Insurers and hospitals actually agree on more than you’d think. They both say AI medical coding has changed how hospital stays are billed. Where they part ways is on why, and that disagreement now has a dollar figure attached.
On September 24, the Blue Cross Blue Shield Association (BCBSA) published a number its actuaries had been chewing on for two years.
Key takeaways
- BCBSA ties $942 million in extra spending to more hospital stays being billed as “medically complex,” with no matching rise in treatment.
- The American Hospital Association says patients really are older and sicker than in 2023.
- Both sides accept that AI medical coding tools and scribes raise coding intensity. The fight is over whether that reflects real complexity or a nudge toward higher-paying codes.
- Three other groups, including PwC and CMS Administrator Dr. Mehmet Oz, point the same way on near-term cost.
What Did the BCBSA Study Find About AI Medical Coding?
Using de-identified inpatient claims from the 31 Blue plans, which together cover about one in three Americans, BCBSA found that the share of hospital stays billed as “medically complex” rose from roughly 37% at the start of 2023 to 40% by the end of 2025. It links that shift to $942 million in additional spending, without a matching increase in the care patients actually received.
About 70% of that added cost, the association says, came from secondary diagnoses pulled from a single lab value. That’s exactly the kind of detail an AI documentation tool is built to catch, and one a rushed clinician might once have left off the chart.
BCBSA’s senior vice president of product and data science, Luke Chalker, summed up the logic: “If patients are truly sicker, we’d expect to see more treatment.” In the data, he says, diagnoses climb while care doesn’t.
The Hospitals’ Response, and Why It Isn’t Unreasonable
The American Hospital Association rejects the insurer’s framing. Its argument is that patients admitted today genuinely are older and carry more chronic conditions than those admitted in 2023, a fact no algorithm invented.
That’s a fair point. Hospitals have argued for years, with some justification, that under-documentation was its own problem. Doctors noticed conditions they never formally coded, and work went unbilled, because a stretched physician typing notes between patients simply ran out of time. If an AI scribe or coding assistant catches a real secondary diagnosis a doctor would once have missed, it isn’t inventing illness. It’s correcting years of undercounting.
Where both sides quietly agree
Here’s the uncomfortable bit. According to reporting on the dispute, both sides privately concede the mechanism. Ambient AI scribes and AI medical coding tools do increase coding intensity. The whole argument is about whether that increase reflects real, previously missed complexity or a systematic push toward whichever code pays more.
What BCBSA does not claim
To its credit, BCBSA doesn’t overreach. Its white paper stops short of proving AI caused the shift. It relies on claims data rather than full patient charts, and it names no specific hospital system or AI vendor.
Is AI Medical Coding Just One Data Point? Three More Sources Say No
Over the past six months, three other independent readings have pointed in the same direction.
Peterson Health Technology Institute
After convening hospitals, payers, vendors and regulators in April, PHTI estimated that AI scribes and coding tools add roughly $1,004 per provider per month in downstream cost.
PwC Health Research Institute
In June, PwC surveyed actuaries at 27 health plans. Nearly 70% named provider-side AI as a top-three cost inflator heading into 2027.
CMS Administrator Dr. Mehmet Oz
On September 23, the day before the BCBSA paper, Oz told an industry summit that AI in healthcare will be inflationary before it becomes a source of savings.
Four organizations, four different incentives and methods, one uncomfortable timeline: AI is adding cost to the system now, with efficiency gains arriving later, if at all.
Why Employers and Business Leaders Should Care
This isn’t only a hospital CFO problem. Every dollar of that $942 million shows up somewhere else eventually: in premiums, in stop-loss pricing, or in a self-insured employer’s own claims experience.
BCBSA’s data shows more than 60% of hospital systems already use AI coding tools, and that share is rising. For anyone managing group health benefits, the spread of AI medical coding means the practical result is that renewal talks over the next two to three years may come with a documented, citable reason for rate increases that has little to do with patients using more care and a lot to do with how hospitals now bill for the same care.
The Number Worth Remembering
Whether AI medical coding is fair, meaning legitimate capture of missed complexity or AI quietly optimizing for the highest-paying code, will take years and probably several more studies to settle. In the meantime, the figure to hold onto isn’t $942 million. It’s the 3-point gap between how much sicker claims say patients got and how much more treatment those same claims show they received.
Frequently Asked Questions
1. What did the BCBSA study find about AI and hospital billing?
Hospital stays billed as medically complex rose from about 37% to 40% between early 2023 and the end of 2025, tied to $942 million in additional spending with no matching rise in treatment.
2. Does the study prove AI medical coding caused higher billing?
No. BCBSA’s white paper stops short of proving that, relies on claims data instead of full charts, and names no specific hospital or AI vendor.
3. What do hospitals say about AI medical coding?
The American Hospital Association says today’s patients are older and have more chronic conditions than in 2023, and that AI can correct past under-documentation.
4. How could this affect employer health costs?
The extra spending can flow into premiums, stop-loss insurance prices and self-insured claims, so upcoming renewals may include rate increases justified by billing changes rather than higher utilization.
Explore more expert insights, leadership stories, and business strategies at GlobeVox Leaders
