The fifth annual Abarca Forward conference, hosted by the pharmacy benefit manager (PBM) Abarca Health at its headquarters in San Juan, Puerto Rico, dug into key aspects of prior authorization (PA). The invitation-only event brought together payers, PBMs, specialty pharmacies and clinicians to discuss how evolving technology and improved processes and communication can return PA to its roots of safeguarding patients while reducing healthcare costs.
Abarca invited five leading PA players to present their companies’ innovations and insights into transforming PA and utilization management today. Across the session, speakers returned to one idea: the challenge is not prior authorization alone, but the fragmented, manual experience surrounding it. Here’s what they had to say.
Akeel Williams, CoverMyMeds
For Akeel Williams, Senior Vice President at CoverMyMeds, the McKesson-owned medication access company, the next era of utilization can be defined by the answers to three questions: which decisions should be instant, which decisions require human review and how do we experience the system when it works well.
The current PA system treats even routine decisions as if they were complex, Williams said, noting that physicians and their office staff spend hours every week navigating the basics of PA processes and systems. “This tells us the system doesn't do a great job of distinguishing between what's obvious and what's uncertain and that causes abrasion,” he said. “The effects show up as delayed care, lost capacity, lost bandwidth and unnecessary costs.”
Williams’s second question is about which decisions truly deserve human review. “Human judgment is the most expensive, valuable asset in healthcare,” he said. “We should spend it where it changes outcomes, not where it arbitrates ambiguity created by core design.”
Williams’s third question, about recognizing and acknowledging when the PA system works well, shows a shift from policy thinking to process thinking, he said: “Clinicians should experience clarity and fewer interruptions, while payers should experience more consistency of care, more ability to forecast and measure outcomes.”
In tying these points together to create the PA system of tomorrow, Williams said we’ll need to shrink the footprint of PA without shrinking its purpose. “The next generation of PA will be defined by how proportional, intelligent and invisible we can make it while protecting affordability and quality,” he said. “The future of PA isn't less control; it's better judgment applied early.”
Andrew Mellin, Surescripts
Next, Andrew Mellin, Chief Medical Information Officer of Surescripts, shared insights on the inefficiencies in current processes, such as the manual handling of PAs, and discussed the potential of AI to improve these processes by making them more transparent and efficient.
Mellin described a collaboration between Surescripts, a large payer, and a health system. In their three part process, they codified clinical policies. “Then, at the time the prescription is signed and before it reaches the pharmacy, a series of real time interactions between the EHR and PBM systems intelligently surface and apply the relevant clinical evidence already present in the EHR,” Mellin said. He noted that in this model, decisions are made at the PBM within seconds. By delivering the right information for the right medication at the right time, patients can go to the pharmacy with fewer barriers and gain access to their medications much more quickly.
Mellin believes the lessons learned from this collaboration, especially those involving AI, have industrywide applicability. “There will be a profound impact on PA but right now, deliberate caution is called for,” he said, noting that a leader of a PBM in the Surescripts collaboration was uncomfortable assuming the risks associated with AI. “Being deterministic at least for a while is really important. That will change over time,” he added.
Mellin also encouraged companies to simplify the questions used in their PA processes. “I personally reviewed thousands of prior auth questions across many payers and many drugs and was struck by the ambiguity of the PA questions,” he said. “When you're working with technology, ambiguity creates errors and problems.”
Similarly, Mellin suggested bringing clarity to policies regarding acceptable data for drug decisions, using as an example the question, Is this for chronic headache or acute headache? “In the prescription, the doctor writes one tablet when you have a headache, with 10 pills for acute headaches, but we see payers that won’t accept that,” he said. “Look for sources of evidence beyond those in your rigid policies.”
Kyle Kiser, Arrive Health
Kyle Kiser, CEO of Arrive Health, emphasized the importance of interoperability and AI in solving foundational challenges in healthcare, advocating for a modern network that supports shared decision-making. To do that, he said, the industry needs to solve several fundamental challenges. “How do we drive full participation and eligibility so that every patient that walks into a doctor's office can be matched with the plan they're associated with?” he said. “A lot of health plans don't participate fully in the eligibility data and if we don't solve that problem, we can't solve the rest of them.”
According to Arrive Health data, in up to 30% to 40% of cases, the EMR does not contain patients’ eligibility data. “We found that about one third of the time it was creating prior authorizations that were unnecessary and not required for that patient. They may be required for that plan, but the patient had already satisfied the criteria,” Kiser said. “Driving towards personalized insights is essential as we want to leverage the full impact of interoperability and AI.”
Kiser singled out the vital role of pharmacists in real-time benefit implementation. “If you don't embed the intelligence of the pharmacist into that transaction, it's going to be wrong,” he said. “The drug compendia configured inside the EMR often speaks a different language than the pharmacist’s system and the pharmacist fills that gap by figuring out what the plan pays and knowing that's the plan to run.”
