key: cord-292335-al6v3b9x
authors: Crotty, Matthew P.; Meyers, Shelby; Hampton, Nicholas; Bledsoe, Stephanie; Ritchie, David J.; Buller, Richard S.; Storch, Gregory A.; Kollef, Marin H.; Micek, Scott T.
title: Impact of antibacterials on subsequent resistance and clinical outcomes in adult patients with viral pneumonia: an opportunity for stewardship
date: 2015-11-18
journal: Crit Care
DOI: 10.1186/s13054-015-1120-5
sha: 
doc_id: 292335
cord_uid: al6v3b9x

INTRODUCTION: Respiratory viruses are increasingly recognized as significant etiologies of pneumonia among hospitalized patients. Advanced technologies using multiplex molecular assays and polymerase-chain reaction increase the ability to identify viral pathogens and may ultimately impact antibacterial use. METHOD: This was a single-center retrospective cohort study to evaluate the impact of antibacterials in viral pneumonia on clinical outcomes and subsequent multidrug-resistant organism (MDRO) infections/colonization. Patients admitted from March 2013 to November 2014 with positive respiratory viral panels (RVP) and radiographic findings of pneumonia were included. Patients transferred from an outside hospital or not still hospitalized 72 hours after the RVP report date were excluded. Patients were categorized based on exposure to systemic antibacterials: less than 3 days representing short-course therapy and 3 to 10 days being long-course therapy. RESULTS: A total of 174 patients (long-course, n = 67; short-course, n = 28; mixed bacterial-viral infection, n = 79) were included with most being immunocompromised (56.3 %) with active malignancy the primary etiology (69.4 %). Rhinovirus/Enterovirus (23 %), Influenza (19 %), and Parainfluenza (15.5 %) were the viruses most commonly identified. A total of 13 different systemic antibacterials were used as empiric therapy in the 95 patients with pure viral infection for a total of 466 days-of-therapy. Vancomycin (50.7 %), cefepime (40.3 %), azithromycin (40.3 %), meropenem (23.9 %), and linezolid (20.9 %) were most frequently used. In-hospital mortality did not differ between patients with viral pneumonia in the short-course and long-course groups. Subsequent infection/colonization with a MDRO was more frequent in the long-course group compared to the short-course group (53.2 vs 21.1 %; P = 0.027). CONCLUSION: This study found that long-course antibacterial use in the setting of viral pneumonia had no impact on clinical outcomes but increased the incidence of subsequent MDRO infection/colonization.

Interactions between viral and bacterial respiratory pathogens have been recognized dating back to the 1918 influenza pandemic [1] . Bacterial pneumonia is a wellrecognized serious complication of influenza infections and coinfections are commonplace [2] [3] [4] [5] [6] [7] [8] [9] [10] . Respiratory syncytial virus (RSV), parainfluenza viruses, rhinoviruses, and adenoviruses have also been linked to bacterial coinfections in humans [11] [12] [13] [14] [15] [16] [17] [18] . Animal studies have suggested synergism between bacterial pathogens and other respiratory viruses [19, 20] . The relationship between viral and bacterial respiratory infections creates a difficult situation for clinicians determining the appropriate use of antimicrobials as they treat hospitalized patients with pneumonia while also trying to minimize the development and selection of resistant organisms.

Respiratory viruses are increasingly recognized as the primary etiology of pneumonia among patients requiring hospitalization (2.7-5 % of pneumonia cases) [21, 22] . Advanced technologies using multiplex molecular assays and PCR improve the diagnostic ability to identify viral pathogens in a timely manner and may impact the use of antibacterials in patients with no bacterial infection identified.

Several studies have investigated the impact of respiratory viral pathogen identification on antibacterial exposure [23] [24] [25] [26] . Decreased antibiotic use was observed in two pediatric studies assessing the impact of rapid viral diagnostic tests for respiratory tract infections; however, these results were not mirrored in similar adult studies [23] [24] [25] . These studies all used immunofluorescent staining as the primary diagnostic technology. To our knowledge, only one study using PCR-based respiratory virus detection has been reported and found no change in antibacterial use with improved diagnoses for lower respiratory tract infections [26] .

