Prior authorization (PA) requirements usually succeed in reducing use of certain procedures, but the evidence on whether patients are harmed as a result is not so clear, according to a , a nonprofit founded by the "Big Three" U.S. automakers and their principal labor union, the United Auto Workers.
"The research generally shows [that PA] reduced use of the targeted care, sometimes with offsetting increased use of preferred treatments," wrote co-authors Ani Turner, Samantha Clark, and George Miller of the Altarum Institute's Center for Value in Health Care. "Evidence also shows reduced spending on the targeted care, although this does not always translate into lower overall health care spending. Positive outcomes have been shown under PA programs for diagnostic imaging, where high rates of use and cost were reduced, and opioid prescribing, where rates of subsequent opioid abuse and overdose were reduced."
The authors conducted their study via an analysis of peer-reviewed literature on prior authorization, as well as "gray literature" including reports by the Medicare Payment Advisory Commission, the Government Accountability Office, and the American Medical Association (AMA). The authors also interviewed insurance and pharmaceutical industry experts.
Cost Estimates for Doctors Vary Widely
Regarding the cost of PA, "Estimates of the cost burden to physician practices vary considerably, from $80,000 annually per physician to between $2,200 and $3,400 annually per physician (2010 dollars)," the authors wrote. "We find the direct cost of PA likely to be closer to the lower-end estimates, which were focused specifically on PA interactions with insurers and were based on staff time requirements recorded in real time, rather than on requirements estimated by providers from memory."
"The American Medical Association has done surveys annually and puts out numbers related to PA from those surveys, and the study that's quoted a fair amount puts the [annual] cost at $80,000 per physician, but that study looks at interactions between physicians and insurance companies more broadly -- maybe a lot is PA but it's not clear it's entirely so," Turner said in a phone interview.
"Other studies had numbers of less than $10,000 per physician -- those were doing tracking of actual time spent in practices relating specifically to PA. That's why we concluded that while there is a wide range, based on the methodology and rigorousness of the studies, the true cost is likely to be at the lower end of those estimates -- in thousands rather than tens of thousands, but it varies based on whether or not the physician is able to do things electronically, and how many insurers the physicians are dealing with, because the requests and method of submitting requests vary."
Furthermore, "some studies show that when physicians groups get together and do centralized activity on PA, that can be a more efficient way of doing it ... So there's not one answer, even on the cost side," Turner said. "It also depends on whether insurers support electronic submission."
Not Going Anywhere
"It's clear to us that PA is not going anywhere in the near future," she added. "If anything, it seems to be increasing, but hopefully it's becoming more automated and more standardized ... and hopefully there's a pull on providers and insurers toward streamlining and automating the process so it is less of a burden."
Turner said she understands that physicians have concerns about the PA process. "While I'm sympathetic with the fear of having the insurer looking over your shoulder, there is the issue of low-value care, of fraud and abuse, and of good treatment of resources," she said. "When physicians are surveyed, a very high percentage believe that patients are negatively impacted by PA, and I don't discount their views on that, but we're looking for what the evidence says, and certainly the evidence says people get less care, and less care that's found to be inappropriate."
For example, prescription drugs have a lot of prior authorization associated with them, especially if they're prescribed off-label. "In the meantime the patient decides to say, 'Forget it; it's not worth the hassle,' so care is delayed and sometimes abandoned altogether," said Turner. "The question is, does that negatively affect their health in the long run, and that's less clear." The only area where some evidence outcomes are harmed is in psychiatry, she added. "Patient adherence can be very difficult if someone has a condition like schizophrenia -- making sure they stay on their medications and constantly adjusting doses can be challenging. So when there are interruptions due to PA, there is evidence in the literature that can interrupt care for people with mental health conditions, and that does put outcomes at risk."
Increasing Use in Public Programs
PA's use, while common among commercial insurers, has until recently been limited in public programs like Medicare and Medicaid. However, the investigators noted, "recent demonstration programs and the growing role of managed care organizations in delivering care to Medicare and Medicaid enrollees are expanding the use of PA for the publicly insured ... In recent years, [federal law] was amended to allow PA to be tested and evaluated under Medicare fee-for-service for several categories of care, including repetitive, scheduled, non-emergency ambulance transport, non-emergent hyperbaric oxygen, home health care, and power mobility devices. Stemming from these demonstrations, systemwide PA requirements have been implemented for power mobility devices; CMS maintains a of specific requirements."
One way that's been suggested to make PA less of a burden is to "gold card" certain providers whose PA requests are always approved, and exempt them from PA requirements. However, one insurance industry representative "made the point that it's not as straightforward as you might think to implement gold carding," said Turner. For instance, "You may have providers within the same practice who do things differently, and it may be difficult to distinguish one doctor from another and just give a gold card to one physician."
In addition, "the systems in place [from the payer side] aren't always easy to configure to allow a waiver of the requirements," she said. "And sometimes the physician may have a high approval rate for particular types of services but not for others, so it's difficult to say they get a pass on all their PA requests. But in principle it makes a lot of sense to not have to go through that hassle if the vast majority of time someone is approved."
Turner noted that "some insurers may say they'll sunset their PA requirements if they find that across all of their providers, the vast majority of PA requests are approved. So it's desirable for insurers to track and regularly update requirements and eliminate them if they're not really serving a useful purpose. That's another way the burden could be reduced."