key: cord-1019186-yx313o2k authors: Mockler, Gretchen L; Samantha, Novotny; Wei, Hou; Yuhang, Liu; Schoenfeld, Elinor title: Patient Self-Report Superior to Electronic Medical Record Abstraction for Identifying Positive COVID-19 Symptoms at Illness Onset date: 2022-05-12 journal: nan DOI: 10.1016/j.focus.2022.100005 sha: 640fe10bd7df723238eaf37324411994194a4bbc doc_id: 1019186 cord_uid: yx313o2k Introduction Most initial COVID-19 research focused on hospitalized patients. Presenting symptomatology in the outpatient setting was poorly characterized, making it difficult for primary care physicians to predict which patients would require hospitalization. Purpose To characterize presenting symptoms of COVID-19 infection and baseline patient characteristics and evaluate for correlation with disease severity, duration, and chronicity in the outpatient setting. Methods 107 adult, English speaking patients with suspected and confirmed COVID-19 cases at the three primary care practices of Stony Brook University Hospital were studied between March and December, 2020. Survey data was collected from patient telephone interviews and electronic medical record (EMR) abstraction. Potential risk factors assessed included participant demographics, medical comorbidities, and number and type of symptoms at illness onset. Outcome measures included symptom duration, hospitalizations, and persistence of symptoms at 12 weeks from study enrollment. Results Patient self-report survey elicited nearly twice as many symptoms described at illness onset vs. those recorded in the EMR (p<0.0001). Higher number of symptoms at illness onset was positively associated with symptom duration and chronicity. The presence of fever and hypoxia at the onset of illness were each positively associated with eventual hospitalization for COVID-19 disease. Conclusions Early in the setting of newly emerging infectious diseases, particularly those such as COVID-19 which involve multiple organ systems, patient self-report of symptoms using a complete review of systems, rather than EMR abstraction alone, may be key for accurate disease identification and characterization as well as prediction of eventual disease severity, duration and chronicity. Grant, $7,500, July, 2020. This sponsor had no role in study design, collection, analysis or interpretation of data, the writing of the report or the decision to submit the report for publication. Presented in part at the North American Primary Care Research Group (NAPCRG) Annual Conference, November, 2021. As COVID-19 ravaged nations around the world in early 2020, the U.S. reported the highest number of infections worldwide. 1 New York State, New York City (NYC) and the NYC suburbs rapidly became COVID-19 epicenters, with an extraordinary burden of disease and mortality. As of February 5, 2022, there have been over 76 million COVID-19 cases and 900,000 deaths in the United States. Suffolk County, NY, where our study was based, has reported over 417,000 cases and 4,200 deaths. 1 Most initial COVID-19 research focused on hospitalized patients. 2, 3 Presenting symptomatology in the outpatient setting and non-respiratory symptoms were poorly characterized. This made it difficult for primary care physicians to predict which patients would require hospitalization or decompensate while being managed at home, and hospital beds were limited. To address these knowledge gaps, we initiated a study using patient interviews and electronic medical record (EMR) review to characterize presenting symptoms of COVID-19 infection in the outpatient setting and evaluate for correlations with disease severity, duration, and chronicity. IRB approval for this longitudinal, observational study was obtained. Patients >18 years of age who presented to any of the three suburban primary care practices of Stony Brook University Hospital with physician-suspected or test-confirmed COVID-19 between February and August, 2020, and who agreed to telephone interview and EMR review by study staff, were eligible. Of the primary care practices involved in this study, all three were faculty practices, one was a patient-centered medical home (PCMH), and one included residents being precepted by faculty. The size of the three practices included in this study totaled 24,463 patients: 15,942 in internal medicine primary care; 7,821 in family medicine; and 700 in preventive medicine. Participants were followed from April through December, 2020. Study staff were trained in interviewing and EMR data abstraction. Data was collected via telephone and EMR review, and recorded into a REDCap database. Patient interview questionnaires were designed in regular consultation with emerging Centers for Disease Control and Prevention (CDC) data regarding COVID-19, and queried demographics, medical comorbidities, medications, prevention of transmission practices, travel, COVID-19 symptom course and treatment, and test results. Parallel EMR data were abstracted from clinician notes and from