Duke 2008 | ASHE | Equity and Efficiency in Health and Healthcare

In hospital service areas such as cardiology, cardiac surgery, and orthopedics, physicians play critical roles in the hospital's purchases of high-cost drugs and devices known as "physician preference items." To address the misaligned physician and hospital incentives in these purchasing decisions, a few US hospitals have implemented gainsharing. Gainsharing uses group-based incentives in which compensation is determined based on savings in predetermined areas relative to the spending in the previous year. Although these programs might violate self-referral and anti-kickback regulations, HHS' Office of Inspector General has permitted them on a case-by-case basis, subject to meeting specific requirements. Although our previous work indicates that the currently approved version of gainsharing reduces costs by about 7%, additional work is needed to understand the mechanisms that generated these savings, the heterogeneity in physicians' responses to gainsharing's incentives, and how alterations to the current version would affect physician decisions.

In this paper, we theoretically and empirically consider the effects of gainsharing on physicians' choices of devices and drugs. The paper develops a multi-stage model of the currently implemented version of gainsharing. In the model, the hospitals' physicians establish agreed-upon appropriate treatment guidelines. Physicians choose drugs and devices by trading off their utility from reimbursement, the gainsharing payout, their patient's utility, and disutility from punishment due to deviating from the guidelines, which occurs due to peer monitoring. With these physician responses, the model indicates how the resulting standardization on drugs and devices influences the hospital-manufacturer bargaining game and the subsequent prices paid. The model establishes that costs reductions can be generated through three primary mechanisms: quantity, via reductions in utilization; substitution of lower-priced items, including delayed adoption of new, high-cost alternatives; and prices via physician standardization, either by increasing hospitals' bargaining power with manufacturers, or by reaching lower-priced tiers due to contract compliance. Comparative statics yield predictions about how the effect of gainsharing varies with physician and group characteristics, as well as predictions about how changes to the current version of gainsharing would alter physician choices.

We apply the model to data from 13 of the 15 completed and ongoing gainsharing programs in cardiology provided by Goodroe Healthcare Solutions. These data include information on all drugs and devices provided to patients including the manufacturer and the price paid by the hospital net of rebates, patient demographics and risk factors, the treating physician, and the physician's group affiliation. With these data, we estimate the role of the substitution, price and quantity effects in generating savings under gainsharing. We also test the model's predictions regarding heterogeneity in physicians' responses to these incentives, including practice size, the influence of the group in the hospital's overall purchasing decisions, and the pre-gainsharing characteristics such as within-group heterogeneity in device utilization. The results provide insights about the mechanisms by which gainsharing generates savings, and offers guidance to policymakers and hospitals in designing and implementing gainsharing programs.