Administrative costs of prior authorizations for glucagon-like peptide-1 agonists

Abstract

BACKGROUND:

Prior authorizations (PAs) help ensure appropriate prescription drug use but can be time-consuming for prescribers, clinic staff, and patients. Having a centralized, pharmacy-led PA process has shown significantly lower time to approval when compared with other decentralized models. However, the administrative impact of processing PA requests for glucagon-like peptide-1 (GLP-1) agonists prescribed for weight loss is unknown.

OBJECTIVE:

To compare the total staff time spent on GLP-1 agonists prescribed for weight loss vs GLP-1 agonists for diabetes and other prescription medications.

METHODS:

This prospective observational study was conducted at UC Davis Health, a large academic health system from October 2024 to April 2025. Six licensed pharmacy technicians from a centralized pharmacy PA team participated in a self-reported time survey. PA tasks were categorized into 3 groups: (1) GLP-1 agonists for weight loss, (2) GLP-1 agonists for diabetes, and (3) top 10 non–GLP-1 agonist medications processed by volume at UC Davis Health. The primary outcome is defined as total hands-on time per PA, measured from PA initiation to removal from the pharmacy technician’s work queue following the final decision by the third-party payer. A sample size of 50 PAs per group was selected based on projected mean PA process times of 10 minutes for weight loss GLP-1 agonists and 7 minutes for diabetes GLP-1 agonists (β = 0.2, α = 0.05). The Mann-Whitney U-test was applied to compare continuous variables (hands-on time, cost), whereas the Fisher’s exact test was used for dichotomous variables (approval rate).

RESULTS:

A total of 150 PAs were collected. A self-reported time survey showed that weight loss GLP-1 agonists were associated with a significantly higher mean time per PA (7.1 minutes; 95% CI = 3.4-10.8; P < 0.0001), lower approval rate at 48% vs 90% (odds ratio [OR] = 0.10, 95% CI = 0.04-0.31; P < 0.0001), and higher mean cost per PA ($6.74; 95% CI = 3.21-10.27) compared with diabetes GLP-1 agonist medications. The PA process for weight loss GLP-1 agonists was longer (7.2 minutes; 95% CI = 3.3-11.1), more costly ($6.77; 95% CI = 3.08-10.46), and less likely to be approved (48% vs 90%; OR = 0.17; 95% CI = 0.07-0.44) compared with non–GLP-1 agonist medications. There were no differences between the diabetes GLP-1 agonist and top 10 non–GLP-1 agonist medications.

CONCLUSIONS:

Total hands-on time spent on GLP-1 agonists prescribed for weight loss was significantly longer compared with GLP-1 agonists for diabetes and other non–GLP-1 agonist medications. Additionally, GLP-1 agonists for weight loss had lower approval rates and higher cost per PA.

Plain language summary

This study compared staff processing times, approval rates, and costs associated with prior authorizations (PAs) for 3 different types of medications: glucagon-like peptide-1 (GLP-1) receptor agonists prescribed for weight loss, GLP-1 agonists prescribed for diabetes, and the top 10 medications prescribed at the study site that required a PA. GLP-1 agonists for weight loss had significantly longer processing times, lower approval rates, and higher costs per PA compared with the other 2 groups.

Implications for managed care pharmacy

This study quantifies the administrative burden associated with PAs for GLP-1 agonists prescribed for weight loss. This concept was demonstrated through longer processing times, lower approval rates, and higher costs per PA. The findings may help guide managed care organizations on formulary decisions and streamline workflows to reduce delays to patient care. Additionally, the results may also encourage health systems to implement workflow changes for other medications that are similarly burdensome.

Prior authorizations (PAs) are intended to ensure appropriate prescription drug use but can often be time-consuming for prescribers, clinic staff, pharmacy teams, and patients. According to a 2024 American Medical Association physician survey, respondents indicated the PA process impacts patient outcomes in a negative way and fuels physician burnout, leading to unnecessary spending and additional office visits. Typically, PAs require detailed medical justification, including documentation of indication, relevant laboratory results, and previous treatments used. Given these requirements and delays, clinicians reported that their patients would have to endure ineffective initial treatments (eg, because of step therapy requirements), require more clinician visits, or face hospitalization. The vast majority of American Medical Association physician survey respondents (88%) reported that PAs led to higher overall utilization of health care resources.¹

