-performed ultrasonography affects patient treatment and disposition in a rural ED with limited radiology services.

Methods: This was a prospective observational study surveying 2 emergency physicians on the effects ultrasonography use has on physician decisionmaking. The study took place at a 7,000-patient-volume single-coverage rural ED. A radiologist is incised, and the skin was peeled back. All were penetrated by a needle driver, and 0.5-cm toothpick ends were inserted into the foreign body group approximately 1.0 cm deep. All 42 thighs were prepared by the same individual, and every effort was made to avert leaving superficial evidence of a foreign body. Radiography of a single thigh with a known foreign body present was performed a priori to assure that the toothpick ends were not visible on radiograph. Three blinded registered diagnostic medical sonographer-certified or -eligible sonographers (emergency medicine ultrasonography director, 2 emergency medicine ultrasonography fellows) were then asked to identify the thighs containing a foreign body by using a 11-9 MHz linear array transducer. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were then calculated.

Results: Experienced sonographers demonstrated proficiency at detecting radiolucent foreign bodies (sensitivity 0.91, 95% confidence interval [CI] 83 to 96; specificity 0.81, 95% CI 74 to 86; PPV 0.80, 95% CI 72 to 85; NPV 92, 95% CI 84 to 97).

Conclusion: Compared with historical data using 8-MHz transducers, new highfrequency transducers, in the hands of trained emergency medicine sonographers, improve the detection of radiolucent foreign bodies.

Romagnuolo L, Tayal VS, Tomaszewski C, Norton J, Saunders T/Carolinas Medical Center, Charlotte, NC; Charlotte Eye, Ear, Nose, and Throat, Charlotte, NC Study objectives: Optic sheath nerve diameter (OSND) has been proposed as a marker for increased intracranial pressure. Ongoing studies in head injury patients and previous work in normal adults have identified an upper limit of 5 mm as a normal OSND. Trendelenburg position is often used in hypotensive patients, and reverse Trendelenburg position (30 degrees head up) is often used in head injury patients. We asked whether there would be any change in OSND in normal human adults between the supine, Trendelenburg, and reverse Trendelenburg positions.

Methods: This was a prospective study using consenting normal adult patients. Three investigators measured the OSND in each eye of 10 volunteers in the supine, Trendelenburg, and reverse Trendelenburg positions. Approximately 1 minute was allowed between position changes to achieve equilibration of intracranial pressure. The degree of angulation in the Trendelenburg and reverse Trendelenburg positions was approximately 30 degrees from the horizontal. Each investigator was blinded to the results of the other investigators. Data were analyzed by using the paired t test.

Results: In the supine position, the mean OSND was 4.6±0.71 (SD) mm in the right eye and 4.5±0.56 (SD) mm in the left eye. In the Trendelenburg position, the mean OSND was 4.4±0.72 (SD) mm in the right eye and 4.7±0.53 (SD) mm in the left eye. In the reverse Trendelenburg position, the mean OSND was 4.4±0.49 (SD) mm in the right eye and 4.8±0.76 (SD) mm in the left eye. There was no significant difference in OSND between positions for either eye by analysis of variance. Interobserver agreement was ±1 mm in at least 90% of participants, regardless of position.

Conclusion: OSND measurement by ultrasonography does not significantly change with Trendelenburg or reverse Trendelenburg position in comparison with the supine position. Further work in abnormal patients with these positions is indicated.

Wong HE, Moore C, Skrupky R, Fox JC, Lambert MJ/Resurrection Medical Center, Chicago, IL; Yale University Medical Center, New Haven, CT; University of California-Irvine Medical Center, Orange, CA Study objectives: Lower extremity duplex ultrasonography is the study of choice for the emergency diagnosis of deep venous thrombosis (DVT) but demands considerable time and resources. In some hospitals, duplex ultrasonography is not universally available on a 24-hour basis in the radiology department. We sought to determine the accuracy of emergency physician compression ultrasonography (EPCUS) and D-dimer in safely excluding DVT at the bedside.

Methods: Eighty patients presenting to 2 academic emergency departments throughout a 9-month period and who had suspected DVT were enrolled on a convenience basis in this ongoing comparative study. Patients who had symptoms for more than 1 week, who had chronic DVT, or who were receiving anticoagulation treatment were excluded. Twenty emergency physicians who had undergone 2 hours of initial didactic and hands-on training in EPCUS performed the studies. EPCUS consisted of 2-point femoral and popliteal compression ultrasonography by abdominal or pelvic pain, cramps, or bleeding. LOS was determined by the time registered to the time discharge instructions were signed by the patient. Results were analyzed by using the Student's t test.

Results: The average LOS for the emergency physician-performed ultrasonography group was 149 minutes less (P<.05) than the obstetrician-performed ultrasonography group and 168 minutes less (P<.04) than the group with ultrasonography performed by both emergency physicians and obstetricians. Excluding patients with hyperemesis, the differences were 156 minutes (P<.04) and 181 minutes (P<.04), respectively. No ectopic pregnancies were missed, as determined by 1-year telephone follow-up.

Conclusion: Focused ultrasonography by emergency physicians reduces average LOS for patients presenting with symptomatic first-trimester pregnancies without missing ectopic pregnancies.

Raio CC, Lee DC, Hormozdi SM, Theodoro D, Ko P, Aujalay N, Trimarco T/North Shore University Hospital, Manhasset, NY Study objectives: We determine whether soft-tissue foreign body removal by emergency medicine residents can be facilitated by bedside ultrasonography.

Methods: We performed a prospective, randomized, controlled, pilot study at an academic hospital with a 3-year emergency medicine residency training program and an emergency ultrasonography fellowship training program. Twelve emergency medicine residents were randomized to 2 separate study groups. Each group of 6 residents was stratified according to their level of emergency medicine training, with 2 residents per year in each group. The experimental group was then given a 15minute lecture by a trained ultrasonographer covering the use of bedside ultrasonography in detecting and removing foreign bodies. Twelve standardized chicken legs were prepared for foreign body insertion. The skin was removed from each leg, and a small 0.5-cm incision was made at the superior aspect of the leg, allowing for the placement of a 0.25-cm metallic foreign body approximately 1 cm deep into the meat of the chicken. All participants were made aware that a metallic foreign body was present in each chicken leg. The experimental group was asked to remove the foreign body with the aid of a Sonosite 180 Plus bedside ultrasonographic scanner with an L38/10-5 MHz linear transducer, in addition to the tools present in a standard suture set (forceps, scissors, needle driver, sterile towels, and sterile gauze). The control group was given only the suture set. Total time was then recorded for foreign body localization and removal by each resident, with a time limit set at 10 minutes. χ 2 statistics were used to evaluate the data.

Results: For the 6 residents in the control group (ultrasonography not used), the mean time to foreign body localization and removal was 101 seconds. For the 6 residents in the experimental group, 1 resident was unable to localize the foreign body. The remaining 5 residents averaged 275 seconds to foreign body localization. All 6 residents in the experimental group were unable to remove the foreign body within 10 minutes (P=.0011).

Conclusion: Previous studies have confirmed that bedside ultrasonography can be used in confirming the presence of a foreign body; however, in our study it does not appear to expedite the removal of soft-tissue foreign bodies when used by emergency medicine residents with limited ultrasonography training. Although there were several key limitations in this pilot study, the use of bedside ultrasonography by novice users had an inverse relationship with time to foreign body localization and removal.

Theodoro D, Ko P, Hormozdi S, Raio C, Trimarco T/North Shore University Hospital, Manhasset, NY Study objectives: Historically, emergency physicians have used ultrasonography transducers in the 8-MHz range to detect radiolucent foreign bodies with some success. We hypothesize that new high-frequency ultrasonography transducers, in the hands of fellowship-trained emergency physician sonographers, significantly improve the ability to detect radiolucent foreign bodies.

Methods: This was a blinded pilot study of an educational animal model at a suburban academic center (census of 60,000). Nineteen of 42 chicken thighs were randomly chosen to contain a foreign body. All thighs (including skin) were initially has increased rapidly throughout the past decade. The purpose of this study was to compare the characteristics of sexual assault in postmenopausal victims of sexual assault versus female victims younger than 50 years.

Methods: We conducted a retrospective matched case-control study to assess epidemiology and anogenital injuries in postmenopausal women (>50 years) presenting to 3 free-standing urban sexual assault clinics during a 5-year study period. Controls were younger victims of sexual assault (aged 13 to 49 years) who presented to the clinic on the same day. Two controls were matched to each case. Genital trauma was documented by using colposcopy with nuclear staining and digital photography. Comparative data from the 2 patient groups were analyzed with χ 2 and t tests.

Results: Thirty-three cases of sexual assault in postmenopausal women were identified, representing 2% of the women presenting to the 3 sexual assault clinics. The age range among victims was 50 to 96 years (mean 63+13.9 years); the age range among the 66 controls was 12 to 42 years (mean 24+9.0 years). Postmenopausal victims were more likely to be assaulted by strangers (48% versus 12%; P<.001) in their own home (76% versus 30%; P<.001) and had a greater frequency of nongenital and anogenital injuries (85% versus 67%; P<.001). In comparison, the younger control group was more likely to have a history of sexual assault (53% versus 24%; P<.001) and reported more alcohol or drug use (53% versus 12%; P<.001). The localized pattern of anogenital injuries was similar in both groups; however, postmenopausal women had a greater incidence of lacerations and abrasions (P=.007).

Conclusion: The postmenopausal woman is not immune to sexual assault. The epidemiology of sexual trauma in this age group is unique, and the severity of anogenital injuries was significantly different compared with that of younger victims.

Bush CM, Jones JS, Rossman L, Wynn B, Alexander C/Spectrum Health-Butterworth Campus, Michigan State University, Grand Rapids, MI Study objectives: Most of the literature about anogenital injuries resulting from sexual assault is limited to victims examined within 24 to 48 hours. The purpose of this study was to compare the frequency and types of anogenital trauma in female victims examined acutely (<24 hours) with those who presented for forensic examination 24 to 120 hours after the sexual assault.

Methods: This retrospective cohort trial evaluated consecutive female patients, aged 13 years or older, presenting to a sexual assault clinic during a 5-year study period. Patient demographics, assault characteristics, and injury patterns were recorded by using a standardized classification system. χ 2 and analysis of variance tests were used to compare anogenital findings in victims examined within 5 different 24-hour periods after assault.

Results: A total of 740 cases met the inclusion criteria; 562 (76%) patients were examined within 24 hours after sexual assault. Victims who delayed seeking medical examination were younger (18.9 versus 23.0 years; P<.001) and more likely to be assaulted by a known acquaintance or family member (83% versus 69%; P=.01) and reported more alcohol or drug use before the assault (59% versus 45%; P=.02). The frequency of anogenital lacerations and abrasions decreased from 64% at less than 24 hours to 41% at more than 48 hours after the assault (P<.001). Ecchymosis was actually more common at 48 to 71 hours, and the degree of localized erythema did not significantly decrease until 96 hours after assault.

Conclusion: One quarter of women presented to an urban sexual assault clinic more than 24 hours after their assault. The frequency and types of anogenital injuries vary significantly, depending on timing of the forensic examination.

Wenker K, Gorchynski J/Middlesex Hospital, Middletown, CT; University of California-Irvine Medical Center, Orange CA Study objectives: Partner violence affects nearly 1 in 3 women in her lifetime; however, detection of partner violence is poor, despite screening in emergency departments (EDs). Data about partner violence are available for states and large cities but are limited elsewhere. This study establishes the point and 1-year prevalence of partner violence among women visiting an ED in Orange County, CA, a suburban area south of Los Angeles, and determines differences in partner violence rates with regard to race, income, and education.

Methods: We conducted a prospective, observational study during 12 consecutive months in 2000 to 2001. An 11-question anonymous written survey was administered to a convenience sample of 370 women at a Level I, university, tertiary refer- . Confirmatory lower extremity duplex ultrasonography was performed in the radiology department on all enrolled patients. EPCUS was performed at the bedside before the confirmatory study. All patients underwent D-dimer testing with the VIDAS rapid enzymelinked immunosorbent assay (bioMerieux, Inc.), with a value of more than 500 ng/mL representing a positive test. Results: Results of 80 EPCUS examinations were compared with the reference standard of radiology-performed lower extremity duplex ultrasonography. Fifteen EPCUS studies were positive, 14 (93.3%) of which had confirmed DVTs. One falsepositive study was noted. Fifty-four EPCUS studies were negative, all (100%) of which were confirmed. No false negative studies were noted. Eleven (13.8%) EPCUS studies were equivocal, of which 81.8% were negative on confirmatory examination. Two equivocal EPCUS studies had DVTs on confirmatory examination. Among nonequivocal examinations, EPCUS was 100% sensitive and 98.2% specific for DVT. D-Dimer test result was positive in all patients with DVT. Fourteen (17.5%) patients had a negative D-dimer test result, none of whom had DVT on confirmatory exam. D-Dimer test was 100% sensitive but only 21.5% specific for DVT.

Conclusion: In this study, EPCUS had a positive predictive value of 92.9% and a negative predictive value of 100% for DVT in nonequivocal examinations. Emergency physicians with didactic training were able to accurately exclude DVT at the bedside. D-Dimer test had poor specificity and was clinically useful only when negative. Duplex ultrasonography may be unnecessary when bedside EPCUS performed by qualified emergency physicians clearly demonstrates no thrombosis, although confirmatory duplex ultrasonography is indicated for positive or equivocal EPCUS examinations.

Jones JS, Rossman L, Wynn B, Alexander C/Spectrum Health-Butterworth Campus, Michigan State University, Grand Rapids, MI Study objectives: We compare the epidemiology and clinical findings associated with sexual assault by gang (ie, multiple offenders, 1 victim) versus individual assailants in a community-based population of women presenting to an urban sexual assault clinic.

Methods: This cross-sectional, retrospective study evaluated consecutive female patients, aged 13 years or older, presenting to a sexual assault clinic during a 5-year study period. The clinic is associated with a university-affiliated emergency medicine residency program and is staffed by forensic nurses trained to perform medicolegal examinations by using colposcopy with nuclear staining. Patient demographics, assault characteristics, and injury patterns were recorded by using a standardized classification system. Data from the 2 patient groups (victims of gang versus individual assailants) were analyzed with χ 2 and t tests.

Results: A total of 1,347 cases met the inclusion criteria; 179 (13.3%) women were victims of multiple assailants. The mean number of assailants involved in gang rapes was 2.9 (range 2 to 15 offenders). The majority of these assailants (97%) were male; however, 6% of the gang rapes involved male and female assailants. The 2 victim groups did not differ by age, marital status, family income, or previous assault history. Victims of gang rape were more likely to be ethnic minorities (34% versus 16%; P<.001) than individual rape victims. Gang rapes were also characterized by more alcohol and drug involvement, a greater proportion of stranger assailants, higher incidence of physical coercion, and more nongenital injuries. The pattern and frequency of genital trauma (68% versus 63%; P=.226) was similar in both groups of women. However, victims of gang rape had a larger mean number of genital injuries (2.5 versus 2.1; P=.001) and had more documented anal injuries (11% versus 23%; P<.001).

Conclusion: Thirteen percent of the sexual assaults in this urban population were committed by gang assailants. The epidemiology of sexual trauma, degree of offender violence, and severity of physical injuries are different when compared with those of women assaulted by a single assailant. Methods: This study was conducted in an urban teaching hospital throughout 6 weeks (October to November 2002). In randomized, 4-hour blocks, all men triaged to the acute care portion of the ED were eligible for enrollment. Men were excluded if they were younger than 18 years, refused to participate, had altered mental status, or needed lifesaving intervention. Data included demographics, the Revised Conflict Tactics Scale (CTS2), the Hurt, Insulted, Threatened, Screamed (HITS) scale, and the Partner Violence Screen (PVS). The CTS2 has population scores for female to male psychological aggression and physical assault. Neither the HITS (4 questions) nor the PVS (3 questions) has been validated for use in the ED for detecting male victims of IPV.

Results: During the study, a total of 116 men presented to the ED; 86 patients were eligible for inclusion and 32 patients refused consent. The mean age was 41.8 years. Race was 73% black, 20% white, 3% Hispanic, and 3% Asian. Thirty-nine percent (21/54) of men had positive scores as on the psychological aggression portion of the CTS2, whereas 20% (11/54) had positive scores on the physical assault portion of the CTS2. When the HITS was compared with the psychological portion of the CTS2, it had sensitivity of 30%, specificity of 88%, and a positive predictive value (PPV) of 60%. When the HITS was compared with the physical assault portion of the CTS2, it had sensitivity of 46%, specificity of 88%, and PPV of 50%. When the PVS was compared with the psychological portion of the CTS2, it had sensitivity of 35%, specificity of 88%, and a PPV of 64%. When the PVS was compared with the physical assault portion of the CTS2, it had sensitivity of 46%, specificity of 85%, and a PPV of 46%.

Conclusion: Neither the HITS nor the PVS accurately detects male victims of IPV in the ED. Further research is needed to develop a valid, brief screening test to identify male victims of IPV in an acute setting.

Results: EMS agencies with high ETI experience (group A) averaged 5.6 intubations per medic a year, and those with low experience (group B) averaged 2.9 intubations a year. For patients in cardiac arrest, ETI success rate was 98% in group A and 95% in group B (P=nonsignificant). In patients with perfusing rhythms, ETI success rate was 95% in group A and 74% in group B (P<.001). For DFI, ETI success rate was 93% in group A and 76% in group B (P<.001). Group A's success rate was similar for arrest, nonarrest, and DFI (P=nonsignificant).