Broad-spectrum antibacterial exposure increases the risk of subsequent infections with multidrug-resistant organisms (MDROs) and leads to a vicious cycle of empiric broad-spectrum antibacterials to combat increasingly resistant organisms [27] . We and others have previously shown that patients with culture-negative pneumonia frequently receive treatment with broadspectrum antibiotics, usually in excess of 5-6 days of therapy despite lack of evidence for a bacterial etiology of infection [28, 29] . It is important to recognize that these studies were performed prior to the availability of rapid viral diagnostics which may have influenced how antibiotics were used during those study periods. Use of new diagnostic technologies for respiratory virus detection could decrease unnecessary antibacterial exposures and subsequent MDRO infections. This study aimed to describe the use of continued empiric antibacterials in patients with known viral pneumonia and to determine the impact of such therapies on subsequent bacterial infections/colonization and clinical outcomes.

This was a single-center, retrospective cohort study of patients with a positive respiratory virus panel (RVP) at Barnes-Jewish Hospital (BJH) (a 1300-bed urban academic medical center in St. Louis, MO, USA) between 1 March 2013 and 7 November 2014. The study protocol was approved by the BJH, Washington University, and St. Louis College of Pharmacy Institutional Review Boards.

Consecutive patients who were ≥19 years of age and admitted to BJH for ≥24 hours were assessed for study inclusion. Patients were identified through a query of an internal database which tracks respiratory viruses. Patient admissions in which a respiratory virus was identified by FilmArray ® respiratory panel (FARP) assay (BioFire Diagnostics, Salt Lake City, UT, USA) were screened for inclusion in this study. Included patients had to meet the study definition of viral pneumonia. Patient admissions were excluded if rhinovirus or enterovirus was identified by nasopharyngeal (NP) swab alone but could also be included if identified from lower respiratory tract specimens. Additionally, patients were excluded if there was a virus identified by RVP within the previous 90 days or if a bacterial pathogen was identified by the respiratory panel. Patients who were transferred from an outside hospital and those who were discharged/died or were made comfort care less than 72 hours after the index RVP report date were also excluded to better evaluate continued empiric antibacterial use in this population.

The FARP assay is a multiplexed nucleic acid test capable of simultaneous qualitative detection and identification of multiple respiratory viral and bacterial nucleic acids. 

Viral pneumonia was defined as identification of a respiratory virus by FARP and a new or progressive radiographic infiltrate within 48 hours of the index RVP. Short-course antibacterial administration was defined as treatment with systemic antibacterials for less than 3 days while long-course therapy was defined as 3-10 days of antibiotics after the index RVP report date with no bacterial infection identified at any point during the admission. MDROs were defined as methicillin-resistant Staphylococcus aureus (MRSA), vancomycin-resistant enterococci (VRE), or a bacterium exhibiting in vitro resistance to at least one drug in ≥3 classes of antibacterials for which the organism is not intrinsically resistant [30] . Subsequent isolation of a MDRO was defined as isolation of a study-defined MDRO or a positive epidemiologic screen (MRSA NP swab; VRE stool specimen) ≥10 days after the report date of the index RVP. Positive epidemiologic screens for MRSA or VRE were only considered to be subsequent colonization of an MDRO if the patient had at least one screen in the previous 180 days and all screening tests were negative prior to the index RVP. Immunocompromised status was defined as a diagnosis of human immunodeficiency virus (HIV), active malignancy (stem cell transplant or receiving chemotherapy), solid organ transplant, or currently on immunosuppressive therapy (prednisone 20 mg/day for at least 30 days or equivalent).

Information regarding the FARP including time of collection/report, type of specimen, patient location at time of collection, and resulting findings were obtained from an internal database. Additionally, all available aerobic and anaerobic bacterial cultures were evaluated based on electronic medical record (EMR) query. In vitro susceptibilities of isolated bacterial pathogens were evaluated as reported per institutional practices. Urine Legionella antigen, direct-fluorescent antibody for Pneumocystis jiroveci, and Clostridium difficile toxin assay were also evaluated.

Data collected to describe patient groups included demographic information, comorbid conditions, and clinical outcomes. Charlson's comorbidity index was used as a summative score of underlying disease states [31] . Antibacterial exposures were calculated using EMR orders for systemic antibacterials. Days of therapy (DOT) and DOT normalized per 1000 patient-days (DOT/1000PD) were calculated as described previously [32] . EMR queries were used to acquire patient information where possible. Manual chart review was used to validate and supplement all EMR queries.

