
Scholar argues that FDA has the authority to regulate coverage algorithms used by health care insurers.
The recent murder of Brian Thompson, the former CEO of UnitedHealthcare—the largest health insurance company in the United States—has sparked widespread debate about the practices of insurance companies.
One practice now under greater scrutiny is health care insurers’ use of artificial intelligence (AI) algorithms to deny patient coverage.
In a forthcoming article, Jennifer D. Oliva, a professor of law at Indiana University Maurer School of Law, argues that use of unregulated coverage algorithms leads to improper claim denials and delays in patient care, resulting in deadly consequences for patients. She contends that recent federal efforts to reform these algorithms are inadequate because they fail to address critical issues, such as the lengthy appeals process, and only apply to specific government programs.
Oliva argues that the U.S. Food and Drug Administration (FDA) has preexisting authority to regulate coverage algorithms and urges the agency to take action.
In the United States, health care providers are compensated through a fee-for-service system, meaning they are paid for each service or procedure they perform. Insurers argue that this model encourages volume and high-cost services, cutting into their profit. As a result, health care insurers have subjected a growing number of medical procedures and treatments to a mandatory review process before care is administered.
These manual reviews, however, are time-consuming, prone to human error, and resource intensive, which has resulted in treatment delays for patients.
To control costs and minimize delays, some insurers have designed AI algorithms to automate this process.
Before care is provided, insurers use these algorithms to determine whether a specific, provider-recommended course of treatment is “medically necessary” and whether that treatment qualifies for coverage under the patient’s insurance plan, explains Oliva.
Oliva notes, however, that the adoption of AI technology has led to a significant increase in claim denials. According to a report by the Senate Permanent Subcommittee on Investigations, UnitedHealthcare’s denial rate for post-hospital care more than doubled between 2020 and 2022 after it implemented algorithms to automate its review process.
Many of these coverage denials are erroneous and illegal, claims Oliva. Oliva highlights a class action lawsuit against UnitedHealthcare alleging that its AI algorithm wrongfully denied claims. The complaint states that approximately 90 percent of UnitedHealthcare’s denials were overturned on appeal by federal administrative law judges.
Oliva explains that these denials are devastating for patients, who are forced to either endure a lengthy and costly appeals process or pay out-of-pocket—an option that is unfeasible for the average American. Insurers have a financial incentive to deny claims since less than one percent of patients appeal and others would not survive long enough to complete the process, Oliva points out.
Yet, recent measures to regulate these algorithms have been inadequate, argues Oliva.
The Centers for Medicare & Medicaid Services (CMS) issued a rule mandating that coverage determinations must “account for” the patient’s individual circumstances, but it permits Medicare Advantage Organizations to use algorithms to “assist” them in making these determinations.
Oliva argues that this rule lacks crucial guidance for insurers implementing coverage algorithms. For example, it does not specify what it means to “use” algorithms or how insurers should “account for” individual circumstances in their review process, states Oliva.
CMS also issued a rule aimed to streamline the prior authorization process and improve transparency. The rule requires insurers to disclose which medical items and services require prior authorization, provide a specific reason for any denial to providers and patients, and report information about prior authorization determinations.
Oliva contends that this rule fails to address many longstanding issues in the review process. For instance, it does not clarify how insurers decide which services require prior authorization, the criteria used for these determinations, or how to address the burdensome appeals process, explains Oliva.
In addition, CMS rules do not apply to employer-sponsored health plans, which covered 60.4 percent of people under 65 in the United States in 2023, states Oliva.
Oliva argues that more robust oversight of these algorithms is needed “to ensure their accuracy, validity, and fairness” before they are used by health care insurers. Oliva contends that FDA already has the regulatory authority to do so.
Under the Food, Drug, and Cosmetic Act, FDA is granted both pre-market approval authority and expanded post-market regulatory powers over “medical devices.” Oliva contends that coverage algorithms qualify as “medical devices” because they make health care treatment decisions by accepting or declining coverage for patients.
Some medical devices, however, are exempt from agency oversight if a provider can “independently review the basis” for the algorithm’s coverage denial. To qualify for this exemption, the manufacturer must describe “the inputs used to generate the recommendation” and “the basis for rendering a recommendation,” among other details. Oliva notes that insurers claim that their algorithms are proprietary, preventing them from satisfying these transparency requirements.
Even if insurers were able to meet the requirements of this exception, they would still be subject to FDA’s authority if the algorithms are “reasonably likely to have serious adverse health consequences.” Given the significant health outcomes patients face from treatment denials, these coverage algorithms would not qualify for exemption, claims Oliva.
Facing some of the highest health care costs in the world, Americans that are denied insurance coverage could face life-or-death consequences. Oliva warns that coverage algorithms may exacerbate the misalignment between profit-driven insurance companies and patients, who depend on timely and appropriate care. Oliva calls for FDA oversight of these algorithms to ensure that they are accurate and fair before being used to make life-changing determinations.