A transfusion rate carries the Blue Cross Blue Shield Association's case against hospital AI coding. At the quarter of hospitals where billed severity for major bowel surgery grew fastest, 13.7% of 2025 cases carried an anemia diagnosis against 9.9% at peers, yet only 16.9% of those patients were transfused, against 19.3% elsewhere (BCBSA white paper, September 2026). Coded diagnoses rose while the treatment for them fell.
BCBSA published the paper on September 24 with a system-wide estimate: Blue plans paid about $942 million more over 2024 and 2025 than 2023 coding patterns would have produced, as more than 60% of hospital systems adopted AI tools that scan lab reports and clinician notes (BCBSA release, September 2026). For a commercial health actuary the figure lands in the least visible part of inpatient trend, the severity mix inside a fixed DRG contract.
Key Takeaways
- $653 million of the $942 million, about 70%, came from 55,158 excess complex cases paid at roughly $11,800 each, all driven by secondary diagnoses that moved claims into higher-paying DRG tiers without more surgery.
- 75.6% versus 65.0%: top-quartile coding-growth hospitals billed major bowel procedures as complex far more often than peers, yet ran 11.5% ICU use against 13.2% and the same 4.0-day median stay.
- About 1.8 points of trend: a March 2026 Blue Health Intelligence brief put roughly 20% of a 9% per-member rise in Blue commercial inpatient cost on coding intensity, concentrated in 10% of hospitals.
- 5.90% coding adjustment: Medicare Advantage carries a statutory offset for coding differences in 2027, while a commercial DRG contract carries no equivalent, so the drift shows up as unit cost trend.
- About 5% case-mix growth from 2019 to 2024 is the hospitals' competing acuity measure, and any real acuity growth after 2023 is counted inside the $942 million, which its baseline method cannot separate from coding.
What the White Paper Measured in Major Bowel DRGs 329 to 331
The paper, titled Hospital Coding Intensity Analysis: Major Bowel Procedures, covers Blue claims from the first quarter of 2023 through the fourth quarter of 2025. Across all severity-tiered DRGs, the share of inpatient cases billed with a complication or comorbidity (CC) or a major one (MCC) rose from roughly 37% to about 40%. Against 2023 rates, hospitals classified 55,158 additional cases as complex, generating $653 million of incremental payment, which the paper puts at 70% of the coding-intensity cost.
Its deep dive is MS-DRGs 329 to 331, colon and bowel surgery split into MCC, CC and no-complication tiers. MCC claims rose from 20.2% to 22.7% between 2023 and 2025 while non-complex claims fell from 36.6% to 32.8%, worth $60.8 million on its own. The fastest-growing "bump codes" are diagnoses a single lab value can generate: unspecified acidosis (E87.20), hyponatremia (E87.1) and acute posthemorrhagic anemia (D62).
Two measures of the same patients' acuity point in opposite directions in the paper's 2025 comparison of the top 25% of hospitals by complexity growth against everyone else.
| 2025, DRGs 329-331 | Top 25% hospitals | Other hospitals |
|---|---|---|
| Cases billed complex (CC or MCC) | 75.6% | 65.0% |
| ICU utilization | 11.5% | 13.2% |
| Transfusion, all cases | 3.6% | 3.9% |
| Anemia diagnosis rate | 13.7% | 9.9% |
| Transfusion rate, cases coded with anemia | 16.9% | 19.3% |
| Median length of stay | 4.0 days | 4.0 days |
BCBSA's Luke Chalker read the gap as evidence that "AI is identifying more billable conditions, not sicker patients" (BCBSA, September 2026). The single-code pattern has a federal precedent: an Inspector General review of Medicare inpatient billing found more than half of stays billed at the highest severity level had only one diagnosis qualifying them for that level (HHS OIG, OEI-02-18-00380, February 2021).
How One Lab Value Reprices a Commercial DRG Claim
A severity-tiered DRG turns on whether at least one qualifying secondary diagnosis appears on the claim. One CC moves a bowel case from DRG 331 to 330; one MCC moves it to 329. Under a contract that pays a negotiated base rate times the DRG weight, the hospital's price never changes and the payment still rises. The paper notes that AI may be set to flag every out-of-range result, catching mild anemias clinicians typically leave untreated, and offers that as the explanation for the 38% higher anemia rate.
Adoption explains the timing. In a fall 2024 survey of 43 non-profit health systems, 24% had medical-coding AI fully deployed and another 21% were deploying it in limited areas (JAMIA, May 2025). By BCBSA's September count the share using AI coding tools exceeds 60%.
