Yolanda Troublefield, MD, JD, attending surgeon and otolaryngologist at Southcoast Physicians Group in North Dartmouth, Mass., and a member of the Physician Payment and Policy Workgroup of the American Academy of Otolaryngology–Head and Neck Surgery, recently saw an eight-year-old child with recurrent otitis media.
Explore This Issue
August 2026It was time, clearly, for an adenoidectomy with ear tube placement.
But when Dr. Troublefield submitted the procedure for prior authorization through the health insurance company, it was denied. The recurring infections were well-documented by both primary care and otolaryngology, complete with an audiogram. What’s more, the patient’s parents are also doctors, so they are familiar with the standard of care.
“We’ve been doing tonsils and adenoids and ventilation tubes for the last 50 years—seems pretty straightforward,” she said. “Every single ‘i’ has been dotted, ‘t’ has been crossed.”
But none of that mattered when it came to the request for insurance coverage. And Dr. Troublefield has little doubt that it has to do with the involvement of artificial intelligence (AI). The large language model being used seems to look at only one part of the note, but not the whole note. Although “recurrent otitis media” was the diagnosis, the assessment-and-plan part of the note read “acute otitis media,” which was meant to identify the need for an antibiotic at that particular visit. So Dr. Troublefield would need to move on to the next step, and possibly to a peer-to-peer conversation, which would disrupt her schedule and cause delays in seeing patients.
Denials at Machine Speed
The dawn of AI in the realm of prior authorizations and clawbacks for simpler services already rendered, otolaryngologists say, has led to denials at an ultra-rapid clip, sometimes in perplexing fashion, obliging physicians to do more work to get patients coverage for the medical care they need, and leaving them more exasperated all the time. Health practices are increasingly using AI to submit documentation to insurance companies and to respond to denials, leading to automated exchanges that will have unknown effects on the healthcare system.
“What we’re seeing is faster denial of claims because of the use of AI,” Dr. Troublefield said. “You can set the dial in different ways. You can set it high, medium, or low. And they’re set high for denials.”
In an American Medical Association physician survey last year, 60% of physicians said they are concerned that AI has already started to increase or will increase prior authorization denial rates. Fifty-five percent said prior authorization delays access to necessary care either always or often, and 79% said prior authorization sometimes leads to abandonment of even pursuing treatment. A notable 26% reported that prior authorization has led to a serious adverse event for a patient in their care (AMA. https://tinyurl.com/5v52dscr).
“I think physicians were very hopeful that AI was going to be a boost, a solution if you will, to reduce the administrative burdens that we all face,” said Bruce Scott, MD, a former president of the American Medical Association and an otolaryngologist at ENT Care Centers in Louisville, Ky.
Unfortunately, that has not panned out, he said.
“Eventually, their bot is going to talk to my bot. Because my bot is going to tell me how to document so I get authorization, and then their AI is going to get even smarter and deny that, so I get a denial letter back. And in the meantime, the physicians and patients are stuck in the middle, and that’s the problem.”
Regulators Enter the AI Fight
A flurry of new state laws has taken effect, or soon will, to regulate the use of AI in the denial of claims. An Alabama law requires insurers to base determinations on a patient’s unique circumstances and not rely on group datasets. An Indiana law prohibits insurers from using AI as the sole basis for downcoding a claim. And a Washington state law requires that determinations only be made by licensed and qualified health professionals. Those are just a few.
But the government will also begin deploying AI to provide its own health insurance through Medicare. The WISeR program—for Wasteful and Inappropriate Service Reduction—will put AI to use, along with human review, to “reduce clinically unsupported care by working with companies experienced in using enhanced technologies to expedite and improve the review process for a pre-selected set of services that are vulnerable to fraud, waste, and abuse,” according to the Centers for Medicare and Medicaid. The program was launched in January in six states: New Jersey, Ohio, Oklahoma, Texas, Arizona, and Washington (CMS. https://tinyurl.com/2wprefe7).
Fighting Automation with Automation
The use of AI by insurers has left physicians with little choice but to use the technology themselves, said Bradford Bichey, MD, a rhinologist at Indiana Sinus Centers who has developed an AI product to streamline physicians’ office operations, including documentation for prior authorization submittals.
A few years ago, Dr. Bichey said, he started noticing massive clumps of clawback attempts, or requests for documentation for cases insurers suggested should not have been covered. Suddenly, he was getting 50 such requests, when the norm used to be about five at a time. And the language was strange, suggesting it was “early experimentation” with the use of AI.
“The verbiage was slightly off,” he said. “I got one letter that was almost hateful.” These letters have since gotten more professional in their tone, he said.
In response, Dr. Bichey developed Blue, an AI product under his corporation, Nemedic, that produces not just a clinical note with AI, based on the clinical encounter with the patient, but also a prior authorization document tailored to the specific insurer for that particular indication.
For instance, once the word “United” is spoken during a clinical visit, the software will take that as a cue to prepare a document tailored to the requirements of United Healthcare, as determined by the insurer’s Web-available documents. The prior authorization submission requires a full justification, with much more detail than is in a clinical note. For a nasal endoscopy, for example, a document reads “routine anterior examination was insufficient to visualize the deeper nasal passage and assess for ongoing mucosal disease. Diagnostic rigid nasal endoscopy was medically necessary to direct visualization of the sinonasal mucosa,” and so on. This note is also careful to make note of exclusions, such as “this procedure was not performed for routine screening, but to inform management.”
