Association of Observation Stays with Clinical Outcomes and Costs in Medicare: An Instrumental Variable Analysis

Background: Observation stays in Medicare have grown over the last 15 years, yet limited research exists on how observation may impact outcomes for older adults. Objective: To examine the relationship of an observation stay with 30-day hospital returns, total acute care days post-discharge, mortality, and out-of-pocket costs, compared to an inpatient admission. Design: Retrospective cohort study using instrumental variable analysis. Participants: A 20% sample of US Medicare beneficiaries admitted to acute care with a length of stay < 5 days between 2009 and 2019. Interventions: Observation stay vs inpatient admission. Main Measures: Unplanned hospital return within 30 days, total 30-day post-discharge acute care days, 30-day mortality, and 30-day acute care out-of-pocket spending. Key Results: A total of 3,958,377 hospitalizations met study criteria, of which 1,656,631 (42%) were observation stays and 2,301,746 (58%) were inpatient admissions. Compared to inpatient admissions, observation stays were associated with a 4.39 percentage point (95% confidence interval [CI] 3.56%, 5.22%) higher rate of 30-day unplanned hospital returns, but no difference in total 30-day post-discharge acute care days (difference − 0.02 days; 95% CI − 0.08, 0.03), no difference in 30-day mortality (difference 0.20 percentage points; 95% CI 0.00, 0.40), and lower 30-day out-of-pocket costs (difference − $552; 95% CI − $561, − $542). Conclusions: Among Medicare beneficiaries hospitalized for fewer than 5 days, observation stay was associated with higher rates of 30-day unplanned hospital returns compared to inpatient admission. However, we simultaneously observed lower out-of-pocket costs among those hospitalized under an observation stay. The mixed results suggest that additional research and engagement with relevant parties is needed to optimize observation stay policy.