Fact Sheet: Artificial Intelligence and Coding Intensity

August 2026

The Issue

Artificial intelligence (AI) tools are helping alleviate administrative burden for providers, particularly in areas like billing, coding and documentation. Ultimately, these tools can reduce documentation time, improve coding accuracy, expand appointment capacity, and enhance staff and patient satisfaction.

Recently, unsubstantiated claims have suggested that the use of these tools has increased coding intensity, leading to higher health care costs. However, these claims not only ignore the numerous underlying drivers of changes in coding intensity, including an aging patient population, but also overlook the legal, ethical and contractual obligations on hospitals to ensure appropriate coding.

AHA Take

Hospitals are increasingly caring for patients with higher acuity, a trend that is appropriately reflected in provider coding practices. Key drivers include an aging population, the increasing prevalence of chronic disease, and the continued shift of less complex care to outpatient settings.

Regulatory guidelines and coding conventions dictate that clinicians determine the level of care based on an evaluation of the patient, which includes incorporating the patient’s medical history. While some AI tools, such as AI scribes, help providers document their findings in the patient’s chart, providers have legal, ethical and contractual obligations to ensure appropriate coding.

In stark contrast, some commercial insurers have used unsubstantiated allegations of provider upcoding to arbitrarily and inappropriately reduce provider reimbursement for medically necessary care that has already been delivered. These “downcoding programs” often use automated edits to unilaterally reduce reimbursement (known as a downcoding or a partial denial) for higher-level care without reviewing medical documentation. In other words, some insurers are using automated tools to partially deny claims, forcing providers to engage in burdensome and costly appeals to be reimbursed. Such actions often result in the plan overturning its own initial partial denial, further evidence of the incredible administrative waste and burden these actions create in the healthcare system.

At the same time, many of the same insurers that have instituted such downcoding programs have themselves argued that their enrollee populations are sicker and in need of more services for purposes of their own payments.^1,2,3, In short, these plans want it both ways — a sicker enrollee population for purposes of health plan risk scores but a healthier one when it comes time to cover healthcare claims.

The AHA has urged the administration and Congress to prevent insurers from systematically reducing reimbursement under the pretense of widespread inappropriate upcoding, which often forces providers to engage in overly burdensome appeals processes simply to be paid for the care their patients need.⁴

Background

AI is transforming care delivery in countless ways, supporting increased access, improved outcomes and reduced costs. It offers tremendous potential to help reduce the more than $1 trillion the healthcare system spends annually on administrative functions.⁵

While more structured clinical notes and accurate coding should be common goals for insurers and providers, some commercial insurers have asserted that AI tools used by providers are inappropriately resulting in higher coding intensity. However, these accusations are unsubstantiated and ignore the multiple factors contributing to growth in coding intensity, including a sicker, higher-acuity hospital patient population.

Aging, Sicker Beneficiary Population. An aging population and the increasing prevalence of chronic disease continue to raise the level of complexity and intensity of hospital care. AHA data show 19% of hospital expense growth from 2019 to 2024 reflects caring for sicker, more complex patients, as hospitals devote more staff time, intensive monitoring and specialized treatment to each case.⁶ Furthermore, a recent AHA/Vizient analysis found that hospital case-mix index — a standard measure of how sick patients are — rose by about 5% between 2019 and 2024, indicating that a larger share of hospital care is devoted to higher-acuity patients with multiple conditions, greater clinical needs and longer stays.⁷ The continued rising prevalence of chronic diseases, such as heart disease, cancer and liver disease, also underscores the rising acuity of patients treated in hospitals.⁸
Continued Shift of Lower-acuity Care to the Outpatient Setting. At the same time, advances in medicine have enabled more routine and lower-acuity care to shift to outpatient settings, leaving hospitals to care for an inpatient population with greater clinical and resource needs. Together, these shifts have created a new normal in which many hospitals are treating a greater share of patients requiring intensive services, specialized staffing and around-the-clock capacity.
AI Tools Support Coding Accuracy. AI can support accurate coding and documentation. Hospitals maintain robust auditing and coding compliance programs to ensure they are appropriately capturing patient acuity levels and meeting regulatory requirements. Coding has always been governed by established coding guidelines, official conventions and regulatory expectations, which define a compliant and consistent approach to code assignment. The use of AI tools enhances precision and consistency within these existing frameworks, while human validation remains essential to ensuring coding integrity, regulatory compliance and accurate representation of patient complexity.
Changes in Coding Guidelines. Updates to evaluation and management coding guidelines, along with transitions in diagnostic coding systems, are altering coding patterns and increasing the reporting of more specific and higher-acuity diagnoses. As providers adapt and data stabilize, these changes contribute to shifts in coding intensity.

Assertions of provider upcoding are particularly striking given the substantial and growing body of evidence involving insurer-driving coding practices. Ironically, lawsuits, investigations and reports from the Medicare Payment Advisory Commission (MedPAC), the Department of Justice (DOJ) and Congress have documented upcoding practices among many of the large insurers, like adding diagnosis codes that may not be supported by clinical care. In 2025, MedPAC reported that upcoding contributed to $40 billion in overpayments to Medicare Advantage plans.⁹ In March 2026, one payer settled a lawsuit with the DOJ on upcoding allegations.¹⁰ The state of Massachusetts filed a lawsuit in May 2026 against another large commercial payer for potentially fraudulent upcoding practices.¹¹ The fact that commercial insurers are now attributing upcoding concerns to providers is at odds with a substantial body of evidence documenting plan-led upcoding practices.

Resources

End Notes

¹ unitedhealthgroup.com/content/dam/UHG/PDF/investors/2025/UNH-Q2-2025-Form-10-Q.pdf

²https://s202.q4cdn.com/665319960/files/doc_financials/2024/q3/ELV-^{USQ_Final-Transcript_2024-10-17.pdf}

³ investors.molinahealthcare.com/news-releases/news-release-details/molina-healthcare-reports-second-quarter-2025-financial-results

⁴ aha.org/testimony/2025-07-22-aha-statement-house-ways-and-means-committee-hearing-medicare-advantage

⁵ “Active steps to reduce administrative spending associated with financial transactions in US health care,” Sahni, N., et. al., Health Affairs Scholar, Volume 1, Issue 5, November 2023, qxad053, doi.org/10.1093/haschl/qxad053

⁶ aha.org/system/files/media/file/2026/03/Costs-of-Caring-2026.pdf
⁷ Ibid.

⁸trillianthealth.com/hubfs/2025%20Trends%20Shaping%20the%20Health%20Economy%20Report%20%7C%20Trilliant%20Health.pdf

⁹medpac.gov/wp-content/uploads/2025/01/Tab-M-MA-status-report-January-2025-SEC.pdf

¹⁰ justice.gov/opa/pr/aetna-agrees-pay-1177-million-resolve-false-claims-act-allegations

¹¹ mass.gov/news/ag-campbell-sues-united-healthcare-for-defrauding-masshealth-out-of-100-million