Riverside University Health System Neurosurgery Residency

Prior authorization (PA) has been the bane of numerous physicians. In theory it has been thought to be a “check and balance” process to evaluate necessity of treatments prior to completion of the treatment.  These prior authorizations have been criticized as a way for payers to “spend less” and in some cases physicians have even noted the need to attempt to evaluate whether something will be covered or not and make treatment decisions based on ability to pay—resulting in lower quality of care. Others have thought that a prior authorization in theory—flag better treatments, improve quality of care and reduce overutilization and overspending to improve care.

Due to the concerns regarding prior authorizations including the possibility of both delayed and poorer quality of care, in their article “Perceptions of prior authorization burden and solutions” authors Sahni et al. sought to survey providers, patients, and private payer employees to understand perceptions regarding prior authorizations and propose possible improvements to the system. This study consisted of 3 separate surveys with specific questions applicable to each of the respondents and focused on questions related to approval, time to approval, level of burden to complete the process, clinical impacts, prior experience with prior authorizations and potential solutions. Artificial intelligence solutions were also investigated.  Surveys were conducted using a web-based approach using free text, multiple choice and Likert scales. All surveys were completed in between April and June 2023.

A noted strength to this study, although survey based was the large sample size.  5766 patients responded to the survey—of which 1005 met criteria to participate. Providers returned 5659 surveys and 1010 met criteria to participate. Private pay employes notably returned fewer surveys with 261 completing screening and 115 meeting criteria to participate. Survey responses were tabulated, and descriptive statistics were utilized. Ultimately statistical analysis was completed both using a multivariate regression and two-tailed t-tests. 25 specialties returned surveys including specialties defined as “spine” as well as neurosurgery. Looking specifically at spine data amongst providers, there was a 67% overall perceived approval rate among providers, 40% perceived time to approval of more than 5 days and only 3% thought that they had no prior authorization burden.  Within neurosurgery providers, 67% there was also a 67% perceived approval rate but 64% of providers thought prior authorizations took greater than 5 days to be approved and only 4% thought they had no prior authorization burden. Within neurosurgery patients estimated their approval at 84%, 40% noted the process took greater than 5 days, and 32% thought their prior authorization burden was none. Within spine patients perceived an approval rating of 85%, 38% percent believed their PAs took greater than 5 days to be approved and 30% believed they had no burden from PA. Overall data suggests that  amongst all specialties within the provider category, providers estimated that there is a 70% approval rate, 27% believe their approvals took greater than 5 days and only 2% thought that the level of burden was none. Patients overall felt they were approved 88% of the time, 37% had PAs take longer than 5 days and 34% felt no PA burden. Private payers estimated 73% approval, only 15% thought PAs took longer than 5 days to approve and 7% felt no burden from PAs. Statistically, patient perceptions of prior authorization approval and perceived approval taking longer than 5 days was significantly different than those of providers and private payers. Similarly, the difference in perception of burden were significantly different amongst all groups. It seems that truly both the providers/physicians as well as the payers seem to be more frustrated with the system than the patients although there is still a plurality of patients at 66% reporting burden with the PA system.

Of note, 51% of patients did not need to assist in the PA process and 39% spent less than 1 hour. Additionally, 38% of patients had “no miscommunication with the PA process. That means an alarming 62% of patients had some issues regarding this. Providers fared far worse with 1% reporting “no miscommunication” and 0% of private payers. Clearly this is an issue that needs to be addressed.  Fifty seven percent of providers noted difficulty in determining if a PA was needed which was significantly different than patients (26%) and private payers (21%). Public insurance was noted to be less likely to be successful in the PA process compared to those with private insurance. Notably insurance type otherwise did not correlate with any other metric. Age also correlated with perceived burden. PAs were also thought to have grown in volume among providers (65%), and payers (79%).

