Efficiency Implications of Physician Integration: Behavioral vs Administrative

In recent years, the physician practice landscape has been characterized by a shift away from small, single specialty physician practices and towards larger, more integrated providers. Responses to this trend have been mixed, with some hailing it as a cost saving cure-all and others warning about the dangers of increased market power and the potential for anticompetitive behavior. This trade-off has been debated by health care professionals, economists and government agencies in boardrooms, academia and courts. The discussion of integration has been impeded by a failure to carefully define terms and distinguish between two distinct components of integration: administrative and behavioral. Administrative, or financial, integration happens when providers merge, or hospitals purchase physician practices. This type of integration is associated with increased bargaining power and higher reimbursements. Furthermore, through profit sharing, financial integration can create an incentive for providers to refer patients to other specialists for more tests or more care, some of which may be unnecessary. In contrasts, behavioral integration refers to doctors working together and coordinating care. It has been associated with decreased waste and more efficient care. Previous work has often used measures of administrative integration, such as the share of physician practices owned by hospitals, to proxy for behavioral integration. Those modeling decisions are understandable as, up to this point, a metric which separately captures behavioral integration in a systematic way has not existed. The lack of a metric has been a hurdle to evaluating these two components separately. In this paper, I use Medicare data on physician patient sharing patterns to develop metrics that capture physician practice integration at the behavioral level. I compare these behavioral integration metrics to a more standard organizational level integration metric. The low correlation, only 0.30, demonstrates that these metrics are distinct. Using all these metrics, I examine the impact of these two types of physician integration on the utilization of medical care. With national data over time, I use changes in integration and utilization within regions to estimate how the different types of integration impact the ability to provide quality care at a low cost, which I refer to as efficiency. As a model of physician behavior predicts, I find that behavioral integration reduces cost while improving quality. In contrasts, financial integration appears to increase cost without having an impact on quality. These results are robust to different measures of behavioral integration and different identification strategies.