Conclusion: Paramedics with widely ranging experience in ETI have similar success rates when performing the procedure on patients in cardiac arrest. Success rate is significantly higher for paramedics who perform ETI more frequently. The group with more experience was successful in more than 90% of patients, whether the procedure was performed in cardiac arrest, perfusing rhythm, or in DFI.

With Acute Coronary Syndrome Diercks DB, Kirk JD, Boghos E/University of California-Davis Medical Center, Sacramento, CA Study objectives: We describe the use of emergency medical services (EMS) in patients who present to the emergency department (ED) with a final diagnosis of acute coronary syndrome (ACS).

Methods: This is a retrospective analysis of all patients who presented to our tertiary center urban ED from January 2002 through September 2002, with a final diagnosis of ACS. Patient demographics including ethnicity, medical history, and cardiac risk factors were recorded. Final diagnosis of ACS was based on diagnostic testing suggestive of ischemia, myocardial infarction as defined by the European Society of Cardiology, angiographic lesions more than 70%, or a cardiac-related death during the initial hospital visit or 30-day follow-up period. Ethnicity was defined as white, Hispanic, black, Asian, and other. Univariate analysis was performed on the following variables to determine their influence on EMS use: age, history of coronary artery disease (CAD), diabetes, hypertension, hypercholesterolemia, tobacco use, family history of CAD, and ethnicity. Significant variables were entered into a logistic regression analysis.

Results: The study cohort comprised 243 patients. Of these patients, 97 (40%) presented to the ED by ambulance. The cohort was composed of 199 (49%) white, 32 (13%) Hispanic, 32 (13%) black, 13 (5%) Asian, and 42 (17%) other ethnicity patients. The regression analysis identified age odds ratio (OR) 1.02 (95% confidence interval [CI] 1.0 to 1.04; P=.017) and black ethnicity OR 0.41 (95% CI 0.17 to 1.01; P<.053) as predictors of EMS use.

Conclusion: After adjustment for potential confounding variables, EMS use increased with age; however, black ethnicity was associated with a decrease in use.

Tale Perez A, Kamin R, Morran M/Hartford Hospital, Hartford, CT Study objectives: Naloxone, an effective medication for the treatment of opioid intoxication, has well-established criteria for administration. In 2002, the United States faced a nationwide naloxone shortage. We believe the indiscriminate use of this medication persists. We investigated paramedics' adherence to indications for administration of naloxone.

Methods: We conducted a retrospective medical record review of patients given naloxone in the field between January 1 and March 31, 2003, in an urban tertiary care setting (emergency department annual volume 70,000). Records were screened for the following criteria: respiratory rate less than 6 breaths/min, pinpoint pupils, evidence of intravenous drug use, Glasgow Coma Scale (GCS) score less than 12, or cyanosis. Patients with at least 3 of 5 elements met criteria for naloxone therapy. An improvement to a GCS score of at least 14 and a respiratory rate of at least 10 breaths/min within 5 minutes of naloxone administration constituted a positive response.

Results: Twenty-two patients received naloxone. Of these, 3 patients met criteria. A respiratory rate less than 6 breaths/min was documented in 2 of 22 patients, a GCS score less than 12 was documented in 17 of 22 patients, and pinpoint pupils were documented in 13 of 22 patients. Two of the 3 who met criteria had a clinically documented response to naloxone. One of 19 who did not meet criteria (respiratory rate 8 breaths/min) had a positive response to naloxone.

Conclusion: Published criteria for naloxone administration is frequently not adhered to. Furthermore, indiscriminate use resulted in no clinical benefit. More events should be prepared to treat a wide variety of injury and illness. Additional investigations are needed to expand the body of literature and further delineate injury and illness patterns during these events. Results: During the study period, 2,327 search and rescue events were recorded; 50% for traumatic injuries, 24% for medical illness, 15% for missing persons, and 11% for assistance of uninjured or stranded individuals. The average age of a person rescued was 32 years (<1 year to 98 years), with males represented 62%. The average time of a search and rescue event was 5 hours 0 minutes. Fifty percent of the traumatic injuries involved the lower extremity, followed by the head (14%), the chest and abdomen (12%), the upper extremity (10%), and unspecified (14%). Thirty percent of these injuries were fractures. Thirty-one percent of these injuries resulted from a fall. The 2 most common medical reasons necessitating rescue were dehydration (25%) and hypothermia (14%). Sixty-nine percent of deaths were a result of traumatic injury, 22% were a result of drowning, and 11% were a result of cardiac arrest. Further analysis of the data is ongoing.

Conclusion: Yosemite National Park encompasses an area approximately the size of the state of Rhode Island, including 13,000-foot peaks, white-water rivers, raging waterfalls, vertical climbing walls, and miles of trails for hiking, camping, and crosscountry skiing. It hosts more than 4 million visitors a year. The National Park Service and Yosemite Search and Rescue are responsible for providing search and rescue services throughout the park. During the 10-year period from January 1990 through December 1999, more than 2,000 search and rescue events were recorded. Fifty percent of these events were a result of traumatic injuries, most commonly from falls, and 24% for medical illness, most commonly from dehydration. These results have important implications for those who provide search and rescue services in terms of planning, staffing, and education of personnel. They also identify several common conditions such as dehydration and lower extremity injuries that may be reduced through educational programs directed at park visitors.

injured in bicycle, moped, and motorcycle crashes on a rural island setting. I attempt to determine areas to be targeted for injury prevention interventions and assess the interventions' effectiveness over time to decrease the number of crashes.

Methods: Patients presenting to the island's hospital emergency department (ED) were prospectively surveyed about the data objectives and type of crash, level of training, potential cause of crash, and expense of care. Study groups were studied by calendar year on an island with a summer seasonal population of more than 105,000. E-codes were used to track potential cases, as was a survey tool used in the ED. Data were entered into a Microsoft Works database to calculate statistics. The study was approved by the hospital review board.

Results: There was a steady decrease in the number of moped crashes during the study period. The climate data suggested a potential cause of crashes because of sand blowing onto the road. Throughout the study period, a number of factors remained constant despite prevention interventions: a minority of moped riders had previously ridden a moped or motorcycle. The moped rider had received an average of fewer than 10 minutes of instruction before operating the vehicle and had planned to stay on the island for fewer than 3 days. There was no significant difference among the groups in regard to intoxicated drivers. Changes noted during the study period included an increase in bicycle and motorcycle crashes caused by off-road use and stunts.

Limitations: Data may be limited because the only patients surveyed were those who presented to the ED by EMS, walk-ins, or private vehicle. Weather during the year may have affected the rate of 2-wheeled vehicle use and, hence, the rate of crashes. One site was surveyed; results may not be generalized to other settings. Injury prevention interventions may be affected by confounding factors.

Conclusion: A significant number of moped crash victims use EMS for transport, potentially causing increased associated costs of care. The majority of moped users are seasonal visitors to the island who plan to stay fewer than 3 days. The interventions used during the study period may be having an effect on decreasing the number of moped crashes. It is recommended that continued educational efforts be targeted toward a structured driver education course and helmet use, particularly for bicycle riders.

Shukla PC/University of Texas Southwestern Medical Center, Dallas, TX Study objectives: Despite an increase in the number of motor vehicles that are equipped with air bags, an increasing number of emergency department (ED) patients report that they were unrestrained occupants at motor vehicle crash (MVC). Knowledge of predictors of use of motor vehicles with air bags (MVWAB) is important to address this discrepancy. I identify predictors of self-reported use of MVWAB in ED patients presenting after MVC.

Methods: This was a prospective, cross-sectional study of consenting patients older than 18 years who presented to the ED of a large tertiary-care teaching institution from June 1, 2001 , to May 30, 2002 . Anonymous data were collected for a convenience sample of ED patients by using a pilot-tested survey instrument of Likert scale-based 12 questions.

Results: Among 3,600 ED patients who were screened, 3,208 (89%) were eligible for the study. The study population included a derivation cohort of 800 patients who had been in an MVC and a validation cohort of 2,408 patients who had not been in an MVC. A bivariate analysis was used to determine the predictive factors with significant association with MVWAB use among patients who had been in an MVC (P<.05). The stepwise multivariate logistic regression model was used to control for the confounders, and it revealed 4 predictors of reporting MVWAB use: older age (odds ratio [OR] 2.3; 95% confidence interval [CI] 1.5 to 3.4), white race (OR 2.6; 95% CI 1.7 to 3.9), regular seat belt use (OR 2.4; 95% CI 1.6 to 3.6), and an attitude that air bags are highly effective (OR 2.9; 95% CI 1.6 to 5.2).

Conclusion: These data suggest that report of MVWAB use by ED patients after MVC is strongly influenced by age, race, use of seat belt, and their attitude toward air bag effectiveness. Health care professionals may use this valuable information in treating such ED patients and as an injury prevention tool in educating ED patients.

Shukla PC/University of Texas Southwestern Medical Center, Dallas, TX Study objectives: Because of recent negative media coverage of air bags, misper-cent of children aged 0 to 3 years were restrained in car safety seats, 83% of children aged 4 to 7 years were restrained in child booster seats or seat belts, and 79% of children aged 8 to 11 years were restrained mostly in seat belts. Ninety-two percent of respondents believed that car and booster seats protect child passengers and that younger and smaller children should be restrained in such devices every time they are in motor vehicles. Eighty-four percent of respondents believed that there should be laws requiring car or booster seat use. When respondents who reported that their children were restrained only "some of the time," "rarely," or "never" were asked about barriers to restraint use, the most common answer was "my child resists" (61%), followed by "I don't think restraints are needed" (17%) and "using restraints is inconvenient" (14%). The most believable sources of information were safety forces and medical professionals (ie, fire or emergency medical services, police, physicians, and nurses).

Conclusion: Children's restraint use declined steadily with increasing age of the child. Campaigns involving fire or emergency medical services, police, physicians, and nurses may be the most effective way to provide parents with information needed to overcome barriers to pediatric passenger restraint use.

Sorondo B, Lane PL, Dominici P/Albert Einstein Medical Center, Philadelphia, PA Study objectives: The purpose of this study is to compare motor vehicle crash (MVC) injury characteristics among occupants of passenger vehicles and light trucks. It was hypothesized that occupants using the 3-point seatbelt would have less severe injuries and lower mortality rates when having an MVC in a light truck compared with those in a passenger vehicle.

Methods: This is a population-based observational study. CODES (Crash Outcome Data Evaluation System) records obtained from the state Department of Health for 1994 through 2000 were used. This system merges information from police reports, emergency medical services (EMS), and hospital discharge information for the State of Pennsylvania by using CODES 2000 software. The Fatality Analysis Reporting System vehicle classification was used. The light truck category includes sport utility vehicles. ICDMAP-90 was used to calculate Injury Severity Score (ISS) and Abbreviated Injury Score.

Results: A total of 23,921 merged files for passenger vehicle and light truck MVC injury cases were used. Of those, 18,018 people wore a 3-point seatbelt: 14,162 of 3,856 involved passenger vehicle and light truck cases, respectively, including 13,908 drivers (D), 3,703 right front seat (RFS) passengers, and 407 rear seat (R) passengers. There were no statistically significant differences between mortality rates among the 3 groups when passenger vehicle cases were compared with light truck cases. There were higher proportions of head and face injuries in light truck-D than passenger vehicle-D crash victims (30.9% versus 28.4%, P=.0069; 7.8% versus 5.9%, P=.0002). There were higher proportions of thoracic and lower extremity injuries in passenger vehicle than in light truck cases (17.7% versus 15.9%, P=.018; 20.9% versus 17.8%, P=.0002). The mean ISS in passenger vehicle-D cases was higher than that in light truck-D cases for the right center impact configuration (passenger vehicle mean ISS 10.94, 95% confidence interval [CI] 9.98 to 11.859; light truck mean ISS 7.73, 95% CI 6.628 to 8.831; P=.0009). There was a higher proportion of thoracic injuries among passengers of passenger vehicles (17.64% versus 13.28%; P=.0042; 95% CI 0.1576 to 0.0713). There were no significant differences in mortality rates, injury severity, and location of the injury between passenger vehicle-R and light truck-R passengers.

Conclusion: Passenger vehicle-D had higher proportion of thorax and lower extremity injuries than light truck-D cases. Passenger vehicle-D had more severe injuries than light truck-D cases in right center impact crashes. Passenger vehicle-RFS passengers experienced more thoracic injuries than light truck-RFS cases.

Discussion: The mass-differential considerations would suggest that light truck drivers should experience less severe injuries and fewer deaths. These data do not support those conclusions. Limitations include the accuracy of the records involved and of the CODES matching process. The study also does not account for human factors involved in MVC causation.

Hirshberg AJ/Martha's Vineyard Hospital, Oak Bluffs, MA Study objectives: I compare demographic data, emergency medical services (EMS) involvement rates, disposition, helmet use, and injury data among the patients Results: To date, follow-up is complete on 100 patients in group A and 31 patients in group B. Baseline regular helmet use rate in group A before intervention was assumed to be 0%. Baseline regular helmet use rate in group B was 35.5% (11 of 31) according to preintervention questionnaires. At 2 weeks postintervention, regular helmet use rate was 56.0% in group A and 38.7% in group B. At 6 months, regular helmet use rate was 66.0% in group A (66 of 100) compared with 45.2% in group B (14 of 31), which was statistically significant (χ 2 =4.3; P=.038). Knowledge of the statewide bicycle helmet law increased for parents from 77.1% baseline in both groups to 91.6% at 6 months, with no significant difference between groups. Knowledge of the helmet law for child patients increased from 48.4% baseline to 80.9% at 6 months, with more group A patients knowing about the helmet law than group B patients at 6 months (88.0% versus 58.1%; χ 2 =13.7; P≤.0005).

Conclusion: Distribution of free bicycle helmets and safety education from an ED to low-income families increases regular helmet use rate in children previously without helmets to a level that surpasses the rate in children who already owned helmets. Also, patients and parents of both groups appear to have an increase in bicycle helmet law knowledge as a result of this safety intervention, with a greater knowledge impact on children who had previously not owned a helmet. EDs are potentially powerful settings for effective bicycle safety programs.

Melniker LA, Briggs WM, Mancuso C/Weill Medical College of Cornell University, New York, NY Background: The Trauma Injury Severity Score (TRISS) estimates the probability of death of patients who have sustained physical trauma. TRISS has been in use for more than a decade for research purposes to allow comparison of survival probabilities from various practice settings. TRISS incorporates the type and severity of injury but not comorbidity.

Study objectives: We proposed to increase the predictive accuracy of TRISS, given that comorbidity has been established as a predictor of mortality and that TRISS does not include the effect of comorbidity on the probability of death. Using the National Trauma Data Bank of the American College of Surgeons, we proposed to develop and validate a revised TRISS model, including comorbidity.

Methods: The National Trauma Data Bank includes demographic characteristics, types of injuries, TRISS, comorbidity, and survival data. Using the generalized linear modeling procedure in SAS software, we compared the probability of TRISS death to the actual death rate. Then, using split halves of the data, we trained the revised model for probability of death by including the Charlson Comorbidity Index. We tested the original TRISS model and the revised model with comorbidity by using the other half of the data. The value of the revised model including the comorbidity variable was analyzed with the Briggs and Rupert Skill Score.

Results: There were 310,136 patients with blunt trauma. For this group, the TRISS probability of death was 0.038 (SD 0.001), whereas the actual death rate was 0.046 (SD 0.001), a 17% underestimation of death. The analysis of probability of death with a comorbidity variable demonstrated that more comorbidity was associated with a higher probability of death (P<.0001). The revised TRISS equation, testing comorbidity, revealed a significant skill score of κ equal to 0.04, with a P value equal to .0001.

Conclusion: In this study, we found an underestimation of the probability of death in patients with blunt trauma by using the original TRISS equation. The revised TRISS, incorporating comorbidity, is a more accurate predictor of death in older patients with blunt trauma and, therefore, is a more useful instrument in the research setting.

Bruns Jr J, Gottesfeld S, Hahn S, Jagoda A/Mt. Sinai School of Medicine, New York, NY Study objectives: Emergency department (ED) management of mild traumatic brain injury (MTBI) remains controversial, and little has been published about clinical practice patterns. The objectives of this study were to determine current US physician practice patterns and to assess the correlation between current ED practice and the American College of Emergency Physicians' (ACEP) recently published MTBI guidelines.

Methods: Physicians attending the 2002 annual meeting of a large emergency medicine professional organization were asked to complete a 2-page survey indicating

A ceptions about the risks and benefits of air bags are likely to occur, which could lead to undermining of public support for the use of this type of protective restraint. Objective: I determine the prevalence of various attitudes toward air bags among emergency department (ED) patients and assess knowledge and attitudes of ED patients toward safety and effectiveness of air bags in preventing motor vehicle crash-related serious injuries.

Methods: This was a prospective cross-sectional study of patients aged 18 years or older who were treated during 2001 at a tertiary care county teaching hospital ED. Anonymous data were collected through use of 22 knowledge-and attitudebased questions in structured interviews of patients during the ED visit.

