Market access in the AI era: Impacts on prior authorizations

By Michelle Kaner, PharmD, Senior Market Access Strategist, Mosaic

The healthcare industry has long invested in technology, continually seeking to boost productivity and foster innovation. The past year was notably marked by the rise in popularity of generative artificial intelligence (AI), capturing the interest of payers, providers, patients and manufacturers alike. KLAS Research reported that over half of healthcare organizations surveyed are gearing up to either adopt or invest in AI technologies within the year.¹ Echoing this enthusiasm on the consumer front, a Deloitte survey of 2,014 adults revealed that 53% of consumers believe generative AI could improve access issues and shorten wait times for medical care.²

At CES 2024, insurance company Elevance Health, which covers 117 million lives, took center stage discussing AI adoption. They highlighted the need to provide members with personalized care recommendations, help them understand their healthcare benefits and simplify the healthcare experience.³ With increased AI adoption among payers, manufacturers must understand the implications of these tools and recognize how they impact the accessibility of their products for patients.³

Generative AI use case: Zooming in to focus on prior authorization (PA) support

Generative AI holds the promise of transforming the PA process, traditionally a bottleneck in patient care. According to the American Medical Association (AMA), more than 90% of physicians report that PA requirements delay patient healthcare, placing a significant burden on healthcare providers and straining health plans with the management of case reviews and appeals. Additionally, one-third of these physicians have observed severe consequences due to such delays, including life-threatening incidents or hospitalization for their patients.⁶In response to this challenge, Doximity, a digital platform for physicians, introduced generative AI software that assists in administrative writing tasks, like drafting PA letters, appeals, and more.⁷

It's important to note that payers have also been employing AI technologies, including machine learning and data analytics, to improve various aspects of PAs for some time. However, tech giants like Google, Microsoft and IBM are at the forefront of exploring ways to leverage generative AI to refine the PA process further, aiming to enhance efficiency and alleviate the financial and time burdens currently placed on payers and providers. Generative AI excels at rapidly analyzing data -- it can both summarize extensive information and can identify crucial details that might be missed in manual data reviews. The integration of these AI-driven approaches is expected to address PA-related administrative challenges and inefficiencies, which are estimated to cost the U.S. healthcare system approximately $25 billion annually. The adoption of more automated solutions like generative AI could lead to significant financial savings, possibly as much as $454 million per year.⁸

Real-world examples of AI adoption by health plans

Health Care Service Corporation (HCSC)

Leveraging AI, HCSC can process PAs 1,400 times faster than with previous methods. Developed in 2021 and piloted the following year with a focus on specialty pharmacy and behavioral health, the HCSC AI tool achieved an 80% approval rate for behavioral health services and 66% for specialty pharmacy requests. By triaging and approving requests with minimal information, the technology enhances efficiency and frees up clinical staff to review more complex cases.¹⁰

Leveraging AI responsibly

“The use of AI in PA can be a positive step toward reducing the use of valuable practice resources to conduct these manual, time-consuming processes. But AI is not a silver bullet.” —AMA trustee Marilyn Heine, MD¹¹

In late 2023, a lawsuit was filed against United Healthcare (UHC) by the families of two deceased Medicare Advantage members, claiming that UHC wrongfully denied coverage for medically necessary post-acute care using an AI-powered algorithm.¹² In a parallel scenario, Cigna is also grappling with legal challenges as members and their families allege the inappropriate use of AI tools to deny claims.¹³

Balancing AI’s capability in healthcare^14-17:

October 2023: President Joe Biden issued an executive order on artificial intelligence that marked a significant effort to enhance accountability in the development and deployment of AI technology. The order established a framework for addressing privacy, security, safet, and equity issues in AI use, providing a pathway for collaboration between the industry and the federal government while fostering innovation
November 2023: House members urged the Centers for Medicare & Medicaid Services (CMS) to evaluate the use of AI by Medicare Advantage (MA) plans to prevent excessive rates of denials. They proposed that MA plans should provide reasons for denials, assess AI’s role and ensure algorithms avoided inappropriate factors. They also called for a comparison of AI rulings to traditional coverage decisions to ensure coverage parity with traditional Medicare
December 2023: More than two dozen healthcare payers and providers, including CVS, Premera Blue Cross and Mass General Brigham made voluntary safety, security and transparency commitments to the White House regarding the use of artificial intelligence. Stakeholders aligned industry action on AI around the FAVES principles stating AI should lead to healthcare outcomes that are fair, appropriate, valid, effective and safe
February 2024: CMS clarified the application of the Medicare rule CMS-4201-F and the Affordable Care Act on AI use in MA coverage determinations. This guidance ensured that MA organizations could utilize AI for coverage decisions, if they adhere to legal standards and prioritize patient-specific circumstances