What risk adjustment does
Risk adjustment sets each Medicare Advantage enrollee's payment based on a risk score derived from their diagnosed health conditions, using CMS's Hierarchical Condition Category (HCC) model. Enrollees with more or more severe diagnosed conditions generate a higher risk score, and plans are paid more to cover them. The model was introduced in 2004 to offset the financial incentive for plans to enroll healthier-than-average beneficiaries, a dynamic known as favorable selection. (See What MA Is, and How It Grew for that history.)
Because a beneficiary's payment depends on which diagnoses are on record, risk adjustment creates an incentive — for both traditional Medicare providers and MA plans — to document diagnoses thoroughly. The two settings, however, differ in how strong that incentive is.
Coding intensity
In traditional Medicare, a provider's payment for a given visit is generally unaffected by how many diagnoses they record. In Medicare Advantage, by contrast, a plan's payment is directly tied to the diagnoses on record for each enrollee. This asymmetry creates an incentive for MA plans to document diagnoses more thoroughly than would occur under fee-for-service care — a pattern researchers call coding intensity.
A 2020 study estimated that MA enrollees generate risk scores 6% to 16% higher than they would if evaluated under traditional Medicare's documentation patterns.1
A simplified illustration
Suppose diabetic beneficiaries carry a risk score of 1.5, non-diabetic beneficiaries carry a risk score of 0.5, and pre-diabetic beneficiaries — a condition that requires active screening to identify — carry a risk score of 1.0. If traditional Medicare does not routinely screen for pre-diabetes, every FFS beneficiary is classified as either diabetic or non-diabetic. As long as MA plans also do not screen for pre-diabetes, both populations are evaluated identically and paid identically.
If MA plans begin screening for pre-diabetes, however, some enrollees who would otherwise have been classified as non-diabetic (risk score 0.5) are reclassified as pre-diabetic (risk score 1.0). The average risk score among MA enrollees rises, and payments to their plans rise correspondingly — even though the underlying health of the population has not changed. The two populations are no longer being evaluated on the same basis, since only one of them was screened.
Health Risk Assessments and chart reviews
Two tools are central to how MA plans identify diagnoses beyond what a routine office visit would surface:
- Health Risk Assessment (HRA)
- An evaluation, often conducted in an enrollee's home, initiated by the plan (or a vendor working on its behalf) rather than by the enrollee's treating physician, used to identify diagnoses that may not have been documented elsewhere.
- Chart review
- A retrospective review of an enrollee's medical records, sometimes conducted by a plan or a coding vendor, that can surface diagnoses not associated with any specific clinical encounter.
A diagnosis identified through an HRA or chart review that is not otherwise linked to a clinical encounter has drawn particular scrutiny. HHS's Office of Inspector General estimated that such HRA-only diagnoses generated $2.6 billion in risk-adjusted payments in 2017 alone, with in-home HRAs responsible for roughly 80% of that total.2 MedPAC estimated in 2025 that unlinked HRA-derived diagnoses were responsible for approximately $10 billion per year in payments to MA plans.3
CMS's April 2026 announcement of CY2027 Medicare Advantage payment policy finalized a rule excluding diagnoses generated by HRAs unlinked to an actual clinical encounter — estimated to account for about 85% of all HRAs — from risk score calculations, beginning in 2027.4
A 2025 industry-commissioned analysis by Milliman, prepared for UnitedHealth Group, estimated that removing in-home HRA diagnoses from risk adjustment would reduce payments by $62 billion.5
The coding-intensity adjustment
Separately from the HRA-specific rule above, the Affordable Care Act established a uniform, across-the-board reduction applied to all MA risk scores to offset coding intensity generally, rather than targeting any specific diagnosis source. This adjustment currently has a statutory floor of 5.9%.