The Federal Government Put a Dollar Amount on Add-Only Risk
In March 2026, Aetna paid $117.7 million to settle False Claims Act allegations tied to its risk adjustment practices. $106.2 million of that settlement covered an add-only chart review program from payment year 2015. The remaining $11.5 million addressed false morbid obesity diagnosis codes submitted from 2018 through 2023. The case was brought by a whistleblower, a former Aetna risk adjustment coding auditor who understood exactly how the program was designed and why it was problematic.
This came on top of Kaiser’s $556 million settlement covering diagnoses that the DOJ complaint alleged lacked clinical support across approximately 500,000 members. Together, these cases represent over $670 million in settlements from just two organizations. And the OIG’s February 2026 Industry-wide Compliance Program Guidance made it clear that add-only chart reviews are now a specifically identified high-risk practice across the entire Medicare Advantage industry.
The message isn’t subtle. Retrospective programs that only add codes and never remove unsupported ones are regulatory targets.
Why Deletion Is as Important as Addition
The fundamental problem with add-only retrospective programs is mathematical. If your process only moves in one direction, your submitted diagnoses will always trend upward regardless of whether your members’ clinical conditions justify that trend. CMS tracks this at the population level. When a plan’s risk scores rise year after year without corresponding improvements in clinical outcomes, the data tells a story that regulators understand.
Deletion isn’t about losing revenue. It’s about ensuring that what you’ve submitted can withstand scrutiny. A code that was valid last year may not be valid this year if the condition isn’t being actively managed. A diagnosis that was added during a retrospective review may not hold up if the chart note lacks adequate MEAT evidence when auditors examine it with fresh eyes.
Two-way review, where coders identify both missed diagnoses and unsupported ones, creates a coding population that accurately reflects clinical reality. The adds capture legitimate complexity. The deletes remove liability. The net result is a risk profile that regulators can’t easily challenge because it’s been validated in both directions.
Building the Two-Way Workflow
Operationalizing two-way retrospective review requires technology that evaluates documentation in both directions simultaneously. When a coder opens a chart, the system should show which conditions are documented with adequate MEAT support (valid adds), which conditions are in the record but lack sufficient evidence (potential deletes or queries), and which previously submitted codes have no current documentation backing them (removal candidates).
Explainable AI makes this practical at scale. The system maps each diagnosis to specific evidence in the clinical note, identifies which MEAT elements are satisfied, and flags where evidence is missing or ambiguous. The coder validates the AI’s assessment rather than searching through the chart from scratch. The audit trail builds automatically because every decision is linked to documented reasoning.
The cultural challenge is real. Coding teams need updated performance metrics that value accuracy over volume. Removing an unsupported code must be recognized as a compliance success, not penalized as a productivity failure. Leadership has to redefine what good performance looks like before the workflow change can stick.
The Model That Survives Scrutiny
Retrospective chart review remains a necessary function. Plans will always need to reconcile clinical documentation with submitted data. The question isn’t whether to do retrospective review. It’s whether the program is designed to withstand the regulatory environment that now exists.
Plans restructuring their programs around Retrospective HCC Coding that validates in both directions, adding legitimate diagnoses and removing unsupported ones, are building the model CMS has explicitly called for. The supplemental data process is a two-way street. Plans treating it that way are aligned with regulatory expectations. Plans still running one-way programs are accumulating the exact type of exposure that generated nine-figure settlements.
ALSO READ: Newslikeyoucom: Complete Guide to the Platform, Features, Content, and Future of Digital News