To streamline this process for clinicians, standardized forms, electronic submissions, and centralized teams are used. A study at UC Davis Health (UCDH) published in 2016 found that a centralized, pharmacy technician–led PA process significantly reduced labor costs and time to approval.² Even with these resources, meeting PA requirements can often place a significant administrative burden on health care physicians, particularly in settings without dedicated staff and sufficient resources. These administrative duties have been correlated to lower levels of career satisfaction and a higher rate of burnout for physicians.^1,3 Previous studies illustrate the additional costs PAs place on patients and the health care delivery system. A 2014 analysis of health care payment claims for patients with schizophrenia and bipolar disorder showed that patients with formulary restrictions, inclusive of PAs, were more likely to be hospitalized and had higher inpatient costs and higher total costs.⁴ Meanwhile, a time series analysis of Medicare beneficiaries with opioid use disorders found that removing PA requirements was associated with an increase in the use of buprenorphine-naloxone, which was subsequently linked to a reduction in emergency department visits and hospitalizations.⁵ To reduce clinician burden and provide better access to care for patients, more than 50 health insurers have pledged to streamline the PA processes.⁶ Little progress has been made to the PA process, despite advocacy for legislative action and changes in payer plans. As a result, stakeholders remain skeptical of improvement or enforceable accountability in the system.^6,7

With the growing volume of popular medications such as glucagon-like peptide-1 (GLP-1) agonists, the full extent of their administrative impact on this PA process remains unclear. The American Diabetes Association’s recommendations have led to decreasing formulary restrictions for GLP-1 agonists used in treating type 2 diabetes, but it is unclear whether the same trend applies to GLP-1 agonists approved for weight management.⁸

The primary objective of this study is to evaluate the administrative burden of the PA process by comparing the total staff time spent on GLP-1 agonists prescribed for weight loss, GLP-1 agonists for diabetes, and non–GLP-1 agonist medications. Secondary objectives include assessing PA approval rates and estimated labor costs.

Methods

This prospective observational study aimed to compare the processing times of PAs at UCDH-affiliated outpatient clinics between October 2024 and April 2025. The institutional review board at the University of California, Davis, determined that this study was exempt from review. Consecutively processed PAs were considered for inclusion to mitigate the potential for selection bias. PAs were divided into 3 groups: (1) GLP-1 agonists prescribed for weight loss (n = 50), (2) GLP-1 agonists prescribed for type 2 diabetes (n = 50), and (3) top 10 non–GLP-1 agonist medications (by volume) that required a PA at the study site (n = 50) (Supplementary Table 3 ^{(853.1KB, pdf)} , available in online article). PA requests that were initially submitted but later determined to be not required were excluded. In the diabetes group, off-label GLP-1 agonist prescriptions for weight management were excluded to ensure outcomes reflected US Food and Drug Administration (FDA)–approved use. Additionally, weight loss GLP-1 agonist PAs processed for Medicare Part D beneficiaries were excluded because, under the Medicare Prescription Drug, Improvement, and Modernization Act of 2003, weight loss drugs are not covered by Part D. Because these requests are categorically nonreimbursable and do not undergo the standard PA review process, they do not reflect the administrative burden this study is aimed at evaluating.⁹

PAs were completed by 6 dedicated pharmacy technicians in a centralized setting following a standardized workflow. Two technicians were assigned to each study arm (weight loss GLP-1 agonists, diabetes GLP-1 agonists, and top 10 non–GLP-1 agonist medications). Each technician completed 25 PA reviews within their assigned category. The PA requests were either automatically transmitted from an outpatient UCDH clinic at the time of prescribing or manually initiated by the pharmacy, which was then sent to the centralized PA team. Once received by the pharmacy technician, the patient’s medical record was reviewed for insurance type, indication of therapy, medications tried and failed, and other information as needed. Upon completion of the medical record review, the technician would submit the completed PA electronically to the insurance for review.

A self-reported time survey was conducted to assess the total time each technician spent processing a PA. Pharmacy technicians systematically collected data, including medication name, time spent completing a PA, and the PA outcome. Technicians recorded their start and stop times throughout each PA submission process. The timer began when the technician initiated the PA and was paused during periods outside their control. This included waiting for a clinician’s response, while on hold with the insurance, or awaiting a response after PA submission. The timer resumed for any follow-up tasks, such as submission of additional documentation requested by the insurance. Employee breaks were also excluded from the recorded time. The total hands-on time was recorded until the PA was removed from the technician’s work queue, which occurred after a final determination letter (such as approval or denial) was received from the insurance, and no further follow-up action was needed.

