Risk Adjustment Overview

Note that Zus Risk Adjustment features are currently in public preview and are subject to change.

Risk Adjustment Factor (RAF) Gaps (also called "risk gaps") help identify potentially missing or incomplete risk-adjustment diagnoses for a Medicare Advantage patient in the current calendar year. The goal is to surface opportunities for providers and care teams to confirm/document conditions and ensure appropriate coding and billing.

A risk gap is calculated per patient and calendar year, for each relevant HCC (CMS-HCC v28). Gaps can come from multiple sources—historical encounter diagnoses, AI-driven suspecting, and payer-provided gap files—and are combined into a single, consistent representation.

Gap types

Gap type	Definition
Recapture	The patient had evidence of an HCC in a prior year, but it's not present this year.
Suspect	No prior or current-year diagnosis exists for an HCC, but there is evidence suggesting the condition exists.
Net-new	The HCC appears in the current year, but did not appear in the prior year.

Gap status

Risk gaps carry a status indicating whether the gap is still open or appears likely/confirmed to close.

Status	Definition
`open-gap`	No evidence of the HCC being documented in the current year; the gap remains open.
`closed-gap`	The HCC is considered closed for the year (e.g., acknowledged by the payer).
`expected-to-close-1st-party`	Current-year documentation exists from the provider/organization, but closure has not yet been confirmed by the payer.
`expected-to-close-3rd-party`	Current-year documentation exists from an external source, but closure has not yet been confirmed by the payer.
`accepted-for-review`	A user has acknowledged the gap and marked it for follow-up or review.
`dismissed`	A user has decided the gap is not actionable (e.g., inaccurate, not relevant, or not appropriate to pursue).

Data sources

Risk gaps may be identified by any combination of:

Data source	What it means
Rules-based evaluation	Evaluation of historical encounter diagnoses (FHIR `Condition` resources).
AI-generated suspects	AI-inferred potential conditions (with confidence + evidence references).
Payer-provided gaps	Payer-supplied open/closed gaps and contributing ICD-10 factors.
User updates	Status changes and workflow actions taken by end users.

RAF Scores

A RAF Score (Risk Adjustment Factor score) estimates a patient's expected healthcare cost based on demographics and diagnosed conditions. Scores are computed using CMS-HCC v28 coefficients and the Community Non-Dual Aged (CNA) model.

We calculate three scores per patient and calendar year:

Score	Which HCCs are included
Potential	All HCCs from non-dismissed risk gaps.
Expected	HCCs from gaps with status `expected-to-close-1st-party`, `expected-to-close-3rd-party`, or `closed-gap`.
Actual	Only HCCs from gaps with status `closed-gap` (acknowledged by payer).

Score components

Each score is the sum of:

Factor	Description
Age/sex	Demographic bracket (e.g., F70–74, M85–89) based on age as of Feb 1 of the payment year.
HCC coefficients	Weight for each HCC included in that score type.
Diagnosis count	Factors D1–D10+ based on total HCC count.
Interactions	Combination factors (e.g., Diabetes + Heart Failure).

Glossary

Term	Meaning
Risk Gap	A potential diagnosis/coding gap for a patient and calendar year, represented as a FHIR `MeasureReport`.
RAF Score	Sum of coefficients for a patient's demographics and HCCs; reflects expected cost. Includes potential, expected, and actual variants.
HCC Code	Hierarchical Condition Category code used for risk adjustment (CMS-HCC v28).
ICD-10 Code	Diagnosis code that may map to one or more HCCs. Risk gaps can track associated ICD-10 codes and provenance.
RAF Coefficient	The risk-adjustment weight associated with an HCC (and other model factors). Risk gaps store the HCC coefficient used in scoring.
Recapture	Prior-year HCC evidence exists, current-year evidence missing.
Suspect	No current/prior-year diagnosis, but evidence suggests the condition exists (rules evidence, AI suspecting, payer file, etc.).
Net-new	Current-year evidence exists, no prior-year evidence.
Encounter Diagnosis	A diagnosis recorded during a clinical encounter (represented as a FHIR `Condition` with an encounter-diagnosis category).
Evidence (evaluated vs related)	"Evaluated" means the resource was used to determine the gap; "Related" provides helpful context for reviewers.