Concerningly 42% of providers note that PAs are a high factor in burn out with follow-ups being the most cumbersome. Even more concerning is 31% of patients felt that PAs negatively impacted their ability to seek treatment with 85% noting stress and 84% noting delay in care. 92% of providers similarly noted delays in care even up to 2 weeks for 14% of respondents. 62% of providers noted more visits and therefore even more costs and in 60% of responses patients had more severe symptoms and conditions— an overall alarming trend.

From a financial perspective payers noted that in 98% of survey responses employees worked exclusively on PAs for which a median of 3 employees worked on each case and were involved in approximately 20 cases a week. In the provider group: Billing and coding specialists spent a median of 9 hours per week working on PAs, 5 hours for practice managers and RNs 2.5 hours. When evaluating AI, 65% of private pay respondents reported plans to incorporate AI to the PA process with noted concerns leveed regarding possible clinical portions or review amongst providers and patients.

Overall, the authors have identified some key trends. Patient care appears to be impacted with concerns about delays in care, but patients do not seem to feel the burden of PAs as much as providers and payers. Therefore—as neurosurgeons and physicians—we may need to help drive these changes as the overall general population may not be as invested in it as we are. Similarly, there is a financial burden from both the requirements of using time and financial investment in the PA process with so many workers putting so many hours into PAs. AI appears to be a possible solution for automating some parts but the potential usage for clinical approval is clearly a concern amongst patients and providers. Although there may be cost savings we may soon be coming to a world where we must ask ourselves—do we want AI to evaluate whether our patients can get the treatments they need?  We may have to ask ourselves— and quickly– how and in what way can we advocate for changes in this PA system that seems to be getting larger year in and year out and how can we use these new technologies, such as AI without introducing some new unforeseen burden or causing harm to our patients or their care.

Alexander Lopez, MD

Prior authorization (PA) is the process by which insurance assesses whether the treatment/workup prescribed by the provider is essential to the patient’s treatment. Research has shown that this process is vital to avoid overutilizing financial medical resources. However, the steps can be cumbersome, delay patient care, and financially expensive, costing 35 billion dollars a year. This study article examines prior authorization for US healthcare using survey data from over 2,000 individuals from payors, patients, and providers.

The survey data showed that more than half of the prior authorizations were completed by providers, and the providers reported significant time and financial costs associated with the approval process. Regarding insurance type, patients with public insurance were perceived to be less successful than those with private payors. Of the patients who participated in the prior authorization process, 88% received prior authorization approval, 37% had to wait 5 days for that approval, and 34% felt burdened by the process.

This study also looked at the efficiency of PA, where the patient respondents believed that the provider (85%) and private payers (76%) handled the process very efficiently. However, providers and private payors report their efficiency levels at much lower percentages, 34% and 62%, respectively. The discrepancy between the patient’s perceptions is likely attributed to the fact that providers are the ones calling for prior authorization instead of patients. The amount of time spent on prior authorization by providers was equivalent to 100,000 full-time nurses per year working on prior approval. Of the individuals spending time on the prior authorization process, physicians spend an average of 1 hour per week, 1 hour for advanced practice providers, 9 hours for billing or coding specialists, 5 hours for practice managers, and 4 hours for medical office specialists.  While prior authorization was meant to offer a check and balance to the medical system, this process contributes to provider burnout, delays in patient care, and increased medical costs.

Interestingly, the burden of prior authorization is different across all medical specialties. Pediatrics had the lowest approval rates from patients, while plastic surgery had the highest. Dermatology had the lowest approval rates for providers, and physical therapy had the highest. Within Neurosurgery, 29% of patient respondents reported a negative impact from PA when seeking treatment. Patients have a higher perceived approval rate (84%) than providers (64%). In addition, neurosurgery providers report overall time to approval taking more than >5 days in 64% of the participants.

