Meredith Rosenthal*

Dissertation Title:  "Risk Sharing in Managed Care"

Managed care continues to evolve in the pursuit of sustainable cost savings. In its early form, the industry relied on "command and control" mechanisms to influence physician practice style and reduce the cost of care. In response to the backlash from physicians and patients against the encroachment of managed care into the doctor-patient relationship, health plans are seeking alternative ways of containing costs. This dissertation explores one of these alternatives that is increasingly prevalent and controversial: risk sharing with providers. There is concern that putting physicians at risk for the cost of treating patients compromises ethical principles and may lead to reductions in the quality of care. To help inform the debate about risk sharing in managed care, the three papers that comprise my thesis explore issues related to the design of risk sharing contracts as well as their impact in an outpatient mental health setting.

The first paper presents an economic model of physician contracting in which physicians affect multiple dimensions of health care utilization. The objective of the paper is to investigate whether the variation in risk sharing contracts observed in managed care may be explained in part by the need for health plans to give incentives to physicians to practice efficiently while minimizing unnecessary risk exposure. My results support the idea that contractual form may be selected so that physicians bear risk only on the margin where they exert substantial control.

The second and third papers of my dissertation evaluate the impact of a natural experiment in which a managed behavioral health plan changed from a fee-for-service to a case-rate system for reimbursing behavioral health groups for outpatient mental health care. The second paper demonstrates that the case-rate reimbursement system reduced visits per episode by approximately 15 percent. In addition, there was evidence that visits that were likely to be of lower value were more likely to be eliminated. Finally, the response to the case rate varied according to how heavily invested in managed care the group was (share of revenue from risk contracts) and how intensively they monitored member clinicians.

In the third paper, I explore whether the reduction in utilization associated with the introduction of the case rate was associated with measurable differences in quality of care. Looking at the process of care, I found that case-rate patients were more likely to be referred to community and self-help programs, perhaps as substitutes for therapy. In addition, they were more likely to be medicated than fee-for-service patients. Finally, in terms of the best measure available in the data of health status improvement from treatment, change in global assessment of functioning (GAF), no difference could be detected.