A priori power analysis was performed and determined that 50 PAs in each group would be necessary to detect a significant difference with an α of 0.05 and a power of 0.80. Owing to limited preliminary data on the outcome’s variance, investigator judgment and literature were used to determine a sample size estimate. This estimate assumed that 95% of outcome times would fall within a specified range of 45 minutes with an SD of 35 minutes. The Mann-Whitney U-test was used to compare the time spent per PA and the cost per PA, whereas the Fisher’s exact test was used to compare the approval rates of PAs processed across the 3 study groups.

Results

There were a total of 16,676 requests generated during the study period. Of these, a total of 6,920 requests were for weight loss GLP-1 agonists (3,613), diabetes GLP-1 agonists (1,734), and top 10 non–GLP-1 agonist medications (2,023). The first 50 eligible PA submissions were identified for each group, yielding a total of 150 PA requests (Figure 1). Across the 3 groups, 6 submissions ultimately did not require a PA and were therefore excluded and replaced to maintain a sample size of 50 per group. The final analysis therefore included 150 PA requests, evenly distributed across weight loss GLP-1 agonists (n = 50), diabetes GLP-1 agonists (n = 50), and the top 10 non–GLP-1 agonist medications (n = 50).

FIGURE 1.

Study Sample Attrition Diagram

GLP-1 = glucagon-like peptide-1 receptor agonist; PA = prior authorization.

Primary and secondary endpoints are shown in Tables 1 and 2. The average hands-on time required to complete a PA request was significantly longer for GLP-1 agonists prescribed for weight loss compared with the other 2 groups. The mean processing time (± SD) was 13.5 minutes (± 12.7) for weight loss GLP-1 agonists, 6.4 minutes (± 4.5) for diabetes GLP-1 agonists, and 6.3 minutes (± 6.1) for the top 10 non–GLP-1 agonist group (Supplementary Figure 2 ^{(853.1KB, pdf)} ). In Figure 2, weight loss GLP-1 agonists were associated with a significantly higher time spent per PA relative to non–GLP-1 agonist medications (mean difference = 7.2 minutes; 95% CI = 3.3-11.1; P < 0.0001). The PA process for weight loss GLP-1 agonists was also significantly longer (mean difference = 7.1 minutes; 95% CI = 3.4-10.8) vs diabetes GLP-1 agonists. There was no difference between diabetes GLP-1 agonists and non–GLP-1 agonist medications regarding mean time spent per PA (0.1 minutes; 95% CI  = −2.0 to 2.2; P = 0.39).

TABLE 1.

Outcome Comparison: Weight Loss vs Top 10 Non–GLP-1 Agonist PAs

	Weight loss (n = 50)					Top 10 non–GLP-1 agonists (n = 50)					P valuea	Estimated between-group difference (95% CI)
	%	Mean	SD	Median	Range	%	Mean	SD	Median	Range
Total staff time, minutes		13.5	12.7	10.8	(1.8-73.2)		6.3	6.1	4.8	(1.2-31.2)	0.0001	7.2 (3.3-11.1)
PA approval rate, %	48					84					0.0001	0.17 (0.07-0.44)
Cost/PA, $b		12.8	12.0	10.2	(1.7-69.2)		6.0	5.7	4.5	(1.1-29.5)	0.0001	6.77 (3.08-10.4)

^a For continuous outcomes (eg, time and cost), P values and 95% CIs were estimated using the Mann-Whitney U-test. The Fisher’s exact test was used to compare dichotomous outcomes (eg, approval rate).

^b Cost/PA was estimated based on median salary and benefits for the pharmacy technician involved in the PA.

GLP-1 = glucagon-like peptide-1; PA = prior authorization.

TABLE 2.

Outcome Comparison: Weight Loss vs Diabetes

	Weight loss (n = 50)					Diabetes mellitus (n = 50)					P valuea	Estimated between-group difference (95% CI)
	%	Mean	SD	Median	Range	%	Mean	SD	Median	Range
Total staff time, minutes		13.5	12.7	10.8	(1.8-73.2)		6.4	4.5	4.8	(1.8-22.8)	0.0001	7.1 (3.4-10.8)
PA approval rate, %	48					90					0.0001	0.10 (0.04-0.31)
Cost/PA, $b		12.8	12.0	10.2	(1.7-69.2)		6.0	4.3	4.5	(1.7-21.6)	0.0001	6.74 (3.21-10.27)

^a For continuous outcomes (eg, time and cost), P values and 95% CIs were estimated using the Mann-Whitney U-test. The Fisher’s exact test was used to compare dichotomous outcomes (eg, approval rate).

^b Cost/PA was estimated based on median salary and benefits for the pharmacy technician involved in the PA.

PA = prior authorization.

FIGURE 2.

