Impact of HMO market structure on physician-hospital strategic alliances

Objective. To assess the impact of HMO market structure on the formation of physician-hospital strategic alliances from 1993 through 1995. The two tre nds, managed care and physician-hospital integration have been prominent in reshaping insurance and provider markets over the past decade. Study Design. Pooled cross-sectional data from the InterStudy HMO Census an d the Annual Survey conducted by the American Hospital Association (AHA) be tween 1993 and the end of 1995 to examine the effects of HMO penetration an d HMO numbers in a market on the formation of hospital-sponsored alliances with physicians. Because prior research has found nonlinear effects of HMOs on a variety of dependent variables, we operationalized HMO market structu re two ways: using a Taylor series expansion and cross-classifying quartile distributions of HMO penetration and numbers into 16 dummy indicators. All iance formation was operationalized using the presence of any alliance mode l (IPA, PHO, MSG, and foundation) and the sum of the four models present in the hospital. Because managed care and physician-hospital integration are endogenous (e.g., some hospitals also sponsor HMOs), we used an instrumenta l variables approach to model the determinants of HMO penetration and HMO n umbers. These instruments were then used with other predictors of alliance formation: physician supply characteristics, the extent of hospital competi tion, hospital-level descriptors, population size and demographic character istics, and indicators for each year. All equations were estimated at the M SA level using mixed linear models and first-difference models. Principal Findings. Contrary to conventional wisdom, alliance formation is shaped by the number of HMOs in the market rather than by HMO penetration. This confirms a growing perception that hospital-sponsored alliances with p hysicians are contracting vehicles for managed care: the greater the number of HMOs to contract with, the greater the development of alliances. The mo dels also show that alliance formation is low in markets where a small numb er of HMOs have deeply penetrated the market. First-difference models furth er show that alliance formation is linked to HMO consolidation (drop in the number of HMOs in a market) and hospital downsizing. Alliance formation is not linked to changes in hospital costs, profitability, or market competit ion with other hospitals. Conclusions. Hospitals appear to form alliances with physicians for several reasons. Alliances serve to contract with the growing number of HMOs, to p ose a countervailing bargaining force of providers in the face of HMO conso lidation, and to accompany hospital downsizing and restructuring efforts.