Characterizing and Forecasting Emergency Department Visits Related to COVID-19 Using Chief Complaints and Discharge Diagnoses

In response to the unprecedented public health challenge posed by the SARS CoV-2 virus (COVID-19) in the United States, we and our colleagues at the Johns Hopkins University Applied Physics Laboratory (JHU/APL) have developed a model of COVID-19 progression using emergency department (ED) visit data from the National Capital Region (NCR). We obtained ED visits counts through targeted queries of the NCR Electronic Surveillance System for the Early Notification of Community-Based Epidemics (ESSENCE). To focus on ED visits by COVID-19 patients, we adjusted the query results for typical ED visit volumes and for reductions in ED volumes due to COVID-19 precautions. With these ED visit data, we fitted a logistic growth model to characterize and forecast the increase in cumulative COVID-19 ED visits. Our model achieves the best fit when we assume that the first NCR visit occurred in early January. We estimate that approximately 15,000 COVID-19 ED visits occurred prior to May 2020 and that approximately 17,000 more visits will occur in subsequent months. We plan to deploy an operational pilot of this model in the NCR ESSENCE environment, assisting local public health authorities as they brace for a second wave of COVID-19. Additionally, we will iteratively assess potential model refinements, aiming to maximize our models relevance for local public health authorities situational awareness and decision-making.