BHI's March brief, covering plans with about 62 million members, found Blue commercial inpatient cost per member up 9% from 2023 to 2024, with roughly 20% of that increase attributed to coding intensity, or about 1.8 points (BHI issue brief, March 2026). That brief found the growth concentrated: the top 10% of hospitals moved from 46.8% to 59.8% complex admissions between mid-2022 and early 2025, a 13.1-point rise, against 4.2 points everywhere else. A plan whose network leans on high-growth systems carries a multiple of the average coding load, and a book-level trend selection averages it away.
Medicare builds an offset into its own formula. CMS applies the statutory minimum 5.90% MA coding pattern difference adjustment to Medicare Advantage risk scores for 2027 (CMS 2027 Rate Announcement, April 2026). A commercial DRG contract has no such factor, so coding drift persists through every renewal until it is audited, renegotiated or carved out. The site's coverage of the 2027 trend surveys shows the same driver surfacing: PwC named provider AI revenue tools, Segal carried the 20% inpatient figure, and Marsh put physician coding at up to a point. The September paper adds the DRG-level discordance test those surveys lacked.
The Baseline Problem Behind the $942 Million
The headline number and the evidence measure different things. The $942 million compares every hospital's 2024 and 2025 case mix with 2023 rates, so it counts all complexity growth since 2023 as coding. The discordance proof covers one DRG family and the top quartile against its peers. An actuary removing coding intensity from a trend selection gets a ceiling from the first figure, while bowel surgery, at $60.8 million, is the only family where the paper pairs dollars with treatment evidence.
The hospitals' counter-measurement fills that gap. The American Hospital Association cites an AHA/Vizient finding that the case-mix index rose about 5% from 2019 to 2024, and says 19% of hospital expense growth over that period reflects sicker, more complex patients (AHA fact sheet, August 2026). Any genuine acuity growth after 2023 sits inside the BCBSA baseline with nothing to separate it out.
The AHA's sharper point lands on the payer's own revenue line: insurers that downcode hospital claims argue their enrollees are sicker for risk scores, which the AHA says amounts to wanting it "both ways." Diagnoses on paid claims feed the HHS risk adjustment model in the ACA markets and encounter-based risk scores in Medicare Advantage. To the extent a bump code maps to a payment HCC, the same entry that raises a Blue plan's claim cost can raise its risk-adjusted revenue, and a downcoding program that strips it cuts both lines at once.
Blue audit evidence suggests how large the recoverable share may be. In the March brief, one Blue plan's review at an outlier system found fewer than 20% of maternity cases coded with posthemorrhagic anemia met clinical criteria. Recoveries of that kind arrive after payment, through disputes the AHA already calls arbitrary, so the paid triangles that set next year's trend carry the full coded severity while any recovery is booked later and contested.
Further Reading
- Provider AI Revenue Tools and PwC's 9% Medical Cost Trend for 2027 – the trend-survey view of provider AI coding before the DRG-level evidence arrived.
- Segal's 2027 Survey Pegs Medical Trend at 9.9% – where the 20% inpatient coding figure first reached a renewal survey.
- Marsh's 8.2% Health Trend for 2027 – physician-side coding intensity and why unit-cost decomposition misses it.
- Cigna Sets a $3 Billion AI Savings Target – the payer side of the AI cost ledger and how the medical loss ratio splits it.
- OIG Extrapolates $180M From Sampled MA Overpayments – how diagnosis audits become recoveries on the Medicare Advantage side.
Sources
- Blue Cross Blue Shield Association, Hospital Coding Intensity Analysis: Major Bowel Procedures (September 2026)
- Blue Cross Blue Shield Association, Analysis examines AI in hospital billing as spike in complex patients adds nearly $1 billion in extra costs (September 24, 2026)
- Blue Health Intelligence and BCBSA, Rising Coding Intensity and Its Impact on Health Care Affordability (March 2026)
- American Hospital Association, Fact Sheet: Artificial Intelligence and Coding Intensity (August 2026)
- CMS, Announcement of Calendar Year 2027 Medicare Advantage Capitation Rates and Part C and Part D Payment Policies (April 6, 2026)
- HHS Office of Inspector General, Trend Toward More Expensive Inpatient Hospital Stays in Medicare Emerged Before COVID-19 and Warrants Further Scrutiny, OEI-02-18-00380 (February 2021)
- Journal of the American Medical Informatics Association, Adoption of artificial intelligence in healthcare: survey of health system priorities, successes, and challenges (May 2025)