Brad Wright, Xuan Zhang, Momotazur Rahman, Keith Kocher

Outpatient observation stays are increasingly substituting for standard inpatient hospitalizations. In 2013, the Centers for Medicare & Medicaid Services adopted the controversial Two-Midnight Rule policy to curb long observation stays and better define the use of hospital-based observation services versus inpatient hospitalizations. We seek to determine the extent to which Medicare beneficiaries exposed to long observation stays (>48 hours) are clinically similar to those with short observation stays (≤48 hours) because this has relevance to the Two-Midnight Rule. Using 100% Medicare claims data from 2008 to 2010, we identified all patients with long observation stays (>48 hours) who were admitted through the emergency department (ED). We report beneficiary characteristics, as well as crude and risk-adjusted 30-day rates of mortality, readmissions, and return ED visits stratified by observation stay length. Seven percent of 2.8 million observation stays were greater than 48 hours. Beneficiaries with long observation stays tended to be older, women, nonwhite, and urban residents, with a greater number of comorbid conditions. Crude rates increased with observation stay length for all 3 outcomes. However, after directly standardizing the rates, we observed the reverse trend because all adjusted rates decreased stepwise with observation stay length greater than 48 hours in a dose-response pattern. Patients with observation stays lasting longer than 48 hours are a clinically distinct population. Our findings support the conceptual underpinnings of the Two-Midnight Rule, but suggest that observation versus inpatient determinations should be based on actual length of stay rather than prospective prediction to reduce the administrative ambiguity this policy has created.