Results: In a convenience sample of 3,208 patients, only 30% of patients reported having a motor vehicle equipped with air bags. Use of an air bag was perceived to be effective (preventing/reducing frequency of motor vehicle crash-related serious injuries) by 20% of patients (versus perceived ineffective by 80% of patients; P<.001). Only 15% of patients believed that air bags are safe (unlikely to result in injuries; versus perceived unsafe by 85% of patients; P<.001). Also, 86% of patients thought that air bags might be harmful to children. Thirty-eight percent of patients reported a favorable attitude toward use of a motor vehicle equipped with air bags (versus negative attitude in 62% of patients; P<.001). In addition to misconceptions, lack of using a car with air bags was also associated with high risk-taking behavior, younger age, and Hispanic race (P<.001).

Conclusion: In this study, the majority of patients did not have a proper understanding of air bag effectiveness and safety. These findings have implications for the design of ED-based education and interventions promoting use of cars with air bags in this at-risk population. If successful, such interventions might lead to important public health benefits.

Kwon KT, Zlidenny AM, Vaca FE/University of California-Irvine Medical Center, Orange, CA Background: Fatal head injuries make up 90% of bicycling deaths. Helmet use has been shown to reduce up to 85% of bicycling-related head injuries. Unfortunately, most studies report low bicycle helmet use in children, with a lower helmet use rate in children from lower socioeconomic backgrounds. Previous interventional studies have shown mixed results on the effects of educational programs with free helmet distribution or subsidies. The effectiveness of a bicycle helmet distribution and safety promotion program based in an emergency department (ED) setting has not been investigated.

Study objectives: We quantify the effect of free bicycle helmet distribution and safety information to children of low-income families presenting to an ED. Our primary objective was to estimate helmet use rate in children receiving free helmets and to compare that rate with that of children of low-income families who already owned helmets. Secondary objectives were focused on assessing changes in knowledge and attitudes of bicycle safety in children and parents as a result of the educational intervention.

Methods: The setting for this prospective interventional study was an urban academic ED with an annual census of 45,000 visits. Children aged 5 to 16 years, with regional Medicaid insurance, and presenting to our ED as patients or related to persons being treated were eligible for enrollment. Ownership or access to a bicycle was required. New bicycle helmets and safety material were provided free by the regional county health care agency. Patients in the study were divided into 2 groups: group A were patients who did not already have a helmet, and group B were patients who already owned a helmet. Patients and parents were given a tested bicycle safety questionnaire before an educational intervention consisting of a 15-minute bicycle safety video and the distribution of bicycle safety brochures. Additionally, patients in group A were given a helmet properly fitted by trained study personnel. Followup telephone interviews of patients and parents were conducted at approximately 2 weeks and again at 6 months after the ED visit by using a 35-question tested survey instrument. Parents and patients were interviewed independently. A patient was considered a regular helmet user if he or she met all of the following criteria: parent stated that child always used a helmet, child stated that he or she always used a helmet each time he or she rode a bicycle in the past month, child stated he or she used a helmet the last time of bicycle use, and child stated that he or she planned to use a helmet at next bicycle use. Differences among groups were compared with Pearson χ 2 analysis.

Hutson HR, Anglin D, Rice P/Keck School of Medicine, University of Southern California, Los Angeles, CA Study objectives: In 1999, 43 million people in the United States had face-to-face contact with law enforcement. Of these people, 500,000 were hit, kicked, pushed, choked, hit with a flashlight or baton, or restrained by a K-9 dog; sustained pepper spray injuries; or were threatened with a firearm. Usually, law enforcement officers use only enough force to control or subdue an individual. At times, they exceed the bounds of necessary force, leading to excessive use of physical force. How often excessive use of physical force by law enforcement officers occurs is unknown. Anecdotally, many emergency physicians have treated patients that they suspected had sustained trauma from excessive use of physical force by law enforcement officers.

Objective: We determine emergency physicians' experience, clinical practice, education and training, and perspectives involving patients injured by suspected use of excessive physical force by law enforcement officers.

Methods: This study consisted of a mailed survey to a random sample of all fulltime academic emergency physicians (2,239) in the United States. The survey was conducted from May 2002 to December 2002 and had up to 4 mailings. The questionnaire contained 36 questions.

Results: Three hundred fourteen (80%) of 393 emergency physicians responded. Seventy-four percent of respondents were men. The race of respondents was white (88%), Asian (6%), Hispanic (4%), and black (3%). Ninety-seven percent of emergency physicians have treated patients they suspected were victims of excessive use of physical force by police. Sixty-seven percent of emergency physicians estimated recognizing more than 2 cases per year. Ninety-two percent of cases involved injuries caused by blunt trauma: batons (32%), flashlights (25%), and fists and kicks (47%). Emergency physicians at public teaching emergency departments (EDs) were 4.2 (95% confidence interval 1.95 to 9.21) times more likely to treat patients they suspected had been subjected to excessive force than emergency physicians at university or community teaching EDs. Seventy-four percent of cases of suspected use of force went unreported by emergency physicians. Ninety-six percent of EDs have no policies or guidelines for managing cases suspected of involving excessive use of force by law enforcement officers. Seventy-one percent of emergency physicians believe that it is within their scope of practice to refer cases of suspected excessive use of force for investigation, and 50% believe that emergency physicians should be legally required to report cases of suspected excessive use of force by police. Ninetythree percent of respondents had no education or training in the management of their usual evaluation, neuroimaging, and disposition practices for several MBTI clinical scenarios. These scenarios included various patient ages, mechanisms of injury, and levels of neurologic impairment. Data were also obtained about the respondents' level of training, practice environment, and familiarity with the MTBI guidelines.

Results: Two hundred one emergency medicine residents and 501 emergency medicine attending physicians completed the survey; 89% practiced in the United States. Although only 44% of respondents were aware of the ACEP MTBI guidelines, approximately 70% of the average physicians' responses were consistent with the guidelines. Attending physicians had more correct (ie, consistent with the guidelines) responses on average than residents. Thirty-six percent reported that the MTBI guidelines had changed their clinical practice. Ninety-six percent correctly responded that normal plain skull radiographs were not predictive of intracranial lesion. Fifty percent do not follow recommended guidelines for discharging patients with a negative computed tomography (CT) scan result. Eighty-six percent thought it was important to identify any intracranial lesion, including lesions not requiring neurosurgical intervention. Sixty-one percent ordered more head CTs than are recommended by the ACEP guidelines.

Conclusion: Emergency physicians' treatment of MBTI patients is generally good, but they obtain more CT scans for MTBI than currently recommended. Increased physician awareness of ACEP MTBI guidelines may facilitate a more selective approach to neuroimaging.

Korn CS, Hollinger M, Franklin A, Henderson SO/Keck School of Medicine, University of Southern California, Los Angeles, CA Study objectives: We describe our experience with all patients presenting to a Level I trauma center with traumatic spinal cord injuries (TSCI).

Methods: This was a retrospective review of all TSCI patients presenting to a Level I trauma center from 1997 to 2002. Data included age, sex, mechanism of injury, emergency department (ED) diagnosis, and the administration of the corticosteroid methylprednisolone in the ED.

Results: One hundred ninety-five patients presented to the ED, and all medical records were available for review. One hundred forty-two of 195 (73%) were male patients, with a mean age of 33.5 years (range 8 to 88 years). The mechanism of injury was divided into 6 categories: falls, 57 of 195 (29%) patients; motor vehicle crashes, 56 of 195 (28%) patients; gunshot wounds, 53 of 195 (27%) patients; crush injuries, 11 of 195 (6%) patients; auto versus pedestrian, 11 of 195 (6%) patients; and other, 7 of 195 (4%) patients, which included penetrating stab wounds. The ED diagnosis had TSCI or rule out TSCI on 81 (42%) of 195 of the records, followed by extremity paralysis 77 (39%) of 195 records and central cord syndrome 19 (10%) of 195 records. The remainder of the patients (18/195; 9%) were diagnosed with spinal shock, complete cord transection, Brown-Séquard syndrome, or anterior cord syndrome. Methylprednisolone was administered to 148 (76%) of 195 of the patients in the ED. Patients with blunt trauma (139/195; 71%) received methylprednisolone 89% (124/139) of the time during their ED stay, whereas patients with penetrating trauma (56/195; 29%) received methylprednisolone 43% (24/56) of the time. Beginning in 2000, the administration of methylprednisolone to penetrating trauma patients dropped to 5 (14%) of 35 patients.

Conclusion: The majority of TSCI patients treated in this population were male patients who had fallen, were involved in a motor vehicle crash, or had sustained a penetrating trauma by gunshot or stabbing. The diagnosis of TSCI was noted on the majority of records, followed by extremity paralysis. Methylprednisolone was given to 76% of the patients in the ED; the majority of the patients were victims of blunt trauma. A decrease in methylprednisolone administration was seen in penetrating trauma victims beginning in 2000.

Jackson RE, Walters B, Sipp S, Scipione C, Freimer G, Swor RA/William Beaumont Hospital, Royal Oak, MI Study objectives: Recent publications have had conflicting conclusions on the risk of Coumadin use on intracranial hemorrhage after blunt trauma. Our objective is to retrospectively estimate the odds ratio (OR) of anticoagulation medication for intracranial hemorrhage (ICH) in emergency department (ED) patients presenting to our American College of Surgeons Level I ED.

Results: A total of 1,210 records were reviewed; 71 (5.9%) patients had orthopedic PI. Of those patients, 10 were found to have evidence of pelvic fracture by radiography but a negative HPE (sensitivity 85.9%). When the fracture patients were divided into the nonintoxicated group versus intoxicated group, the recalculated sensitivities were 89.1% and 75.0%, respectively, with no statistically significant differences (P=.15). Sensitivities were also calculated by comparing patients with a GCS score less than 13 with patients with a GCS score of 13 or greater. In the group with a GCS score of 13 or greater, the HPE picked up injuries with a sensitivity of 91.5% (54/59). For patients with GCS score less than 13, the sensitivity was 58.3% (7/12; P=.002). The logistic regression model showed a correlation between higher GCS score and having a positive HPE in the presence of PI.

Conclusion: In these data, there was a statistically significant decrease in sensitivity of the clinical examination to detect PI in patients with GCS score less than 13 versus those with GCS score of 13 or greater. The logistic regression model did not find a significant correlation between BAL and the sensitivity of the HPE for fracture or dislocation.

Davila A, Laurora M, Fulton J, Jacoby JL, Heller MB, Reed J/St. Luke's Hospital, Bethlehem, PA Study objectives: Urushiol exposure from Toxicodendron (poison ivy, poison oak, or poison sumac) is the most common cause of allergic contact dermatitis (ACD). To the authors' knowledge, no study shows that application of any agent after exposure effectively prevents binding of urushiol to the skin and ameliorates ACD. We investigated whether a new topical agent, Zanfel, a mixture of alcohol solubles and anionic surfactants, prevents or treats experimentally induced ACD.

Methods: This was a randomized, double-blind, crossover design, experimental study of 24 consenting volunteers. With each subject serving as his or her own control, both forearms received three 2.5-cm 2 applications of 1.5 µg of urushiol. At 3 intervals after exposure (48, 96, and 144 hours), the sites were treated with placebo or Zanfel. Pretreatment and posttreatment ACD reaction scores and subjective itch scores were compiled each time. ACD was scored with a 0-to 6-point Likert scale according to erythema, induration, and vesiculation; itch was quantified with a 10cm visual analog scale. Comparisons were by paired t test, with α set at .05.

Results: At all times after urushiol exposure, Zanfel significantly improved the Toxicodendron-induced ACD compared with placebo. The difference in ACD scores at 48, 96, and 144 hours all favored the Zanfel group versus placebo (-0.76 versus -0.24, 0.32 versus 0.74, and 0.55 versus 0.96, respectively; all P<.05). Itch scores improved for both groups at all 3 times, with a trend favoring the Zanfel group at 96 hours (P=.058).

Conclusion: The new topical agent Zanfel, when applied after exposure to urushiol, ameliorates or prevents Toxicodendron-induced allergic contact dermatitis.

Moreno R, Lowe RA, Brooks HS, Lapidus J/Oregon Health and Science University, Portland, OR Study objectives: Emergency department (ED) management of nausea and vomiting has recently become challenging because of multiple drug shortages and warnings against droperidol. We sought to determine baseline patterns and predictors of ED antiemetic use.

Methods: We used the most recently available (2000) National Hospital Ambulatory Medical Care Survey, a national probability sample of US ED visits. All visits with a chief complaint or final diagnosis of nausea or vomiting were analyzed. National estimates and variances were calculated using SUDAAN software. Predictors of antiemetic use were examined by using multiple logistic regression.

Results: Inclusion criteria were met by 1,873 ED patients, representing 8,061,301 estimated visits nationally. Mean patient age was 32 years, with 62% female patients and 55% non-Hispanic white patients. The most common diagnoses were gastroenteritis (13%) and abdominal pain (11%). Seventeen percent of patients were admitted. Pharmacologic antiemetic therapy was used in 42% of visits. The most commonly prescribed antiemetics were promethazine (25% of visits), prochlorperazine (9%), and droperidol (3%). In multivariate analyses, patients receiving intravenous fluids were more likely to receive antiemetics (odds ratio [OR] 2.4; 95% confidence inter-

A Conclusion: This study shows that the vast majority of emergency physicians have treated patients they suspected of being injured because of excessive use of physical force by law enforcement officers. The majority of cases went unreported, although most emergency physicians believe that these cases should be referred for investigation. This study also points out the need for education and training in the treatment of patients with injuries from suspected excessive use of force. The American College of Emergency Physicians, the Society for Academic Emergency Medicine, and the American Academy of Emergency Medicine should develop treatment guidelines and educational programs for patients with injuries from suspected use of excessive physical force by law enforcement officers.

Allen TL, Carter JL, Harker CP, Stevens MH/LDS Hospital, Salt Lake City, UT Study objectives: Venous thromboembolic (VTE) disease remains a significant risk factor for trauma patients. Many trauma patients experience injuries that preclude anticoagulation. Inferior vena cava (IVC) filters that can be retrieved might prove beneficial in the trauma patient at high risk for VTE and bleeding complications because of anticoagulation.

We review the procedural, VTE, and pulmonary embolism (PE) complication rates of a group of trauma patients treated at a single institution with an IVC filter that is designed to allow retrieval.

Methods: Trauma patients at risk for VTE were identified and treated by institutional protocol. A subset of these patients, who had contraindications to chemical and mechanical prophylaxis, were treated with an IVC filter that could be retrieved within 14 days or left in situ permanently. A retrospective review of medical records was used to identify these patients, indications for IVC filter placement, and complications associated with the use of the filter.

Results: From January 2001 to March 2003, 34 patients received the retrievable filter. Twenty-two filters were placed prophylactically, whereas 12 were placed because of demonstrated VTE. Retrieval was attempted in 19 cases and was successful in 18 of 19 cases. One attempted retrieval was unsuccessful for technical reasons. There was no attempt at retrieval in 15 cases. There were no complications of bleeding, device migration, or infection and no episode of PE occurred with the filter in place.

Conclusion: An IVC filter that can be retrieved appears safe and efficacious for the prevention of PE in the high-risk trauma patient who cannot receive anticoagulation therapy. Routine use of such a filter may help prevent the morbidity associated with permanent IVC filter placement.

Williams E, Fasano C, Bryant E, Czincila R, Derzie A, Roloff C, Lam A, Sorondo B/Albert Einstein Medical Center, Philadelphia, PA Study objectives: Studies have shown that combination of medical history and physical examination (HPE) is sensitive for detecting pelvic orthopedic injury (PI) in the alert, nonintoxicated patient. For the blunt trauma patient who is intoxicated or with depressed sensorium (Glasgow Coma Scale [GCS] score <13), there is conflicting literature on whether the clinical examination is still sensitive for PI. The purpose of the study is to evaluate the sensitivity of the HPE for detecting patients with depressed sensorium or blood alcohol level (BAL) of 100 mg/dL.

Methods: This was a retrospective, observational study. Records from blunt trauma patients admitted to an urban Level I trauma center from January 1, 1997, to November 30, 2002, were reviewed. The hospital trauma registry database was used to identify patients with International Classification of Diseases, Ninth Revision, Clinical Modification codes consistent with pelvic fractures and dislocations (PI). Inclusion criteria were age 16 years or older and a documented BAL and GCS score. Sensitivity of the HPE to diagnose PI in patients with GCS score less than 13 and GCS score of 13 or greater was assessed and compared. The same analysis was performed in patients with BAL less than 100 mg/dL (nonintoxicated) or 100 mg/dL or greater (intoxicated). Logistic regression model was used to assess the correlation between BAL and GCS score with the sensitivity of the HPE to diagnose PI.

