Administrative Burden and Costs of Prior Authorizations in a Dermatology Department

Key Points

Question

How common and costly are prior authorizations (PAs) in 1 dermatology practice?

Findings

In this cross-sectional study at a large academic department spanning 11 clinical locations in Utah, PAs were found to impose a substantial burden, with median administrative costs per PA of $6.72 overall, and $15.80 for biologics. Most PA appeals were subsequently approved.

Meaning

Reforms to the PA system might increase practice efficiencies, decrease administrative costs, and improve patient outcomes.

This cross-sectional study examines the administrative burden and costs of prior authorizations in a large academic dermatology department over a 30-day period.

Abstract

Importance

Insurance companies use prior authorizations (PAs) to address inappropriate prescribing or unnecessary variations in care, most often for expensive medications. Prior authorizations negatively affect patient care and add costs and administrative burden to dermatology offices.

Objective

To quantify the administrative burden and costs of dermatology PAs.

Design, Setting, and Participants

The University of Utah Department of Dermatology employs 2 full-time and 8 part-time PA staff. In this cross-sectional study at a large academic department spanning 11 clinical locations, these staff itemized all PA-related encounters over a 30-day period in September 2016. Staff salary and benefits were publicly available. Data were analyzed between December 2018 and August 2019.

Main Outcomes and Measures

Proportion of visits requiring PAs, median administrative time to finalize a PA (either approval or denial after appeal), and median cost per PA type.

Results

In September 2016, 626 PAs were generated from 9512 patient encounters. Staff spent 169.7 hours directly handling PAs, costing a median of $6.72 per PA. Biologic PAs cost a median of $15.80 each and took as long as 31 business days to complete. The costliest PA equaled 106% of the associated visit’s Medicare reimbursement rate. Approval rates were 99.6% for procedures, 78.9% for biologics, and 58.2% for other medications. After appeal, 5 of 23 (21.7%) previously denied PAs were subsequently approved.

Conclusions and Relevance

Prior authorizations are costly to dermatology practices and their value appears limited for some requests. Fewer unnecessary PAs and appeals might increase practice efficiency and improve patient outcomes.

Introduction

Insurance companies use prior authorizations (PAs) to address inappropriate prescribing and unnecessary variations in care, commonly for expensive medications and procedures. However, PAs often delay and sometimes prevent essential and appropriate treatments and are associated with poor patient outcomes.^1,2 Prior authorizations add administrative burden to physician offices and appear to be increasing over time.^1,2,3,4,5 The American Medical Association has called for PA reform.³

Dermatology has been disproportionately affected by price advances in biologic medications and substantial price increases for previously inexpensive generic medications^6,7,8,9; we suspect that this has had an effect on the occurrence of PAs. Some health insurance plans even require PAs for diagnostic skin biopsies, despite nearly universal approval of these PAs.¹⁰

Data on the administrative burden and costs of PAs in dermatology are lacking. Thus, we sought to describe staffing costs required for PA management in our large academic center during September 2016.

Methods

In September 2016, the University of Utah’s Department of Dermatology employed 42 board-certified dermatologists and 9 advanced practice clinicians (2 nurse practitioners and 7 physician assistants), who staffed 9512 outpatient visits (general, pediatric, and surgical dermatology) across 11 clinic locations in Utah. On staff were 10 full-time employees who managed dermatology-related PAs, 8 who worked on PAs part-time, and 2 who managed PAs full-time.

For 30 days in September 2016, PA staff members itemized all PA-related encounters, recording patient identifiers, prescriber’s names, medication or procedure names, PA processing start and end times and dates, authorization status (approved, pending, or denied), and reason(s) for denial. Criteria for classification as a PA-related encounter included all telephone-based or computer-based (PA-related websites and emails) contact with pharmacies, insurance companies, patients, and prescribers. Detailed data-entry training was performed by 1 author (A.M.S.) before study initiation. Prior authorization staff assessed the database for face validity and provided weekly feedback on ease of use and recommendations for improvement.

