Hospitals' growing use of artificial intelligence coding tools was linked to nearly $1 billion in additional healthcare spending for Blue Cross and Blue Shield companies between 2023 and 2025, according to a new analysis by the Blue Cross Blue Shield Association.

The study examined changes in hospital secondary diagnoses using AI-assisted coding systems and found that more patients were classified as having complex medical conditions. Those changes resulted in an estimated $942 million in additional medical costs for Blue Cross and Blue Shield insurers over the two years.

The findings raise concerns that AI coding systems may be increasing healthcare costs by identifying additional conditions that affect how hospitals are reimbursed, rather than reflecting a corresponding increase in patients' medical needs.

Researchers examined diagnoses such as anemia following major bowel surgery. BCBSA Senior Vice President of Product and Data Science Luke Chalker said hospitals would normally be expected to provide additional treatment if patients were genuinely becoming sicker.

Instead, the analysis found substantially more anemia diagnoses without a corresponding increase in blood transfusions, according to Chalker.

“The disconnect between diagnoses and treatment suggests that AI is identifying more billable conditions, not sicker patients,” Chalker said.

Of the $942 million in additional costs identified in the study, approximately $650 million came from secondary diagnoses that moved claims into higher-reimbursement categories.

BCBSA Senior Vice President of External Affairs David Merritt said the findings highlight the need to examine how AI tools are being used as healthcare costs continue to rise for families, employers and taxpayers.

The analysis comes as healthcare affordability remains a major concern for Americans and a prominent issue heading into the November midterm elections. A September survey found that voters trusted Democrats over Republicans on healthcare by a 22-point margin.

The BCBSA study does not establish that AI coding tools were solely responsible for the additional spending, but the association said the findings warrant further examination of how the technology is affecting medical billing and healthcare costs.