Feature
Article
Learn how prior authorization began, why it hurts doctors and patients, and how new CMS rules and AI could finally streamline the process in 2025 and beyond.
prior authorizations are a burden for both physicians and patients
Prior authorization has morphed from a cost-control checkpoint into one of medicine’s biggest choke points. Doctors now juggle an average of 43 authorization requests a week—about 12 staff hours that could be spent on patient care, an American Medical Association survey shows. In addition, physicians
to their bottom line as well. A whopping 97% of physicians say they should be reimbursed for prior authorization, according to a Physicians Practice survey.
Patients feel the delay, too, and sometimes suffer the consequences. According to the AMA survey, nearly one in four physicians reported that prior authorization has led to a serious adverse event for a patient in their care, including hospitalization, permanent impairment, or death.
Regulators are responding. A January 2024 Centers for Medicare & Medicaid Services (CMS) rule will force Medicare Advantage, Medicaid and Affordable Care Act plans to answer routine requests within seven days and urgent requests within 72 hours starting in 2026, while also opening fast, FHIR-based electronic pipelines by 2027.
Whether those deadlines — and the rise of artificial-intelligence tools that auto-assemble paperwork — will ease the logjam or simply accelerate denials is the next question facing physicians, payers and policymakers.
Medical Economics spoke with Heather Bassett, MD, chief medical officer of Xsolis about the current state of prior authorizations and whether technology can make them less of a headache for physicians and their patients.
Prior authorization (PA) is a cost-control tool in which a clinician must secure advance approval from a health plan before it will pay for a medication, test, treatment or piece of equipment. CMS defines the process as a request “submitted for review before the service is rendered … and before a claim is submitted for payment.”
Physicians now spend nearly two full workdays each week wrangling prior authorizations. The American Medical Association’s 2024 survey found doctors handle a median of 43 requests weekly— paperwork that siphons about 12 staff hours from direct patient care and fuels burnout.
For patients, the delays can be dangerous. Nine in 10 physicians told the AMA that authorization hold-ups harm clinical outcomes, and 29 percent said a patient in their care suffered a serious adverse event because treatment was stalled. Not surprisingly, many patients simply give up; most doctors report that prior-auth barriers often lead people to abandon recommended therapies altogether.
The financial bite is real, too. Medical Group Management Association analyses show practice spending on prior-auth staffing jumped 43 percent between 2019 and 2024, even as reimbursement lagged. Some groups now employ full-time specialists whose only job is chasing approvals—overhead that small offices can ill afford.
Adding to the frustration, criteria for approval remain opaque and inconsistent from one payer to the next. A recent Guardian report detailed how algorithms at a major insurer could deny hundreds of thousands of claims in minutes, leaving physicians guessing at the logic and scrambling to appeal. Critics warn that, without transparency and human oversight, artificial intelligence may simply speed up denials rather than streamline access.
Delayed chemotherapy infusions, postponed joint replacements and abandoned insulin starts have turned PA into a national access issue. In July 2024, nine in 10 physicians told AMA pollsters that PA negatively affects clinical outcomes. Public frustration erupted after investigative reports showed algorithms at one large payer rejected 300,000 claims in two months—one every 1.2 seconds.
Mounting evidence that prior‐authorization delays harm patients and drain practice resources has pushed regulators and lawmakers onto the offensive. The federal government struck first. In January 2024 the Centers for Medicare & Medicaid Services finalized its Interoperability and Prior Authorization Rule, requiring Medicare Advantage, Medicaid and Affordable Care Act exchange plans to answer urgent requests within 72 hours and routine requests within seven days, and to publish their approval and denial metrics beginning in 2026. The rule also forces those payers to connect with clinicians through FHIR-based application programming interfaces by 2027—an attempt to drag the process out of the fax era.
Congress is trying to lock in further guardrails. The bipartisan Improving Seniors’ Timely Access to Care Act, re-introduced in the 119th Congress, would codify real-time electronic prior authorization in Medicare Advantage, add a “specific reason” requirement for every denial and create a gold-card program for clinicians with high approval rates. The measure has drawn broad support from the American Medical Association, American College of Physicians and more than 250 medical specialty societies.
States, meanwhile, have become laboratories of reform. Texas’ Gold Card Act (HB 3459), enacted in 2021, exempts physicians who achieve at least a 90 percent approval rate for a given service from future PAs for that service—a model other legislatures are studying. At least 35 states now cap response times or streamline appeals, according to an AMA law survey. Some are going further: a 2025 Connecticut bill would bar insurers from using artificial-intelligence algorithms to make coverage determinations after a ProPublica investigation found an AI tool denied more than 300,000 claims in two months.
Even payers feel the pressure. UnitedHealthcare said it will eliminate prior authorization for about 20 percent of its procedure codes and roll out its own gold-card program in 2024, moves the insurer credits to state action and pending federal mandates.