We quantified the total number and type of PAs received, their final disposition, proportion of clinic visits requiring PAs, time required to complete each PA in minutes, business days needed to finalize a PA (either approval or denial after appeal), and administrative cost per PA. Time required to complete each PA (from PA request initiation to final decision) was summarized as median number of days required to process a PA, including time spent waiting for the decision and median time (in minutes) a staff member directly spent to process each PA. For PAs pending final decision at study termination, the total time spent until that point was used. Categories of PAs assessed herein included the following: (1) medications requiring a PA due to formulary status, (2) medications requiring step therapy due to formulary status, (3) medications requiring approval for quantity-limit or dose-limit exceptions, and (4) procedures requiring a PA. All of these trigger administrative work needed to help patients obtain either the prescribed or the on-formulary medication.

Type of PA was categorized as procedures (Mohs or excisions), medications (nonbiologic and biologic), light therapy (blue light and narrow-band ultraviolet B), and other. Other included laser surgery, nail avulsion, laser hair removal, iontophoresis machine, aero-spacer, and unspecified procedures.

To approximate direct PA costs, publicly available staff salary and benefits were used to calculate each employee’s minute-rate based on a 40-hour work week ([annual salary + benefits]/100 800 min). This minute-rate was multiplied by the time required to complete each PA.

For the most time-intensive and expensive PA, we identified the associated clinical encounter to calculate the proportion of PA cost as a function of visit reimbursement, based on the Medicare Physician Fee Schedule for calendar year 2016.¹¹ While physician time is costlier and adds to the overall administrative burden of PAs, physician involvement is more varied and tracking physician time was beyond our scope. The University of Utah’s Institutional Review Board approved this study. Informed consent was not required owing to use of retrospective, deidentified medical record data. Statistics were descriptive and performed with STATA, version 14.2 (StataCorp).

Results

In September 2016, 626 PAs for 615 patients were received from 9512 patient visits, equating to 0.07 PAs per patient encounter. A clinician seeing 35 patients per day would receive roughly 2.3 PAs per day (0.07 × 35). Medications accounted for 314 of 626 PA requests (50.2%), 235 procedures (37.5%), 62 light therapies (9.9%), and 15 other procedures (2.4%) (Figure 1). Among medications, 245 of 314 (78.0%) were for nonbiologics, and among procedures, 135 of 235 (57.5%) were for excisions (Table).

Figure 1. Histogram Showing Frequency of Prior Authorization Costs .

Light therapy and other prior authorization categories were of too low frequency to be visually comparable on the same axis but similarly demonstrated left-shifted histograms. USD indicates US dollars.

Table. Dermatology Prior Authorizations (PAs) by Type and Outcome Performed Over a 30-Day Period at a Single, Large Academic Dermatology Department.

PA type	Total, No. (%)	Median (IQR) [range]			No. (%)			Rate (%)a
		Business days to process 1 PA	Time to process 1 PA, min	Cost to process 1 PA, $	Approved	Denied	Pending	Approved	Denied
Total	626 (100)	1 (1-1) [1-31]b	12 (6-21) [1-148]	6.72 (3.54-11.80) [0.47-79.92]	398 (64)	76 (12)	152 (24)	84.0	16.0
Procedures	235 (37.5)	1 (1-1) [1-13]	9 (3-15) [1-51]	5.20 (1.44-9.12) [0.47-31.85]	225 (96)	1 (0)	9 (4)	99.6	0.4
Mohs surgery	100 (16.0)	1 (1-1) [1-13]	5 (2-14.5) [1-47]	2.40 (0.96-6.96) [0.47-22.56]	95 (95)	0	5 (5)	100.0	0
Excisions	135 (21.6)	1 (1-1) [1-12]	11 (6-17) [1-51]	6.50 (3.36-9.75) [0.47-31.85]	130 (96)	1 (1)	4 (3)	99.2	0.8
Medications	314 (50.2)	1 (1-2) [1-31]	13 (7-25) [2-148]	7.67 (4.13-14.30) [1.18-79.92]	115 (37)	69 (22)	130 (41)	62.5	37.5
Nonbiologic medications	245 (39.1)	1 (1-2) [1-16]	11 (6-18) [2-102]	6.49 (3.54-10.93) [1.18-60.18]	85 (35)	61 (25)	99 (40)	58.2	41.8
Biologic medications	69 (11.0)	1 (1-3.5) [1-31]	30 (19-46) [7-148]	15.80 (8.80-23.01) [3.96-79.92]	30 (44)	8 (12)	31 (45)	78.9	21.1
Light therapy	65 (10.4)	1 (1-1) [1-6]	12 (8-23) [1-80]	7.80 (5.20-14.95) [0.65-52.00]	48 (74)	6 (9)	11 (17)	88.9	11.1
Blue light therapy	40 (6.4)	1 (1-1) [1-6]	12.5 (8-20) [1-45]	8.13 (5.20-13.00) [0.65-29.25]	29 (72)	3 (8)	8 (20)	90.6	9.4
Narrow-band UVB therapy	25 (4.0)	1 (1-1) [1-6]	11.5 (9-28) [5-80]	7.48 (5.85-18.20) [3.25-52.00]	19 (76)	3 (12)	3 (12)	86.4	13.6
Otherc	12 (1.9)	1 (1-2) [1-20]	30 (10-38) [5-71]	17.39 (4.70-22.09) [3.25-33.37]	10 (83)	0	2 (17)	100.0	0