One promising solution is leveraging artificial intelligence (AI) to assist in this process. AI would remove administrative burdens like data entry and facilitate discussions among payors and providers. Evidence-based treatment guidelines and insurance policies can be programmed into AI to assist with the approval process.  In addition, AI can streamline the work by triaging cases, thus improving efficiency.  Insurance companies have favored this transition as 65% of payors look towards integrating AI into their prior authorization process in the next 3 years. However, only 11% of provider organizations plan to do the same. Providers are concerned that this technology will minimize burdens or create new problems and factor these costs into their budgets. In contrast, payors worry about cybersecurity and technical infrastructure to support this transition.

AI, in its current state, can be very useful in relieving some of the headaches associated with insurance policies and approving simple authorizations. However, I would be very cautious about its implementation in medicine. AI largely depends on the cohort on which the algorithm was trained. If certain patients have a specific economic background, medical diagnosis, or even race, the AI algorithm can misclassify patients and limit patient care. This was seen historically in 2019 with an insurance company using AI algorithms limiting care to individuals with mental health and other algorithms that had inherent racial bias within them. AI can undoubtedly solve specific issues, but proper oversight is crucial. Often, people will blindly believe the answer produced by an AI tool. With AI being more prevalent in medicine, there should be some transparency on how these tools are created so that providers can understand their limitations. Comprehensive patient populations will need to be used to train this AI algorithm, and these algorithms should be audited frequently to avoid generating skewed biases. Even with progress in AI, this technology will likely still necessitate human oversight for complex cases but significantly reduce the burden.

In conclusion, while prior authorization can be a headache for many providers, the financial utility of previous authorization is likely to ensure its place in the medical field. Between the frustrations of doctor-to-doctor phone tag, I often remind myself that we are trying to care for patients within this process. Most of the burden from PA is placed on providers, and there is an enormous amount of medical funds directed toward this process. AI technology will likely solve the issues with the current PA process. This notion of bringing AI begs further questions for providers and patients.

Are patients okay with an algorithm making decisions about their treatment? Would patients rather have a human involved if it will take longer? The goal of healthcare is to make things better for everyone, empowering both patients and providers.

Anthony DiGiorgio, DO, MHA

I hate to admit it, but prior authorization (PA) remains a necessary component of healthcare, as some level of utilization control is essential. Healthcare, like any other service, is an elastic commodity to some extent—subject to both overutilization and underutilization. As neurosurgeons, we encounter unnecessary tests and procedures regularly, and while we condemn these practices, we must also acknowledge the legitimacy of payers questioning services that may lack clear indications.  How many needless MRI have you seen?  How many questionable spinal fusions? 

The article by Sahni et al. underscores the frustrations associated with PA, but it also highlights a broader systemic issue: the financial pressures driving PA. As long as large hospital conglomerates continue to charge exorbitant prices PA will persist. Large healthcare systems routinely charge 30x the cost for an MRI over independent imaging centers.  Of course insurance companies will questions scans with that level of markup!  This dynamic burdens physicians disproportionately. While hospital systems increase prices, it is physicians, not administrators, who are saddled with the time-consuming administrative burden of justifying care.

The findings of the article emphasize the diverging perceptions between patients, providers, and payers. Patients often report less burden, possibly because they engage less directly with the process. In contrast, providers face the brunt of administrative complexity, contributing to burnout and lost time.

Dr. Brazdzionis and Dr. Lopez provide excellent commentary on the article.  It is important to clarify that AI is not intended to make clinical decisions autonomously, at least not in the near term. Instead, AI is expected to streamline administrative tasks. However, the article accurately reflects providers’ concerns that these technologies may introduce new issues, such as bias, without appropriate oversight.

The challenge moving forward lies in balancing the need for utilization control with reducing administrative burden. PA is not inherently a bad practice—it is a necessary check in a system prone to overuse. However, the burden should not fall exclusively on physicians. As the article suggests, reforms must focus on streamlining processes, possibly with AI, and addressing hospital pricing strategies that inflate costs and drive the need for PA in the first place. Our role as physicians is to advocate for a system that ensures high-value care, minimizes unnecessary barriers, and places the burden where it belongs—on those profiting most from inflated healthcare costs.