In April 2026, the Centers for Medicare & Medicaid Services (CMS) announced excluding diagnosis codes from unlinked chart review records (CRRs) from risk score calculations in Medicare Advantage (MA) and Part D for calendar year (CY) 2027. Let’s dive into the details of how, why, and what plans and providers should prepare to meet this change in the coming year and beyond.
Unlinked CRRs exclusion finalized for CY 2027
Beginning with payment year 2027, CMS will exclude diagnosis information from unlinked CRRs from MA risk score calculations. A chart review record is considered “unlinked” when it is not tied to a specific encounter data record (EDR) documenting an actual item or service delivered. Plans may continue to submit unlinked CRRs but those diagnoses may no longer count toward risk scores, unless a member changes between two different MA organizations.
CMS first proposed this change in the CY 2027 Advance Notice, released January 26, 2026. It was carried through and confirmed in the CY 2027 Rate Announcement, published April 6, 2026, making the policy final and effective for the 2027 payment year. The requirement applies broadly across MA and, in parallel, to Part D risk adjustment; PACE organizations are exempted from this specific exclusion for CY 2027 as they continue their model transition.
Why CMS is excluding unlinked CRRs
This policy reflects CMS’s broader effort to strengthen the integrity of risk adjustment data by ensuring diagnoses are supported by real, documented care. CMS grounded the policy in program integrity findings from the HHS Office of Inspector General (OIG) and the Medicare Payment Advisory Commission (MedPAC), both of which connected unlinked chart reviews and health risk assessments to a substantial share of the coding pattern differences between MA and Original Medicare. Several other findings helped drive this shift:
- Diagnoses without a corresponding encounter lack independent evidence that a service was actually rendered, raising data integrity concerns.
- OIG estimated unlinked CRRs contributed billions of dollars in potential overpayments in a single payment year.
- Nearly six in 10 MA contracts submitted unlinked CRRs as of 2022, indicating the practice is fairly widespread rather than isolated or irregular.
- CMS views encounter data as sufficiently mature to serve as the primary source of truth for risk-adjustment-eligible diagnoses.
Per CMS, the goal of this ruling is to shift competition among plans toward the value and quality of care delivered, rather than toward who can most effectively mine charts for additional diagnoses after the fact.
What this means for health plans
As noted in the 2027 Medicare Advantage and Part D Rate Announcement, CMS actuaries project the exclusion will reduce aggregate MA payments by roughly 1.5%, or more than $7 billion, in 2027 alone. The magnitude of impact will vary widely by plan, falling hardest on organizations that have historically relied on chart review programs to identify and report diagnoses outside of a documented encounter. Other possible predictions include:
- Risk scores may decline for members whose only diagnosis evidence came from unlinked CRRs, directly affecting plan revenue and margin.
- Retrospective chart review workflows must be re-architected to prioritize and demonstrate a link to a qualifying encounter.
- Operational teams need a scalable way to reconcile large volumes of chart review output against encounter data submissions, often across disparate source systems and vendors.
- Plans that cannot efficiently link charts to encounters risk losing legitimate, well-supported diagnoses simply because the link was never established or documented.
The practical challenge is not just accepting the new rule—it's building the infrastructure to link chart-derived diagnoses to encounters at scale, accurately, and fast enough to matter for submission deadlines.
How to meet this challenge
The CY 2027 linkage requirement is now final, not speculative. Plans that get ahead of it are treating chart-to-encounter linkage as an operational capability to have now, rather than a compliance box to check later. Three practices stand out for closing the gap at scale:
Automate matching, but keep thresholds configurable
Charts should be matched to corresponding encounters through clear, pre-determined matching logic—with thresholds that plans can tune themselves. That configurability is what lets each plan set its own balance between match precision and match volume, rather than accepting a fixed, generic standard that may not fit its data quality or risk tolerance.
Route ambiguous matches to human review
When more than one encounter meets match criteria for a given chart, the safer path is to surface those candidates for review rather than force an automated best guess. Staff should be able to view competing matches side by side and manually confirm the correct link, preserving accuracy on the cases that matter most.
Decide deliberately what happens to unmatched charts
Not every chart will have a qualifying encounter available at the time of processing. Plans need a deliberate policy for these cases: hold the chart for a later matching attempt as new encounter data arrives, or release it to continue through existing workflows. This helps avoid unnecessary bottlenecks while still working to maximize the number of compliant, linked diagnoses.
Together, these practices let health plans keep the diagnoses their chart review programs surface, without the administrative burden of manually reconciling every chart against encounter data, and while staying fully aligned with CMS's linked-encounter standard for CY 2027 and beyond.
