An Electronic Health Record-Based Screening Tool to Support Safe Discharges of COVID-19 Patients in the Emergency Department

Emergency departments (EDs) experience a large volume of patients that present with a wide range of COVID-19 symptoms. Patients with COVID-19 often present to EDs early in their disease course, when symptoms are highly variable and the risk of future decompensation is highest. ED clinicians need to make fast decisions around patient dispositions, often with limited data. The large variation in outcomes--from uneventful recovery to hospitalization, ICU admission, and death--has meant that some patients discharged from the ED will later return requiring more intensive care. Thus, there is an urgent need to quickly assess patient risks and allow for a more informed decision-making process in the ED.

This research will develop and validate the COVID-19 ED return screening tool (CERST) that will predict a COVID-19 patient’s risk of a return to the ED, and for those identified at risk, predict the risk of hospitalization and in-hospital mortality. The tool will be developed using natural language processing, machine learning (ML) algorithms, and predictive modeling. One of three risk stratifications will be shown: high risk, those at risk of ED return with associated in-hospital mortality and/or hospitalization; moderate risk, those at risk of ED return who would benefit from observation or additional targeted resources to support outpatient management; and low risk, those unlikely to return to the ED and require only education prior to ED discharge.

The specific aims of the research are:

• Iteratively develop a concept map using mixed methods, which will serve as the ontology categorizing predictive factors for COVID-19 ED returns and inform ML model development.
• Develop and evaluate ML algorithms predictive of ED return risk for COVID-19 patients.
• Prospectively validate a CERST using real-time data.

This research will use databases from two sites--MedStar Health and the University of North Carolina Health--to ensure a robust and diverse dataset. Fast Health Interoperability Resources (FHIR) standards will be used to assist with model interoperability between the two sites, given that they use two different electronic health records platforms. The primary outcome for the study will be returns to the ED, with two separate models predicting risk of return at 72 hours and at 9 days. The secondary outcomes will be ED returns that result in hospitalization or in-hospital mortality. CERST has the potential to improve morbidity and mortality from COVID-19 infection by prompting earlier hospitalization of those most at risk from the disease.