AI Healthcare Billing Costs: What BCBSA’s $942 Million Estimate Shows

AI is changing hospital billing by helping identify diagnoses that can increase reimbursement. Blue Cross Blue Shield Association (BCBSA) estimates that AI-assisted coding contributed to $942 million in additional costs for its health plans between 2023 and 2025, but the association did not attribute the entire increase to AI, and the American Hospital Association disputes the analysis’s context.The estimate and competing positions

What BCBSA’s estimate counts

BCBSA said about $653 million, roughly 70% of the billing it identified, was tied to additional diagnoses without a recorded change in care. The association’s senior vice president told CNBC that 60% of hospital systems began using AI coding tools during a period when complex coding grew, while cautioning that several factors affect coding intensity.BCBSA’s figures and attribution

Those numbers describe the insurer association’s analysis, not a settled measure of AI’s causal effect. The American Hospital Association argued that patients are older and more clinically complex and that AI can help providers capture conditions for care planning; it said BCBSA’s analysis lacked the context to assess quality, access, or spending.The AHA’s response

A diagnosis can change payment without changing treatment

Medical coders translate diagnoses and procedures into standardized codes used on claims. An additional diagnosis can move a patient into a higher-paying reimbursement category. BCBSA says some secondary diagnoses may be identified from a single lab value, a pattern that can be amenable to automated detection.How coding affects reimbursement

That does not make every added diagnosis improper. Health economist Christopher Whaley told CNBC that some conditions are legitimate and were previously missed. Others may not influence care even if they support another billing code. The distinction matters: “no change in care” is not the same as “false diagnosis,” but it raises a question about whether the coding reflects clinically meaningful information.Expert views on additional diagnoses

The stakes also reach beyond a single claim. Diagnoses can enter a patient’s medical record, and a revenue-cycle strategist interviewed by CNBC argued that AI-generated coding should be audited and should not operate without human judgment.The case for auditing AI-assisted coding

Automation on both sides can multiply work

Hospitals use AI to support documentation, coding, and appeals; insurers also use it to review claims. BCBSA said its companies use AI in claims review and that a qualified human clinician always reviews clinical denials. The result is automation at several points in the same dispute, rather than a simple handoff from manual paperwork to faster processing.AI across claims review and appeals

There are potential gains: AI can reduce physicians’ administrative work and help providers manage more documentation requests and appeals. But if each side uses tools to scrutinize or contest the other’s decisions, the system may spend more on coding, review, and rebuttal without improving care. The hard question is whether added coding captures conditions that matter to patients, or mainly strengthens each side’s position in a payment negotiation.Benefits and risks of the administrative arms race