OBJECTIVE Determine risk factors for mortality among COVID-19 patients admitted to a system of community hospitals in the United States .
DESIGN Retrospective analysis of patient data collected from the routine care of COVID-19 patients .
SETTING System of more than 180 acute care facilities in the United States .
PARTICIPANTS All admitted patients with positive identification of COVID-19 and a documented discharge as of May 12 , 2020 .
METHODS Determination of demographic characteristics, vital signs at admission, patient comorbidities and recorded discharge disposition in this population to construct a logistic regression estimating the odds of mortality, particular for those patients characterized as not being critically ill at admission .
RESULTS A total of 6180 COVID-19+ patients were identified as of May 12 , 2020 . The majority of COVID-19+ patients (77.8% , 4808) were admitted directly to a medical/surgical unit with no documented critical care or mechanical ventilation within 8 hours of admission . After adjusting for demographic characteristics, comorbidities, and vital signs at admission in this subgroup, the largest driver of the odds of mortality was patient age (OR : 1.07 , 95% CI 1.06-1.08, p <0.001). Decreased oxygen saturation at admission was associated with increased odds of mortality (OR : 1.09 , 95% CI 1.06-1.12, p <0.001) as was diabetes (OR : 1.57 , 95% CI 1.21-2.03, p <0.001).
CONCLUSIONS The identification of factors observable at admission that are associated with mortality in COVID-19 patients who are initially admitted to non-critical care units may help care providers, hospital epidemiologists, and hospital safety experts better plan for the care of these patients.