laboratory and radiology reports. Participants were surveyed weekly until two weeks after COVID-19 symptoms resolved, or a maximum of three months. Questionnaire responses were used to characterize potential risk factors including participant demographics, medical comorbidities, travel, and number and type of symptoms at illness onset including a priori five "key" initial COVID-19 signs and symptoms commonly used in the existing literature as surrogates for disease severity: fever, cough, shortness of breath, hypoxia and confusion [2] [3] [4] [5] . Fever, including low-grade fever, was defined as temperature ≥ 99ºF. 6, 7 Outcomes included increased disease severity, for which hospitalization was used as a marker; increased symptom duration; and symptom chronicity, defined as persistent symptoms at 12 weeks after study enrollment. Patient characteristics were described using medians (range) and frequencies (percentages) when appropriate. Percentages of pre-existing medical comorbidities, number of symptoms at onset of illness, COVID-19 testing, and escalating medical care were calculated. Percentages of each individual symptom were calculated at onset of illness and at 12 week follow up. To evaluate the association of risk factors with hospitalization (ever vs. never) and symptom chronicity (persistent vs. resolved), univariable and multivariable analyses were performed. To evaluate for association of risk factors with duration of symptoms, Spearman correlation was used. For univariable analyses, Chi-square tests were used for categorical risk factors (e.g. medical comorbidities, travel history). Wilcoxon rank sum tests were used for numeric risk factors (e.g. number of symptoms at illness onset). For multivariable analyses, hospitalization (ever vs never) and symptom resolved status (persistent vs resolved) were analyzed as the outcomes in logistic regression. Explanatory variables included number of symptoms, presence of 5 key signs, symptoms at onset of illness, medical comorbidities, history of travel and demographic characteristics (i.e. age, gender and race). All analyses were hypothesis driven and were performed using SAS v9.4 (the SAS Institute, Cary, NC). 178 patients were deemed eligible for the study; 146 were successfully contacted and 107 were enrolled. 14 participants were lost to interview follow up, but EMR follow up was completed. Among participants, the largest group by age was 40-64 years old, and the majority were female and Caucasian. Half had traveled either domestically or internationally during the pandemic period (after January 1, 2020). Nearly one fifth of participants had no pre-existing medical conditions. Close to 50% of participants had a diagnosis of obesity and/or cardiovascular disease, approximately one third had chronic pulmonary disease, and one tenth were diabetic ( Table 1) . Participant self-report at telephone interview yielded nearly two times as many COVID-19related symptoms at illness onset vs. EMR abstraction (median=7 vs. 4, p<0.0001). The most frequent symptoms at onset of illness were fatigue, cough, myalgias and headache. Eighteen percent (19/107) of study participants had persistent symptoms at 12 weeks after study enrollment. In order of prevalence, these symptoms were fatigue, headache, myalgia, and shortness of breath ( Table 2 ). Presence of fever or hypoxia at illness onset was positively associated with hospitalization in the multivariable logistic regression model (Table 3a) . Higher number of symptoms at illness onset was positively associated with increased symptom duration (R=0.2556, p=0.0279) and with chronic symptoms at 3 months from study enrollment (p=0.04) (Table 3b ). This longitudinal study collected patient-reported and EMR data from three primary care practices to describe COVID-19 symptomatology, disease progression and hospitalization in an epicenter of the pandemic's first wave. Patients self-reported nearly twice as many symptoms at illness onset than were recorded in the EMR. This gap in documentation of patients' symptoms has been previously noted in other disease states 8-10 but has particular urgency for our current circumstances. Rapid characterization of the full range of symptoms and presentations of newly emerging diseases such as COVID-19 is essential for accurate diagnosis and management. In the case of COVID-19, higher number of distinct symptoms in the first five days of illness has been associated with eventual need for supplemental oxygen. 4 At the beginning of the COVID-19 pandemic, defining symptoms included cough, dyspnea, fever, and fatigue. 