​​Physician-Led Tech Firms Use AI to Counter Drug Denials and Augment Appeals

Prescription drug denials are rising, complicating treatment for millions of patients and adding new burdens for already stretched providers and hospital staff. According to data from Komodo Health, pharmacy claim denials increased by 16% between 2018 and 2024, reversing a previous downward trend and affecting access to essential medications for chronic conditions like diabetes, asthma, and blood clots. Pharmacy claim denials initially declined between 2016 and 2018, but data since then have indicated a rising trend of denials.

The challenges posed by these denials reflect broader systemic issues that come up in the relationships between hospitals, pharaceutical companies, and pharmacy benefit managers (PBMs), where practices like step therapy often conflict with clinical guidelines. The American Journal of Managed Care® (AJMC®) spoke with experts about the importance of optimizing data to keep up with these changes and implementing artificial intelligence (AI) to reduce the burden on providers when confronted with increasing denials and prior authorizations while ensuring their patients’ access to care and treatment.

This transcript has been lightly edited; captions were auto-generated.

Transcript

Arif Nathoo, MD, CEO and cofounder of Komodo Health:

There's no question that we're seeing an increased rate of denials for even common medications. Now, on the branded-drug side, you mentioned something like Eliquis (apixaban; Bristol Myers Squibb, Pfizer), and so these are the kind of bigger battles and contracts of relationships that PBMs have with pharma companies.

Arif Nathoo, MD, Komodo Health

Data from Komodo Health patient records show that claim denial rates increased by 16% from 2018 to 2024.

This particular case, there's a lot of switching of patients that were supposed to get Eliquis to another therapy based on a certain contract between a pharma company and a PBM. And so these sorts of things are actually driving patients to different therapeutic options, and then the real question is, are patients getting the right standard of care, or by making that switch, are patients feeling worse outcomes? And so there's a lot of research to suggest that not all of these switches are in the best interest of the patient. They're in the best interest of the pocketbook.

Brand name and generic drugs for asthma and allergy, diabetes, hypoparathyroidism, and blood clots were among the most rejected.

The question for us is always, when we see that happening at scale, can we actually trace the outcomes and see whether these patients are having better morbidity or mortality effects, or are they having worse ones as a result of the switch? And so there's a lot of research that's being done there right now, but even just looking at these denials, it tells you that you know there is work to be done when we see that switching at scale.

The increase in denied claims has an impact on physician productivity as well as disease management and outcomes for patients.

Christopher Gold, DO, internal medicine physician at Mount Sinai:

Clinically: delays and initiations of treatments that we're trying to start, or any lapses in treatments if they're already on something, and either there's a denial for the medication already on, or switches in medications can lead to suboptimal management of these medical conditions. It can lead to exacerbation of symptoms, and could potentially lead to complications that could have been avoidable, or emergency visits and more frequent medical office visits.

Christopher Gold, DO, Mount Sinai

Financially, it can affect patients if there are higher costs for them, whether that's tier pricing or out-of-pocket costs for things that they were referred, or things that they need, obviously, we'd have an increased financial burden for patients. If patients do have to pay a higher cost, and they don't have the money to do so sometimes they do try to ration medications, potentially, where they try to use them less often, or at suboptimal dosing to make them last longer and space them out.

Again, it's just going to lead to a suboptimal management of a patient's condition. It can lead to increased risk for complications. I think the idea behind the denials, the prior authorizations, and the insurance things is to lower costs in the short term, but sometimes, treatment delays or lapses can potentially increase costs with preventable complications and more frequent emergency and primary care office visits.

Physician-led tech companies, like Komodo Health and SmarterDx, aim to utilize data and AI to identify and address barriers brought on by increasing denial trends.

Michael Gao, MD, a physician-turned-health-tech entrepreneur, cofounder of SmarterDx:

The way that our software works is when [physicians] get a denial letter, they literally just upload it, and through very secure, encrypted connections, we pull the relevant data and then, effectively, pregenerate the evidence-based arguments that the physician or appeal specialist might otherwise make. So, I think with our clients, they've really reported 3 separate benefits.

Michael Gao, MD, SmarterDx

The first, as you might expect, is, "Wow, this used to take me 2 hours, and now it takes me 10 minutes, because it's all there, and I'm just editing an essay." For a lot of hospitals, what happens is they don't even have enough staff to get to all possible denials. And so historically, they might have ranked the denial say, highest dollar on the top of the list, lowest dollar on the bottom, and then at the bottom you might have denials that are $1000, $2000, or something that nobody ever even gets to because you just run out of time and then the window for replying expires. Because, again, the default is no payment, those were arguments that were lost just because you ran out of time. With this technology, now people are able to get through their entire stack.

And then the third is that even in the high-dollar cases, where it's well worth it for somebody to spend an hour or 2 for a case that's maybe $50,000 or $100,000 of reimbursement, it turns out that editing an essay is a whole lot easier than writing one from scratch. More than that, it often leads to a better product because you're spending more time crafting the argument rather than searching for lab values and copying and pasting somewhere into a letter.

At least subjectively, what some of our clients have told us is, "We feel like the quality of the output of our work has also increased." I think this is really important because a lot of people equate AI with automation. But of course, it's not just that AI can play chess faster than I can, it's that it can play chess a whole lot better than I can as well. And being able to, not automate staff, but actually augment staff with these tools, can actually lead to better output and more accurate output, not just time reduction.