Drivers of Reduction of Inpatient Mortality in Hospitalized COVID-19 Patients

Abstract

Inpatient mortality is an important indicator of the COVID-19 pandemic, both to understand the pandemic’s trajectory and to improve clinical outcomes. A reduction in inpatient mortality from COVID-19 has been described in the United States and the United Kingdom. We analyzed 21,037 patients admitted from March 1, 2020, to November 30, 2020, as part of the American Heart Association’s COVID-19 registry, and performed logistic regression on person-level hospital records. The outcome variable was in-hospital death, and covariates included demographic, clinical, and constructed hospital-level factors. Constructed variables included hospital load and time between diagnosis and hospitalization. We tested the hypothesis that hospital-level burden was associated with inpatient mortality. Of 21,037 patient admitted between March and November, 3,246 (15.4%) died in-hospital. Mortality decreased from 18.7% in March-April to 10.7% in September-November. Adjusted odds ratios for each time period were 0.68 (95% CI 0.59-0.78, P < .001) for May and June, 0.59 (0.49-0.71, P < .001) for July and August, and 0.60 (0.48-0.75, P < .001) for September through November. Hospital load was not significantly associated with inpatient mortality, and so we fail to reject the null hypothesis of this thesis. Further explanations of time effect may include mortality displacement, non-invasive oxygenation techniques, or other unmeasured factors.