To the Editor: Incipient pay-for-performance plans are beginning to improve clinical outcomes in the general medical sector (
1), but they have not yet begun to influence quality improvement efforts in the behavioral health care arena. The research reported by Gates and colleagues (
2) in the November 2005 issue of Psychiatric Services is valuable in conceptualizing pay-for-performance initiatives in behavioral health care.
The authors note that their performance-based contracting model reimbursed providers more heavily the longer the patient remained employed. This contingency provided incentives that encouraged providers to increase the amount of time spent with patients each week in order to receive initial compensation with minimal delay as well as to place patients in good jobs to ensure retention—and additional reimbursement. This strategy exemplifies some of the behavioral engineering principles articulated by Skinner (
3) nearly 40 years ago. Skinner taught that effective reinforcement contingencies maximize the occurrence of desired behaviors (that is, outcomes) by providing incentives after a variable number (rather than a fixed number) of responses and minimizing delay between response and reinforcement.
The desired behavioral outcomes in the study by Gates and colleagues included mental health providers' assisting patients to initially secure and then to retain employment and patients' working successfully. The model used by these authors provided variable (that is, sequential) reinforcement (monetary compensation) contingent on providers' and patients' attainment of incremental steps toward the goal of sustained employment. Each step was reimbursed at a different rate, with the later objectives (sustaining employment for three, six, and nine months) weighted more heavily than the earlier ones (life skills assessment, initial placement, and job skills acquisition) to provide increasingly potent incentives to help patients keep their jobs.
Gates and colleagues' outcomes-based compensation model might tempt providers to encourage patients to accept inadequate or temporary jobs in order to avoid delays in receiving reimbursement. But their preliminary data suggest that the model promotes successful outcomes by rewarding providers for working intensively with patients early in the placement process, assisting them in finding and keeping appropriate jobs, and coordinating more effectively with social service agencies.
The implications for designing behavioral pay-for-performance plans are clear. Providers' compensation should be weighted to differentially reinforce progressive improvements in patients' recovery and sustained functioning. Instead of compensation that is contingent on simply keeping patients in treatment, which subtly reinforces clinical deterioration, incrementally weighted reinforcement contingencies should explicitly link providers' and patients' use of evidence-based treatment guidelines and self-management plans with quantifiable measures of continuing clinical improvement and sustained recovery of functioning.