PA Process Time of GLP-1s vs Comparator (Minutes)

DM = diabetes mellitus; GLP-1 = glucagon-like peptide-1 receptor agonist; PA = prior authorization.

As shown in Figure 3, the PA approval rate for GLP-1 agonists for weight loss was the lowest at 48%, compared with 84% for the comparator medications (P < 0.0001) and 90% for GLP-1 agonists for diabetes (P < 0.0001). No significant difference was found in approval rates between PAs for GLP-1 agonists used for diabetes and non–GLP-1 agonist medications (odds ratio = 1.71; 95% CI = 0.56-5.0; P = 0.55). Based on the total hands-on staff time, the difference in mean cost per PA was $6.77 (95% CI = 3.08-10.46; P < 0.0001) for weight loss GLP-1 agonists and $0.03 (95% CI = −1.95 to 2.01) for diabetes GLP-1 agonists, compared with the non–GLP-1 agonist medications, respectively (Supplementary Figure 3 ^{(853.1KB, pdf)} ). The difference in mean cost per PA was $6.74 (95% CI = 3.21-10.27; P < 0.0001) when comparing PAs for weight loss and diabetes GLP-1 agonists with each other (Supplementary Figure 4 ^{(853.1KB, pdf)} ).

FIGURE 3.

PA Approval Rate (%)

DM = diabetes mellitus; GLP-1 = glucagon-like peptide-1 receptor agonist; OR = odds ratio; PA = prior authorization.

Discussion

This is the first study to directly compare the administrative burden of the PA process for different indications within the same drug class. The main finding demonstrates that hands-on time by pharmacy technicians to complete a PA request for weight loss GLP-1 agonists was more than double the time required to process diabetes GLP-1 agonists and non–GLP-1 agonist medications. Requests for weight loss GLP-1 agonists were also associated with a higher degree of variability in processing time, with a significant proportion taking nearly 30 minutes to complete.

Several potential factors likely contributed to these disparities in processing time. Follow-up tasks and requests for additional information from clinicians and/or payers were much more common in processing weight loss GLP-1 agonists, based on anecdotal observation. Hands-on time to complete clerical tasks (eg, additional phone calls and messaging) reportedly was the primary driver of additional delays. In addition, incongruity and unfamiliarity with the criteria for PA between agents and payers likely also contributed to the disparities observed. PAs for diabetes GLP-1 agonists and non–GLP-1 agonist medications followed well-established workflows. In contrast, weight loss GLP-1 agonists involved newer, less standardized criteria that varied across health plans. These requirements included specific restrictions for body mass index and comorbidities, as well as prerequisite lifestyle modifications.¹⁰ Processing PAs for these agents involved assiduously navigating a range of evolving queries. Finally, a lack of clarity and consistency across payer plans further compounded the volume of follow-up requests, exacerbating delays.

At the clinic level, these bureaucratic hurdles and delays may be magnified by a limited or fragmented capacity to gather, compile, document, and relay information. Patient enthusiasm to initiate therapy may translate into follow-up calls to clinics and pharmacies, generating additional touchpoints, workflow friction, and frustration. In this study, dedicated staff were well trained, equipped, and positioned to handle these requests. Therefore, these findings may underestimate the impact of this process for teams (eg, everyday clinic staff) that are not exclusively focused on these types of PAs. The administrative burden for organizations without a centralized team is unknown but likely even more substantial.

Existing studies have primarily evaluated overall PA turnaround time across various care teams. Carlisle et al reported that medication PAs in a large dermatology department required a median PA processing time of 11 minutes and 30 minutes for nonbiologics and biologics, respectfully.¹¹ Similarities in turnaround time between Carlisle et al and this study may reflect comparable measurement of active staff time, dedicated staff, and use of structured workflows for PAs. A study performed at a urology practice observed a mean time of 23.1 minutes spent per PA request. The authors also described associated delays in the PA process, noting a median of 2 days to receive an initial coverage decision and an additional median of 10 days when an appeal was involved. These observations highlight additional elements in the overall PA timeline that were not examined in the current study.¹²

The resources necessary to process newer GLP-1 agonists may compel organizations to reassess the value and opportunity cost of facilitating patient access to these medications. The cumulative constraint of ballooning administrative costs and slumping revenue are energetically juxtaposed with powerful marketing and social pressures driving demand. Potential solutions appear limited and resource intensive. Foremost, it is essential to set expectations that there will likely be delays ahead with staff, clinicians, and patients. Creating a core, centralized team of pharmacy technicians that is dedicated to weight loss GLP-1 agonists affords multiple advantages to expediting processing times. However, this tactic obviously requires an economy of scale that may be beyond the reach of smaller institutions and health systems. Artificial intelligence (AI) may offer the potential to lessen administrative burdens for staff in the intermediate and long term. However, implementation of AI could also introduce new bureaucratic layers and procedural complexities (eg, oversight, validation, compliance). In the near term, automation through electronic PA systems (eg, prepopulated required fields) may help streamline submissions and decrease processing times.¹³

