An Incentive Mechanism for Using Risk Adjuster to Reimburse Health Care Providers

Presenter: Charles Yan, Institute of Health Economics

Abstract

Background:

Diagnosis-based risk adjustment method has been broadly employed in health care market as the means of reimbursing care providers. Given that the necessary information for the method is the diagnostic codes that are routinely recorded in the care practices, this approach suffers from quite a few problems, like incomplete or inaccurate coding of diagnostic information. One concern is the potential of upcoding, which may occur when health care providers engage in strategic behaviors designed to increase risk-adjusted payment.

Objective:

The paper is to analyze how tendencies towards upcoding by health care providers alter the nature of optimal contracts between providers and purchasers, as upcoding behavior is essentially non-verifiable and hence non-contractible.

Assumptions/model setting:

Because of the uncertain nature of health care, the risk adjuster is attributed to a random variable with the distribution which the upcoding behavior could alter. This leads to an assumption that the factors of resulting in a specific distribution are unobservable, though its realization is measurable. Although the upcoding behavior has the potential to increase the realization of risk adjuster, which in turn brings the provider a pecuniary reward, the behavior could carry serious penalties including the threat of prison sentences once it is caught up. The trade-off between potential reward and punishment is quite crucial in our model. Purchasers use the carrot and stick approach to induce honest recording of diagnostic codes.

This model considers two payment methods. Firstly, the provider is reimbursed based only on the observed realization of the risk adjuster, where the risk of punishment is the instrument used to alleviate moral hazard. Secondly, the remuneration depends not only upon the risk adjuster but also upon the treatment intensity. This approach requires that the provider must deliver the treatment package that is relevant to diagnoses which provides another channel to reduce the motivation for recording diagnostic codes dishonestly.

Results:

In the sense of expected utility increase, under both payments, honest providers receive the same rewards and dishonest ones receive nothing. From the incentive point of view, the distortion in transfer payment and ex-post utility effect under the first payment contract intends to lead the resources being allocated to the patients with worse medical conditions, while the second contract is found to be accompanied by the reduction in the treatment intensity which may hurt the benefit of consuming health care services. In the sense of implementing contracts, however, the model shows that the second payment is superior.

Implications:

This model shows that the motivation of exerting upcoding can be alleviated with proper arrangement of an incentive mechanism. In particular, the provider is indifferent between true recording and manipulating the diagnostic codes if a payment scheme is properly designed.