Among inpatients and outpatients, H influenzae (n=1,239), ampicillin resistance was more than 30%; however, all isolates were susceptible to ceftriaxone and levofloxacin. Resistance rates were less than 5% for amikacin, cefepime, and ceftriaxone among LRTI for inpatients and outpatients among K pneumoniae (n=4,776 and 821) and E coli (n=3,160 and 550), whereas fluoroquinolone resistance rates were more than 6% among K pneumoniae and more than 21% among E coli. For CNS, the incidence of key pathogens isolated was coagulase-negative staphylococci (45.2%; n=5,169, inpatient and outpatient), Staphylococcus aureus (10.6%; n=1,214), S pneumoniae (5.7 %; n=650), E coli (2.4%; n=276), Streptococcus agalactiae (1.1%; n=123), H influenzae (0.7%; n=79), and Neisseria meningitidis (which have no interpretive criteria; 0.5%; n=59). Of S aureus, 72.4% (n=1,130) were oxacillin susceptible. Among oxacillin-susceptible isolates, 99.5% of isolates tested were susceptible to ceftriaxone (n=197). Of S pneumoniae, 85.2% (n=318; 3.5% resistant) were susceptible to ceftriaxone, and 100% (n=37) of S agalactiae were susceptible to ceftriaxone. S pneumoniae and S agalactiae were 54.6% susceptible (n=449; 18.3% resistant) and 100% susceptible (n=100) to penicillin, respectively. Of E coli, 96.7% were susceptible to ceftriaxone and 100% were susceptible to imipenem. For H influenzae, 26.2% (n=61) were ampicillin resistant but 100% (n=52) ceftriaxone susceptible.

Conclusion: Despite widespread resistance to penicillin derivatives, for S pneumoniae and H influenzae, resistance to ceftriaxone remains uncommon. After 18 years of clinical use, ceftriaxone retains high efficacy against organisms causing LRTI and CNS infections, with the exception of methicillin-resistant staphylococci. Antimicrobial surveillance data offer guidance to physicians for empiric prescribing, for which ceftriaxone remains a good choice, with broad-spectrum coverage against the predominant pathogens. However, treatment regimens for patients with serious infections should be developed after consideration of symptoms, laboratory findings, and relevant medical history and in the context of appropriate local and regional antimicrobial resistance trends.

Gums JG/University of Florida, Gainesville, FL Study objectives: Staphylococcus aureus bacilli frequently are isolated from skin and skin structure specimens. Infection severity determines antimicrobial treatment, with initial empiric therapy guided by reported susceptibility patterns of isolated etiologic bacteria. The Centers for Disease Control and Prevention recommends targeting definitive therapy to known pathogens to prevent antimicrobial resistance.

Methods: The ongoing Antimicrobial Resistance Management (ARM) program was initiated in 1997 with establishment of a surveillance database. To date, 251 US hospital laboratories from 6 regions (northeast, north central, southeast, south central, northwest, and southwest) have submitted more than 17 million inpatient and outpatient isolates representing 16 organisms and 44 antibiotics. Data from antibiograms and sensitivity reports of S aureus, methicillin-susceptible S aureus, and methicillin-resistant S aureus (MRSA) isolates from 1990 to 2002 were reviewed by using a Web-based analysis tool to determine resistance to commonly prescribed antibiotics, including fluoroquinolone (ciprofloxacin, levofloxacin) and other antibiotics (vancomycin, clindamycin, erythromycin, and nafcillin/oxacillin).

Results: Total number of isolates and percentage of isolates resistant to each antibiotic were determined nationally and regionally. Nationally, S aureus isolates were more resistant to levofloxacin (41.4%; n=123,868) than to ciprofloxacin (38.7%; n=256,178), with greater resistance to levofloxacin seen in north central (47.9% versus 39.4%) and northeast (52.4% versus 41.9%) regions. These data suggest cross-resistance between ciprofloxacin and levofloxacin. Resistance to ciprofloxacin was greater than to levofloxacin in south central (34.2% versus 20.0%) and southwest (32.2% versus 23.0%) regions. S aureus isolate susceptibility to erythromycin nationally was 51.1% (n=274,873). Compared with erythromycin, S aureus isolates had a much greater susceptibility to clindamycin (73.6%; n=165,683). This difference was seen in every region, with the smallest comparative difference noted in the north central (52.5% versus 54.3%) region. Susceptibility to nafcillin/oxacillin nationally was 64.9% (n=360,460), which ranged from 62.2% in north central and northeast regions to 72.8% in southwest regions. Most of the change in susceptibility to nafcillin/oxacillin and ciprofloxacin throughout the past decade has occurred in the past 5 years (1998 to 2002) , with ciprofloxacin sensitivity declining with increasing levels of MRSA. Percentages of S aureus isolates remaining susceptible to vancomycin were similar (range 99.8% to 99.9%). val [CI] 1.8 to 3.4). Pediatric patients aged between 6 months and 13 years were less likely to receive antiemetics (OR 0.65; 95% CI 0.43 to 0.99), as were black patients (OR 0.63; 95% CI 0.44 to 0.91).

Conclusion: Antiemetics are administered to a minority of ED patients with nausea or vomiting, with most patients receiving promethazine. Multivariate analyses suggesting that pediatric and black patients receive antiemetics less frequently require further study. Further studies will track changes in antiemetic use in subsequent years.

Davis D, Grossman K, Vilke G, Kiggins D, Chan T/University of California-San Diego School of Medicine, San Diego, CA; San Diego State University, San Diego, CA Study objectives: The clinical presentation of aortic dissection (AD) may be similar to that for cardiac ischemia and pulmonary embolus; however, anticoagulation may be detrimental in patients with AD. This retrospective cohort analysis investigates the emergency department (ED) anticoagulation of patients ultimately diagnosed with AD.

Methods: Patients admitted through the ED with a discharge diagnosis of nontraumatic AD were identified by using International Classification of Diseases, Ninth Revision codes. Data about demographic characteristics, clinical presentation, ECG and radiograph results, and hospital course were abstracted from ED and hospital medical records. Patients administered anticoagulants (thrombolytics, heparin, or antiplatelet therapy) in the ED were compared with nonanticoagulated patients with regard to presenting symptoms, chest radiograph and ECG findings, and outcome.

Results: A total of 44 ED patients with nontraumatic AD were identified throughout a 4-year period. Anticoagulants were administered in the ED to 8 (18%) patients. ED-anticoagulated and nonanticoagulated patients were similar with regard to the incidence of chest pain (75% versus 67%; P=.647) and back pain (13% versus 42%; P=.121). ED-anticoagulated patients were more likely to have ST elevations or depressions on ECG (88% versus 7%; P<.001) and less likely to have a widened mediastinum on chest radiograph (0% versus 66%; P=.005). Two ED-anticoagulated patients died from uncontrolled hemorrhage before operative repair, 1 patient required a second surgery for bleeding complications, and another experienced a stroke after reversal of anticoagulation.

Conclusion: There is a clinically significant incidence of anticoagulation administration to ED patients ultimately diagnosed with AD, especially in the presence of ambiguous ECG and radiographic findings. Future studies should focus on better screening strategies to avoid anticoagulation-related complications. Limitations include the retrospective cohort design and relatively small sample size.

Volturo G, Jones ME, Draghi DC, Karlowsky JA, Thornsberry C/University of Massachusetts Memorial Health Care, Worcester, MA; Focus Technologies, Herndon, VA Study objectives: Severe community-acquired pneumonia, often caused by Streptococcus pneumoniae and Haemophilus influenzae; hospital-acquired pneumoniae, often caused by Klebsiella pneumoniae and Escherichia coli; and infections of the central nervous system (CNS) are infections that require urgent treatment. These infections require immediate empiric parenteral antibiotic therapy, often β-lactams such as ceftriaxone. Because emerging antibiotic resistance can complicate empiric drug choice, we reviewed recent in vitro susceptibility for ceftriaxone and other clinically relevant comparators.

Methods: Data from The Surveillance Network Database USA, an electronic surveillance system that collects routine susceptibility test results as reported to physicians from more than 300 sites in the United States, were studied. Only data from unrepeated isolates of S pneumoniae, H influenzae, K pneumoniae, and E coli were included from inpatient and outpatient lower respiratory tract infection (LRTI) specimen sources from 2002 and for prevalent organisms derived from CNS sites from 2000 to 2002 . NCCLS (2003 breakpoints were used to interpret as susceptible, intermediate, or resistant (including ceftriaxone breakpoints for meningeal and nonmeningeal isolates of S pneumoniae).

Results: Resistance rates for LRTI among S pneumoniae inpatients and oupatients (n=2,764 and 1,034, respectively) for vancomycin (0, 0%), levofloxacin (1.0, 1.9 %), and ceftriaxone (1.7, 1.8%) were the lowest of the agents tested. For both patient groups, penicillin and macrolide resistance in S pneumoniae was more than 18%.

(65%). Forty-six percent of the nesiritide-treated patients were admitted to the ICU/cardiac care unit, and 14% were mechanically ventilated, compared with 22% and 8%, respectively, of patients not treated with nesiritide. Patients (29/37 treated with nesiritide) typically received intravenous diuretics within 1.3 hours (median; range 0 to 17 hours) of admission, whereas median time to administration of intravenous vasoactive therapy was 8.0 hours (range 0 to 251 hours) for the 37 patients. Nesiritide-treated patients also received the following intravenous vasoactive drugs: dopamine (22%), dobutamine (19%), nitroglycerin (14%), nitroprusside (3%), and milrinone (3%). During index hospitalization, 54% (20/37) of the nesiritide-treated patients received dialysis treatment, whereas 82% (1691/2059) of the nonnesiritide-treated patients required dialysis during hospitalization. At discharge, 30/37 of the nesiritide-treated patients had improved symptoms. Of the remaining 7 patients, 6 died (4 of these patients received dialysis at index visit) during hospitalization, and data were missing for 1 patient. Paired weight data, available from 24 of the nesiritide-treated patients, showed a median weight change of -3.3 kg during hospitalization and -2.5 kg for those not treated with nesiritide. The median inpatient length of stay for nesiritide-treated patients was 9.1 days (range 1 to 27 days), whereas the non-nesiritide-treated patients had a median length of stay of 3.7 days (range 0 to 69 days).

Conclusion: Chronic dialysis patients who received nesiritide represent a population with a high morbidity, which may require long hospital stays, ICU admissions, and mechanical ventilation. Despite the severity of illness in this group, the nesiritidetreated patients tended to not need dialysis (relative to non-nesiritide-treated patients), experienced substantial symptomatic improvement, and were able to achieve appreciable weight loss (even though only 54% received dialysis during hospitalization). Chronic dialysis patients with acute heart failure represent a particularly ill group of patients who could benefit from nesiritide therapy. Study objectives: Recently, the Food and Drug Administration warned of a prolonged QT interval and torsade de pointes (TdP) as a complication of droperidol use. We sought to detect a change in the corrected QT interval (QTc) and identify episodes of TdP in noncritical emergency department (ED) patients who received droperidol.

Methods: The EmSTAT ED electronic database was searched from January 1, 1997 , to November 30, 2001 , for all noncritical patients who received droperidol and an ECG. Data were analyzed in 3 groups: patients with an ECG only before droperidol treatment, only after droperidol treatment, and before and after droperidol treatment. ECGs were reviewed for QTc intervals. ECGs with atrial fibrillation or flutter, right or left bundle-branch block, or paced rhythm were excluded from Conclusion: Nationally and regionally, the majority of S aureus and MRSA isolates remain sensitive to vancomycin. Nafcillin/oxacillin, ciprofloxacin, and levofloxacin show similar S aureus resistance rates, suggesting that even in methicillinsusceptible infections, resistance to fluoroquinolones is increasing. Therefore, fluoroquinolones may not be a reasonable choice for empiric therapy of skin and skin structure infections.

Bear TC, Stewart LJ, Eichhorn L, Duldner Jr JE/Case Western Reserve University School of Medicine, Cleveland, OH; Akron General Medical Center, Akron, OH; Northeastern Ohio Universities College of Medicine Study objectives: Data describing the emergency department (ED) and subsequent inpatient antibiotic use in meningitis is limited. We evaluated ED and inpatient antibiotic use in patients with suspected meningitis.

Methods: This was a multihospital retrospective review. Patients with a discharge International Classification of Diseases, Ninth Revision diagnoses of aseptic, septic, or fungal meningitis from January 1, 1996, to December 31, 2000, were included. Variables related to ED antibiotic use, inpatient antibiotic use, and outcome were studied.

Results: Of 319 patients, 43.8% were female patients. Median age was 20.5 years (range 1 week to 85 years). Race distribution was 32.9% black patients, 46.7% white patients, and 9.7% Hispanic patients. Major meningitis subtypes included aseptic (83.7%), bacterial (10.6%), and fungal (3.1%). Fourteen (4.4%) patients were HIV positive. Antibiotics were administered to 209 (65.5%) patients and differed according to meningitis etiology (P<.001). Ceftriaxone was the most commonly prescribed antibiotic (80.9%). Only 56 patients received a second antibiotic, of whom 3 also received acyclovir. Three patients with septic meningitis did not receive antibiotics in the ED. Time from ED triage to antibiotic administration was 270.1±11.8 (mean±SEM) and was administered earlier in septic versus aseptic etiology (P=.008). Differences were noted between inpatient and ED antibiotic use. Fifty-seven (51.8%) patients who did not receive antibiotics in the ED did so after admission, and antibiotics were not continued in 18 of 209 patients who received antibiotics in the ED (P<.001). In HIV-positive patients, 21.4% (3/14) received ceftriaxone in the ED, but 57.1% (8/14) received antifungal agents after admission. Survival rate to hospital discharge was 99.3% and was not related to etiology (P=.56).

Conclusion: Antibiotics are commonly prescribed in meningitis. Discrepancies between ED and inpatient antibiotic use exist despite microbial etiology and particularly in HIV-positive patients.

Nesiritide (B-Type Natriuretic Peptide): Data From the Acute Decompensated Heart Failure National Registry Peacock F, Emerman CL, Heywood JT, Abraham WT, Chang SF/The Cleveland Clinic, Cleveland, OH; Case Western Reserve University, Cleveland, OH; Loma Linda University, Loma Linda, CA; The Ohio State University, Columbus, OH; Scios, Inc., Sunnyvale, CA Study objectives: Nesiritide, a rapid-acting systemic arterial, venopulmonary, and pulmonary dilator, has been shown to improve dyspnea and to lower filling pressures in acute heart failure patients with moderate to severe renal insufficiency, including those who are oligoanuric. We describe here the characteristics of chronic dialysis patients from the Acute Decompensated Heart Failure National Registry (ADHERE) database.

Methods: The ADHERE registry is an ongoing retrospective study in which more than 260 US hospitals are contributing baseline, therapeutic, and outcomes data from patients with a primary diagnosis of acute heart failure during the index hospitalization. Characteristics of patients who have complete data as of April 2003, reported to be receiving chronic dialysis treatments before admission, were used to compare those who received nesiritide during their index hospitalization with those who did not. Data were analyzed descriptively.

Results: Of 46,599 patients in the data set, 2,096 were receiving chronic dialysis. Eighty-five percent of the chronic dialysis patients presented through the emergency department. In general, chronic dialysis patients who received nesiritide (n=37) during their index visit had symptoms longer (before enrollment) and were more ill than those who were receiving chronic dialysis and did not receive nesiritide (Table) .

Baseline medication use of the 37 patients who received nesiritide included diuretics (84%), angiotensin-converting enzyme inhibitors (30%), and β-blockers

Felegi WB, Silverman ME, Allegra JR/Morristown Memorial Hospital, Morristown, NJ Study objectives: We determine whether a written pain management guideline with accompanying educational material that recommends decreased use of meperidine curbs the use of the drug in a hospital emergency department (ED) setting.

Methods: This was a retrospective cohort study of ED patients by using computerized billing data for meperidine charges to measure the use of the drug throughout a 33-month period and to determine changes in meperidine use after distribution of a written guideline. The study was conducted in a university-affiliated suburban hospital with an ED volume of 65,000. A written guideline was distributed to all emergency physicians in March 2001, suggesting that the use of meperidine be limited because of potential accumulation of metabolites and adverse effects. Distributed guidelines included printed educational material with appropriate indications, contraindications, and inappropriate indications as deemed by a hospital-wide pain management committee. The drug remained available in the ED, and no restrictions to individual physician orders were made. The mean monthly use was compared by using a 2-tailed Student t test. All tests of statistical significance were conducted with α set at .05.

Results: There was a statistically significant marked decrease in the use of meperidine 2 months after the pain management guidelines were released. The average monthly meperidine use the year before March 2001 was 142±27 times per month, whereas in the 18 months after June 1, 2001, the average use was 42±19 (P<.001).

Conclusion: The distribution of written pain management guidelines with accompanying educational material can change physician prescribing behavior in the ED setting and significantly decrease the use of meperidine in the ED.

Thomas H, Brunett PH, Mullins M/Oregon Health Sciences University, Portland, OR Study objectives: We determine the characteristics and expectations of a cohort of senior medical students applying to an emergency medicine residency.

Methods: All applications through the Electronic Residency Application Service include an e-mail address. These addresses were used to send each applicant a single 3-page confidential survey. Applicants who interviewed were given another opportunity to complete the survey on the day of their interview. No member of the residency selection committee had any knowledge of who chose to complete the survey or what he or she may have said. The institutional review board of our institution approved this study.

Results: Five hundred eighty-nine people applied to our 3-year, university-based residency for the match to begin in July 2003. Of these applicants, 494 were senior medical students in schools within the United States who formed the population for our study. Useable surveys were returned by 243 (49%). When applicants were asked what specialty they would enter if not emergency medicine, 22.6% chose surgery, followed closely by internal medicine subspecialties at 22.2%. Family practice was chosen by 21.2%, whereas 12.3% chose a surgical specialty (eg, orthopedics, otolaryngology). Internal medicine was chosen by 11.3%, and only 7.5% chose pediatrics. Of these students, 7.1% report an average student debt of more than $200,000; 11.8% report a debt level of $150,000 to $200,000. The most frequent response was analysis. Medical records of patients were reviewed for medical and medication history and cardiac arrhythmias. Data were analyzed by using descriptive statistics.

