CMS Stars & Risk Adjustment

Problem

In Medicare Advantage, how much a health plan gets paid — and how well it performs on Stars quality ratings — depends heavily on the accuracy and completeness of clinical documentation. CMS uses Hierarchical Condition Category (HCC) coding to model patient risk, and Stars uses a battery of quality measures to assess preventive and chronic care. Both systems are only as good as the data feeding them.

The gap between what clinicians actually do and what gets documented, coded, and submitted is substantial and consequential. A patient with well-managed diabetes whose chart is missing key chronic condition codes will appear less sick than they are — depressing the plan’s risk-adjusted revenue and understating the true cost of caring for that population. A preventive screening completed in the office but not documented against the right measure will not count toward Stars. Neither failure is about the quality of care; both are about the quality of information.

The transition from HCC V24 to V28 sharpened this problem. V28 introduced new condition categories, retired others, and changed how conditions map to risk scores — requiring health systems to update coding practices, CDI workflows, and risk stratification models simultaneously.

Approach

The work centers on building the informatics infrastructure that closes the loop between clinical care and accurate risk and quality data. In practice this means operating across three layers:

• Documentation gap analysis — using EHR data to identify patients with suspected but undocumented HCC conditions, surfacing these gaps to care teams through targeted outreach lists and point-of-care prompts ahead of annual wellness visits and chronic care encounters
• Coding accuracy and recapture — building workflows that ensure conditions documented in previous periods are recaptured annually, supporting clinical documentation integrity (CDI) programs that bridge the physician-coder gap
• Stars measure performance — mapping quality measure specifications (HEDIS, CMS-defined) to EHR data fields, identifying where compliant care is being delivered but not captured, and building data pipelines that reliably extract and submit measure-compliant evidence
• V28 transition planning — auditing existing HCC code sets against V28 mappings, identifying affected patient panels, and updating risk stratification models to reflect the new coefficient weights

All of this sits at the intersection of clinical knowledge and data engineering. Getting it right requires understanding both what the clinical guidelines say and what the EHR actually records — and knowing where those two things diverge.

Results

Outcomes in this space are measured at the program level over annual cycles, tied to CMS payment reconciliation and Star rating announcements. Key performance indicators span both financial and quality dimensions.

Lessons

The most important insight is that risk adjustment and quality measurement are not separate programs — they are two views of the same underlying problem: health systems generate enormous amounts of clinical data, most of which is never used to accurately represent the population being cared for. CDI, Stars gap closure, and HCC recapture are all symptoms of that broader failure.

Fixing it requires more than better coding. It requires informatics infrastructure that makes complete documentation the path of least resistance for the clinician — where the prompt arrives at the right moment, the abstraction is already done, and the clinician is confirming rather than originating. The closer you get to making documentation feel like care rather than paperwork, the better your data gets.