AI failures · Incorrect automated decision
Cigna PXDX — Doctors Deny 300,000+ Payment Claims in Batches, Averaging 1.2 Seconds per Case
Verified Checked against its sources on 2 sources quoted
- Where it happened
- United States
- Organization
- Cigna
- Date
- March 25, 2023 (ProPublica / The Capitol Forum report; denials covered two months of 2022)
- Function
- Health insurance claims processing
- Where it happened
- Post-service claim payment review (medical necessity)
- System
- PXDX (procedure-to-diagnosis) automated claim-review algorithm
- Technology
- Rules-based automation (not AI)
- Type of failure
- Incorrect automated decision
- Caught by
- A journalist
Summary
ProPublica and The Capitol Forum reported that Cigna used a system called PXDX in which an algorithm flagged claims whose diagnosis did not match the tests or procedures Cigna considered acceptable, and company medical directors signed off the denials in batches without opening patient files. Internal documents showed more than 300,000 payment requests denied this way over two months in 2022, at an average of 1.2 seconds per case. The denials were of payment after care was given, leaving patients with unexpected bills. Cigna called the reporting biased and incomplete. A putative class action (Kisting-Leung v. Cigna, E.D. Cal.) followed in July 2023; in March 2025 the court granted Cigna's motion to dismiss in part and denied it in part.
Consequence
- Putative class action Kisting-Leung v. Cigna (E.D. Cal., filed July 24, 2023) over the PxDx review; on March 31, 2025, the court granted the motion to dismiss in part and denied it in part, so some claims proceed.
Who caught it A journalistFound outside the organization
Analysis
What could have been done
A review that can actually catch the error
The reviewer has the time, information and expertise to catch the error, not only to sign off.
Internal documents reported by ProPublica show company medical directors signed off more than 300,000 algorithm-flagged payment denials over two months in 2022, in batches and without opening patient files, at an average of 1.2 seconds per case.
Error rates, overrides, complaints and drift are monitored, with thresholds that trigger escalation.
A review time of about one second per denial was visible in the company's own documents, yet the practice became public through journalists' reporting on those documents.
What would likely have prevented it
- Define what a reviewer must open and check before an automatically flagged claim is denied, and do not allow batch approval of flags that deny payment to individuals.
- Treat the rules engine's flag as a recommendation and require a case-level decision recorded with the evidence the reviewer considered.
What would likely have caught it earlier
- Monitor reviewer time per case and the rate at which reviewers overturn automated flags, with thresholds (for example seconds per case or a near-zero override rate) that escalate to compliance.
- Re-review a monthly sample of signed-off denials against the full case file.
Key dates
- March 25, 2023Event
- October 16, 2026Checked against sources
- October 18, 2026Checked against sources
- October 21, 2026Checked against sources
Developments are added when the record is reviewed. Next review: October 16, 2027.
Evidence
Each passage is quoted from its source. The quoted passages are re-checked against the live pages every month.
-
Over a period of two months last year, Cigna doctors denied over 300,000 requests for payments using this method, spending an average of 1.2 seconds on each case, the documents show.
Supports: who, what, scale, 1.2 seconds, date of report
-
Defendants' motion to dismiss (Doc. No. 46) is GRANTED in part and DENIED
Supports: class action over PxDx; ruling on motion to dismiss
Suggested citation. Eminence Scan, “Cigna PXDX — Doctors Deny 300,000+ Payment Claims in Batches, Averaging 1.2 Seconds per Case”, Eminence Growth Solutions, last verified October 16, 2026, https://eminencegrowthsolutions.com/tools/eminence-scan/record/fail-014/.
Using our research: you may read and quote Eminence Scan with credit to “Eminence Scan, Eminence Growth Solutions” and a link; all other rights reserved. Terms: Using our research.
Spotted an error, or a development we have missed? Report a correction. How we check records: research methodology.