The primary endpoint was subsequent isolation of a MDRO ≥10 days after the index RVP report date. Inhospital mortality and readmission at 30, 90, and 180 days were secondary endpoints.

Dichotomous variables were compared using the chisquare test or Fisher's exact test as appropriate. Continuous variables were compared using the Mann-Whitney U test. All tests were two-tailed and P <0.05 was considered significant. Univariate analyses were performed to compare the group that received long-course antibacterials and the group that received short-course antibacterial therapy. Kaplan-Meier survival analysis was used to compare risk of in-hospital mortality between comparator groups while censoring for patient discharge. All statistical analyses were performed using SPSS software (IBM SPSS Statistics, version 22.0; Chicago, IL, USA).

A total of 174 consecutive patients (long course, n = 67; short course, n = 28; mixed bacterial-viral infection, n = 79) were included in this study (Fig. 1 ). Demographic and clinical characteristics are summarized in Table 1 . The median age was 57 years and 46.0 % were male. A majority of patients in the cohort were deemed to be immunocompromised (56.3 %), with active malignancy the most common etiology (69.4 %). Only 5 of 174 patients had been admitted within the 90 days prior to the index admission. Figure 2 provides a breakdown of the subgroup of patients screened in the emergency department setting according to the presence of pure viral and mixed infections.

Rhinovirus/enterovirus (23.0 %), influenza (19.0 %), and parainfluenza (15.5 %) were the viruses most commonly identified in the cohort ( Table 2 ). Of the total 174 patients, 11 had multiple respiratory viruses identified by RVP: RSV (5 of 11), rhinovirus/enterovirus (5 of 11), influenza (4 of 11), human metapneumovirus (3 of 11), parainfluenza (2 of 11), coronavirus (2 of 11), and adenovirus (2 of 11). Specimens resulting positive for a respiratory virus in the cohort included NP swabs (40.8 %), bronchoalveolar lavage (25.9 %), tracheal aspirates (18.4 %), bronchial washes (9.8 %), and sputum samples (4.0 %). The RVP identified a respiratory virus from multiple specimens in 25 patients. Among the 25 patients with multiple positive specimens, the most common were bronchoalveolar lavage (19 of 25 patients) and tracheal aspirates (13 of 25 patients).

No significant differences in comorbidities between groups were identified ( Table 1) . Patients with mixed viral-bacterial infection had statistically greater APA-CHE II scores and were more likely to require vasopressor support compared with patients with pure viral infection. There was no significant difference in the number of immunocompromised patients between groups, although numerically more patients who received long-course antibacterials had an active malignancy or solid organ transplant compared with patients receiving short-course therapy. Virus types identified by FARP assay were similar in all three patient groups ( Table 2 ). The bacterial coinfecting organisms identified in patients with mixed viral-bacterial infection are presented in Table 3 . Respiratory coinfection with a bacterial pathogen was most common, with S. aureus, Streptococcus pneumoniae, and Pseudomonas aeruginosa accounting for the most frequent respiratory coinfecting bacteria.

A total of 13 different systemic antibacterials were used as empiric treatment in patients with viral pneumonia without bacterial coinfection for a total of 466 DOT. Vancomycin (50.7 %), cefepime (40.3 %), azithromycin (40.3 %), meropenem (23.9 %), and linezolid (20.9 %) were the most frequently used empiric antibacterials in patients with viral pneumonia without bacterial coinfection (Fig. 3) . The most common regimens used in viral pneumonia without bacterial coinfection were vancomycin plus cefepime (28.4 %) and vancomycin plus meropenem (13.4 %). A total of 44 (65.7 %) patients with viral pneumonia without bacterial coinfection received empiric MRSA coverage with vancomycin or linezolid. Empiric antibacterial therapy was continued for a median of 4.1 days (interquartile range, 2.5-6.1 days) in viral pneumonia without bacterial coinfection, with most (69 %) being days on intravenous antibacterials.