2, 3 Only later were gastroenterological, neurological and dermatological manifestations recognized. [11] [12] [13] Time constraints due to increased patient volume during periods of emergency can prohibit extensive review of systems (ROS) documentation. For research purposes when investigating new diseases, therefore, EMR data abstraction alone may be insufficient. Supplemental patient interviews with full ROS may be necessary to obtain comprehensive data and define pathognomonic symptoms of new diseases in real time. This method might also help to characterize new viral variants such as the Delta or Omicron variants of COVID-19 as they emerge, as well as "atypical" presentations particular to specific patient groups (e.g. elderly, pediatric, pregnant, immunosuppressed). In order to make this feasible, a full ROS questionnaire might be sent electronically to patients prior to their clinic visits or obtained by clinic support staff via telephone, both for physician review at the time of the patient's clinical visit and for research purposes. In the case of larger studies, automated natural language processing might be employed to maximize extraction of data from free text notes. Notably, real-time supplemental patient interviews are an important tool used in public health case investigation; thus, close collaboration between clinical and public health sectors may play a key role in the rapid and thorough characterization of new diseases. The longitudinal aspect of this study allowed for analysis of associations between presenting symptoms and disease course, demonstrating that the presence of hypoxia or fever at the time of illness onset were each associated with a higher risk of hospitalization, and that a higher number of symptoms at illness onset was associated with increased symptom duration and with symptom chronicity at 3 months from study enrollment. The symptoms which our study found to have Study Limitations: Our patient sample was limited to primary care patients at a single hospital system in Suffolk County, NY during the initial wave of COVID-19. It may not be generalizable to other populations. Small sample size limited our statistical power to analyze correlations between potential risk factors and disease outcomes. As this study occurred during a time when outpatient and inpatient resources were severely limited, many patients were managed from home via telehealth. This limited the ability of clinicians to obtain vital signs or a complete physical exam. Finally, COVID-19 testing was not widely available for the initial period of our study, and test type varied widely later during the study period, so we relied on clinician diagnosis alone when necessary. This method may be needed again in the case of other new infectious diseases for which testing is being developed and refined while the disease itself is still being clinically defined. In our present times of rapid population growth, global warming, high potential for further pandemic diseases and increased virtual assessment of patients, it may be wise to return to a comprehensive review of systems to evaluate for multiple organ system involvement in a newly emerging disease such as COVID-19. The outpatient primary care setting may offer the ideal opportunity for this approach, helping to rapidly characterize novel diseases and save patients' lives earlier in the trajectory of a future pandemic. **Hypertension did not remain significant in the multivariable logistic regression model. An interactive web-based dashboard to track COVID-19 in real time Clinical features of patients infected with 2019 novel coronavirus in Wuhan Clinical Characteristics of 138 Hospitalized Patients With 2019 Novel Coronavirus-Infected Pneumonia in Wuhan Symptom clusters in COVID-19: A potential clinical prediction tool from the COVID Symptom Study app Symptom clusters in Covid19: A potential clinical prediction tool from the COVID Symptom study app medRxiv Low-grade fever: how to distinguish organic from non-organic forms Prevalence and characteristics of fever in adult and paediatric patients with coronavirus disease 2019 (COVID-19): A systematic review and meta-analysis of 17515 patients Longitudinal symptom dynamics of COVID-19 infection Agreement of Ocular Symptom Reporting Between Patient-Reported Outcomes and Medical Records Low Concordance of Patient-Reported Outcomes With Clinical and Clinical Trial Documentation Digestive Symptoms in COVID-19 Patients With Mild Disease Severity: Clinical Presentation, Stool Viral RNA Testing, and Outcomes Altered mental status in 71 deaths due to COVID-19 Cutaneous manifestations in COVID-19: a first perspective Post-acute COVID-19 syndrome 11 symptomatic participants were lost to follow up prior to 12-week interview  Nearly one-fifth of patients with COVID-19 had no pre-existing medical conditions  Patients self-reported nearly twice as many symptoms as recorded in medical records  More symptoms at illness onset was associated with symptom duration and chronicity The authors have no conflicts of interest to report. Grant, $7,500, July, 2020. This sponsor had no role in study design, collection, analysis or interpretation of data, the writing of the report or the decision to submit the report for publication. Presented in part at the North American Primary Care Research Group (NAPCRG) Annual Conference, November, 2021. None.