While artificial intelligence promises exciting efficiency benefits, its usage in insurance prior authorizations demands wider and more holistic oversight from state and federal bodies, a trio of experts argued this week.
Long-term care providers and consumer advocates alike have increasingly raised the alarm about insurers using AI to determine whether patients can receive care. Denials of Medicare Advantage prior authorizations and allegedly premature termination of coverage have drawn scrutiny and even class action lawsuits over the last year.
Writing in an article for Health Affairs Monday, legal and ethics experts from the sector joined the call for more government oversight. While acknowledging the efforts of some insurers to self-regulate, they argued that standardized oversight will be necessary to ensure fair usage of AI across the board.
“While partnership with industry is crucial for these efforts, self-regulation will not be enough to protect patients, and regulators must come up with clearer rules of the road that encourage innovation but put in place guardrails against abuses,” wrote the coauthors — Carmel Shachar, assistant clinical professor of law at Harvard Law School; Amy Killelea, senior associate in the Health Policy and Management Program at Johns Hopkins University; and Sara Gerke, associate professor of law at the University of Illinois.
‘Garbage in, garbage out’
The authors noted that AI can eliminate up to 75% of tasks needed to process prior authorization requests. This significant efficiency improvement could help both payers and providers save time and money, they explained.
However, since AI results are limited by the quality of the data being input, mistakes or bias in that data can have significant negative effects. Some algorithms have been shown to discriminate against marginalized communities — for example, assigning sick Black people for less additional care than similarly sick white people.
Such biased outputs from algorithms could open up insurers to lawsuits under the anti-discrimination provisions of the Affordable Care Act, the researchers noted. Especially if AI is directly making choices about a patient’s coverage.
The efficiency savings of AI could also have the additional side-effect of creating “review creep,” the authors said.
“Because AI software can process claims quickly and is cheap to use per claim,” they wrote, “insurers can extend PA [prior authorizations] and coverage review to lower-cost procedures and treatments. This suggests that we may see ‘review creep’ as insurers increasingly use their new tools on an ever-growing catalogue of services and treatments.”
In short, AI’s greater efficiency could give insurers the option to review — and reject — more appeals for coverage.
A collaborative roadmap
The researchers noted that past advocacy pressure and policymaker concern on this issue has resulted in some regulatory steps already. Those steps are inconsistent, however, and still leave a great deal of gray areas open.
“There are still many regulatory gaps and opportunities for variability across insurance products and states, especially when it comes to regulating more substantive elements of PA,” the authors explained, later adding that “there needs to be more proactive federal and state oversight of AI used by payers, particularly use of AI to make utilization management and other coverage determinations.”
The authors pointed to Colorado as an example that other states can follow. The state has already regulated AI use in life insurance and is currently in the process of considering how to apply regulations to health insurance.
California’s Senate has also passed a bill that would require physicians to review AI usage in prior authorizations.
In April, five federal agencies pledged to review AI’s potential effects on civil rights — including, most relevantly for providers, the Department of Health and Human Services. This collaborative approach from regulators is an important first step to a complex problem, the authors noted — one that promises to continue to impact senior care as the benefits and pitfalls of AI continue to reveal themselves.