Real-time benefit implementations need to result in an easier process to be successful, Kiser explained. “It has to be fewer clicks, and it has to be right nearly every time,” he said. “And if you don't meet those criteria, then physicians will not use it.”
The work of increasing access to real-time, accurate information is happening across many stakeholders, but Kiser stressed the need to keep the patient involved. “We need to create the right connectivity so that when patients are seeking a PA status, they're seeking it from you as adjudicator no matter where they are,” he said.
Diana Benli, Cognizant
Diana Benli, Chief Product Officer, TriZetto Products Group at Cognizant, discussed the concept of "shifting left" in healthcare, which involves moving intelligence and collaboration upstream in the patient journey to improve outcomes and reduce costs. She highlighted the importance of interoperable platforms and AI in achieving what she called a “10 times” improvement in healthcare systems. “It's not about being 10% better, it's being 10 times better, 10 times more predictive, having 10 times less administrative friction and being 10 times more resilient,” she said.
Benli believes that TriZetto’s approach points to one way the PA process could move closer to that 10-times improvement. “We combine the workflows between payers and providers, and we do the integration layer between those organizations,” she said. “Friction doesn't live in one single organization. It lives at the seams between PBMs and health plans, for example.”
Benli described TriZetto’s approach to supporting smoother automated PA workflows, providing value at the point of care: “It shows users, here's your optimal pathway, here's the fastest route to therapy, and here's the alternatives. That's not automation, that's collaboration and intelligence at the point of care,” she said. “If prior authorization becomes predictive and embedded, upstream it becomes value.”
Returning to the theme of her remarks, Benli explained that shifting left is not just about technology. “It really is about a philosophical change moving from being reactive to proactive, gatekeeping to guidance, and incremental improvements to ‘10 times’ improvements,” she said. “To use a plumbing metaphor, if we redesign the water, or data, flowing through the pipes, or infrastructure, for our care models, prior authorization becomes a bridge, not a barrier.”
Sri Somasundaram, Latent Health
Sri Somasundaram, co-founder of Latent Health, presented a vision for using AI to address the inefficiencies in the PA process, noting that the common perception of the PA process as the villain in the story is critically wrong. “The system is designed with checks and balances, but they’re not the problem,” he said. “We contend that it's actually a ‘compute problem’.”
What he means by that, Somasundaram said, is that humans on both the provider side and the payer side sift through data to build a case. When things go well, both sides reach the same conclusion, but it’s been a duplicative task. When it doesn’t go well, there is no clinical agreement, and even more resources are needed to achieve resolution. “This adversarial dynamic that exists between providers and payers that we have all just come to accept, I argue it's actually a symptom of this manual, inefficient compute problem rather than the system,” he said.
Somasundaram argued that the answer is to have “clinical agents with superhuman accuracy”—i.e., an AI “clinical reasoning system”—to handle complex clinical questions and improve the overall system, Somasundaram said. After the AI agent answers complex clinical questions using the research record and applying drug criteria, then “the idea is you then communicate this information outward to all the relevant parties,” he said.
The demand is immense, said, noting that the company is now working with some 50 health systems Somasundaram, up from just a handful a year ago. “This is where healthcare wants to move,” he said, adding that companies must resist the urge to apply AI to existing systems. “A lot of current approaches have been quite rudimentary when what we’re talking about is something fundamentally different.”
Putting it all together
Finally, to make sense of it all, two speakers took the stage to provide context. Greg Ryslik, Chief Data and AI Officer of Stellarus, and Alfredo Bird, partner at Xtillion, highlighted several uniting themes in the presentations.
Everyone seems to feel a need to bring more humanity and transparency to the PA process, noted Bird. Other themes include data exchange and interoperability and the need to build trust while reshaping the PA experience into one that is less fragmented and easier for patients and providers to navigate. “PA doesn't need to feel like a game of ping pong. It can be a more collaborative, conversational experience, and it is our belief that technology can play a role there,” he said.
For Ryslik, the promise of AI in improving PA is clear. “Today, right now is the worst that these AI models will ever be. Tomorrow they will be better, and every day after that,” he said. “But then they co-exist with payers processing faxes by hand and running a mainframe that is older than most of the developers on my team.”
The effects of that dichotomy will be felt for years to come but will eventually recede, Bird said, noting that even organizations still running on a mainframe, may not want to wait years to adopt AI. “That's where very simple architecture and integration choices can make a huge difference,” he said. “Something as simple as adding a universal data layer that sits on those old applications but provides more modern means to interact with it will make a big difference.”
The editorial staff had no role in this post's creation.