The 85% Failure Rate Nobody Talks About
CMS’s CY 2027 Final Rate Announcement, published April 6, 2026, included a data point that should alarm every plan relying on unlinked chart review diagnoses for risk score. Of the 88.8 million chart review reconciliation (CRR) diagnoses submitted in 2023, 85% couldn’t be matched to encounters. That’s roughly 75.5 million diagnosis submissions that failed the encounter linkage test.
Starting in CY 2027, CMS will exclude unlinked CRR diagnoses from risk score calculations, with a narrow exception for MA-to-MA switchers. The payment impact is estimated at negative 1.53%. For plans that built their retrospective programs around submitting chart-review-only diagnoses without verifying encounter linkage, this exclusion directly reduces the revenue those programs generate.
The 85% failure rate reveals the scale of the problem. The vast majority of chart review diagnoses submitted to CMS in 2023 were unlinked. These weren’t edge cases. They were the standard output of retrospective programs across the industry. The programs found codes in historical charts and submitted them without confirming that the diagnoses connected to documented encounters.
Why Encounter Linkage Was Never Built Into the Workflow
Traditional retrospective programs were designed for a simple workflow: receive charts, identify diagnoses, submit codes. The question “does this diagnosis link to a submitted encounter?” was never part of the coding process because CMS didn’t require it for payment. Chart review diagnoses generated risk score credit regardless of encounter linkage.
That changed with the CY 2027 rate announcement. The exclusion creates a new step that most retrospective programs don’t have: pre-submission encounter verification. Before a chart-review diagnosis is submitted, the system needs to confirm that the diagnosis connects to an encounter already in CMS’s system. If it doesn’t, the code will be excluded from risk score calculation and generate no payment value while still creating potential audit exposure.
Plans running fragmented systems face this challenge acutely. If coding happens in one system and encounter data lives in another, verifying linkage requires cross-referencing across disconnected databases for every submitted diagnosis. At scale, that’s impractical without automation.
What Encounter-Verified Retrospective Review Looks Like
The fix is integrating encounter verification into the chart review workflow. When AI identifies a potential diagnosis during retrospective review, the system checks the encounter database before recommending the code. If an encounter exists and the documentation supports the diagnosis with MEAT evidence, the code proceeds through the standard validation process. If no linked encounter exists, the system flags the diagnosis as ineligible for CRR submission under the 2027 rules.
This pre-submission filter prevents plans from wasting coding effort on diagnoses that won’t generate payment value. It also prevents plans from accumulating unlinked submissions that create audit exposure without financial benefit. The filter operates as a quality gate that ensures every submitted chart review diagnosis meets the new encounter linkage requirement before it reaches CMS.
The Deadline Is 2027, but the Work Starts Now
Plans have one year to integrate encounter verification into their Retrospective Risk Adjustment HCC Coding workflows before the CY 2027 exclusion takes full effect. Programs that don’t add this step will keep submitting unlinked diagnoses that generate zero risk score credit, wasting coding resources on work product that has no payment value. The 85% failure rate from 2023 shows how much of the industry’s retrospective output falls into this category. Plans that close the encounter linkage gap now protect both their coding investment and their revenue under the new rules.