Association for Public Policy Analysis & Management

Background: Financial constraints have led Medicare to reduce payments made to hospitals in the long-run, and payment rates have consistently fallen since the late 1990s. Prior studies have shown mixed evidence regarding whether patient health outcomes are impacted when hospitals face financial constraints. This study examines hospital length of stay as an intermediate mechanism that may explain how payments are linked to patient health outcomes.

Data: The primary data source is the Healthcare Cost and Utilization Project Nationwide Inpatient Sample (1997, 2001, 2005), which consists of all-payer discharge-level data of hospital inpatient stays. Other data used include national-level service line-specific markups (the ratio of average Medicare reimbursement over cost) from the Medicare Hospital Cost Reports.

Methods: We conducted discharge-level regressions, modeling hospital length of stay among patients aged 45 years and older for all payers, adjusted for risk (Charlson comorbidity index, age, income, admission type, disposition), Medicare share, interactions between hospital and service line fixed effects, and interactions between time trends and service line fixed effects. We simulated the effect of a 20% reduction Medicare markup, which is the national average across all service lines and years.

Results: A 20% decline in Medicare markup is associated with a 0.30-day reduction in hospital length of stay (6% reduction from mean) among Medicare patients in all hospitals. The magnitude of this effect is larger among urban hospitals, with a reduction of 0.47 days. We observed no statistically significant spillovers into non-Medicare patients.

Conclusions: Hospitals may respond to lower payments by reducing hospital length of stay. Policies aimed at reducing the growth of health care spending should give consideration to the potential effects on hospital operations and patient health outcomes.