Independence Blue Cross (IBX), the dominant Medicare Advantage insurer in the Philadelphia area, agreed on October 1, 2026 to pay $22.5 million to resolve a False Claims Act case brought by a whistleblower and joined by the Department of Justice. The allegations didn’t center on coders submitting bad diagnosis codes in the first place — they centered on what happened after a chart review program flagged that some of those codes couldn’t be substantiated, and the codes stayed on the claims anyway.
For medical coding and compliance teams, that distinction is the real story. A chart review program is supposed to be the safety net. This case alleges the net had a hole in it, and that the hole was known.
What the DOJ Says Happened
According to the Department of Justice and the unnamed whistleblower’s complaint, IBX ran an internal chart review program in which nurses checked whether medical records actually supported the diagnosis codes the plan had submitted to CMS for risk-adjustment payment. The alleged conduct spans 2017 through 2021 (IBX’s own public statement cites 2016 through 2020; the company has not admitted wrongdoing).
The chart review program itself
The program worked roughly as designed when the news was good for IBX: when a nurse’s review supported additional, previously unreported diagnoses, the DOJ alleges those findings were used to pursue more risk-adjustment revenue. When a review instead failed to substantiate a code already submitted and paid, the government alleges that finding often didn’t trigger a deletion or a payment adjustment. CMS requires plans to self-report and return overpayments once they’re identified — a one-directional use of chart review results is exactly what that rule is meant to prevent.
Where the process broke down
Assistant Attorney General Brett A. Shumate’s statement on the settlement, as reported by HealthExec, frames this as part of a broader DOJ posture on Medicare Advantage risk-adjustment integrity: chart review and coding validation programs are now something the government expects to see evidence of actually being followed, not just maintained on paper. A program that exists but whose negative findings don’t flow through to code deletions and overpayment refunds doesn’t satisfy that expectation, in the DOJ’s view, even if the underlying diagnosis coding guidance was followed correctly at the point of initial submission.
Why This Isn’t Only a Medicare Advantage Plan Problem
It’s tempting to read this as a payer-side story, since IBX is the MA plan and CMS payment integrity was the specific mechanism at issue. But the structural problem the DOJ is describing — a validation step that produces findings nobody is required to act on — is not unique to health plans. Hospitals, health systems, and RCM vendors running internal coding audits, second-level reviews, or AI-assisted coding QA face the identical exposure if negative findings from those reviews don’t have a defined, auditable path to a correction. A validation program that only gets used to find additional revenue, and never to catch overpayment, is the pattern regulators are now naming explicitly.
Where Chart Review Programs Typically Fail
Compliance teams evaluating their own chart review or coding validation workflows against this case should look for:
- No defined, time-bound process for converting a “code not substantiated” finding into an actual claim correction or self-disclosure.
- Asymmetric use of review results — findings that increase revenue get acted on faster or more reliably than findings that decrease it.
- No audit trail showing who reviewed a flagged chart, what they decided, and what action followed.
- Review capacity that can’t keep pace with claim volume, so a growing backlog of unresolved findings sits unaddressed.
- No mechanism to aggregate review findings into pattern-level alerts (the same code or same provider failing validation repeatedly).
How Agentic AI Changes the Chart Review Equation
This is precisely the gap agentic coding and validation systems are built to close. Rather than producing a static flag that depends on a human remembering to act on it, an agentic system can tie a “not substantiated” determination directly to a workflow step — routing it to a coder for correction, generating the documentation needed for a self-disclosure, or escalating a repeated pattern for compliance review — with every step logged. That closes exactly the loop the IBX complaint alleges was open: findings that exist but go nowhere. The goal isn’t just faster initial coding; it’s a provable, bidirectional audit trail from validation finding to resolution, which is the specific evidence DOJ investigators are now asking health plans and health systems to produce.
Coding teams don’t need to wait for a settlement of their own to test this. The IBX case is, in effect, a free audit checklist: pull a sample of your own chart review findings from the last two years, and confirm that every “not substantiated” result has a documented resolution. If that trail doesn’t exist end-to-end, that’s the finding worth acting on before a whistleblower does it for you.
Closing that loop reliably, at claim volume, is exactly where Medikode’s automated medical coding platform is built to help — surfacing coding and documentation gaps with a full audit trail from flag to resolution, so validation findings turn into corrections instead of liabilities.