When patients present to emergency departments too sick to be sent home, hospitals increasingly provide observation services, meaning extended outpatient visits expected to last under 48 hours during which patients are treated and reassessed. However, clinical issues may fail to resolve within 48 hours, leading to hospital admission, prolonged observation, or death.


Our objective was to identify patient and hospital characteristics associated with failed observation visits, particularly admission. Using Healthcare Cost and Utilization Project (HCUP) data from eleven states and multinomial logit models, we compared patient and hospital characteristics between successful and failed observation visits. Study subjects included aadults placed on observation status after evaluation in an emergency department. Outcome measures included observation visits resulting in admission, prolonged observation (>48 hours), or death.


Of 855,432 visits, 23.3% resulted in admission, 7.9% prolonged observation, and 0.3% death. In multivariate analyses, factors significantly associated with admission included Medicare insurance, age, primary diagnosis, and receiving care at hospitals that were private, non‐Critical Access, located outside metropolitan areas, or high or low providers of observation services. Significant factors were generally similar for prolonged observation and death. Overall, primary diagnosis had the largest effect on failure risk. Gender, race/ethnicity, income, hospital size, case mix, and local market competition were not consistently associated with visit outcomes.


Many observation visits fail. Risk factors for failure include Medicare insurance, patient age and diagnosis, as well as hospital ownership, location, and experience providing observation services. Policies regarding criteria for observation should be re‐evaluated, particularly for certain diagnoses, and financial incentives for hospitals should be monitored.