Results: Of noncritical patients, 15,374 received 18,020 doses of droperidol in the ED, and 2,431 of these patients also had an ECG. Of the patients with complete data for analysis and no exclusion criteria, 376 had an ECG before droperidol treatment, 1,518 had an ECG after droperidol treatment, and 170 had an ECG before and after droperidol treatment. Nine hundred seventy-three (51.4%) patients in the group with an ECG before or after droperidol were receiving cardiac or psychiatric medications, potentially prolonging the QT interval. One hundred nine (64.1%) patients in the group with an ECG before and after droperidol were receiving such medications. The mean QTc in patients with an ECG before droperidol treatment was 421.3+32.7 ms. The mean QTc in patients with an ECG after droperidol was 421.0+29.1 ms. In the group with ECGs before and after droperidol treatment, the mean QTc was 424.3+30.5 ms and 427.6+201.9 ms, respectively. Sixteen patients had a history of cardiac arrhythmias, and 6 patients had a history of prolonged QT. The mean ratio of the QTc before to after droperidol treatment was 1.009 (95% confidence interval 0.99 to 1.02). No patient had documented TdP.

Conclusion: We detected no significant difference in the QTc in noncritical ED patients with ECGs before and after droperidol administration. No events of TdP were identified.

Kim S, Hohrmann JL, Clark S, Radeos MS, Camargo Jr CA/Massachusetts General Hospital, Boston, MA Study objectives: Many Americans have used a complementary or alternative medicine (CAM) at some point. CAM includes pharmacologically active substances, which has raised concerns about potential interactions between prescription and CAM medications (eg, ginseng-warfarin, ginseng-digitalis, St. John's wort-paroxetine).

Methods: For two 24-hour periods, we interviewed consecutive patients aged at least 18 years and presenting to 4 Boston emergency departments (EDs). Exclusion criteria included severe illness and emotional disturbance (n=224). Trained interviewers assessed use of prescription medications and of any CAM in the past year. Data analysis used χ 2 .

Results: Of 754 eligible patients, 530 (70%) were interviewed. Of these 530 patients, 37% (33 to 42) reported any CAM in the past year, whereas CAM medications per se were used by 21% (17 to 24) of patients. During the past year, the most frequent CAM medications were ginseng and Echinacea (both 7%). Among patients receiving anticoagulants (n=18), ginseng was used by 2 (11%). Among patients receiving digitalis (n=10), ginseng was used by 1 (10%). Among patients receiving antidepressants (n=56), St. John's wort was used by 1 (2%). Use of these 3 prescription medications was not associated with CAM medication usage (P=.71).

Conclusion: Although any CAM usage is common among ED patients, one third does not involve CAM medications. Furthermore, relatively few patients are receiving prescription medications with known interactions with CAM. These results suggest that potential drug interactions between prescription and CAM medications occur but that the frequency is probably low.

Klotz SG, Preval H, Southard R, Francis A/State University of New York Stony Brook, Stony Brook, NY Study objectives: Injectable atypical neuroleptics may supplant benzodiazepine or butyrophenone alternatives. Published studies of intramuscular ziprasidone excluded severe psychiatric agitation and that from alcohol or other substances. Thus, we evaluated ziprasidone (20 mg intramuscular) in patients with psychiatric agitation that was toxicology negative (PSYCH) or was alcohol-(ETOH) or substance-induced (SUBS) in our psychiatric emergency service. Duration of restraint use, as well as improvement in agitation, was assessed.

Methods: We determined Behavioral Activity Rating Scale agitation scores (minimum 1, maximum 7) and duration of physical restraints in a naturalistic nonrandom study of intramuscular sedatives in our psychiatric emergency service during a 3month period. Dosages were 20 mg for ziprasidone and varied for conventional intramuscular sedatives (86% received haloperidol, lorazepam, or both).

Results: Of the patients evaluated, 69 received ziprasidone (20 mg intramuscular)

Methods: This prospective, observational study involved medical students rotating through a university emergency department and its 3 clinical affiliates during the fall of 2002. For each patient encounter, students were asked to document whether their history and physical examination (H&P) was directly observed by either senior emergency medicine residents or attending physicians. Observation of H&Ps was categorized by students as follows: entire or majority, brief or limited, or not observed. The same information was requested for 11 commonly performed procedures.

Results: Forty-eight medical students recorded a total of 3,106 patient encounters. Documentation of H&P observation occurred in 2,830 cases (91.1%). The entire H&P or majority of each H&P was observed in 53 cases (1.9%). In 2,620 cases (92.6%), the H&Ps were not directly observed. A total of 752 procedures were recorded. Certain procedures were commonly observed (entire or majority and brief or limited): pelvic examination (98.7%), arthrocentesis (96.2%), lumbar puncture (94.1%), and incision and drainage (82%). Some procedures were variably observed: splinting (67.3%), laceration repair (66.7%), arterial blood gas test (61%), and nasogastric tube insertion (54.5%). Other procedures were less likely to be observed: phlebotomy (22.4%), urethral catheter placement (19%), and intravenous access (17%).

Conclusion: In this study, we found that few medical student H&Ps were directly observed, and commonly performed procedures were variably observed. Although an emergency medicine clerkship provides students with numerous patient encounters and the opportunity to perform procedures, evaluation of these clinical skills by direct observation may be underused.

Kabrhel C, Liu SW, Thomas SH/Massachusetts General Hospital, Harvard Medical School, Boston, MA; Brigham and Women's Hospital and Massachusetts General Hospital, Boston, MA Study objectives: Despite the importance of evidence-based medical practice and education, a busy emergency department is not always conducive to timely retrieval of medical literature. Having frequently cited studies (eg, the National Emergency X-Radiography Utilization Study [NEXUS] , Prospective Investigation of Pulmonary Embolism Diagnosis [PIOPED]) easily available could be helpful. The objective of this project was to create and analyze resident use of an online collection of select articles relevant to emergency medicine.

Methods: Attending emergency physicians were surveyed to create a compendium of significant primary research articles. A paper collection of 145 articles was assembled, and an online version was added to our medical school's Web-based resource system. Links were created between the online version and the digital resources of the medical school library, enabling medical students, residents, and faculty to access the articles. After a month of resource availability, 52 emergency medicine residents were surveyed as to their use of the collection and its utility. To assess whether training level was associated with use of the literature collection, responses of residents were compared with risk ratio and 95% confidence interval analysis.

Results: All 52 residents responded, and all indicated that the collection was likely to prove useful. Either the paper or online version was used by 16 (31%) residents. Postgraduate year (PGY)-4 residents (9 of 13; 69%) were more than 5 times more likely (risk ratio 5.2; 95% confidence interval 1.4 to 19.8; P=.0025) to use the collection than PGY-1 residents (2 of 13; 15%). Thirty-six residents who had not used the collection cited time (n=15; 41%), unawareness of the collection (n=5; 13%), preference for other resources (n=4; 11%), lack of online passwords (n=3; 8%), and personal disorganization (n=2; 5%). Forty-nine (94%) residents stated that the collection was likely to be useful and practical; 3 (6%) stated that it would be useful but not practical.

Conclusion: As digital resources increase, educators may use a literature collection such as ours to teach evidence-based emergency medicine at the bedside. Further study should focus on optimizing availability and use of these resources.

Faber J, Duggan S, Williams T/Synergy Medical Education Alliance, Saginaw, MI Study objectives: A recent pilot study surveying emergency medicine residents suggests that resident physicians do not feel they receive adequate training and are not comfortable performing sexual assault examinations. Because more sexual

McNinch D, Jones JS, Emery M, Nicely JD/Spectrum Health-Butterworth Campus, Michigan State University, Grand Rapids, MI Study objectives: Medical education has adopted the use of digital photography and other computer technology, which has changed the face of the classroom. However, obtaining quality medical images for presentations is problematic. We compare the effectiveness of Internet image search engines for retrieving high-resolution (>400 dpi) photographs or illustrations for 25 topics in emergency medicine.

Methods: We conducted a cross-sectional search for high-resolution medical images on 25 medical topics by using 20 Internet search engines (including 10 specialized medical search engines). Medical topics were randomly chosen from an emergency medicine text database. Three reviewers independently evaluated each search engine for relevant content of retrieved digital images, currentness of information, and the image quality (resolution). Data were analyzed with χ 2 analysis, Student t tests, and the κ reliability test.

Results: High-resolution digital images were available for all of the emergency medicine topics; the mean number of images per topic was 435+221. Overall, 82% of retrieved images were considered relevant to the medical topic, with an interrater reliability more than 0.80. The proportion of relevant images varied significantly by search engine (53% to 85%; P<.001). Thirty-five percent of these digital images were high-resolution and suitable for teaching purposes. Overall, 3 general search engines (eg, Google, Lycos, and Yahoo) significantly outperformed all other search engines in retrieval of relevant, high-resolution images, including fee-based and specialized medical search engines.

Conclusion: High-resolution digital photographs and illustrations were freely available on the Internet for all of the emergency medicine topics evaluated. Although a number of medical search engines have been developed, they are not as effective as basic search engines that now allow users to search for images on the Internet.

Wald DA, Barrett J, Lafferty K/Temple University Hospital, Philadelphia, PA Study objectives: The Liaison Committee on Medical Education states in its accreditation standards that medical students should "demonstrate on direct observation the core clinical skills, behaviors, and attitudes that have been specified in the school's educational objectives." Medical interviewing, physical examination skills, and procedural competency are skills that may require direct observation to evaluate. However, little information is available on observation of medical students during an emergency medicine clerkship. The purpose of this study was to determine the extent of observation of students during an emergency medicine clerkship.

Oster NS, Chaffee MW/New York Methodist Hospital, Brooklyn, NY Study objectives: We examine the level of emergency preparedness in 22 hospitals in the US Navy Medical Department and identify hospital emergency preparedness strengths and vulnerabilities through the use of the Hospital Emergency Analysis Tool (HEAT). Methods: To assess the level of hospital emergency preparedness in the US Navy Medical Department, a study was instituted using HEAT, an instrument that gathers data through assessment of 100 critical preparedness factors. Each critical preparedness factor is considered important in a hospital's ability to respond effectively in an emergency situation. The HEAT is administered through the use of structured, faceto-face interview of hospital personnel involved in emergency preparedness and response.

Results: Data gathered in the first of 3 phases of assessment indicate that a number of significant strengths are resident in US Navy hospitals, as well as vulnerabilities that could detract from the effectiveness of emergency response in these facilities. Factors that contribute to effective hospital emergency response include use of the hospital incident command system, antidote and emergency drug caches, physical security and facility lock-down procedures, and basic health care professional knowledge of chemical, biologic, radiologic, nuclear, and high-yield explosive emergency response. Weaknesses noted in the first phase of assessment include potential for communications systems failure, lack of coordinated biosurveillance, inadequate inpatient surge capacity during the initial phase of emergency response, and potential for inadequate isolation patient capability.

Conclusion: Although the study is not yet complete, early results indicate that this is a powerful tool that can assist hospitals in the identification of critical strengths and weaknesses in their emergency planning efforts. The instrument used in this study, HEAT, permits the establishment of emergency preparedness benchmarks. Critical factors found to be absent or deficient can be remedied to strengthen each facility's preparedness. Follow-up survey of each facility is planned to evaluate progress. More hospitals need to be examined in the military health system and in the civilian health sector to allow comparison of like levels of care and emergency department patient volume.

Nufer KE, Wilson-Ramirez G, Hughes CE, Shah MB, Crandall CS/University of New Mexico, Albuquerque, NM Study objectives: With our nation on high terror alert, disaster planning has procedures in humans. Few reports evaluate resident performance of invasive procedures. We evaluate procedural proficiency after participation in a procedure laboratory.

Methods: A monthly animal-based procedural training laboratory was offered to residents during their emergency medicine residency training. This laboratory included more than 15 procedures using a live animal model. Senior residents who had taken the laboratory at least once were silently critiqued during an invasive resuscitation laboratory. A case preceded each procedure. Two emergency medicine attending physicians not affiliated with either residency and blinded to the residents' training profiles evaluated each resident and procedure with a standardized score sheet of "critical points." No feedback or instruction was offered during the evaluation. Each evaluation laboratory was not longer than 2 hours.

Results: Each resident (N=10) received a cumulative score for all procedures completed. Residents who had completed 3 or fewer laboratories had a mean score of 41.6 (median 34; SD 13.5). Residents who completed the laboratory 4 or more times had a score of 65.3 (median 79.5; SD 26.1; t statistic 0.87).

Conclusion: Residents who attended an animal-based procedural training laboratory 3 or more times appeared to have an improved score over those who took it 3 or fewer times. Because of limited numbers, these scores may not be statically significant.

Pines JM, Hilton JA, Gilbert K, Amory A, Parker C, Ghaemmaghami CA/University of Virginia, Charlottesville, VA Study objectives: Novel instructional techniques for endotracheal intubation (ETI) have been reported. Multiple publications support the use of a head-mounted video camera as a useful tool in ETI training. As a further, independent evaluation of the technology, we hypothesized that video-assisted instruction on ETI would be superior to traditional teaching methods that do not use video equipment in the training of novice intubators.

Methods: A randomized trial of first-and second-year medical students having no previous ETI experience was conducted. Students were given a 10-minute presentation on the mechanics and equipment used for ETI by 1 of 3 instructors. Students were then paired with instructors and randomized into 2 groups: video-assisted instruction (VID) or standard instruction (No VID). Students in both groups wore a head-mounted video camera. Student-instructor pairs were given a maximum of 3 minutes to perform the ETI on a standard airway-training mannequin. Video images from the head-mounted camera were displayed for the instructors in the VID group and were blinded to the instructor in the No VID group. An in-room observer recorded the following primary endpoints: rates of successful ETI, time to successful ETI, number of attempts required, and number of simulated dental injuries. Data were analyzed with t tests and Fisher's exact test.

Results: Seventy-three student-instructor pairs were randomized as follows: VID (n=43), No VID (n=30). Rates of successful ETI were 26 (60%) of 43 and 20 (67%) of 30 (P=.63) in the VID and No VID groups, respectively. Of groups with successful ETI, mean times to ETI were 126.7 seconds (VID) versus 117.5 seconds (No VID; difference of 8.8; 95% confidence interval -13.0 to 30.9; P=.42). Mean number of attempts was 1.37 (VID) versus 1.4 (No VID; P=.89). Mean number dental injuries was 1.23 (VID) versus 1.03 (No VID; P=.64).

Conclusion: No benefits were observed in ETI success rates, time to ETI, number of attempts, or dental injuries on a model mannequin between the novice intubators receiving video-assisted instruction and those receiving standard instruction. A larger, sufficiently powered independent study should be conducted to reduce the role of potential confounders found in our investigation.

Hansen LD, Char DM/Washington University School of Medicine, St. Louis, MO Study objectives: Disaster medical assistance teams (DMATs) are organized through the National Disaster Medical System (NDMS) to respond to large events. Multiple teams often respond to a single disaster. A survey was undertaken to assess how teams define operational roles and use triage and clinical protocols during training and on deployments.

Methods: We mailed a 10-question multiple part survey to all 37 DMATs and followed up by telephone. We asked about formal job descriptions, triage criteria, clin- gency nurses. We administered scenario-based, 25-question, multiple-choice, precourse and postcourse written examinations to the learners to assess cognitive gains. We constructed these examinations by using established principles of test design. Results: Among all course participants (n=68), 62 (90%) learners completed the precourse and postcourse examinations: 5 emergency medical technicians, 56 paramedics, and 1 registered nurse/paramedic. We excluded 7 learners from the study because of personal scheduling conflicts that prevented them from completing the examinations. Mean pretest and posttest cognitive scores were 53.2%±10.6 and 80.8%±11.4, respectively. The pretest and posttest score difference was highly significant (t=-16.2; P<.0001).

Conclusion: An interactive, simulation-enhanced curriculum of emergency response to terrorism training for out-of-hospital responders can result in significant cognitive gains for participants. As the course continues, additional studies will address skill performance and retention.

Martens KA, Hantsch CE, Stake CE/Loyola University Medical Center, Maywood, IL Study objectives: Nuclear, biologic, and chemical (NBC) accidents or terrorist acts can result in emotional and physical devastation. NBC events may create prolonged mass casualty disaster situations. Development of hospital emergency management policies for NBC events requires a reevaluation of resources, including 1 of the most crucial components, personnel. Employee participation may be affected by conflicts between personal and professional obligations. An assessment of employee attitudes may aid in planning efforts. In addition, providing for the care and security of employees and their families, as well as patients, may increase participation and consequently improve outcomes. The purpose of the study was to determine (1) anticipated availability of emergency department (ED) staff during various disaster scenarios, including NBC events; (2) importance of personnel support factors that may enhance participation during a prolonged disaster; (3) reliance of employee dependents on the medical center for basic needs during extended staff work hours; and (4) willingness of dependents to act as volunteers.

Methods: Approval of the department chair, ED director, and manager was obtained. A survey and explanatory cover letter were made available to each ED employee working at least half time. Survey completion was voluntary. There was no unique identifying marker linking employees to their surveys. Positions were recorded to quantify response rate and to use in data analysis. Respondents indicated whether they would or would not be willing and able to work during various mass casualty disaster scenarios if the event occurred while they were already at work or at home. Sixteen support services were rated on a scale of 1 (no importance) to 5 (essential) with respect to their importance in facilitating the ability of employees to work during a disaster event. Specific questions were included about dependent care needs.

