Risk adjustment

Risk adjustment is a set of statistical methods that normalize comparisons of cost or outcomes across people or organizations by accounting for differences in underlying risk — age, diagnoses, prior utilization, or other factors known to predict cost or health need. Without it, a hospital or health plan serving a sicker population would look worse than one serving a healthier one, even with identical quality of care.

In healthcare, risk-adjustment models such as CMS-HCC (Hierarchical Condition Categories) assign each patient a risk score from their diagnoses and demographics, which is then used to set expected spending, adjust quality-measure comparisons like the hospital readmission rate, or determine capitated payments to health plans. In insurance more broadly, risk adjustment transfers funds between insurers based on the health risk of the members they enrolled, so insurers aren't penalized for attracting sicker customers. This differs from underwriting analytics, which prices an individual policy, and from actuarial modeling, which projects aggregate losses.

Risk adjustment matters because it underpins fair comparison in population health management and value-based payment, and because getting it wrong — through incomplete diagnosis coding or model gaming — directly moves money. A well-documented pitfall is "upcoding," where diagnoses are coded more severely than clinically warranted to inflate risk scores and payments.