“You have to know [the insurer’s] inclusion and exclusion,” Dr. Bichey said. “And it’s almost impossible for someone who does billing to constantly keep up with all these plans because we see hundreds of different types of insurance.”
Handling an appeal is typically simple, he said, because it’s usually a matter of pointing to something that was already in the documentation that was submitted but was overlooked or ignored.
“When we do this, we just resubmit the same note and say, ‘You’re wrong—look at paragraph three,’” Dr. Bichey said. He said he hasn’t had to have a peer-to-peer call in two years, due to the dramatically improved documentation up front.
Not an AI Problem, but a Governance Problem
While it can seem that AI itself is leading to problems with unnecessary denials, that is not really the case, said Matthew Crowson, MD, assistant professor of otolaryngology–head and neck surgery at Harvard Medical School and director of clinical informatics and artificial intelligence at Massachusetts Eye and Ear.
“It’s not so much that AI is the problem,” he said. “It’s more the governance around this stuff, or lack thereof.”
He said that AI hasn’t resolved the dynamic of provider versus payer.
“It’s turning into a battle of the bots,” he said. “The application of AI by providers and payers is cutting both ways. It doesn’t resolve the fundamental tension between approving and denying based on medical necessity. It’s scaling up existing decision-making workflows faster.”
In some respects, AI has improved administrative efficiency in ways that help patients, he said. For instance, if a certain ear drop is recommended for an infection and it is not on the formulary for a given health insurance plan, a nurse or medical assistant would have needed to dig up the relevant supporting information and fill out a form to appeal for coverage. Using AI, a contextualized appeal can be drafted very quickly, which may get medications into patients’ hands more quickly.
“What would take a human process maybe seven business days to do, now you can do it in one,” he said. “I think administratively, it’s massively speeding up our ability to respond quickly.”
Efficiency Gains—or Just More Activity?
A report by the Peterson Health Technology Institute earlier this year—based on workshops involving senior leaders from healthcare systems, health plans, technology developers, investment firms, and federal agencies—painted an unclear picture of the effects of AI on the health of the U.S. medical landscape itself (Peterson Health Technology Institute. https://tinyurl.com/yc8d8dm9).
For one thing, the report says, AI may reduce the cost for individual organizations to get prior authorizations done, but it has not reduced overall costs across the system. It also says that the deployment of AI by providers is increasing “billing intensity,” with more severe diagnoses and advanced treatments identified, with accompanying higher costs.
Workshop participants, the report says, worried that optimized AI use on both the insurance and provider side runs the risk of making the whole process “more activity-intensive,” but not really more efficient.
“Bots don’t get tired of asking questions, so my review queue keeps growing,” said one healthcare provider, according to the report.
Patients Caught in the Middle
Meanwhile, the prior authorization process continues to strain the relationship between physicians and their patients, otolaryngologists said.
Dr. Troublefield said that complex conversations with patients regarding insurance coverage requirements and hurdles are a major factor with patients every day.
“It takes more time—I don’t think any of us runs on schedule at all. I would say five years ago, we probably could run on schedule. Not anymore. It never, ever, ever happens,” she said. “We’re finding that we’re constantly having to apologize, apologize, apologize. And that’s not the way people want to be treated.”
Dr. Scott said a patient with a tumor growing in the maxillary sinus had been hesitant about surgery but had finally accepted that there was a need for it. Then the patient received a letter from the insurer saying that the procedure was denied because she had not yet been on an antibiotic, which Dr. Scott said was a preposterous assessment of the situation.
But then, the patient “called and said, ‘The insurance company said I might get better on an antibiotic, so shouldn’t we try an antibiotic?’
“Now I’ve got to go explain to this patient, and I’m thinking about the trust relationship that I had with this patient and having to re-establish that trust.”
He said physicians should appeal more often when claims are denied—in his practice, every denial is appealed. But he understands that burnout leaves physicians feeling they cannot go to battle every time.
He said he and the AMA are also pushing for physicians to be involved in the development of regulations and the processes surrounding AI.
“Will it work in my practice? Will I get sued? Will I get paid?” he said.
A Fork in the Road for Prior Authorization
Dr. Crowson said the field is approaching a fork in the road: One path is that providers and payers both have AI to process mundane paperwork and accomplish basic tasks faster, with humans then having more time to focus on complex cases. The other path is an “escalating compliance and gaming war.”
He hopes for a “pragmatic, middle-of-the-road approach—use it for what AI is good for: automating administrative monotony and other simple tasks that can be safely and efficiently scaled.”
Dr. Troublefield said she is worried that using AI will just add “another hoop” for physicians to jump through if they need to use certain buzzwords in conversations with patients. But she said that software that automatically creates templates based on insurer requirements is an attractive idea if it is practical enough to use.
She said she is somewhat hopeful that technology could be an answer and that smart legislation—such as regulating AI involvement in insurance decisions—could also be helpful. More patient voices—and even celebrity voices—would be a major help in the legislative effort as well, she said. But, without question, it will be up to physicians to change with the times.
“The system is changing, and either you have to be part of the change, or you’re going to get left behind,” Dr. Troublefield said. “It’s the same thing as when the automobile happened, right? You’re not stopping it, so you have to figure out ways within the system to make sure that what you want, and what your goals are, are actually achieved.”
Thomas R. Collins is a freelance medical writer based in Florida.
Leave a Reply