Results: Response rates were 96% of 21 emergency physicians and 4 trauma surgeons (MD/DOs), 67% of 42 nurses (RNs), 58% of 19 flight crew (RNs and paramedics), 63% of 19 technicians, and 67% of 6 secretaries. Overall participation was 72%. For a nuclear event while they are at work, 88% of MD/DOs, 54% of RNs, 91% of flight crew, and 50% of other staff anticipated being willing and able to work extended duty. For a biologic event while at work, 92% of MD/DOs, 50% of RNs, 91% of flight crew, and 56% of other staff stated they would stay. For a chemical event while at work, 96% of MD/DOs, 57% of RNs, 91% of flight crew, and 56% of other staff stated they would stay. Sixteen percent of respondents were "unable to answer" about their ability to stay during an NBC event. Overall, there was a 22% further decrease in staff willing to come in from home for an NBC event. Eighty-five percent of all ED staff indicated they would commit to work in a disaster occurring as a result of multiple trauma or weather. The importance of support factors, listed in descending order of perceived priority, for all employees were security at work, training, shelter for self, medical care, shelter for family, clarification of role, food, scheduled relief, compensation, security at home, stress debriefing, transportation, child care, pet care, home or farm care, and elder care. The relative importance of these factors was only minimally related to the employee position. The most common limitation for employees who stated that they had other obligations that made it impossible for them to work was child care or family commitments. Thirty-four percent of respondents thought that they would rely on the hospital for care of their dependents. For these respondents, care or accommodations for an average of 2.6 dependents per employee would be necessary. The ages of dependents were 6% taken on a new importance and urgency. The New Mexico Disaster Medical Assistance Team responded to hurricanes Andrew (HA) and Iniki (HI), the Northridge, CA, earthquake (NC), and tropical storm Allison, which caused the Houston flood (TS). To help with future disaster planning, we looked back at the patient encounters after each of these disasters. We also assess whether patient needs changed with time after the disaster.

Methods: This was a retrospective review of all patients treated by the New Mexico Disaster Medical Assistance Team after each of these disasters. Data analyzed include age, medical history, chief complaint, diagnosis, treatment, triage category, disposition, and time to presentation after the disaster. The time frame of the data recorded was from day 1 to day 14 after the event. According to this time and a median of 7.5 days, the patient encounters were divided into an early phase and a late phase. We defined early as patients presenting within 7 days of the disaster and late as those presenting 8 or more days after the disaster. Data were analyzed by using χ 2 .

Results: Two thousand two hundred fifty-five patients were treated. Subtotals were HA 712, HI 344, NC 621, and TS 578. Median ages were HA 33, HI 36, NC 22, and TS 32. The majority of patients had no medical history (53.6%). The most common chief complaints were upper respiratory infection symptoms (22.4%), wounds (17.3%), and musculoskeletal pain (11.6%). The most common diagnoses were wounds (15.6%), musculoskeletal pain (9.0%), and upper respiratory infection (7.7%). Nineteen percent of patients received no treatment. The treatments rendered most often were antibiotics (20.2%), pain relievers (17.1%), tetanus toxoid (15.5%), and wound care (14.2%). Triage levels were 62.6% green, 10.8% yellow, and 2.5% red. Of the patients, 24.1% had no triage level documented. Only 9.8% of patients required transport to a hospital. There were a total of 922 patients treated in the early phase and 1,143 treated during the late phase, which does not equal 2,255 because not all patients had date of service recorded. In comparing the needs of patients presenting early with those presenting late, there were no significant differences in chief complaint. The only difference among the diagnoses was that of upper respiratory infection (5.7% early versus 7.9% late; odds ratio [OR] 0.69; 95% confidence interval [CI] 0.48 to 0.99; P=.04). Wound care was less frequently performed in the early phase (6.7% versus 9.7%; OR 0.67; 95% CI 0.48 to 0.94; P=.01), and pain medications and tetanus vaccinations were given more frequently in the early phase (23.3% versus 17.5%, OR 1.43, 95% CI 1.15 to 1.79, P=.001; and 24.5% versus 14.7%, OR 1.88, 95% CI 1.50 to 2.37, P<.001, respectively). There was no difference in antibiotics dispensed. Additionally, there were no differences found when the patient dispositions were compared.

Conclusion: Disaster medical assistance teams should be prepared to see high volumes of lowacuity patients. They should also be well stocked with antibiotics, suture materials, supplies for irrigation, bandages, and tetanus toxoid. Additionally, quarterly training should stress wound care. There were some differences found in patients presenting early versus late, but whether or not these differences are clinically significant is debatable because the number of upper respiratory infections diagnosed is probably not significant when a disaster is planned for. Also, the difference in frequency of wound care, tetanus vaccinations, and pain medication given is probably not clinically significant because each of these treatments still represented a large portion of treatments given in the early and late phases. Disaster planning will likely always remain daunting, but these data may help guide disaster planners. Study objectives: An effective emergency response to acts of terrorism requires the acquisition of knowledge and skills new to the majority of out-of-hospital providers. We sought to assess the cognitive knowledge gained from a highly interactive, simulation-enhanced curriculum developed to prepare first responders for such a task.

Methods: We developed a 16-hour, 2-day, multimedia-and simulation-enhanced course that places learners in realistic situations by using the equipment required to respond to various chemical, biologic, radiologic, and explosive acts of terrorism. Small-group sessions and large-group drills were conducted. Errors in skills performance were corrected immediately, and then skills were repeated to achieve mastery. Participants included emergency medical technicians, paramedics, and emer-Nufer KE/University of New Mexico, Albuquerque, NM Study objectives: Since the events of September 11, 2001, and repeated high terror alerts, disaster planning has taken on a new urgency. There are little to no data published on pediatric disaster victims. Also, there is no requirement for pediatric advanced life support or any other pediatric training for disaster medical assistance team (DMAT) participation. We assess whether the needs of pediatric disaster victims differ from those of adult disaster victims.

Methods: This was a retrospective review of patients 18 years or younger cared for by the New Mexico DMAT after 2 hurricanes, 1 flood, and 1 earthquake. We compared the following variables between the pediatric and adult disaster victims: chief complaint, diagnosis, treatment rendered, triage category (green, yellow, or red), and disposition. Data were analyzed by using χ 2 .

Results: Six hundred sixty-one (29.3%) patients were aged 18 years or younger. Median age was 4 years. The proportion of patients with chief complaint of wounds was similar (pediatric versus adult, 16.3% versus 17.7%; odds ratio (OR) 0.91; 95% confidence interval (CI) 0.70 to 1.16). The pediatric patients were more likely to present with upper respiratory infection symptoms (40.5% versus 15.3%; OR 3.77; 95% CI 3.04% to 4.66%), abdominal complaints (abdominal pain, vomiting, or diarrhea; 8.9% versus 5.6%; OR 1.66; 95% CI 1.15% to 2.37%) and rashes (8.9% versus 5.0%; OR 1.86; 95% CI 1.28% to 2.68%) but less likely to present with musculoskeletal pain (5.0% versus 9.9%; OR 0.48; 95% CI 0.31% to 0.71%). The 2 groups were equally likely to be diagnosed with wounds (14.7% versus 16.0%; OR 0.90; 95% CI 0.70% to 1.17%). The pediatric patients were more likely to be diagnosed with ear infections (14.3% versus 2.4%; OR 6.63; 95% CI 4.46% to 9.98%), upper respiratory infection (13.7% versus 5.0%; OR 3.00; 95% CI 2.16% to 4.15%), and gastroenteritis (6.0% versus 1.9%; OR 3.20; 95% CI 1.94% to 5.32%) but less likely to be diagnosed with musculoskeletal pain (2.5% versus 11.7%; OR 0.19; 95% CI 0.11% to 0.32%) and stress (1.6% versus 5.0%; OR 0.31; 95% CI 0.15% to 0.59%). The pediatric patients were more likely to receive antibiotics (24.3% versus 14.7%; OR 1.87; 95% CI 1.49% to 2.33%). Wound care was used with similar frequency. The overall acuity in both groups was low, with triage categories as follows: pediatric 57.6%, 9.5%, and 2.0% and adult 64.8%, 11.7%, and 2.7% green, yellow, and red, respectively. Pediatric patients were less likely to be triaged as green (OR 0.74; 95% CI 0.61 to 0.90); however, they were triaged as yellow or red with similar frequency. Thirty-one percent of pediatric cases had no documented triage level. One child required intubation, a 12-year-old child who had experienced a closed head injury. Fewer pediatric patients were sent to a hospital (5.9% versus 11.6%; OR 0.48; 95% CI 0.32% to 0.69%).

Conclusion: Pediatric patients represent a large proportion of disaster victims. The needs of the pediatric patients differ in several important ways. The chief complaints of abdominal pain and rashes and diagnosis of gastroenteritis may be viewed as more challenging medical treatment problems than musculoskeletal pain and stress. Also, the only patient requiring intubation was a pediatric patient. These observations may demonstrate that pediatric patients have special needs and that there may be a need for more pediatric training among DMAT members. Although some team members had pediatric training through an emergency medicine or family practice residency, others who practice as psychiatrists and internists had little to none. Although the overall acuity of pediatric encounters during DMAT deployments is low, the high volume and different needs of pediatric patients cared for should prompt a reevaluation of the pediatric training or background required for DMAT participation.

Pepe PE, Swienton RE, Schwartz RB, Dallas CE, Lillibridge SR, Tang N, Carlton PK, Blanck RR, James JJ/University of Texas Southwestern Medical Center, Dallas, TX Study objectives: After the events of fall 2001, a federal appropriation, managed by the Centers for Disease Control and Prevention (CDC) through 1 of its 8 public health emergency preparedness centers, was earmarked to have several academic institutions and federal agencies assimilate their preexisting "all-hazards" advanced disaster life support training programs into a standardized, consensus-driven course for national distribution. In addition to all paramedic, nursing, and medical students, targeted audiences would include all applicable health care providers who may have to deal with disaster situations and homeland security issues. younger than 1 year, 20% aged 1 to 5 years, 20% aged 6 to 10 years, 30% aged 11 to 20 years, 12% aged 21 to 65 years, and 12% older than 65 years. Fourteen percent of dependents could volunteer during a disaster event.

Conclusion: This study indicates that a minimum 17% reduction in the ED workforce should be anticipated by hospitals in planning for NBC events and other disasters. A disaster occurring during "off" shifts may result in further reduction because of a reported increased unwillingness to come to work as opposed to remain at work. The rate of participation may be even less, depending on the ultimate choice of the ED workforce currently undecided. The significant number of "unable to answer" responses demonstrates the importance of raising these issues for individual consideration and for departmental and institutional preparation. A nuclear event caused the most concern among employees with respect to their perceived ability to work. Policy development ensuring adequate security, training, and accommodations for staff may have the greatest effect on promoting employee participation. An appraisal of variables affecting the ability of health care providers to spend prolonged time at work during NBC events will assist organizations in developing emergency management policies.

Chasm R, Jerrard D, Van Wie D, Wegner S/University of Maryland, Baltimore, MD Study objectives: Great fear has been expressed about the panic and mayhem that would ensue in a populace subjected to a chemical-biologic attack. The resultant frenzied stampede to a city's or the nation's emergency departments (EDs) may be overwhelming. The goal of the Department of Health and Human Services was to have vaccinated 500,000 emergency health care workers against smallpox by a target date of March 2003. Considering that less than one tenth of this number have availed themselves of the opportunity to be vaccinated, it is likely that they themselves may have concerns about their own vulnerability to some of these agents, the victims of which these emergency providers would be called on to render care to in the event of an attack. Current models do not exist about health care workers' attitudes and concerns about exposing themselves to chemical and biologic toxins in the event of a mass incident. We assess the attitudes of ED health care workers about their willingness to expose themselves to victims of a chemical-biologic attack and the agents responsible.

Methods: An observational cross-sectional study design was selected for an academic emergency medicine residency program based in an urban center treating 90,000 patients annually. Attending emergency physicians, emergency medicine residents, nurses, and technicians were questioned anonymously about such things as whether or not they had chosen to obtain the smallpox vaccine and the reasons for that decision. In addition, this population was queried about their willingness to respond to a mass exposure to various biologic or chemical toxins (anthrax, radiation, sarin, smallpox, VX). Staff perception of their personal susceptibility or that of their family was assessed. Measurement of staff intent to participate wholly in the care of victims of an exposure and not succumb to a desire to leave the premises to minimize risk to themselves or close personal contacts was also analyzed. Institutional review board approval was obtained. Data are reported as proportions and 95% confidence intervals (CIs).

Results: Eighty-two staff member responses were tallied. Of 82 staff, 3 (3.6%; 95% CI 1% to 9%) had opted to obtain the smallpox vaccine. Nearly half the respondents believed moderately or strongly that the fear of contracting the disease was their main reason in forgoing the vaccine. Assessment of fear revealed that 23 respondents (28%; 95% CI 19% to 39%) had great concern for personal safety if exposed. Twenty-one of 76 respondents (27%; 95% CI 18% to 39%; 6 did not answer this question) stated that they would leave the ED if called to care for the victims of any of the biologic or chemical toxins listed. Of physicians, 4 of 43 (9%; 95% CI 3% to 20%) would not expose themselves or their family members and would vacate the ED.

Conclusion: The general expectation of the medical community and the lay public is that should an emergency arise, emergency health care personnel will be available for treatment. As we prepare for a terrorist attack by increasing inventories of antibiotics, antidotes, and vaccines, the potential for emergency workers in numbers sufficient to affect care who refuse to expose themselves to victims of a chemical-biologic attack is a scenario that may need to be accounted for when constructing models for response to such events. Further large-scale studies should be performed. years and admitted from the ED to the OU for treatment of asthma from August 1999 to August 2001 were included. Patients with bronchiolitis or pneumonia were excluded. The IA group was defined as patients who were admitted from the OU to an inpatient ward or who were subsequently readmitted within 2 days of OU discharge. Main outcome measures were historical and clinical factors identified during the ED encounter that were associated with IA.

Results: One hundred seventy-five patients aged 2 to 20 years (median 4.0 years) met inclusion criteria; 55% were male patients. Of these patients, 46 (26%) subsequently required IA. When patients who were discharged from the OU within 24 hours were compared with the IA group, there was no statistical difference in historical factors (ie, age, duration of symptoms, medications before ED presentation, past hospitalizations for wheezing) or clinical factors (ie, vital signs, retractions, wheezing), except for oxygen saturations at OU admission. More children in the IA group had oxygen saturations less than 88% or were receiving supplemental oxygen at OU admission (32 versus 13%; χ 2 ; P=.01; odds ratio 3.2; 95% confidence interval 1.25 to 8.11). Chest radiographic findings were not predictive of need for IA.

Conclusion: The need for supplemental oxygen or oxygen saturations less than 88% at ED patient transfer to the OU were associated with the need for subsequent IA from the OU. Prospective studies are needed to confirm this finding and to identify other factors predictive of IA. OU admission is an efficient and effective hospitalization option for pediatric ED patients with asthma.

Kondamudi N, Merchant S, Kim J, Huq M, Kumar S/The Brooklyn Hospital Center, Brooklyn, NY Study objectives: Acute asthma accounts for almost 200,000 admissions to hospitals every year. Hospitalization rates are variable and are influenced by geographic location, type of patient population, environment, and access to medical care. Identification of factors amenable to intervention can potentially be targeted to reduce hospitalization rates in the community. We determine factors associated with asthma hospitalization. We also seek to determine the use of New York State recommended asthma action plan among the study population.

Methods: The study was conducted in a pediatric emergency department (ED; annual census 18,000) in a 600-bed inner-city community hospital. Parents of children (aged 6 months to 18 years) with physician-diagnosed asthma attack were interviewed by using a pretested questionnaire. The treating physician then completed the remainder of the questionnaire to capture all data points. The study was conducted between July 1, 2000, and June 30, 2001. Hospitalized patients were compared with those discharged by using SPSS statistical software.

Results: Hospitalization occurred in 167 of the 727 asthma ED visits, for a rate of 23%. Age of the child, type of insurance, sex, time at presentation, nebulizer use at home, and environmental variables were not associated with hospitalization. History of prematurity (odds ratio [OR] 1.7), atopy (OR 1.2), ED visits (OR 1.4), and ICU admission (OR 1.9) were not associated, but family history of asthma (OR 1.82; 95% confidence interval [CI] 1.2 to 2.7) and onset of wheeze in infancy (OR 0.4; 95% CI 0.3 to 0.7) were significantly associated. No maintenance medication use was reported for 80% of patients, and the use of β 2 -agonists (OR 1.6, 95% CI 1.0 to 2.4) and nonuse of inhaled steroids (OR 0.38; 95% CI 0.19 to 0.76) was significantly associated with hospitalization. Asthma education within previous 6 months and use of asthma action plan were not associated with hospitalization but occurred in only 20% (n=142) and 10% (n=73) of patients, respectively. Peak flow measurement was done in only 14% (n=101) of patients. Pulmonary index score of more than 7 was strongly associated with hospitalization (OR 45.8; 95% CI 27.6 to 76.2). Logistic regression revealed independent association of pulmonary index score and onset of wheeze in infancy with hospitalization.

