Use of patient health survey data for risk adjustment to limit distortionary coding incentives in Medicare

Abstract: A core problem with the current risk-adjustment system in Medicare Advantage and accountable care organization (ACO) programs—the Hierarchical Condition Categories (HCC) model—is that the inputs (coded diagnoses) can be influenced for gain by risk-bearing plans or providers. Using existing survey data on health status (which provide less manipulable inputs), we found that the use of a hybrid risk score drawing from survey data and a scaled-back set of HCCs would, in addition to mitigating coding incentives, modestly lessen risk-selection incentives, strengthen payment incentives to deliver efficient care, allocate payment across ACOs more efficiently according to markers of population health that are not as affected by practice patterns or coding efforts, and redistribute payment in a manner that supports equity goals. Although sampling error and survey nonresponse present challenges, analyses suggest that these should not be prohibitive. Overall, our proof-of-concept analysis suggests that using survey data to improve risk-adjustment performance is a promising strategy that merits further development.

McWilliams, J. M., Weinreb, G. G., Landrum, M. B., Chernew, M. E. (2025). Use of patient health survey data for risk adjustment to limit distortionary coding incentives in Medicare. Health Affairs, 44(1), 48-57.
Download Paper | Podcast interview