AI Coding Dispute Intensifies as Payers Challenge Rising Hospital Charges

The clash between payers and providers over AI-driven coding is intensifying, with insurers arguing that documentation tools are inflating reimbursement without corresponding care.

AI is quickly becoming the latest battleground in the long-running financial tension between payers and providers.

Hospitals are deploying ambient listening and autonomous coding tools to streamline workflows and capture patient acuity while payers have deployed AI to automate claim audits and issue denials.

Now, the Blue Cross Blue Shield Association (BCBSA) has issued a research brief suggesting that AI-assisted coding is a driver of commercial inpatient spending. For revenue cycle leaders, the report signals how payers may work to craft a narrative that favors them.

The Data Behind the Trend

Commercial inpatient costs increased by approximately 9% per member between 2023 and 2024 across a subset of participating plans, according to the BCBSA analysis of claims data. BCBSA researchers estimate that about 20% of this increase is directly attributable to rising coding intensity without a corresponding increase in provided care.

The report suggests that inpatient DRG reimbursement methodologies are vulnerable to this coding intensity. Admissions documented as more complicated through secondary diagnoses qualify for higher-severity DRG tiers, which are reimbursed at higher levels.

The analysis highlights "discordant care," defined as instances where the severity of coded conditions increases, but the performed treatments do not change in ways consistent with that severity. Coding intensity increased over time and was concentrated among 10% of hospitals in the sample, suggesting shifts in documentation practices rather than broad changes in patient acuity, according to the report.

Maternity care is cited as a primary example of this discordance. High-growth hospitals increased their coding of postpartum anemia from 4.0% of maternity admissions in the second quarter of 2022 to 12.3% by the first quarter of 2025, according to BCBSA data. Despite this 8.3 percentage point rise in coded anemia over this period, blood transfusion rates remained relatively flat, increasing by only 0.4 percentage points.

The Provider Perspective: Accuracy vs. 'Discordancy'

While the BCBSA report frames this trend as an affordability issue, revenue cycle leaders argue that the data requires clinical context. Keisha Downes, Vice President of Mid-Revenue Cycle at Beth Israel Lahey Health and HealthLeaders Exchange member, noted on LinkedIn that the payer definition of "discordant care" may oversimplify clinical reality.

Responding to the report's focus on transfusion rates, Downes pointed out that per ICD-10 CM coding guidelines, a diagnosis should be captured if it requires clinical evaluation, therapeutic treatment, diagnostic procedures, an extended length of stay, or increased nursing care and monitoring. "A transfusion is one possible intervention, not the diagnostic threshold," she wrote, arguing that the absence of a transfusion does not mean the clinical condition is absent.

Additionally, rise in complexity may be a correction of past inaccuracies.

"The real question isn't 'why is coding intensity going up?'" Downes wrote. "It's 'why was it underrepresented for so long, and what are we going to do now that we can see it more clearly?'"

Next Steps for Revenue Cycle Leaders

The BCBSA brief indicates that as AI adoption expands, payment models will increasingly reward intensive coding unless plan safeguards or adjustments to the DRG payment system keep pace.

To prepare, revenue cycle leaders must maintain internal audit processes, validating that AI-assisted output is reviewed with the same rigor applied to manual coding. As Downes noted, governance, validation, and compliance infrastructure matter. Organizations must ensure that new tools tie coded severity to objective clinical indicators so payment reflects the complexity of delivered care.