Conclusion: The objective pulmonary index score and onset of wheeze in infancy are strongly associated with childhood asthma hospitalization. The use of a written asthma action plan in this study group is low.

Tarsi J, Sterner L, Sterner A, Lewis L/Washington University School of Medicine, St. Louis, MO Study objectives: We examine patterns of relapse and recidivism in pediatric asthma patients and determine risk factors associated with repeated use for acute Methods: Content experts and educational specialists, along with representatives from multiple academic, military, and federal agencies, established a curriculum and developed educational material by using the requested "all-hazards" approach. Similar to the practical training scenario aspects of advanced cardiac life support (ACLS) and advanced trauma life support (ATLS), the new advanced disaster life support (ADLS) course is offered in a didactic and skills laboratory (1 day) template using similar student targets, coordinators, scheduling, and space requirements. The typical course includes the use of high-fidelity mannequins, virtual simulations, and actual decontamination procedures using protective clothing and respirators. In addition, ADLS has an introductory 12-hour modular, didactic component, basic disaster life support (BDLS), which may be presented in lectures or through distance learning and computer simulation. Certification and course credit can be given for completion of BDLS alone. BDLS-certified persons are eligible for the more sophisticated ADLS course (involving the decontamination procedures and the actual training scenarios using interventions such as antidote administration in response to the scripted scenario and "physical findings" on the high-fidelity mannequins).

Results: ADLS has undergone beta testing with extremely positive feedback across all health care provider groups. Targeted for completion and release in early 2004, ADLS and BDLS will receive validation and Continuing Medical Education accreditation through the CDC and the American Medical Association.

Conclusion: The ADLS course and its introductory BDLS component are federally funded (CDC) initiatives meeting many of the needs outlined by federal officials and professional organizations (eg, American Medical Association, American Public Health Association, the American College of Emergency Physicians) for training the medical community at large for disaster preparedness and homeland security by using a standardized, expert-consensus, "all-hazards" approach.

Hardasmalani MD, Bithoney WG, Vincent D, Gold N/St. Joseph's Hospital and Medical Center, Patterson, NJ Study objectives: We evaluated the efficacy of levalbuterol versus albuterol in acute exacerbation of asthma in the pediatric population. A secondary objective was to evaluate whether levalbuterol compared with albuterol decreases emergency department treatment time or hospital admission rates.

Methods: The study used a prospective double-blinded randomized research design. Seventy patients completed the study. Baseline oxygen saturation, respiratory rate, and peak flow rates were measured. Patients were then randomly treated with either albuterol or levalbuterol, along with ipratropium bromide. Oral steroid was administered to all patients after the second treatment. Oxygen saturation, respiratory rate, and peak flow rate were measured after each treatment. At the end of 3 treatments, decision about additional treatments or admission was made according to national guidelines by the treating physician.

Results: All patients showed improvement in oxygen saturations, respiratory rates, and peak flow rates. No statistically significant differences were observed in the 2 groups, nor were any statistically significant differences found in the need for additional treatments or hospital admissions.

Conclusion: We conclude that there is no statistically significant benefit of levalbuterol over albuterol in terms of efficacy or pharmacoeconomics.

Miescier MJ, Nelson DS, Kadish HA/Primary Children's Medical Center, Salt Lake City, UT; University of Utah, Salt Lake City, UT Study objectives: Short-stay observation units (OU) serve patients who require more evaluation or treatment than possible during an emergency department (ED) visit and who are anticipated to stay in the hospital for a short defined period. Asthma is a common admission diagnosis in a pediatric OU. Two studies have examined factors in pediatric asthma management that might predict discharge from a 12-hour OU. To the authors' knowledge, no study has examined these factors for an OU with a 24-hour maximum length of stay. We describe characteristics of patients with asthma treated in a 24-hour length of stay pediatric OU and identify clinical factors that might predict the need for inpatient admission (IA) after OU treatment.

Methods: This was a retrospective medical record review conducted in Primary Children's Medical Center, a tertiary care children's hospital. Children older than 2 respectively). A greater percentage of male patients reported that medication unavailability precipitated their asthma exacerbation compared with women (26% versus 17%). A larger number of female patients did not have an asthma precipitant recorded on their record compared with men (23% versus 16%). The Table lists the relative frequencies of asthma precipitants.

Conclusion: Precipitating etiologies for acute asthma in the ED were generally similar in men and women during the 4-month study period; however, important differences exist between the male and female patients of our study. Male patients were more likely to report that medication unavailability contributed to their ED visit compared with female patients. Also, female patients were less likely to have an asthma precipitant recorded on their record compared with male patients. However, the largest etiology of asthma precipitant was not readily identified by either the patient or physician. Study objectives: Previous studies have looked at environmental factors affecting acute exacerbation of bronchial asthma. These studies have looked at the effects of cold temperatures and determined that asthma exacerbations could be triggered by these factors. However, no study has ever evaluated the effects of a heat emergency on acute asthma exacerbation. During a 2-week period in 1995, temperatures in the Chicago area created a heat-disaster situation. The purpose of our study is to determine whether the incidence of asthma exacerbation varies during a heat emergency.

Methods: This was a retrospective incidence study comparing relative frequency of emergency department (ED) asthma cases during the 2 weeks of the 1995 Chicago heat disaster with 2 weeks from the preceding year. The sample consisted of all ED encounters from 23 randomly selected metropolitan hospitals. The study population was stratified by age and gender. Statistical inferences were made by using χ 2 and the Breslow-Day test.

Results: There were 55,551 patients reviewed, with an increase in ED asthma cases of 29% during the heat weeks (95% confidence interval 14% to 45%). Pediatric patients showed an 83% increase in incidence during heat weeks and adults a 16% increase, whereas the elderly showed a 41% decrease. The differences between age groups were significant (Breslow-Day P=.0012 and P=.0087). There was no difference between male and female patients (Breslow-Day P=.1577).

Conclusion: During the Chicago heat disaster of 1995, the relative incidence of asthma treated in EDs increased 29% from the previous year. This increase was highest among children, whereas the elderly actually experienced a decrease in ED asthma visits.

Tarsi J, Holmes J, Lewis L, Zimmerman NA, Castro M, Grayson M/Washington University School of Medicine, St. Louis, MO Study objectives: We identify risk factors associated with relapse after treatment for acute asthma in a large, inner-city emergency department (ED). Although asthma relapse has been examined in the past, few studies have examined outcomes specifically in this high-risk population. asthma of the pediatric emergency department (ED). Previous studies have observed that there is a subset of asthma patients treated repeatedly in EDs. These patients have poorer outcomes and account for a greater percentage of health care resources. By examining demographic characteristics of repeated versus nonrepeated ED users, we hope to identify factors related to ED recidivism. Methods: This was a case-control, retrospective, medical record review study of 1,272 patients treated for acute asthma in a Level I pediatric ED from January 1999 through April 2002. Eligible patients were those who met the criteria of relapse or recidivism for asthma. Relapse was defined as a return asthma visit to the ED within 20 days of index visit. Recidivism was defined as 2 or more asthma visits to the ED within 365 days of the index visit (ie, ≥3 visits per year). Controls patients were those treated only once for asthma in a 365-day period.

Results: Two thousand two hundred fifty-eight total visits were examined, with cases making up 1,307 (58%) of these. Data were collected on 319 (25%) study patients and 957 (75%) controls. The 988 repeated-visit intervals were grouped according to relapse or recidivism: 21% of patients met the criteria for relapse (0 to 20 days); 79% met the criteria for recidivism (21 to 365 days). χ 2 analysis yielded significant differences between cases and controls about race (P<.001) and sex (P<.05). An independent Student's t test showed significant differences in mean age (P<.05). Adjusted odds ratios (ORs) indicated that black race and male sex were at higher risk for recidivism (OR 4.2 for race; 95% confidence interval [CI] 2.6 to 6.7; OR 1.4 for sex, 95% CI 1.1 to 1.8). Younger age was associated with an independently increased risk, but in the multivariate logistic regression model, this increase was not found to be significant.

Conclusion: A disproportionate number of children make up the majority of the visits to the ED, with 25% of the patients accounting for 58% of the visits for asthma. These repeated users of the ED historically have poorer outcomes, including higher asthma-related mortality. Black race and male sex are 2 important predictors of increased visits in this study population. Further research is needed to discern other factors that increase the risk of repeated ED visits and to guide the development of interventions that could improve outcomes in this high-risk population.

Singh A, Bou CI, Chen L/New York Medical College/Metropolitan Hospital Affiliation, New York, NY Study objectives: Acute asthma exacerbation is a common presentation to the emergency department (ED), accounting for approximately 2 million ED visits and nearly 470,000 hospitalizations each year. Previous studies have reported that sex differences exist in asthma symptom reporting, admission rates, and length of hospital stay between men and women. At particular risk from asthma are African Americans (AA) and Hispanic Americans (HA) living in inner-city conditions. Epidemiologic studies have identified exposure to environmental triggers as 1 of the reasons why this population has a disproportionately high rate of asthma morbidity and mortality. The objective of this study was to identify sex differences in asthma precipitants in AA and Puerto Rican Hispanic patients in an urban ED.

Methods: This was a retrospective cohort of consecutive adult patients presenting to our institution with the diagnosis of acute asthma exacerbation. Metropolitan Hospital is a city hospital located in the southeast region of Spanish Harlem in New York City and serves primarily an AA and HA population. During a 4-month period from July 2002 through October 2002, consecutive medical records were obtained that contained the diagnosis of acute asthma exacerbation. A 2-physician team abstracted biographic and epidemiologic data from the records. Records were excluded if no biographic information about race could be identified from the record or if the self-reported race of the patient was neither AA nor HA.

Results: A total of 369 patients had the discharge diagnosis of acute asthma exacerbation during the study period. Eleven records contained missing biographic data about patient race and were excluded from analysis. Twenty-four records were of patients who were neither identified as AA or HA nor included in the analysis. A total of 334 patients were included in the analysis. The average age for men and women was similar (men 42 years, range 18 to 88 years; women 42 years, range 18 to 82 years); however, compared with the female group, more men were identified as HAs (78% versus 69%) and, also compared with women, fewer men were identified as AAs (22% versus 31%). AA and HA male and female patients had similar precipitants of asthma during the study period. The largest category of precipitating etiology was "unknown asthma precipitant" for men and women (29% and 32%, dren in more urban counties; however, this pattern was inconsistent. The adjusted odds of an ED visit for asthma versus all other diagnoses by county of residence was calculated by using multiple logistic regression to control for demographic factors known to affect asthma (age group, race, sex, and payer status), again using Baltimore City as the reference county. Unexpectedly, when adjusted for these factors, ED visits by children living in affluent suburban Howard County had a greater likelihood of being for asthma than ED visits by children from Baltimore City. Conclusion: Asthma remains a critical health problem for many Americans, particularly children. In this study, we try to isolate the influence of county level geographic variability on ED asthma use. From an overall public health perspective, it is clear that Baltimore City (an urban area with high rates of poverty and a predominantly black population) has the highest burden of disease from asthma, as measured by children seeking care in EDs. However, this study illustrates that geographic variability in ED use for asthma is not based solely on an urban environment. Further research is needed to better refine our understanding of environmental and socioeconomic factors associated with asthma exacerbations and ED use.

Methods: This was a prospective telephone survey performed on a convenience sample of adult patients after being treated for acute asthma in a large, urban ED (>80,000 visits in 2002). Patients (aged 18 to 64 years) treated in the ED and with a primary diagnosis of acute asthma exacerbation and discharged to home were included. Patients with concurrent respiratory disease or psychiatric illness were excluded. Relapse was defined as a return visit to any ED or an unscheduled visit to a clinic or primary care physician (PCP) for asthma within 4 weeks of the index visit.

Results: Of the 99 patients who met selection criteria, 62 (63%) were available for follow-up, with 1 patient refusing to participate. The demographic characteristics of the remaining 61 study patients were black (82%), female (66%), and mean age 38 years (SD±12). These demographic characteristics did not differ significantly from those of patients for whom no follow-up was possible. Twenty percent of the study patients reported having a relapse within 4 weeks. The number of patients reporting an admission to the hospital for asthma in the previous year was significantly higher in the relapse than nonrelapse group (P<.001; unadjusted odds ratio 12.9; 95% confidence interval 2.3 to 72.9). Conversely, patients currently using oral steroids were less likely to have a relapse (P<.05; odds ratio 0.20; 95% confidence interval 0.05 to 0.84). Other variables examined included having a PCP, smoking exposure, principal site for asthma care (PCP, clinic, ED), current use of short acting β 2 -agonists, and inhaled corticosteroid therapy. Patients without relapse were more likely to list the ED as their usual site of care, 21% versus 0%, and not have a PCP, 50% versus 30%. These differences were not statistically significant.

Conclusion: In this high-risk, inner-city population, 20% of patients discharged from the ED for asthma can be expected to have a relapse within 4 weeks of index visit. Admission to the hospital for asthma in the previous year was the only significant predictor of relapse in this population, with current oral steroid usage showing a protective effect. Further research is needed to assess predictors of outcomes in this high-risk population to help develop prevention strategies that decrease recurrent ED visits.

Hirshon JM, Weiss SR, LoCasale R, Levine E, Blaisdell CJ/University of Maryland, Baltimore, MD; National Aeronautics and Space Administration, Goddard Space Flight Center, Greenbelt, MD Study objectives: Asthma is a significant clinical and public health problem in the United States, especially for children. In 1999, asthma led to an estimated 478,000 hospitalizations for all ages and 1,997,000 emergency department (ED) visits nationwide. It has been estimated that the total direct and indirect monetary cost for asthma in 1998 was $11.3 billion dollars. Regional variation has been noted on a national level for asthma and at national and local levels for other respiratory diseases. Little is known about the variation of asthma within smaller geographic areas. This study analyzes ED use data to determine whether geographic variability of pediatric asthma cases exists for smaller jurisdictions within a state.

Methods: We obtained nonconfidential ED use records for children aged 0 to 17 years for the entire state of Maryland, for services rendered from the second quarter of 1997 through the first quarter of 2001 (inclusive). These data included a primary diagnosis made at discharge according to the International Classification of Diseases, Ninth Revision, Clinical Modification at the institution where the clinical care was delivered. For this study, asthma ED visits were defined as any discharge with an International Classification of Diseases, Ninth Revision, Clinical Modification code equal to 493.XX. Visits with a diagnostic code for a newborn birth and those without a valid discharge diagnostic code were excluded from this analysis. Additionally, we categorized children by county of residence by merging the reported ZIP code with a commercially available ZIP code database (CD-Light's ZIPlist5 Database of US ZIP Codes, May 1998). Records without a valid Maryland state ZIP code, including residents of other states, were excluded.

Results: There were a total of 1,154,806 ED visits made by Maryland children during this 4-year period, of which 47,973 (4.2%) were for asthma. Baltimore City had the most ED visits in this period for all causes (252,685) and for asthma (16, 409) and the highest average annual use rate for asthma (254.4/10,000) compared with all other Maryland counties. Baltimore City had the highest number of pediatric visits overall, as well as the largest number of ED asthma visits as a percentage of total visits for all age groups and both sexes compared with all other counties. Children in rural counties tended to have less ED use for asthma than chil-

We analyzed 2 models: the logarithms of assault patients per day and assault patients per month. Results: There were 10.3+4.9 assaults a day, 2.6+2.4 gunshot wounds a day, and 3+2.5 stab wounds a day. There was a significant (P<.0001) difference in assaults by day of the week

Time series analysis did not show additional effects of temperature, precipitation, or airborne pollutants on the day-to-day difference in number of assault patients admitted to these hospitals. Conclusion: We found significant differences in incidence of assaults by day of the week and season. The decrease in assaults on

Sexually Transmitted Disease Prevalence in Patients Presenting for Initial Evaluation After Sexual Assault

Standard protocols include testing for gonorrhea, chlamydia, and syphilis and providing prophylaxis for these diseases. The objective of this study was to determine the prevalence of sexually transmitted diseases in victims of sexual assault at initial presentation to the ED. Methods: This was a retrospective medical record review of patients presenting to a Level I urban trauma center that served as a rape treatment center for the city during 2002. Demographic information and the results of gonorrhea cultures, chlamydia testing, and rapid plasminogen serology were recorded. Patients were advised to receive follow-up evaluation and cultures at their primary care physician, obstetriciangynecologist, or city health clinic. Results: Three hundred fifty-two patients were treated after sexual assault. There were 342 (96%) female patients and 10 (2.8%) male patients. The average age was 22.8 years (range 2 to 84 years). There were 225 (63%) black patients, 85 (24%) white patients, 11 (3%) Hispanic patients, 6 (3%) Asian patients, and 28 (8%) other or unknown patients. A total of 948 tests were obtained; only 272 tests were available for analysis because of improper culture preparation. Eight (2.9%) of the 272 tests were positive and included 3 gonorrhea, 4 chlamydia, and 1 rapid plasminogen. Conclusion: Only a small percentage of sexual assault victims had a sexually transmitted disease at presentation, but these tests are still important in the overall evaluation

Optimal Experience Level for Paramedics to Maintain Proficiency in Performing Endotracheal Intubation in Patients With Perfusing Rhythms and Those in Cardiac Arrest

