Widespread expansion of managed care has caused dramatic evolution in the structure of the American health care delivery system, most notably in relationships between hospitals and physician organizations. Reducing health care cost growth without sacrificing quality requires increased coordination of care and better alignment of incentives between physicians and hospitals. Hospitals that fail to develop requisite health network infrastructure and managed care contracting skills will be disadvantaged in cost and quality competition and risk their very survival. Whether rapidly evolving forms of physician- hospital organizational arrangements (POAs) are capable of improving hospital performance is unclear. Federal and state regulators are concerned about implications of MSOs, PHOs, and other POAs for quality and cost, payment, insurance licensing, provider solvency, antitrust, and Medicare/Medicaid risk contracting. Proposed research consists of a conceptual and empirical analysis, through which I will investigate five research questions: (1) What are the salient aspects of incentive and control (selective contracting, payment mechanisms, utilization management) and physician/hospital ownership (vertical integration vs. contractual arrangements) of POAs? The research will develop a conceptual framework of POA strategy based on transaction cost economics and agency theory. Empirical analysis will investigate hospital and market structure drivers of (2) differences between POA strategy adopters and nonadopters, and (3) change in adoption patterns over time. Hospital performance in managed care contracting (HMO and PPO admission market shares) will be compared for (4) various POA strategies, and (5) POA strategy vs. market structure. Econometric models will control for hospital characteristics, market structure (hospital, physician, health plan), regulation, and self-selection bias. Research will use a unique federal/state dataset of 2,300 acute care hospitals nationwide in 1993 and 1996 and multistate subsample; panel data estimation of simultaneous equations will address serious methodological limitations of prior research.
