[Policy Alert] State Insurance Commissioners Investigate Algorithmic Prior Authorization Denials

#Policy #Alert #State #Insurance #Commissioners #Investigate #Algorithmic #Prior #Authorization #Denials

AI in health insurance claims, why prior authorization is so difficult and how AMA fights to fix it by American Medical Association AMA

Title: AI in health insurance claims, why prior authorization is so difficult and how AMA fights to fix it
Channel: American Medical Association AMA

[Policy Alert] State Insurance Commissioners Investigate Algorithmic Prior Authorization Denials

[How-To] How To Conduct Automated Performance Testing On Synthetic Datasets Created For Model Training

[Policy Alert] State Insurance Commissioners Investigate Algorithmic Prior Authorization Denials

A major regulatory shift is unfolding across the United States. State insurance commissioners and federal regulators have launched formal investigations into major health insurance companies over their use of automated algorithms and artificial intelligence (AI) to review and systematically deny prior authorization requests.

For years, prior authorization was a slow, manual administrative process. Today, insurers use proprietary software to render decisions in seconds. However, mounting evidence suggests these algorithmic tools are being used to bypass human clinical review, resulting in high rates of erroneous denials for necessary medical care.

This policy alert breaks down the current regulatory landscape, the technologies under scrutiny, and what these investigations mean for healthcare providers and patients.

The Rise of AI and Algorithmic Prior Authorization

Prior authorization is a health plan cost-control process requiring providers to obtain approval before delivering specific treatments, tests, or medications. Historically, this required human medical directors to review clinical charts.

To cut costs and accelerate decision-making, major payers integrated AI-driven systems into their utilization management workflows.

How Algorithmic Decision-Making Works in Healthcare

In theory, AI prior authorization tools match a patient’s clinical data against established medical necessity criteria.

Data Input: The provider submits clinical notes, diagnoses, and treatment plans.
Algorithmic Analysis: The software compares this data against proprietary datasets and predictive models.
Output: The system either approves the care instantly or flags it for denial/further review.

The Human Cost: Automated Denials Without Clinical Review

The core issue driving regulatory investigations is not the technology itself, but how it is applied. Reports and class-action lawsuits allege that insurers use algorithms to reject claims en masse without human clinical oversight.

In many cases, the algorithms predict how much care a typical patient should need (e.g., a set number of days in post-acute rehabilitation) rather than evaluating the actual, individual medical needs of the patient. When patients require care beyond the algorithm's prediction, the system triggers an automatic denial.

Why State Insurance Commissioners Are Stepping In

Insurance regulation in the United States is primarily managed at the state level. State insurance commissioners are tasked with protecting consumers from unfair claims settlement practices.

Following high-profile media exposés and class-action lawsuits, regulators are investigating whether automated systems violate state laws requiring individualized medical reviews.

Key Investigations and Regulatory Actions by State

| State | Primary Regulatory Action / Investigation | Target of Investigation / Focus | | :--- | :--- | :--- | | California | Department of Managed Health Care (DMHC) investigation | Investigating the use of algorithms to deny care; evaluating compliance with state-mandated clinical review laws. | | Texas | Senate Committee hearings and Department of Insurance inquiries | Examining whether AI-driven denials bypass state laws requiring Texas-licensed doctors to perform reviews. | | New York | Department of Financial Services (DFS) circular letters | Warning insurers that AI tools cannot replace human clinical judgment and must be audited for bias and accuracy. | | National (NAIC) | National Association of Insurance Commissioners Model Bulletin | Establishing a regulatory framework for insurers using AI, focusing on transparency, governance, and accountability. |

Legal and Compliance Violations Under Scrutiny

Regulators are evaluating whether insurers using AI prior authorization tools are violating several key legal standards:

Duty of Good Faith and Fair Dealing: Insurers are legally obligated to thoroughly investigate claims. Batch-denying claims in seconds suggests a lack of individualized investigation.
Unauthorized Practice of Medicine: State laws often dictate that only licensed medical professionals can deny coverage based on medical necessity. If an algorithm makes the final determination, the insurer may be in violation.
Fiduciary Duty Breaches: Especially under Medicare Advantage plans, where insurers receive government funding to manage care, arbitrary denials may violate federal and state fiduciary responsibilities.

Inside the Tech: Prominent Tools Under Fire

Two specific algorithmic systems have become the focal points of recent regulatory and legal actions:

1. Cigna's PXDx (Procedure to Diagnose Exemplar)

An investigation by ProPublica revealed that Cigna used an algorithm called PXDx to review and reject thousands of claims at a time. The system allowed Cigna's medical directors to sign off on batches of denials in seconds, spending an average of just 1.2 seconds per claim. This bypassed the manual review process for low-cost tests, shifting the burden of appeal onto the patient and provider.

2. UnitedHealth's nH Predict (NaviHealth)

A class-action lawsuit alleges that UnitedHealth Group utilized the "nH Predict" AI tool to prematurely cut off post-acute care for elderly patients under Medicare Advantage plans. The lawsuit claims the tool had a known high error rate but was used to systematically deny rehabilitation and nursing home stays based on rigid, algorithmic projections rather than the patients' actual physical progress.

What This Means for Providers, Patients, and Payers

As state insurance commissioners intensify their investigations, all stakeholders in the healthcare ecosystem must adapt to changing compliance standards.

Actionable Strategies for Healthcare Providers

Healthcare organizations must defend their patients against arbitrary denials while protecting their revenue cycles.

Document for Human and Machine Audiences: Write clinical notes that clearly address standard medical necessity criteria. Use highly structured data, as algorithms struggle to parse vague or disorganized clinical narratives.
Establish an "AI Denial" Tracking System: Track which denials originate from automated platforms (like PXDx or NaviHealth). Spotting patterns helps in drafting targeted, highly effective appeals.
File Prompt Complaints with State Insurance Commissioners: Do not just appeal to the insurer. If an automated denial appears to violate state laws regarding clinical review, file a formal complaint with your state’s insurance department. This feeds regulatory investigations with real-world evidence.

Rights and Recourse for Patients Facing AI Denials

Patients have legal protections when dealing with algorithmic insurance denials:

Request the Internal Criteria: Under the Affordable Care Act (ACA) and state laws, patients have the right to request the exact clinical guidelines and algorithmic data used to deny their claim.
Utilize the External Review Process: If an internal appeal is denied, patients can request an independent external review. Independent medical examiners—not the insurer's algorithm—make the final decision, and insurers must abide by their ruling.

The Future of AI Regulation in Healthcare Insurance

The era of unregulated AI in healthcare administration is drawing to a close.

At the federal level, the Centers for Medicare & Medicaid Services (CMS) issued a final rule for 2024 clarifying that Medicare Advantage organizations cannot make coverage determinations based on algorithms or AI that override localized or national coverage determinations.

At the state level, expect to see more insurance commissioners issuing strict audit requirements. Insurers will likely be forced to prove that a qualified medical professional reviewed and signed off on every single medical necessity denial.

The takeaway: While AI has the potential to streamline administrative workflows, it cannot replace human clinical judgment. Providers and patients who understand their rights under these emerging state and federal policies will be best positioned to fight back against arbitrary automated denials.

[Case Study] Academic Medical Center Slashes Clinical Trial Protocol Amendment Delays Via Ehr Pre-Screening