Taken together, these efforts signal a rare alignment of physicians, patients and policymakers: curb the paperwork, restore clinical autonomy and ensure that when prior authorization is used, it moves at the speed of modern medicine — not the pace of a busy fax line.
In the decades after World War II, commercial insurers—led by regional Blue Cross plans—began scrutinizing hospital admissions and lengths of stay to curb spiraling costs. By the early 1960s, more than 60 Blue Cross plans were already reviewing claims for “medical necessity,” and many required physicians to certify the need for continued hospitalization after a set number of days. These early utilization-review programs introduced the idea that a payer could insist on approval before care was delivered—the conceptual forerunner of modern prior authorization.
Congress cemented the movement in 1973 with the Health Maintenance Organization Act, which offered federal grants and, crucially, required large employers that provided insurance to give workers the option of an HMO plan. The statute’s emphasis on gatekeeping and cost containment normalized pre-admission certification, second-opinion mandates and other mechanisms that soon evolved into formal prior-authorization requirements.
As HMO enrollment surged—topping 50 million members by the mid-1990s—payers pushed prior authorization well beyond hospital stays. Advanced imaging, elective surgeries and the newest brand-name drugs all fell under scrutiny, and entire third-party “utilization-management” vendors sprang up to process requests. By decade’s end, few physicians could order a costly test or prescribe a specialty medication without first securing a payer’s green light.
Electronic health records became nearly universal after the HITECH Act, yet prior authorization remained a stubbornly manual workflow. CAQH tracking shows that as late as 2022 only about 28 percent of medical PAs moved through the industry-standard X12-278 electronic transaction; most still relied on faxes, phone calls or payer web portals—driving up staff time and denial frustration.
The Centers for Medicare & Medicaid Services took its biggest swing at the problem in January 2024, finalizing the Interoperability and Prior Authorization rule. Beginning in 2026, Medicare Advantage, Medicaid and ACA marketplace plans must answer urgent PA requests within 72 hours, standard requests within seven days and expose decisions through a FHIR-based API by 2027. Commercial payers are already piloting AI triage engines to auto-approve low-risk requests, while health-system bots extract chart data and draft submissions in seconds. Whether those tools lighten workloads or simply accelerate automated denials will shape the next chapter in this long-running hurdle.
Artificial intelligence is racing into prior authorization on two very different tracks. Health plans are deploying machine-learning models to sift requests in milliseconds, promising near-instant answers and lighter workloads. Yet three in five physicians fear the opposite is happening: an AMA survey released in March 2025 found 61 percent believe payers’ use of unregulated AI is increasing denials and worsening patient harm.
That skepticism is landing in courtrooms. A federal judge in February allowed a class-action lawsuit to proceed against UnitedHealthcare after plaintiffs alleged the insurer’s algorithm wrongly denied medically necessary care to Medicare Advantage members. Legislators are weighing in, too. After reporters uncovered an AI tool that rejected hundreds of thousands of claims in minutes, Connecticut lawmakers introduced a 2025 bill that would bar insurers from making coverage decisions with AI alone.
Providers, meanwhile, see a different promise. Health systems are piloting AI “agents” that read the chart, pull required documentation and submit electronic PAs without human keystrokes. The Medical University of South Carolina says the technology reclaimed more than 5,000 staff hours each month, while early adopters report first-pass approval rates topping 95 percent.
Federal regulators are watching both currents. Beginning in 2026, CMS will require payers to give a specific reason for every AI-assisted denial and to publish aggregate approval data—an attempt to keep black-box algorithms from overruling clinical judgment. Whether artificial intelligence ultimately streamlines prior authorization or simply speeds up bad decisions will depend on that transparency, human oversight and the willingness of doctors and insurers to keep the patient—not the model—in charge.
After half a century of incremental tweaks, prior authorization is finally at an inflection point. Federal time-limit mandates, a patchwork of state gold-card laws and a swelling chorus from organized medicine have pushed the once-invisible paperwork routine onto the policy center stage. By 2026, Medicare Advantage and Medicaid plans must answer most requests in under a week and publish their approval data; by 2027, they must exchange those decisions through real-time FHIR interfaces.
At the same time, artificial intelligence is challenging the very definition of “medical review.” Done right—under transparent rules and with human oversight—AI could lift a crushing administrative weight and let clinicians redirect thousands of hours back to patient care. Done poorly, it risks automating denial at scale. Early state bills and pending federal proposals signal that lawmakers will not give the technology a free pass.
For physicians, the takeaway is clear: engage now. Track your own authorization metrics, adopt electronic workflows where possible, vet AI tools for auditability and lend your data and stories to advocacy efforts. The AMA survey numbers tell the stakes—12 staff hours a week, one serious patient harm in three practices—and they give reformers the ammunition they need.
Whether the coming rules and algorithms create a smoother path or a faster bottleneck will depend on how forcefully clinicians, payers and regulators keep patients—not paperwork—at the center of the process. The next two years will show whether prior authorization finally becomes the streamlined safety check its designers promised or remains a costly detour on the road to care.
