The Insurer Is Paid by Diagnosis, So Diagnoses Multiply
Medicare pays private plans more for sicker members, which is necessary to stop plans from avoiding expensive patients. It also means a plan is paid more for documenting a condition, and documentation is something a plan can influence.
Why Risk Adjustment Has to Exist
Start with the problem it solves, because the mechanism is indefensible without it. If a government paid private health plans a flat amount per enrolled member, every plan would compete to enroll healthy people and avoid sick ones. That competition would be far more profitable than competing on care quality, and the sickest patients, the ones the program exists for, would be the least wanted.
Insurers call this adverse selection management, and left unchecked it destroys the purpose of the program. Risk adjustment is the fix: pay each plan according to the expected cost of the specific members it enrolls, so a plan that takes on a diabetic patient with heart failure receives substantially more than one enrolling a healthy member.
How the Payment Is Actually Calculated
In Medicare Advantage, each enrollee receives a risk score built from demographics and from diagnoses recorded during the previous year. Diagnoses map into groupings that carry defined weights, the weights are summed, and the resulting score multiplies a county benchmark payment.
A member with a score of 1.0 generates the benchmark. A member with a score of 1.6 generates sixty percent more, every month, for the following year.
| Input | Effect on Payment |
|---|---|
| Age, sex, eligibility status | Baseline component |
| Documented chronic conditions | Adds weight for the following year |
| Number of qualifying conditions | Additive, with interactions |
| Conditions present but not documented | No payment effect |
That last row is the entire issue. A condition the member has generates no revenue unless it appears in a claim or an accepted medical record. Revenue therefore depends not on how sick the member is, but on how thoroughly that sickness is written down.
The Predictable Response
Plans invested heavily in making sure conditions get documented, through several channels.
In home health assessments send a clinician to a member residence for an annual visit, often at no cost to the member, during which a comprehensive review may identify and record conditions. Retrospective chart review employs coders to examine existing medical records for documented conditions that were never submitted on a claim. Provider education and incentives encourage physicians to code to the highest level of specificity supported by the record.
Each of these has a defensible version. Undiagnosed conditions in elderly populations are real, home visits do identify genuine problems, and physicians do under document. Each also has a version in which the activity generates revenue without generating any change in care.
The test that separates the two is simple to state and hard to satisfy: did the identified condition lead to any treatment, referral, or change in management? A diagnosis that produces a payment and nothing else has not helped the patient.
What Oversight Has Found
Government auditors have examined this repeatedly. Reports have found that a substantial share of diagnoses generating payments came only from health risk assessments or chart reviews, with no corresponding clinical service delivered during the year, and that certain condition categories appeared far more frequently in Medicare Advantage than in traditional Medicare for comparable populations.
Enforcement has followed under the False Claims Act, on the theory that submitting diagnosis codes known to be unsupported constitutes a false claim for payment. Several large settlements and ongoing cases involve exactly this allegation. The legal question is generally not whether documentation programs are permissible, since they are, but whether specific submitted codes were supported by the record, and whether the plan ignored evidence that they were not.
The Audit Mechanism and Why It Was Contested
The government verifies coding through risk adjustment data validation audits, which pull member samples and check whether medical records support the submitted diagnoses. The contested element has been extrapolation: applying the error rate found in a sample to the plan entire population, which converts a modest sampling finding into a very large repayment.
Rules finalized in 2023 permitted extrapolation for payment years from 2018 forward, and were promptly challenged by the industry. The dispute matters financially because the difference between recovering errors in a sample and extrapolating across a book of business spans billions of dollars.
Reading the Sector
For anyone analyzing these companies, several disclosures carry unusual weight. Year over year risk score growth well above the rate of underlying population aging invites the question of whether it reflects better documentation rather than sicker members. The medical loss ratio, meaning the share of premium spent on care, is the counterweight, since a plan raising revenue through coding without corresponding cost will show it there. Reserves and disclosures related to government audits and investigations quantify the tail risk. And any change to the risk model itself, which regulators periodically revise to remove condition categories judged too easily influenced, can shift revenue across an entire industry without any change in operations.
The Bottom Line
Risk adjustment is necessary, well intentioned, and structurally exposed, because it makes revenue a function of a variable the payee controls. The honest framing is that the program bought protection against cherry picking and paid for it with an incentive to document aggressively. Reform proposals mostly aim at the same target, which is to weaken the link between a recorded diagnosis and a payment when no care follows from it. Until that link is broken, the response to a payment formula based on documentation will continue to be more documentation.