Better Information, Better Care: How Coding Supports Today’s Complex Patients

Correctly assessing the clinical status of a patient is imperative to delivering high-quality care. Patients today are older and more clinically complex, and new artificial intelligence tools are helping providers appropriately capture their patients’ conditions to aid in care planning.

A report recently released by the Blue Cross Blue Shield Association suggests that hospitals’ use of AI-enabled documentation tools is driving higher healthcare costs through increased coding intensity. Its analysis estimates that coding-related changes resulted in $942 million in additional spending for BCBSA plans between 2023 and 2025 and implies that this spending may be wasteful. However, BCBSA’s analysis lacks the context needed to meaningfully assess how these tools impact healthcare quality, patient access or spending.

Hospital Patients Are Getting Sicker and More Complex

BCBSA’s analysis starts with the observation that a growing share of hospital inpatients are being classified as clinically complex. It then largely attributes that trend to coding behavior while overlooking several well-documented factors affecting the patients that hospitals treat every day.

Americans are living longer and managing more chronic conditions than in the past. At the same time, advances in medicine and changes in coverage policies have shifted many lower-acuity procedures to hospital outpatient departments and other ambulatory care sites.¹ As a result, the patients who remain hospitalized are increasingly those with greater clinical and resource needs.

Those trends are reflected in hospital case-mix data and broader population health indicators, including data from the federal government.² They help explain why hospitals are documenting and reporting more complex patient conditions and why increased complexity should not be assumed to reflect inappropriate coding.

Better Documentation Supports Better Care

Technology can help clinicians comprehensively and accurately capture and organize information already documented in the medical record, reducing administrative burden while supporting care planning and coordination. These technologies can also help identify conditions that may previously have been underreported despite being supported by clinical documentation. As a result, they can improve medical record accuracy and provide a more complete picture of a patient’s severity of illness, resource needs and clinical risk.

BCBSA points to a perceived disconnect between increasing coding complexity and relatively stable treatment patterns to suggest the coding changes are inappropriate. But the absence of an additional procedure does not establish that a diagnosis is invalid. A comorbidity can affect clinical decision-making, monitoring, nursing resources, medication management, prognosis and risk without necessarily resulting in a transfusion, surgery or other easily identifiable service. Ultimately, the appropriate test for whether a patient has been correctly coded is not whether reimbursement increased; it is whether the patient’s medical record supports the diagnosis.

Importantly, compliance safeguards exist. Hospitals continue to rely on provider documentation requirements, coding standards, clinical review processes, audits and other mechanisms to ensure that reported diagnoses are clinically supported and meet applicable reporting requirements.

BCBSA’s Analysis Does Not Show That Diagnoses Were Unsupported

The central question is not whether hospitals are reporting more diagnoses. The question is whether the patient’s medical record supports those diagnoses.

BCBSA’s analysis insinuates conclusions about AI-enabled coding without reviewing the underlying medical records to determine whether reported diagnoses were clinically supported. It also does not appear to identify which claims involved AI tools rather than traditional documentation and coding processes. Nor does it fully account for changes in patient morbidity and the continued migration of lower-acuity care away from inpatient settings.

Without answering those questions, the analysis does not demonstrate that hospitals coded patients incorrectly. It may simply reflect more complete documentation of increasingly complex patients.

The $942 Million Figure Lacks Important Context

Even if accepted at face value, BCBSA’s estimate appears to account for only a small fraction of commercial healthcare spending and premium growth. Employer health insurance premiums have been increasing roughly 5% to 7% annually. BCBSA’s September white paper does not say how large its $942 million estimate is relative to Blue plans' overall spending; the figure is a cumulative total over two years. However, BCBSA's own earlier analysis provides a way to gauge its scale.

BCBSA’s March 2026 issue brief estimated that per-member commercial inpatient costs among participating Blue plans rose about 9% between 2023 and 2024, and attributed roughly 20% of that increase (about 1.8 percentage points) to rising coding intensity.³ Because inpatient facility care accounts for about 17% of employer-sponsored health spending,⁴ even BCBSA’s own estimate implies inpatient coding intensity contributed only about 0.3 percentage points to total commercial spending growth, compared with a 7% increase in employer family premiums in 2024.⁵ That is likely why, when discussing premium increases, BCBSA itself recently pointed to other cost drivers, in particular, prescription drugs.⁶

Clinical Documentation Should Be the Standard for Both Insurers and Hospitals

There also is an important question of consistency. Blue plans themselves participate extensively in the Medicare Advantage program, where more complete diagnosis capture can increase the payments health plans receive through risk adjustment. Plans have been known to contract with third parties to conduct in-home health risk assessments,⁷ encourage providers to collect and submit complete diagnoses using technologies such as “voice translation software to improve accuracy and efficiency,”⁸ and conduct retrospective medical-record reviews to identify additional conditions affecting beneficiary risk scores.⁹

Federal auditors have repeatedly identified unsupported diagnoses submitted by individual Blue plans for Medicare Advantage risk adjustment. In a recent audit of BlueCross and BlueShield of Alabama, for example, the Department of Health and Human Services Office of Inspector General found unsupported diagnoses in 247 of 271 sampled enrollee-years.¹⁰ Separately, the Department of Justice has alleged that Anthem — now part of Elevance Health, a BCBSA member — used retrospective chart reviews to identify diagnoses that increased Medicare Advantage payments while failing to delete diagnoses when reviews did not substantiate them.¹¹

These examples do not establish that all insurer diagnosis capture is inappropriate. Likewise, increased diagnosis capture by hospitals should not be presumed improper simply because it affects reimbursement. The appropriate standard in both cases is whether the diagnosis is clinically supported by the medical record.

Conclusion

Accurate diagnosis capture is not a coding problem; it is a patient care imperative. As providers adopt technologies that help clinicians document patient conditions more completely, the key question remains whether diagnoses are clinically supported, not whether they increase measured complexity. BCBSA’s analysis raises concerns about coding trends, but it does not demonstrate that hospitals are coding patients inaccurately or that AI tools are driving inappropriate spending.

Notes

Cost Management

Making Healthcare More Affordable