Outcomes Data Published from Large, Diverse COVID-19 Patient Study in NYC
Sema4 scientists released results from an analysis of patients tested for SARS-CoV-2 at five hospitals in the Mount Sinai Health System. Spanning more than 28,000 patients, this is one of the largest and most racially diverse COVID-19 studies performed in the United States to date. The study findings were posted as a preprint on medRxiv.
The team at Sema4 analyzed de-identified electronic medical records from 28,336 patients tested for SARS-CoV-2 at Mount Sinai Health System hospitals in Manhattan, Brooklyn, and Queens between February 24 and April 15. Of those patients, 6,158 tested positive for the coronavirus and 3,273 were admitted to the hospital.
Consistent with other reports, this study confirmed that COVID-19 prevalence in African Americans and Hispanics is disproportionately high in New York City. However, for patients admitted to the hospital, the analysis found no differences in mortality rates based on ethnicity, indicating that inpatient care helps to address this health care disparity.
The team also identified several risk factors linked to increased mortality rates for COVID-19 patients, including age, oxygen levels, body mass index, and elevated creatinine, among others. Asthma was associated with longer hospital stays, but did not appear to be linked to increased mortality rates. The study, which found higher mortality rates at hospitals where patients had more severe cases upon entry, can also be used to guide neighborhood-based testing for SARS-CoV-2.
Li Li, MD, Sema4’s Vice President of Clinical Information and an Assistant Professor at the Icahn School of Medicine at Mount Sinai in the Department of Genetics and Genomic Sciences, who was one of the lead scientists on the project, commented: “In a thorough review of published studies investigating COVID-19 mortality rates, we found unintentional biases, such as small sample size, that limited the broader utility of the data. By including all patients tested at these five member hospitals and performing advanced statistical modeling including multivariate analyses, we have removed that bias and generated findings that should be more useful in improving the understanding of which clinical features track with disease progression and associated outcomes.”