Accuracy of the Simplified HOSPITAL Score in Predicting COVID-19 Readmissions—Exploring Outcomes from a Hospital-at-Home Program

SUMMARY Goal: As strategies emerge to off-load hospital systems and prevent readmissions amid the COVID-19 pandemic, pragmatic assessments of readmission risk become increasingly important. The simplified HOSPITAL score is an extensively validated tool that predicts 30-day potentially avoidable readmission (PAR). Scores of 0 to 4 predict a 30-day PAR risk of 6.4%, while scores ≥ 5 predict a 30-day PAR risk of 17.3%. Its role in patients with COVID-19 is unknown. Our goal was to assess the simplified HOSPITAL score’s accuracy in patients with COVID-19 and explore outcomes related to a hospital-at-home program. Methods: Patients discharged following an admission for clinically symptomatic COVID-19 from two hospitals belonging to the same healthcare system in the Midwest were included. Those who died, discharged to hospice or an acute care hospital, whose length of stay was < 1 day, or who discharged against medical advice were excluded. The simplified HOSPITAL score was tabulated for included patients to predict their 30-day PAR risk. The Brier score was calculated to compare the observed rates of 30-day readmission with rates predicted by the simplified HOSPITAL score. Prediction models with a Brier score <.25 are considered useful. Principal Findings: Among 612 patients, the overall 30-day PAR rate was 10.1%. Most patients (n = 522 [85.3%]) had simplified HOSPITAL scores of 0 to 4, and 41 (7.8%) of these patients were readmitted. Among the 90 patients (14.7%) with scores ≥5, 21 (23.3%) were readmitted. The Brier score was 0.088, indicating very good accuracy between the predicted readmission risk and observed readmissions. In patients with scores 0 to 4, readmissions were highest in those discharged to acute or subacute rehabilitation (10.4% [8/77]), intermediate in those discharged home (8.1% [32/394]), and lowest in those discharged to hospital at home (1.9% [1/51]). However, these differences did not reach statistical significance. Application to Practice: The simplified HOSPITAL score was accurate in patients with COVID-19 and can be used to direct resources toward those predicted to be at increased risk for readmission and to assess outcomes from readmission reduction strategies. Hospitals at home may be a promising strategy to decrease readmissions in patients with COVID-19.