Why clinic RAF averages can mislead ACO coding reviews
A clinic’s average risk adjustment factor reflects its patient mix as well as coded diagnoses. Looking at the full distribution can help distinguish a healthier panel from a possible documentation gap.

A clinic’s average risk adjustment factor (RAF) score is not, on its own, evidence that the practice is undercoding. RAF estimates the expected cost and clinical risk of a patient population, and its score reflects both demographics and documented diagnoses. Comparing clinic averages without accounting for differences in who they treat can lead accountable care organizations (ACOs) to misread coding performance, according to a Medical Economics analysis.
That distinction matters in value-based care oversight, where RAF is used to assess expected patient risk. A clinic serving younger, healthier Medicare beneficiaries may have a lower average than one treating more older, dual-eligible, disabled or institutionalized patients. The average alone cannot show whether a lower score reflects a healthier panel or conditions that were not captured in coding.
What RAF scores include
At the patient level, RAF combines demographic factors, diagnosis-based factors and applicable interactions. Demographic factors include age, sex, Medicaid or dual status, original disability entitlement, and community or institutional status. Diagnosis factors depend on conditions documented and coded in a way that maps to hierarchical condition categories, subject to model rules such as hierarchies and interactions.
As a result, two patients with low RAF scores may have different profiles: one may be relatively healthy with few relevant conditions, while another may have chronic illnesses that were not assessed, documented, coded or submitted in a way that affects RAF. The score by itself does not establish which explanation applies.
Why the full distribution matters
The analysis argues that ACOs should examine patient-level RAF scores across a clinic’s panel, rather than relying only on its mean. A distribution may show a low-score cluster, a broader middle group and a long high-score tail. Those patterns can help reviewers ask whether a low-score group reflects a genuinely healthier population or a possible diagnosis-capture gap.
A low-RAF cluster is not inherently a coding problem, the analysis says. It may reflect healthier or newly Medicare-eligible patients. A high-score tail may include people with greater medical complexity, frailty, disability or institutional status, for whom specific, compliant severity documentation matters. No clinic-level data or benchmark is provided in the article, so it does not establish whether any particular practice is undercoding.
This report draws on information from medicaleconomics.com.
Editorial checks passed Facts, figures and quotes checked against 1 source
Add Clinic Near Me as a preferred source on Google