Methods: Data were collected prospectively during 2002 on all endotracheal intubation (ETI) attempts in an advanced life support system staffed by paramedics with varying ETI experience. Mean number of intubations per paramedic was determined. Paramedics operated under standing orders for ETI in cases of respiratory arrest or failure or loss of airway protective reflexes. Drug-facilitated intubation (DFI) was permitted. Successful intubation was defined by verified endotracheal placement by using capnography and a variety of clinical methods. Emergency medical services (EMS) reports were abstracted

Methods: This is a 5-year prospective survey of bystanders to adult OHCA vicemergency medical services (EMS) documents. Criteria for definite, possible, and non-CCI delay were prospectively developed. Two paramedic reviewers were trained on these criteria, reviewed bystander and EMS data, and scored each for possible delay. For analysis, definite delay and possible delay were combined into a single variable. We collected common bystander and EMS OHCA demographics. Outcome was survival to hospital discharge. Results: During the study period, we identified 1,004 OHCA, of which 779 bystanders completed interviews. Of these interviews, 239 were excluded because of disagreement among raters

Delay was associated with decreased survival (4.7% versus 9.7%; P=.036). For witnessed emphasis needs to be placed on clinically relevant criteria (respiratory rate, GCS) for

Methods: This was a retrospective review of medical records, airway data forms, billing, respiratory therapy, and out-of-hospital records at an academic Level I trauma center with an annual census of 65,000. We compared the total number of nontrauma intubations during 2 consecutive 18-month periods (representing prediversion and diversion periods) with the numbers of closure hours, ambulance runs, and potential confounders (ED monthly census, admissions, mean length of stay, use of bilevel positive airway pressure, and out-of-hospital intubations). Results: There were 1,573 total intubations for the study period (791 during the prediversion period and 782 during the diversion period)

Injuries and Illness During a Multiday Recreational Charity Cycling Event: The 2001 Heartland AIDS Ride

Study objectives: We describe the incidence and type of injury and illness during a multiday recreational charity cycling event. We describe our experience during the 2001 Heartland AIDS Ride and suggest strategies for medical support for similar events

participants rode 519.1 miles from

It is apparent from the study that injury and illness occur during these events. Providers of medical support for class III Results: There were recorded 306 childhood traffic accident patients in Klaipeda (with approximately 20,000 population) during 5 years. When children involved in traffic accidents from 1994 to 1998 were compared, a twofold increase was observed. One hundred eight children in traffic accidents of 306 (35.62%) children were children not crossing at pedestrian crossings. Fifty-five (17.97%) children crossed between stopped or parked vehicles. Seventeen (5.56%) children were injured because of not paying attention to traffic lights, 50 (16.34%) children were injured on the pedestrian crossing, 32 (10.46%) children were injured while riding on motorcycles, bicycles, or mopeds, 25 (8.17%) children were injured as passengers, and 18 (5.88%) children were classified as others. Of the children in traffic accidents, 81.37% were pedestrians. Spring and summer are the most common seasons for injury (195 [63.73%] of 306 children in crashes), along with the afternoon hours (noon to 8 PM, 68.89%). Most casualties of child traffic injuries were in the first to fourth forms. Conclusion: Means for traffic injury prevention and traffic safety should be

What Is the Impact of Child-Specific Motor Vehicle Restraints on Pediatric Morbidity and Mortality from Motor Vehicle Crashes in Pennsylvania From

Methods: The study cohort consisted of patients aged 0 to 10 years admitted to Pennsylvania trauma centers from 1992 to 2001. The Pennsylvania Trauma System Foundation provided data for crash characteristics, hospital complications, and outcomes. Statistical analyses included Pearson's χ 2 , Student's t test, and analysis of variance; significance was set at P value less than .01. Results: Legislation in 1993 mandating child seats for children up to age 4 years resulted in an increase in the reported use of child seats from 24% of patients to 37% of the pediatric victims of MVC in 1994 (P=.02). Overall, use of restraints for children younger than 15 years increased as well from 45% in 1993 to 66% in 1994 (P=.02)

Pediatric Passenger Restraints: What Do Parents Say?

The Children's Hospital at The Cleveland Clinic, Cleveland, OH Study objectives: Motor vehicle crashes are the leading cause of unintentional injury-related deaths among children aged 14 years and younger in the United States. Rates of restraint use decline steadily with increasing age of the child. We examine parental attitudes, beliefs, and barriers to the use of pediatric passenger restraints in motor vehicles for children aged 11 years or younger. Methods: An independent market research firm conducted a random sample telephone survey of parents living in an urban county. Results: Eight hundred respondents were surveyed (682 parents or foster parents, 100 other relatives, and 18 nonrelatives.) Among the parental respondents

Conclusion: This method identified out-of-hospital delay as infrequent but not rare in OHCA and is associated with worsened outcome

Tissue Plasminogen Activator for Acute Stroke: The Los Angeles Prehospital Stroke Study

Methods: This was a 3-year, prospective, consecutive case series in Los Angeles, CA. Participants were patients older than 40 years and determined by Los Angeles Fire Department paramedics to have had an acute stroke by meeting Los Angeles Prehospital Stroke Screen (LAPSS) criteria (previously described). Patients meeting LAPSS criteria were transported to any of 60 hospitals in the greater Los Angeles area. Results: Six hundred forty (80%) of the 800 patients who met LAPSS criteria during the study interval had complete hospital records available for review. Five hundred thirty-eight (84%) patients had an emergency department diagnosis of stroke or transient ischemic attack. Of these patients, 26 (4.8%) of the final diagnosis stroke patients received t-PA. The mean delay from reported onset of symptoms until accessing 911 was 18 minutes (range 0 to 48 minutes). The mean elapsed time from onset of symptoms to administration of t-PA was 138 minutes (range 0 to 224 minutes). Three patients were administered t-PA outside of the recommended 3-hour interval, 2 of whom died. Two (8%) patients who received t-PA subsequently developed intracranial hemorrhage. One of these patients died and the other was left vegetative. Altogether, 4 (15.4%) t-PA patients died. The mortality for patients who did not receive t-PA was 10.9%. Conclusion: Although a small percentage of patients with acute stroke received t-PA

Analysis of Child Traffic Injuries and Means for Prevention and Safety Within a Safe Community Program

Studies show that in 2000, 90% of all traffic fatalities occurred in low-and middle-income societies. The majority of the victims of these incidents were people who will never be able to afford a car: pedestrians, cyclists, and users of public transportation. The human suffering for victims of traffic-related injuries and their families is incalculable. The costs associated with surgery, prolonged hospitalization, and long-term rehabilitation for such victims, in addition to their lost productivity, represent tens of billions of dollars each year. Traffic injury is still an unsolved problem in Lithuania and other Baltic countries. In Lithuania, approximately 5,000 people die annually due to injuries, including 400 children and adolescents. In traffic accidents, there are approximately 700 deaths annually (22/100,000 population) including 70 adolescents and children. The injury rate in 2001 was 111.0 per 1,000 population; the child injury rate was 109.9 per 1,000 population. Disability; deaths caused by injuries; and social, humanitarian, and economic harm to society show the importance of injury-prevention activities. Study objectives: The aim of the investigation is to analyze the causes of childhood traffic injuries and to predict means for traffic injury prevention and safety promotion

Inhospital mortality occurred in 24 patients (4.2%; 95% CI 2.8% to 6.1%). History of Coumadin use did not increase the risk of ICH (OR 1.2; 95% CI 0.69 to 1.97) or MORT (OR 2.1; 95% CI 0.8 to 5.5). The use of antiplatelet medication did increase the risk of ICH

95% PT more than 14.9, the risk of bleed was not increased

Conclusion: Antiplatelet medications increase the risk of ICH, whereas use of Coumadin increases the risk of neither ICH nor MORT. However, a GCS score of

Baseline Behavioral Activity Rating Scale scores were high for PSYCH (n=40), ETOH (n=10), and SUBS

Ziprasidone decreased agitation scores rapidly for PSYCH, ETOH, and SUBS, respectively (means 5.7, 5.3, and 5.6 at 15 minutes and 3.2, 3.3, and 3.0 at 45 minutes, respectively; P<.01]); at 2 hours, scores were 2.5, 2.1, and 2.3, respectively. For conventional sedatives (n=7), Behavioral Activity Rating Scale scores were 6.4 at baseline, 5.4 at 15 minutes, 3.3 at 45 minutes, and 2.7 at 2 hours (P=nonsignificant from ziprasidone)

none had prolonged corrected QT interval; 1 dystonic reaction occurred with ziprasidone. Conclusion: Intramuscular ziprasidone appears effective in treating severe agitation, including that associated with ETOH-or SUBS-induced intoxication

Simple linear regression analyses to determine the relation between medicolegal concern and clinical suspicion for individual tests were not significant; however, for all resident tests ordered, (clinical suspicion) -1 predicted medicolegal concern (P<.001). The predictive value of this model was low (adjusted R 2 =0.025). For medicolegal concern more than 50 (high concern), we did find a significantly lower clinical suspicion than for low-risk tests for attending physicians (P=.02); for residents, there was a trend towards lower clinical suspicion, but the results were not significant (P=.22). Conclusion: Concern about medical malpractice is often a significant factor in the decision to order diagnostic tests in the ED, and often the perceived medicolegal concern outweighs clinical suspicion. This study found that residents perceive higher levels of medicolegal concern and clinical suspicion in test ordering than attending physicians. Chest radiographs and troponin were most frequently ordered when medicolegal concern outweighed clinical suspicion

were given log books to record specific aspects of their experience, including observations of medical errors and how these were dealt with. A definition of medical error was not included, but examples were given. Books were returned at the end of each rotation. Reported medical errors were classified as medication, diagnostic, procedural, management, communication or documentation, and detrimental delays. The frequency and type of error, subsequent serious adverse events (SAE), and whether the patient or family was informed of the error (revealed) were determined and analyzed with descriptive statistics. Results: Thirty-five (49%) of 71 students reported 53 errors, with 16 (31%) revealed. Nineteen SAEs (34%) occurred, with 6 (32%) revealed. Of 13 (25%) medication errors, 2 were revealed; 4 medication-SAEs occurred, with 1 revealed. Of 12 (24%) diagnostic errors, 8 were revealed; the 2 diagnostic-SAEs were not revealed. Of 10 (19%) management errors, 5 were revealed; 6 management-SAEs occurred, with 4 revealed. Of 7 (14%) communication or documentation errors, none were revealed, including 1 communication or documentation-SAE. Of 6 (12%) detrimental delay errors, none were revealed

Procedural Proficiency After Participation in Invasive Procedure Laboratories

Methods: This study, approved by the institutional review board, included physicians at a postgraduate year (PGY)-1 to PGY-3 emergency medicine residency program located in a small Midwestern city. A preintervention survey was distributed to 20 emergency medicine PGY-1 to PGY-3 residents. The survey included questions about the number of examinations performed, confidence in performing examinations, and general knowledge about care of sexual assault victims. First-year residents served as internal controls and did not receive intervention. The secondand third-year resident participants were given a collection of self-study materials consisting of articles and textbook excerpts. These residents also observed and assisted with examinations performed by SANE practitioners. The SANE program cares for an average of 60 sexual assault patients annually. A posttest was then given to the residents to determine whether knowledge and confidence in performing sexual assault examinations had improved. Results: After intervention, 70% of residents had performed at least 1 examination in comparison with 40% before intervention. Sixty percent of residents after intervention answered that they were either somewhat or very confident in performing the examination compared with 40% in the pretest

The Effect of Perceived Malpractice Risk on Diagnostic Test Ordering in Academic Emergency Medicine Practice

Methods: A prospective observational study was performed in a suburban academic ED with approximately 60,000 patient visits per year. ED patients 18 years or older who had 1 or more prespecified diagnostic tests ordered by the treating emergency physician were eligible for inclusion. Diagnostic tests included troponin, chest radiograph, extremity radiograph, head computed tomography (CT), chest CT, abdominal or pelvis CT (ACT), cerebrospinal fluid analysis, abdominal ultrasonographic scan, pelvic ultrasonographic scan, and ventilation/perfusion scan. Surveys were administered to resident and attending physicians in the ED immediately after tests were ordered. The surveys used visual analog scales (VAS) from 0 to 100 to estimate the physicians' clinical suspicion that ordered tests would yield abnormal results and to estimate the influence of medicolegal concern on the decision to order each test. VAS scores were then analyzed by using statistical methods (means, SDs, t tests, and simple linear regression). Results: Two hundred thirty-two patients had a total of 314 tests performed throughout a 6-month period. Residents' perceived medicolegal concern exceeded clinical suspicion in 33.3% of cases, whereas attending physicians' medicolegal concern exceeded clinical suspicion in 15

Medicolegal concern outweighed clinical suspicion most frequently among residents for chest radiograph ical guidelines, and the degree of integration of these tools into field operations. Using a 5-point Likert scale, we solicited opinions about patient care and team interoperability according to the development of field guidelines. Results: Surveys were completed by 27 DMATs (73%)

61% of nurses, and 46% of physicians but only 23% of ancillary staff. Most job descriptions reflect local or state emergency medical service regulations (78%)

Use of triage tools varies, depending on the setting (77% actual deployment, 88% disaster drill, 44% local aid event)

treatment (62%), and disposition (37%) issues. The purpose of protocols varies; 30% provide minimal guidance, 60% provide general recommendations, and 10% provide detailed algorithms. Protocols have been field tested by 7 (77%) of 9 teams. When asked whether creation of national field protocols would improve patient care and enhance interaction between teams, 16 (60%) of 27 agreed, 5 (18%) were neutral, and 5 (18%) disagreed. Development of standardized protocols is favored by 62% of teams. Conclusion: Most DMATs support the development of standardized field protocols; however, few (33%) have such clinical tools. Although the majority of teams surveyed are familiar with NDMS-sponsored triage guidelines, their application is variable, with many relying on local criteria. Scope of practice guidelines within DMATs is limited to out-of-hospital providers and nurses

Chan YF, Low R, Meehan S, Dun V, Ghandhi A/University of Medicine and Dentistry of New Jersey, New Jersey Medical School, Newark, NJ; New York City Health and Hospital Corporation, New York, NY Study objectives: We relate the incidence of assault in city hospitals to environmental factors and the September 11, 2001, attack. Methods: Assault patients were identified from billing records of 6 of 11 city hospitals operated by New York City Health and Hospital Corporation. Pollution data were obtained from the Environmental Protection Agency and weather data from the National Weather Service. Pollutants studied included particles less than 10 parts per million (oxides of nitrogen, nitrogen dioxide, ozone, and sulfur dioxide). This is an observational study covering 2,636 days beginning January 1, 1995. Data were analyzed with Auto-Regressive Integrated Moving Average (ARIMA) modeling by using SAS version 9. We did not include in our models complex interaction ral center ED in Orange County. Eligible women were at least 18 years of age and were in an intimate relationship at the time of the survey or during the previous year. Exclusion criteria were altered mental status, acute psychiatric illness, prison residence, severity of illness preventing participation, and inability to obtain consent or administer the survey privately.Results: All surveys were included in statistical analysis. The point prevalence for physical, emotional, and sexual partner violence was 6.8%. The 1-year prevalence was 37.0%. Emotional abuse was reported most often; sexual abuse was reported least. Women with emotional partner violence in the past year were 49.7 times more likely to have current emotional abuse than women without past emotional partner violence (P<.0005). Those with physical partner violence in the past year were 3.8 times more likely to report current physical partner violence (P<.0005). Income correlated inversely with partner violence. We found no correlation between partner violence and race or education.Conclusion: Data for partner violence in Orange County are consistent with data elsewhere. Emergency physicians should be vigilant in screening all women to avoid future abuse and delaying needed services. Limitations of this study include a convenience sample, single investigation site, and brief survey instrument.

Deitch K, Sorondo B, Lane P, Fisher J, Mallalieu J, Mallalieu R, Nguyen T/Albert Einstein Medical Center, Philadelphia, PA Study objectives: The purpose of this study was to identify characteristics associated with violence recidivism in the youth population of an inner-city emergency department (ED).Methods: This was a descriptive study conducted at an inner-city Level I trauma center. Trained research assistants identified patients aged 10 to 24 years during January 2001 to June 2002 who presented to the ED as a result of a violent injury. Patients were screened and enrolled at presentation to the ED by using a structured data collection tool. Up to 6 attempts were made to contact all patients by telephone by using a standardized questionnaire 4 to 12 months after their initial visit. Comparisons of means and χ 2 analyses of the demographic variables were performed. A logistic regression model was constructed to assess the characteristics associated with repeated violence.Results: Of 741 patients treated in the ED, 545 patients consented and were enrolled. One hundred ninety-six patients were successfully contacted for followup. The mean age for the uncontacted group was 18.58 years (95% confidence interval [CI] 18.28 to 18.88 years) and for the contacted group was 17.59 years (95% CI 17.06 to 18.02 years). The age difference was significant, with P=.002. There were no significant differences in sex, ethnicity, or type of injury between the contacted group and the uncontacted group. Of the 196 contacted patients, 41 (20.92%; 95% CI 15.11% to 26.73%) reported at least 1 repeated episode of violence. The mean age for the repeat group was 16.83 years (95% CI 15.67 to 17.99 years); 68.29% were male. There were no significant differences in age, sex, or ethnicity between the repeated violence group and the non-repeated violence group.Conclusion: Youths who presented to the ED and had experienced a violent injury were a difficult group to follow up and were at risk for repeated violence. There was no association between repeated violence and demographic characteristics. Further analyses of follow-up systems and risk factors for repeated violence are needed.