BCBSA says AI-assisted hospital coding added about $942m to its plans' costs

Blue Cross Blue Shield Association has put a price on hospitals' use of AI-assisted coding: about $942 million in extra spending by its plans over two years, according to research the insurer released in late September 2026 ¹. The coding vendors named in the debate, Solventum and Codametrix, reject the idea that their tools are inflating bills improperly ¹.

What BCBSA found

BCBSA's case is that hospitals are billing more inpatient stays as complex, with more diagnoses attached that attract higher reimbursement, while treatment has not risen to match ¹. Luke Chalker, BCBSA's senior vice president of product and data science, pointed to "significantly more anemia diagnoses" at the hospitals concerned without a matching rise in transfusions, and concluded that "AI is identifying more billable conditions, not sicker patients" ¹.

The mechanism runs through secondary diagnoses such as anaemia, malnutrition and low sodium, which can move a stay into a higher-paying diagnosis-related group whether or not the condition was treated ². Automated documentation tools scan a patient's chart and prompt physicians to record these conditions ². BCBSA does not allege fraud; the dispute is over whether these findings justify higher payment ².

BCBSA is not alone. PwC analysts attribute a 9% increase in insurers' medical costs next year to AI-backed billing tools ¹.

The vendors' answer

Solventum's coding platform is used by more than 80% of US hospitals, and Codametrix automates coding for over 500 hospitals and health systems ¹. Hamid Tabatabaie, Codametrix's chief executive, accepts that autonomous coding is raising payer costs but argues the tools mostly capture care that providers were previously failing to record ¹. Dr Travis Bias, deputy chief medical officer of health information systems at Solventum, said higher payments follow from a fee-for-service system that pays "for volume, not for value", while noting he had not studied the BCBSA analysis closely enough to confirm or dispute it ¹. Both companies point to compliance guardrails built into their products ¹.

What it means for the market

Revenue cycle AI is now the subject of an open fight between payers and the hospitals that buy it. For vendors, the commercial pitch of capturing under-documented care is exactly what insurers are now framing as a cost problem ¹. For hospital buyers, Bias warned of more audits, denials and appeals, saying large health systems already waste "billions of dollars" on rebutting and appealing denials ¹. For payers, the costs flow on to premiums paid by employers and patients ¹.

The rules underneath are old. The list of conditions that can raise a payment has not been overhauled in nearly two decades; the last substantial change to diagnosis-related groups came in 2007 ². CMS proposed in 2019 to downgrade or remove many of these codes after finding they were not linked to higher costs, but the effort stalled after hospitals objected ². Leah Pierson, a resident physician writing in STAT, argues that some documentation tools default to adding secondary diagnoses and that hospitals should build tools letting clinicians say whether a finding is clinically significant, with regulators ensuring they do ². Value-based payment, which does not reimburse every code, is the longer-term answer both vendors point to, though it has not overtaken fee-for-service ¹.

Caveats

The $942 million figure is BCBSA's own estimate from its own claims, and the vendors contest its interpretation rather than the rise in costs ¹. The STAT piece is an opinion article by one physician, not an independent analysis of the data ². The open question named in the reporting is still the gap between rising diagnosis codes and flat treatment patterns ¹.