Total antibacterial exposure differed between the longcourse and short-course groups at 2116 and 484 DOT/ 1000PD, respectively ( 

The number of patients with subsequent MDRO colonization or infection was not significantly different between groups (Table 4 ). However, in instances of subsequent infection or colonization, where a single patient could have more than one organism, there was a higher rate of MDRO identification among isolates from the group that received long-course antibacterials compared with the group receiving short-course therapy (53.2 vs. 21.1 %; P = 0.027) ( Table 4) In-hospital mortality was statistically higher for the mixed-infection group compared with the long-course therapy group (Table 4) . Kaplan-Meier survival analysis showed that the mixed-infection group had the lowest overall survival, but these differences were not statistically ICU admit was determined to be oncology related if the patient fit the study definition for active malignancy *Statistically significant difference (P <0.05) between long-course and mixed-infection groups **Statistically significant difference (P <0.05) between short-course and mixed-infection groups BMI body mass index, CCI Charlson's comorbidity index, COPD chronic obstructive pulmonary disease, ESRD end-stage renal disease significant (Fig. 4) . ICU mortality was also significantly higher for patients in the mixed-infection group compared with the long-course therapy group. Patients receiving long-course therapy or those with mixed infection had statistically longer ICU length of stay compared with patients receiving short-course therapy. Hospital readmission rates were similar between groups at 30, 90, and 180 days after index hospitalization discharge.

This study compared a cohort of 174 patients with viral pneumonia and mixed viral-bacterial infection based on exposure to continued empiric antibacterials after respiratory virus identification. More of the subsequent infecting or colonizing bacterial isolates from the group with pure viral pneumonia who received continued long-course antibacterials were defined as MDROs compared with the short-course group (P = 0.027). These findings suggest that more prolonged exposure to broad-spectrum antibacterials in patients with viral pneumonia may have promoted resistance in these patients.

No benefit of continued empiric antibacterials for patients with pure viral pneumonia was seen in this study. The risk of bacterial coinfection in the setting of viral pneumonia, especially with influenza, creates a challenging situation for clinicians. The potential detrimental impact of not treating a bacterial pathogen weighs Data expressed as number (% of total) or median (interquartile range) LOS length of stay, MDRO multidrug-resistant organism, MRSA methicillin-resistant Staphylococcus aureus, VRE vancomycin-resistant enterococci *Statistically significant difference (P <0.05) between short-course and long-course groups **Statistically significant difference (P <0.05) between long-course and mixed-infection groups ***Statistically significant difference (P <0.05) between short-course and mixed-infection groups heavily on the decision process and downstream effects of such therapies may be disregarded. Our findings of similar clinical outcomes between patients with pure viral pneumonia who received long-course antibacterials after virus recognition and those who did not may suggest opportunity for de-escalation of empiric antibacterial therapy when viral pneumonia is identified. A previous randomized controlled trial by Oosterheert et al. [26] evaluated implementation of real-time PCR rapid diagnostics for respiratory pathogen identification. They found increased diagnostic yield with the assay but no difference in antibiotic use, and hypothesized that reluctance to change treatment based on testing results may have inhibited cost-effectiveness from being demonstrated. In our study, systemic antibacterials were discontinued following identification of a respiratory virus by RVP for several patients; however, whether virus identification directly led to discontinuation of antibacterials cannot be determined. The willingness of prescribers to de-escalate and stop antibacterials in this setting may suggest increased recognition of the role of viral pathogens in pneumonia. Additionally, the expanded panel of viruses detected may have factored into how results were perceived, as prescribers may have been more likely to attribute pneumonia to newly detectable viruses such as human metapneumovirus. However, it is not possible to definitively determine the rationale for stopping antibacterial therapy.

Timely antibiotic administration is crucial for treating hospitalized patients with suspected pneumonia [33] . Antimicrobial de-escalation attempts to balance the use of these essential drugs up front with the emergence of resistance [34] . The optimal strategy for de-escalation of antibacterials in the setting of viral pneumonia without an identified bacterial coinfection is unclear. Our study found no difference in clinical outcomes based on antibiotic duration of therapy in patients with viral pneumonia despite significantly different total antibacterial exposure (DOT/1000PD) between groups. Byington et al. [24] found previously that improved diagnostic technologies enhancing detection of respiratory viruses decreased antibacterial use at a children's hospital. The authors concluded that improved diagnostics are an important tool in decreasing unnecessary antibacterial prescribing. Our study similarly illustrated the potential impact of respiratory virus diagnostics on antibacterial use in an adult population.

