LAUSR

Study Objectives: COVID-19 symptom severity varies between patients, and some remain asymptomatic. During early April 2020, 70% of patients admitted to the emergency department (ED) of a major hospital in New England had COVID-19, many of whom required treatment in the intensive care unit. As the volume of COVID-19 cases presenting to the ED increased, it became essential to develop accurate triage protocols to separate COVID-positive from COVID-negative patients. This study assessed which of three different clinical screening tools – a nursing triage screen (NTS), an ED clinician Review of Systems (ROS), and a standardized ED attending physician COVID-19 probability assessment – was best at identifying patients who had COVID-19 (based on subsequent PCR confirmation). Methods: All 748 patients admitted from the ED between April 27, 2020, and May 17, 2020 were included. Sensitivity, specificity, and positive and negative predictive values were calculated for each screening tool. Logistic regression was used to assess each tool’s performance. A principal components analysis (PCA) was performed;the resulting factors were used to model COVID-19 positivity. Results: The emergency physician’s probability assessment yielded higher sensitivity (0.62, 95% confidence interval [CI] 0.53-0.71, Table 1) than the NTS (0.46, 95% CI 0.37-0.56), and had higher specificity (0.76, 95% CI 0.72-0.80) than the NTS (0.71, 95% CI 0.66-0.75) and the emergency clinician ROS (0.62, 95% CI 0.58-0.67). Categorization as moderate or high probability on the emergency physician’s probability assessment was also associated with the highest odds of having COVID-19 in regression analyses (adjusted odds ratio=4.61, 95% CI 3.01-7.06). Moderate agreement (kappa 0.41-0.60) was observed between the NTS and ED clinician ROS for fever, cough, shortness of breath, and diarrhea;fair agreement (kappa 0.21-0.40) for sore throat, headache, abdominal pain, and vomiting;and poor agreement (kappa 0.00-0.20) for myalgias and chills. The 323 patients who had a response recorded for every symptom were included in the PCA. Only Factor 1 (fever, chills, fatigue, sore throat, rhinorrhea, and cough) was associated with increased odds of testing positive for COVID-19. Conclusion: While the emergency physician’s probability assessment had higher sensitivity and specificity than the other two tools, none of the tools evaluated in this study was sufficiently accurate enough to replace a COVID-19 PCR test on a patient entering a clinical setting where transmission control is crucial. These findings suggest that hospitals not rely on symptom or probability assessment in determining infection status but continue to utilize widespread testing. We recommend that providers in other countries experiencing COVID-19 surges consider the relevance of these findings and that as the pandemic develops (with the potential for continued new variant strains), diagnostic testing efforts should supersede the use of clinical screening tools. [Formula presented]