Primary sources: NAIC Regulatory Framework (B) Task Force materials, November 16, 2026, Dallas. Blue Health Intelligence issue brief, March 2026. BenefitsPRO, Allison Bell, November 18, 2026.

Blues Counted at Least $5.5 Billion in Coding Intensity. The Hospitals Say the Patients Are Sicker.

On November 16, 2026, in Dallas, Luke Chalker of the Blue Cross Blue Shield Association told state insurance regulators that coding intensity increased costs for patients and employers by at least $5.5 billion in 2025, extrapolated across the Blue commercial book. If you run coding or CDI, that is a payer pricing the same secondary diagnoses a CMI lift is built from. It is BCBSA's estimate. It is not a finding that the care was upcoded.

At least $5.5 billion

BCBSA's 2025 figure, on slides in the NAIC Regulatory Framework (B) Task Force packet. At least $942 million is inpatient. At least $4.6 billion is outpatient and professional. The deck is marked proprietary and confidential for Blue Plan use only. It was posted as public meeting material.

What the November 16 slides count

Agenda item 3, "Discuss Claims Coding Issues," ran from 3:00 to 4:00 p.m. at the Dallas Convention Center. The agenda names Chalker, senior vice president of product and data science at BCBSA; Jennifer Holloman of the American Hospital Association; and Eric Ellsworth of Consumers' Checkbook. The confidentiality line is on the deck. The packet is still a public NAIC file.

BCBSA's 2025 lineWhat the slide says it is
At least $942 million, inpatient admissions$653 million inpatient complexity and $289 million inpatient DRG migration.
At least $4.6 billion, outpatient and professional$1.1 billion in E&M leveling on emergency and office visits, $899 million in NCCI edits, $2.5 billion in endoscopic unbundling, and $309 million in add-on code billing.
55,158 cases, national Blue System data, 2025 against 2023Cases that "may not have been coded as complex under 2023 baseline rates."

The September 2026 whitepaper in the same packet drops the hedge: 55,158 additional complex cases, $653 million, $11,800 each, and 70 percent of the incremental inpatient coding-intensity cost. Same count. Both wordings are BCBSA's.

The DRG picture is labeled as an illustration, not an average paid claim. A respiratory-infection admission, DRG 179, weight 0.755, is shown at a typical $20,000. The same stay, once generative AI surfaces sepsis evidence and the case moves to DRG 872, weight 1.0233, is shown at $33,000. For major bowel procedures, DRGs 329–331, hospitals in the top quarter of coding growth coded 75.6 percent of 2025 cases as complex, against 65.0 percent elsewhere, with lower ICU use (11.5 percent versus 13.2 percent) and the same 4.0-day median stay. BCBSA reads that as coding without more care.

A separate March 2026 Blue Health Intelligence brief, which the slides cite, looks at Plans with about 62 million members and finds an approximately 9 percent per-member rise in inpatient costs from 2023 to 2024, with about 20 percent of that rise attributed to coding intensity without more care. The brief says it is not a finding on appropriateness or intent. That is not the $5.5 billion. Do not add them.

The deck says that, with an early focus on identifying coding patterns, BCBSA is monitoring broader market shifts toward proactive surveillance, empowered local decision making, and more real-time interventions, through claims-level clinical evidence. That is a monitoring sentence. It is not an audit protocol.

The hospital answer in the same packet

The AHA slides say patients are older and more clinically complex, that hospitals are obligated to code completely, and that AI tools help capture conditions for care planning. Their 2019–2024 expense split for inpatient prospective-payment hospitals is 36 percent more patients, 19 percent sicker patients, and 45 percent higher input costs, using AHA admissions and expenses plus CMS IPPS impact files. BenefitsPRO reported on November 18 that an AHA executive said patients really are sicker. A longer quotation was not on the page retrieved here, and none is added. The $5.5 billion stays an allegation.

RevCycleAI analysis: the CMI slide and this $5.5 billion are the same codes

This is RevCycleAI's read. Payers are willing to put AI coding on a cost line in front of regulators, and the lines they chose are the audits you should expect. Inpatient complexity and DRG migration are clinical validation and DRG validation. The $1.1 billion is E&M level downcoding. The $899 million is NCCI enforcement. The $2.5 billion is endoscopic unbundling. The $309 million is add-on codes. A vendor's CMI-lift claim is the mirror image of this number. One side books found revenue. The other books cost. Neither figure tells you whether the chart supports the code.

Ellsworth's slides, on the same agenda, list what a consumer advocate thinks payers may do: more edits, automatic downcoding, physician sign-off on AI-coded claims, more chart audits, and more prior authorization on the tests that feed higher-intensity codes. That list is his, not a BCBSA work plan. The operating response is narrower. Document the clinical support for the secondary diagnoses that move the DRG, sepsis included, and track DRG-downgrade denials as their own queue. A code that exists because a model found a lab value, with no treatment and no note, is the case the bowel comparison is built to flag.

The same chart shows up on the Medicare side in the CRUSH proposal and in the fiscal year 2026 improper-payment fact sheet, which says most Part C improper payments are diagnosis data the record did not support. Those are federal measurements. This $5.5 billion is a commercial Blues estimate.

The RCAI signal

At least $5.5 billion is BCBSA's 2025 extrapolation, not a proven upcoding total. Inpatient is at least $942 million. Outpatient and professional is at least $4.6 billion. The AHA's answer, in the same packet, is that patients are sicker. Keep both.

RCAI take

Treat the $5.5 billion as the payer's cost case, and treat the secondary diagnosis as the work. Still open: a hospital-level list, a paid-claim distribution behind the $20,000 and $33,000 illustration, and any review of the 55,158 cases that is not BCBSA's. Your denial log will show which story your payers are acting on.

Sources: NAIC Regulatory Framework (B) Task Force materials, 2026 Fall National Meeting, Dallas, November 16, 2026, 3:00–4:00 p.m. (PDF text opened). Agenda item 3 names Luke Chalker, Jennifer Holloman, and Eric Ellsworth. The BCBSA deck, "The Impact of AI Technologies on Coding Intensity and Healthcare Costs," is marked proprietary and confidential for Blue Plan use only. The $5.5 billion, the $942 million and $4.6 billion split, the $653 million and $289 million, the $1.1 billion, $899 million, $2.5 billion, and $309 million component lines, the 55,158 "may not have been" wording, the DRG 179 / 0.755 / $20,000 and DRG 872 / 1.0233 / $33,000 illustration, the bowel comparison, and the monitoring sentence are the slides'. The September 2026 whitepaper in the same packet, "Hospital Coding Intensity Analysis: Major Bowel Procedures," is the source for the "additional cases" wording, the $653 million, the $11,800 average, and the 70 percent. The AHA slides are the source for the sicker-patient language and the 36 / 19 / 45 split. Ellsworth's potential payer responses are his slides. Blue Health Intelligence issue brief, March 2026 (PDF opened). The 62 million members, the approximately 9 percent, the about 20 percent, and the sentence declining to determine appropriateness or intent are the brief's. BenefitsPRO, Allison Bell, November 18, 2026, 3:55 p.m. (page opened). The summary box says an AHA executive said patients are really sicker. A longer quotation was not retrieved, and none is added. An InsuranceNewsNet item on this meeting was not located and is not used. The RevCycleAI analysis is our read.