Impacts of Rural Hospital Payment Reform: Evidence from the Pennsylvania Rural Health Model

Committee:

Lindsay Sabik, Department of Health Policy and Management, University of Pittsburgh School of Public Health (Committee Chair)

Eric Roberts, University of Pennsylvania, Perelman School of Medicine and Leonard Davis Institute of Health Economics

Coleman Drake, Department of Health Policy and Management, University of Pittsburgh School of Public Health

Jeremy Khan, Department of Critical Care Medicine, University of Pittsburgh School of Medicine

Bruce Jacobs, Research Chief, Division of Health Services, Department of Urology, University of Pittsburgh School of Medicine

Abstract

Rural hospitals have long faced financial challenges, risking service reductions or closures. The Pennsylvania Rural Health Model (PARHM) is a unique payment model tailored for rural hospitals. Developed through collaboration between the Center for Medicare & Medicaid Innovation and Pennsylvania, it operates from 2019 to 2024. PARHM aims to assess if global budgets for hospitals—a fixed amount covering all services—combined with hospital-specific transformation plans, can sustain rural patient care access, enhance care quality, and boost financial stability for rural Pennsylvania hospitals. Sixty-five hospitals were eligible, with cohorts joining in 2019 (5 hospitals), 2020 (8 hospitals), and 2021 (5 hospitals), forming a natural policy experiment.

This study evaluates PARHM's impact on potentially avoidable utilization of inpatient hospital services, access to and utilization of low-margin service lines, and rural hospital bypass for elective surgeries. Employing a difference-in-differences approach with multiple time periods, we compare relative changes between PARHM-participating and non-participating eligible hospitals. Data sources include the Pennsylvania Health Care Cost Containment Council Database for hospitalizations, the American Community Survey for demographic data, and the American Hospital Association Annual Survey for hospital characteristics.

Overall, relative differences between participating and non-participating hospitals were not significantly different across any outcome when pooled across cohorts. However, heterogeneous treatment effects were observed among individual PARHM cohorts. For potentially avoidable utilization rates we found a statistically significant differential decrease in among Cohort 2 hospitals. For low-margin service lines we found a statistically significant differential increase in substance use disorder care among Cohort 1 hospitals. Our results will help determine the success of PARHM and inform the design and implementation of future alternative payment models in rural settings.