The institutional operating expenses of processing PAs were also substantial. In operational terms, roughly one-quarter of total pharmacy resources allocated to PAs were used to process weight loss GLP-1 agonists. In contrast, less than 7% of labor was expended on PAs for diabetes GLP-1 agonists, despite accounting for nearly half as many requests as weight loss GLP-1 agonists. For example, the attributable cost of processing PAs for weight loss GLP-1 agonists could easily exceed $625,000 per year for a large academic health system, based on the proportion of time-on-task and assuming a budget of $2,500,000 (18 full-time equivalents). Parity in PA processing efficiency, in this hypothetical scenario, would free up more than 2.3 full-time equivalents and $300,000 for other pressing operational demands.

The results of this study highlight the administrative burden that PAs for weight loss GLP-1 agonists could have on health care staff. The low approval rating compared with other medications indicate these medications may have more complex requirements that are challenging to document for payers. These burdens have impacts on staff supporting the clinics and physicians prescribing these medications that were largely unknown. Future directions for this work include measuring PA processing time for organizations without a central team dedicated to this work. In addition, the incorporation of automation tools and AI to assess their impact on workflow efficiency should be evaluated. As coverage policies continue to evolve, it will be important to assess how these changes affect PA approval rates and the resulting administrative burden on staff and patients. Lastly, more studies are also needed to evaluate the impact of weight loss GLP-1 agonist PAs across institutions of varying sizes and resources to assess generalizability.

One of the key strengths of this study is its prospective design, which enabled real-time monitoring of the data collection process by the technicians. This approach led to a more adaptive and robust design, ensuring the data collected were true to the study’s methodology. Another strength is how this study was designed to consider FDA-approved indications. In the diabetes group specifically, GLP-1 agonist prescriptions written off-label for weight management were excluded to avoid confounding PA approval rates and ensure that the PA outcomes reflect standard, FDA-approved use. Lastly, the study compared 3 distinct groups: weight loss GLP-1 agonists, diabetes GLP-1 agonists, and a comparator group of the top 10 non–GLP-1 agonist medications. By examining these groups, significant differences were identified between weight loss GLP-1 agonists and the comparator group, as well as diabetes and weight loss GLP-1 agonists. This distinction highlights that the increased time burden on staff is associated with the weight loss indication, rather than the GLP-1 agonist class as a whole.

LIMITATIONS

This study has several limitations to be considered. First, the study relied on the technicians to self-report the time they spent processing the PA. Although a standardized survey was used and training was provided, self-reported time studies are subject to estimation error and recall bias. Moreover, the observational nature of the design may have introduced a Hawthorne effect, wherein pharmacy technicians temporarily behaved more fastidiously than was typical of their practice. This may have led to an underestimation of typical processing times as compared with routine, unobserved workflows. Second, the results reflect the PA workflow of a single-site academic medical center with an established and experienced pharmacy-led PA team. This limits the generalizability to other settings without the same resources or workflows. Third, time and costs per PA were calculated solely based on total technician time spent per PA. However, this did not account for additional clinic staff who may have been involved in the process. Future studies may be necessary to quantify the full burden for other health care staff. Lastly, the study did not control for payer type or specific insurance plan criteria. Widely varying coverage policies and documentation requirements among payers likely contribute to variability in both processing time and approval rates. For example, GLP-1 receptor agonists for weight loss, which often have more extensive criteria, may have lower approval rates owing to stricter payer requirements, plan exclusions for weight management therapies, or patients not meeting diagnostic thresholds. These variations in payer criteria are not limited to the weight loss indication particularly but can be considered a compounding factor when comparing across many other therapeutic indications.

Conclusions

This study highlights the likely administrative impact that weight loss GLP-1 agonists have on PAs and reinforces the time and cost burden for health care staff. These findings highlight the growing operational strain imposed by complex insurer requirements, particularly when access to high-cost medications is restricted based on indication. As the landscape of GLP-1 agonist medications continues to evolve with possibly new indications in the future, it is important to consider the financial and administrative impact this may have on institutions.

Disclosures

The authors report no disclosures.