Practice-style spillovers in surgeon care: evidence from lower extremity joint replacement surgery

• In Event: Upfront Costs, Downstream Savings, and Financial Burden in Health Care

Abstract

Physicians often practice across multiple sites and collaborate within informal care teams, creating pathways for knowledge exchange. This study examines whether Medicare’s mandatory, randomized Comprehensive Care for Joint Replacement (CJR) Model generated spillovers through physician mobility and peer networks in New York State. While evaluations of payment reforms typically focus on effects within treated markets, physician mobility and collaborative practice patterns may transmit incentives and behavioral changes beyond the boundaries of directly exposed providers. Using 2013–20117 HCUP State Inpatient Databases (SID) for New York (SID–NY) and hospital–physician affiliation files, I construct physician network groups that capture cross-market practice and collaboration. First, I define a “primary MD” group—physicians who practiced in CJR-treated metropolitan statistical areas (MSAs) and also affiliated with hospitals located in control MSAs. These physicians serve as potential conduits of policy exposure across geographic markets. Second, I identify a “secondary MD” group consisting of physicians who practiced exclusively in control MSAs but who co-treated patients with the primary MDs, representing direct peer-network exposure. To capture broader diffusion through collaboration networks, I further construct third- through fifth-degree physician groups based on successive co-treatment links. To create a comparable comparison sample, I remove the primary MDs, secondary MDs, and their extended network connections from the baseline comparison group of physicians practicing in control MSAs. This approach helps isolate spillover effects from the direct treatment effect of the CJR model. Additionally, I investigate the diagnoses most frequently treated by primary MDs and identify clinically related procedures, constructing a “related-diagnosis” physician group to examine whether behavioral changes diffuse across clinically similar services beyond the focal joint replacement procedures. The primary outcomes are discharge dispositions following joint replacement surgery, including discharge to home, to a skilled-nursing facility (SNF), and to a home-health agency (HHA). I estimate difference-in-differences models with physician and hospital fixed effects to compare changes in discharge patterns across treatment exposure groups before and after CJR implementation. Relative to comparable control-region physicians, primary MDs practicing in control MSAs adopt behavior consistent with that observed in treated markets: they reduce discharges to skilled-nursing facilities and direct home discharges while increasing the use of home-health agencies. Importantly, these changes occur for both Medicare beneficiaries and privately insured patients, suggesting broader practice pattern adjustments rather than payer-specific responses. Secondary MDs and physicians in the related-diagnosis group exhibit similar changes in discharge patterns, indicating that behavioral responses diffuse through collaboration links and clinical practice environments. Evidence of effects extending to higher-degree physician networks further suggests that information transmission and peer influence propagate the reform’s incentives beyond directly exposed providers and geographic markets. Finally, re-estimating the CJR treatment effect using a “clean” New York sample that excludes physicians exposed through cross-market mobility and collaboration yields larger treatment estimates than those obtained using the full sample. These findings provide causal evidence that mandatory payment reforms propagate beyond treated markets through physician networks.

Author

• Ke Zeng, NORC at the University of Chicago
  Presenting Author