C. difficile infection is a major cause of morbidity and mortality in US hospitals and has been directly linked to exposure to broad-spectrum antibiotics [35, 36] . In a cohort of hospitalized adult patients, Shiley et al. [37] found that significantly more patients who continued to receive antibacterials after diagnosis of a viral respiratory tract infection developed C. difficile infection. One patient in our study who was treated with long-course antibacterials after identification of a respiratory virus also developed C. difficile infection. Strategies to best limit the use of unneeded antibacterials are important to curtail against the growing issues of C. difficile and resistance, and may be aided by de-escalation approaches using enhanced viral diagnostic technologies.

Limitations of this study should be noted. First, this was a small retrospective cohort study of patients at a single institution and may not be representative of all settings. It is important to note that BJH is a regional specialty referral hospital and not a community hospital. This accounts for the case mix with a high prevalence of immunosuppressed patients and the low prevalence of narrow spectrum empiric antibiotic utilization. The small number of patients meeting inclusion criteria did not allow for definitive conclusions to be made regarding group comparison as a lack of statistically significant differences being found could be due to the lack of sample size. Second, patients were determined to have viral pneumonia based on virus identification and radiographic findings but other markers of illness, such as white blood cell count and fever, were not considered and the retrospective nature of the study did not allow evaluation of what drove continuation of antibacterials in some patients but not others. Moreover, we did not attempt to identify risk factors associated with pure viral pneumonia. Third, although coinfecting bacterial pathogens were not identified in patients with pure viral pneumonia, it is impossible to prove that they were not present. Receipt of antibacterials prior to obtaining bacterial cultures could have limited the diagnostic yield of bacterial cultures in some cases and yield from bacterial cultures is not perfect. Finally, all of the viral pneumonia cases occurred in a 20-month period. Viral epidemiology during this time may not be representative of all seasons. Influenza H1N1 p2009 was the primary influenza virus identified in our study (85 %). Incidence rates of bacterial coinfection and coinfecting organisms may differ from year to year and from virus type to virus type, which may hinder application of de-escalation strategies using the results of this study.

It is not possible to directly link the development of subsequent MDRO infections/colonization and C. difficile infection seen in our study to the continued empiric antibacterials administered. All of the patients included in the cohort received antibacterials at some point during their index hospitalization and infection control measures were not directly assessed in these patients. Additionally, hospitalization itself probably increases the risk of these patients being colonized with MDROs. Use of cephalosporins and vancomycin, two of the most commonly administered empiric agents in our study, have been implicated as increasing the prevalence of VRE, the most commonly identified subsequent MDRO in this study [38, 39] . Decreasing exposure to broad-spectrum antibacterials such as third-generation and fourth-generation cephalosporins and vancomycin would be expected to lessen the incidence of VRE and other MDROs as was seen in this study, but the risk of development and transmission of resistance in the hospital cannot be completely eliminated. Antibacterials are extraordinarily important in the treatment of many hospitalized patients and their use is often warranted. Decreasing unnecessary use may help curb acquirement of resistant organism in healthcare settings but even appropriate use can lead to the development of resistance. Only through multifaceted efforts of infection control and antimicrobial stewardship can the spread of MDROs between patients, clinicians, workers, and visitors be diminished.

This study highlights the potential benefits of improved diagnostics for respiratory viruses, primarily the potential for decreased antibacterial exposure and thus decreased selective pressure for resistant bacterial isolates. Antibacterial exposure applies selective pressure and promotes colonization/infection by resistant organisms including MRSA and VRE [40, 41] . Halting this process is essential to maintain effective therapeutic options in the future and may be aided by discontinuation of antibacterials in cases of viral pneumonia. In our study, patients with viral pneumonia exposed to long-course antibacterials had more occurrences of subsequent infection or colonization with MDRO isolates. In contrast, the number of patients with subsequent MDRO infection or colonization was not different between groups although this may be due to the small number of patients in each group. No differences in clinical outcomes, including in-hospital mortality and readmission rates, were observed between patient groups. In the setting of viral pneumonia and no coinfecting bacterial pathogens, discontinuation of antibacterials is reasonable in many if not most cases, and may allow for decreased overall antibacterial use. Enhanced diagnostic technologies can potentially be incorporated into antimicrobial stewardship practices to allow for de-escalation of unnecessary antibacterials. These findings warrant further investigation to determine the applicability of an antibacterial deescalation approach in the setting of viral pneumonia.

