Outcomes-adjusted reimbursement in a health-care delivery system

This paper considers a health-care delivery system with two noncooperative parties: a purchaser of medical services and a specialized provider. A dynamic principal–agent model that captures the interaction between the two parties is developed. In this model, patients arrive exogenously, receive periodic treatment from the provider, suffer costly complications that require hospital care, and eventually exit the system in death. The provider chooses the intensity of treatment in each period, incurs an associated cost, and is reimbursed by the purchaser according to observed patient outcomes. The purchaser's problem is to determine a payment system that will induce treatment choices maximizing total social welfare. The optimal payment system, referred to as the outcomes-adjusted payment system, is identified. It consists of a prospective payment per patient and a retrospective payment adjustment based on adverse short-term patient outcomes. This system induces the most efficient delivery of medical services by combining the immediate ‘threat’ of a retrospective payment adjustment with the future reward of prospective payments generated by surviving patients. A numerical example is provided in the context of Medicare's End-Stage Renal Disease program. The example compares the optimal system to systems that are currently in place. The results suggest that the purchaser can achieve significant gains in patient life expectancy by switching to the outcomes-adjusted payment system, but this requires accurate information about treatment technology, patient characteristics, and provider preferences. The life-expectancy gains do not involve increased medical expenditures.