AI Coding Tools Added $942M to Hospital Bills Without Care Changes

Hospitals' adoption of artificial intelligence for medical coding has driven up insurance claims by nearly $1 billion over two years without corresponding changes in the care patients actually received, according to a new analysis from the Blue Cross Blue Shield Association.

The share of inpatient stays billed to Blue plan members as medically complex rose from 37% at the start of 2023 to 40% by the end of 2025, BCBSA reported. That shift translated to an estimated $942 million in additional costs borne by member plans—costs the association says reflect documentation changes rather than sicker patients or more intensive treatment.

Why it matters

The findings spotlight a growing tension in healthcare finance as both payers and providers deploy AI tools in an escalating claims battle. While hospitals frame the technology as improving coding accuracy and recovering legitimate revenue, payers argue it enables billing inflation that ultimately raises premiums and out-of-pocket costs for employers, patients, and taxpayers. The dispute centers on whether AI is correcting past undercoding or enabling systematic upcoding—a question with major implications for healthcare spending growth.

Secondary diagnoses driving complexity creep

About 70% of the coding intensity increase stemmed from more than 55,000 additional cases where secondary diagnoses pushed claims into higher-severity, higher-reimbursement diagnosis-related groups (DRGs). These secondary diagnosis upgrades alone accounted for $653 million of the total increase, averaging $11,000 per excess complex case.

"Critically, what we found is underneath all of that data [was] no change in corresponding care for a more complex patient," Luke Chalker, BCBSA's senior vice president of product and data science, told reporters. "We now see that coding has materially changed... But we find no evidence of a corresponding change in care."

The association pointed to major bowel procedures as one example: claims at the highest complexity level jumped from 10.2% to 22.7%, while non-complex cases fell from 36.6% to 32.8%, together accounting for nearly $61 million in incremental costs.

Treatment intensity doesn't match billing complexity

To test whether increased coding reflected genuinely sicker patients, BCBSA examined hospitals in the top quartile for complex DRG cases. These facilities showed similar or lower treatment intensity than peers—measured by ICU utilization, transfusions, reoperations, and length of stay—despite billing 65% of cases as completed DRGs.

The analysis also highlighted posthemorrhagic anemia, which BCBSA called "a common bump code." Top-quartile hospitals for anemia diagnosis (13.7% versus 9.9% for others) actually had lower transfusion rates among diagnosed patients (16.9% versus 19.3%).

"The consistent inverse relationship between diagnosis-based complexity and both aggregate resource utilization and diagnosis-specific procedural intervention is the strongest indicator that coding escalation reflects documentation practice changes rather than actual patient acuity shifts," the association's white paper stated.

The AI arms race in healthcare billing

BCBSA attributed the coding changes to "systemic" adoption of AI revenue cycle management tools, referencing a June survey showing more than 63% of healthcare organizations now use AI in revenue cycle workflows. Hospital leaders have defended the technology as helping them code more accurately and efficiently while responding to increased payer scrutiny and claim denials.

Chalker rejected the framing that AI tools simply secure long-deserved payments. "We should be reimbursing for care delivered, [that] is more critical than anything else because that's supposed to be the design of how it all works," he said.

The association acknowledged its analysis relies on claims data rather than clinical documentation, which would more directly show whether patients were genuinely sicker. However, Chalker said Blue plans with chart access have reinforced the findings in their own reviews.

The details were first reported by Fierce Healthcare. BCBSA said it plans to release additional analyses covering outpatient care and other service areas.