(155) Discharged Where? The Importance of Understanding Discharge Dispositions When Estimating In-Hospital Mortality

Publication Number: 486

Background: Choices for classifying hospital discharge dispositions as censoring events, competing risks, or outcomes of interest may have large impacts on study results. However, correct handling of certain dispositions, such as hospital transfer or transfer to hospice, is not obvious, particularly for outcomes such as mortality.

Objectives: To assess the impact of misclassifying certain discharge dispositions as censoring or competing events on the estimation of in-hospital mortality incidence in patients hospitalized with COVID-19.

Methods: Using hospital chargemaster data from 297 hospitals in the United States, we identified adults admitted with a COVID-19 diagnosis between May 01, 2020 and August 15, 2021. Hospital discharge disposition was classified as: home, hospice, hospital transfer, long-term care facility, or death. Our outcome of interest was in-hospital mortality due to COVID-19 infection. We considered patients transferring hospitals to still be at risk of severe disease, and patients discharged to long-term care facilities to no longer be at severe risk. We estimated the cumulative incidence of in-hospital mortality at 30 days under the following scenarios: (1) censor all discharges with dispositions other than death; (2) treat all discharges with dispositions other than death as competing events; (3) treat discharge home as competing event, censor at hospital transfer, long-term care facility, and hospice; (4) treat discharge home or to a long-term care facility as competing events, censor at hospital transfer and hospice. We then repeated (3)-(4) defining the outcome as death or transfer to hospice.

Results: We identified 223081 adults admitted to the hospital with a COVID-19 diagnosis. Of these patients, 65% were discharged home, 14% to a long-term care facility, and 13% died. 2% and 3% were transferred to another hospital and hospice, respectively. The 30-day cumulative risk of in-hospital mortality in each scenario is as follows: scenario (1), 45%; scenario (2), 12%; scenario (3), 16%; scenario (4) 13%. Repeating scenarios (3) and (4) including hospice as part of the mortality outcome, the estimates increased to 18% and 15%, respectively.

Conclusions: Treating all discharges as censoring events overestimated the risk of in-hospital mortality. Treating all discharges as competing events underestimated the risk. Omitting hospice from the mortality outcome led to lower risk estimates. Censoring at transfer to a long-term care facility resulted in a higher estimates. Scenario (4), with hospice, best classified discharges to accurately estimate in-hospital mortality for COVID-19 patients. Proper handling of discharge disposition is essential for valid cumulative incidence estimation for hospitalized patients.