Tying Incentives to Information: Diagnosis-Contingent Contracts and Biased Beliefs in Healthcare

Unnecessary treatment reflects a gap between what patients believe about their health and their true condition; people often seek and consume care for diseases they suspect, but do not know that actually have. The value a patient places on testing therefore depends on their priors about being sick, and when those beliefs are biased, the resulting testing rate might be sub-optimal and leave welfare on the table. We study diagnosis-contingent incentive contracts, which tie treatment incentives to diagnostic results. For malaria, the contract subsidizes rapid diagnostic tests (RDTs) and antimalarials (ACTs) only for patients who test positive. In a cluster-randomized experiment across 140 Kenyan pharmacies, both patient subsidies and provider incentives roughly tripled testing and sharply reduced unnecessary ACT purchases among test-negative patients, working through distinct channels: patient subsidies through lower prices, provider incentives through information and advice. A model of patient choice implies these contracts raise social welfare well above their cost. Estimating the model structurally, we find that because patients substantially overestimate their infection risk, loading incentives onto diagnosis-contingent treatment discounts delivers four times the welfare gains of free testing programs.