Get Full Government Meeting Transcripts, Videos, & Alerts Forever!
Get email alerts on the Claims Integrity topic
No spam. Unsubscribe anytime.
Blue Cross tells Tennessee lawmakers it does not use AI to adjust claims; clinicians review outliers
Summary
Blue Cross Blue Shield of Tennessee outlined a claims-integrity program that flags outlier providers for clinician review, saying adjustments are made by human reviewers and not by artificial intelligence; the insurer provided data on scope and impact and fielded lawmaker questions about vendor processes and provider appeals.
Blue Cross Blue Shield of Tennessee told the House Insurance Committee on Feb. 3 that it uses a vendor-driven crosswalk algorithm to identify provider outliers and relies on clinician reviewers — not AI systems — to recommend claim-level adjustments.
"Very simply, the answer is no," Russell Martie, vice president of state government relations, said when asked whether Blue Cross used artificial intelligence to make adjustments. Martie and Dr. J.B. Sobel, the plan's subject-matter expert, said the program targets providers whose claims disproportionately code at the highest-intensity E&M levels and that nurse and physician reviewers make any downstream edits.
Dr. Sobel described how electronic medical record tools and embedded AI features can prompt diagnoses or documentation that do not always reflect clinical justification. He showed committee data indicating an increase in higher-intensity E&M billing at some provider groups and said Cotivity — the vendor used for first-line review — applies a diagnosis-to-level crosswalk and that clinicians then review flagged claims.
Blue Cross provided program metrics in committee: the insurer said it had processed about 52,000,000 claims across Tennessee providers since April, that 14% of its providers (roughly 4,000 of 39,000) initially met the outlier criteria, and that some 552,000 adjustments had been made with an average impact of $31 per adjusted claim. The company said appeals have overturned a portion of those edits and that no payments are stopped while an edit is applied.
Lawmakers pressed the witnesses on process details, including whether Cotivity uses AI, whether first-level reviews are specialty matched, and how quickly providers are removed from the outlier cohort if their coding patterns change. Blue Cross said the vendor's algorithm dates to 2006, uses no machine-learning model for the edit described, and that removal from the cohort can take months as practices change.
The committee did not take formal action on the presentation; members asked for follow-up data specific to TennCare and signaled plans to hear managed-care organizations and other payers at a later session.
