Karandeep Singh, Assistant Professor, Learning Health Sciences and Medicine, University of Michigan
Validating a Widely Implemented Deterioration Index Model Among Hospitalized COVID-19 Patients
The coronavirus disease 2019 (COVID-19) pandemic is straining the capacity of hospitals and healthcare systems across the United States. This strain has led to an expansion of both advanced ICU capabilities within hospitals as well as the opening of lower-acuity hospital beds (e.g., field hospitals) with limited resources for more stable patients. For example, Michigan has one of the largest number of COVID-19 caseloads in the country and recently approved over 4,000 new hospital beds to provide care to COVID-19 patients. Accurately identifying subgroups of COVID-19 patients at high- and low-risk for adverse outcomes using automated risk-score systems could help to alleviate this strain. Recent work, however, has demonstrated a paucity of validated high-quality models in COVID-19 patients.
The Epic Deterioration Index (EDI) is a proprietary early-warning prediction model that assesses the probability of hospitalized patients requiring a rapid response, resuscitation, intensive care unit (ICU)-level care, or dying. Although there is no published research on the validity of the EDI, ubiquitous use of the Epic EHR across the U.S. has led to growing anecdotal reports of its value in critically ill patients. Implemented in over 100 hospitals across the U.S., advocates have further promoted its use among COVID-19 patients despite the absence of published data on its validity in this population. In this talk, we will present preliminary results from our evaluation of the Epic Deterioration Index at Michigan Medicine. Our findings have potential implications for how this widespread tool may be validated and used by other healthcare systems during the COVID-19 pandemic.