In this single-center retrospective cohort, patients with viral pneumonia who continued to be treated with systemic antibacterials 3 days after virus identification were more likely to have a subsequent infection or colonization with a MDRO than were patients in whom systemic antibacterials were stopped. In-hospital mortality based on Kaplan-Meier survival analysis and readmission rates were not different between groups based on antibacterial exposure after respiratory virus identification. 

All authors declare that they have no competing interests.

Pandemic 2009 influenza A in Argentina: a study of 337 patients on mechanical ventilation

Predominant role of bacterial pneumonia as a cause of death in pandemic influenza: implications for pandemic influenza preparedness

Critically ill infants and children with influenza A (H1N1) in pediatric intensive care units in Argentina

Critically ill patients with 2009 influenza A (H1N1) infection in Canada

Community acquired respiratory coinfection in critically ill patients with pandemic 2009 influenza A (H1N1) virus

Critical illness from 2009 pandemic influenza A virus and bacterial coinfection in the United States

Critical care services and 2009 H1N1 influenza in Australia and New Zealand

Bacterial coinfections in lung tissue specimens from fatal cases of 2009 pandemic influenza A (H1N1)-United States

Pulmonary pathologic findings of fatal 2009 pandemic influenza A/H1N1 viral infections

Lung pathology in fatal novel human influenza A (H1N1)infection

Epidemiology and clinical characteristics of community-acquired pneumonia in hospitalized children

Viral etiology of severe pneumonia among Kenyan infants and children

Incidence of respiratory pathogens in persons hospitalized with pneumonia in two provinces in Thailand

A preliminary study of pneumonia etiology among hospitalized children in Kenya

Etiology of community-acquired pneumonia in hospitalized children in northern Taiwan

Viral coinfections in children with invasive pneumococcal disease

Temporal association between rhinovirus circulation in the community and invasive pneumococcal disease in children

Clinical and molecular characterization of rhinoviruses A, B, and C in adult patients with pneumonia

Infection with human metapneumovirus predisposes mice to severe pneumococcal pneumonia

Viral enhancement of nasal colonization with Haemophilus influenzae type b in the infant rat

Epidemiology and predictors of multidrug-resistant community-acquired and health care-associated pneumonia

Pneumonia pathogen characterization is an independent determinant of hospital readmission

Cost-effectiveness of rapid diagnosis of viral respiratory tract infections in pediatric patients

The effect of rapid respiratory viral diagnostic testing on antibiotic use in a children's hospital

Clinical and financial benefits of rapid detection of respiratory viruses: an outcomes study

Impact of rapid detection of viral and atypical bacterial pathogens by real-time polymerase chain reaction for patients with lower respiratory tract infection

Antibiotic-resistant bugs in the 21st century: a clinical super-challenge

A comparison of culture-positive and culture-negative health-care-associated pneumonia

Epidemiological differences between sepsis syndrome with bacteremia and culture-negative sepsis

Multidrug-resistant, extensively drug-resistant and pandrug-resistant bacteria: an international expert proposal for interim standard definitions for acquired resistance

A new method of classifying prognostic comorbidity in longitudinal studies: development and validation

Measurement of adult antibacterial drug use in 130 US hospitals: comparison of defined daily dose and days of therapy

Infectious Diseases Society of America/American Thoracic Society consensus guidelines on the management of community-acquired pneumonia in adults

Strategies to prevent antimicrobial resistance in the intensive care unit

Burden of Clostridium difficile infection in the United States

Clostridium difficile infection

The use of antimicrobial agents after diagnosis of viral respiratory tract infections in hospitalized adults: antibiotics or anxiolytics

The effect of vancomycin and third-generation cephalosporins on prevalence of vancomycin-resistant enterococci in 126 U.S. adult intensive care units

The problem with cephalosporins

Does antibiotic exposure increase the risk of methicillin-resistant Staphylococcus aureus (MRSA) isolation? A systematic review and meta-analysis

Vancomycin-resistant enterococcal infections

Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color figure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www

Authors' contributions MPC and STM conceived the study. MPC, STM, DJR, MHK, RSB, and GAS participated in the study design. NH and SB performed the electronic medical record queries. MPC and SM conducted data collection and managed the data, including quality control. MPC and STM analyzed the data and interpreted the results. MPC, STM, and DJR drafted the manuscript, and all authors contributed substantially to its revision. All authors read and approved the manuscript.