Insurers say hospitals' AI billing tools added $942 million in costs
By EnkiEdited by VK, Editor
Published
Reporting from TechCrunch, The Decoder
A Blue Cross Blue Shield Association analysis says hospitals using AI coding tools recorded patients as sicker without changing their care, adding $942 million in spending over two years as insurers answer with AI of their own.
What it means for founders
- Revenue share pricing sells, and invites scrutiny. Paying a vendor a cut of recovered revenue makes the purchase easy for a hospital, but it is also the structure insurers now point to as the cause of inflated bills. If you price on outcomes in a regulated market, expect to defend the outcome itself.
- Every automated claim needs its evidence. Any product that assigns a billing code should show the clinical record behind it. That trail is what protects your customer when a payer challenges the claim.
- Both sides are buying. Audit, review and appeal tools for payers and providers now have a documented dollar figure to sell against.
- Watch for a response from Medicare or state regulators, and for insurers adding AI coding terms to contracts. Either would reshape demand in this category quickly.
The story
Hospitals that adopted AI tools for medical coding added $942 million to US healthcare spending over two years, according to an analysis by the Blue Cross Blue Shield Association, the body that represents the Blue Cross insurers. It found a steep rise in patients recorded as having complex conditions, with no sign that the treatment those patients received had changed.
How the money moves
In US healthcare, the codes attached to a hospital stay largely decide what an insurer pays. The analysis says AI assisted coding led hospitals to classify cases as more severe and to add secondary diagnoses, and it puts the average gain at close to $12,000 per case, as The New York Times reported. For hospitals the case is easy to make. The finance chief of one health system has said its coding vendor brings in about $1 million a month, and the vendor takes a share of the extra revenue.
Insurers are automating too, using AI to search records for claims they can deny. Because each round of a dispute now costs little for either side, claims go back and forth more often than before. TechCrunch reports that Abridge founder Shiv Rao conceded the risk of software fighting software on both sides, while arguing the same tools could also lower friction and cost. Writing in Health Affairs, former federal health official David Brailer said he expects AI to raise total health spending, much as the move to electronic health records did.
One side of a long fight
The figure comes from the insurers' side of a payment dispute that long predates AI, and the analysis is framed as a case against hospital coding practices. It arrives as the price of AI itself becomes a budget question elsewhere: The Decoder, in a wider piece on AI costs, reports, citing The Washington Sun, that the NSA has told Congress its bill for testing advanced models will run to billions of dollars this year. In healthcare, though, the concern is less what the software costs and more what it bills.
What we don't know yet
There is no published hospital response to the method, and it is unclear how much of the rise reflects conditions that were previously under recorded rather than overstated. Nor is it known whether Medicare or state regulators will change coding rules, or whether insurers will write limits on AI coding into hospital contracts.
Sources
Enki Daily
Get stories like this every weekday morning.
The day's AI stories for founders, each with what it means for your company. Free.