Payment for Diagnosis Under Limited Outcome Visibility

Abstract

Conventional and emerging physician payment schemes specify reimbursements largely based on the physician’s diagnosis. Yet, correctly identifying a diagnosis can be a challenging task even for skilled physicians, and diagnostic errors are not rare. Measuring the precise extent of diagnostic errors is hard because they are only occasionally brought to light. Medical diagnosis thus presents a setting with limited outcome visibility in that a physician’s diagnostic accuracy is partially observable. In this paper, we develop a parsimonious model to analyze the effect of a diagnosis-based payment scheme under limited outcome visibility. Our model captures how a physician’s decision to exert diagnostic effort interacts with the ultimate diagnostic decision and with the welfare of the payer, physician, and patients. We show this type of payment scheme may induce a higher diagnostic effort but bias the diagnosis. We find outcome visibility influences whether it is optimal to pay a flat amount or different amounts depending on the diagnosis. Our results demonstrate better outcome observability supports the use of flat payment schemes. Despite the inefficiency resulting from limited outcome visibility, surprisingly, better outcome visibility does not necessarily lead to better diagnostic accuracy or a lower expected payment to the physician, so it can be detrimental to the payer and to the patients.