Abbreviations: IQR, interquartile range; UVB, ultraviolet B.

^aPercentages calculated after excluding prior authorizations with a pending status.

^bMedian number of encounters (contacting insurances, pharmacies, physicians, and patients) to complete a PA was 1 (IQR 1-1, [range 1-5]).

^cIncludes laser surgery (n = 7), nail avulsion (n = 1), laser hair removal (n = 1), iontophoresis machine (n = 1), aero-spacer (n = 1), and unspecified (n = 1) procedures.

Prior authorization staff documented 10 184 minutes handling PAs and spent a median (range) of 12 (1-148) minutes per PA (IQR, 6-21 minutes) (Table). Medication PAs were completed in a median (range) of 13 (2-148) minutes (IQR, 7-25 minutes); procedural PAs were completed in a median (range) of 9 (1-51) minutes (IQR, 3-15 minutes) (Figure 2). Prior authorizations were completed in a median (range) of 1 (1-31) business day.

Figure 2. Histogram Showing Frequency of Time Required to Complete Each Prior Authorization.

Light therapy and other prior authorization categories were of too low frequency to be visually comparable on the same axis but similarly demonstrated left-shifted histograms.

For the month studied, the administrative costs of PAs were $3454.15 for medications, $1456.00 for procedures, $690.30 for ultraviolet blue-light therapies, and $222.53 for other PAs. Overall median cost to complete a PA was $6.72. Median PA cost for Mohs surgery was $2.40, whereas biologics had a median cost of $15.80 (Table). The most expensive (time-consuming) PA took 148 minutes to complete for a biologic, omalizumab, at an administrative cost of $79.92, or 106% of the Medicare reimbursement rate for the associated clinic visit (Current Procedural Terminology code 99213).

By the end of the month, 398 of 626 (64%) of PAs were approved overall, 76 (12%) were denied, and 152 (24%) were pending. Of the denied PAs, 23 of 76 (30.3%) were appealed. Of these, 5 of 23 (21.7%) were later approved, 4 of 23 (17.4%) were ultimately denied, and 14 of 23 (60.9%) remained pending. Procedural PAs had 99.6% approval rates: 100% (95 of 95) for Mohs surgery and 96% (130 of 131) for excisions. Medication PAs were approved 62.5% of the time: 58.2% (85 of 146) for nonbiologics and 78.9% (30 of 38) for biologics (Table and eTable in the Supplement).

Discussion

Excluding physician time, our institution required 170 staff hours addressing PAs over a 30-day period, costing $5822.98 in nonreimbursable staff hours. Based on the estimated 6000 visits per year for US dermatologists,¹² a single dermatologist may receive approximately 420 PA requests annually (0.07 × 6000). Costs likely vary substantially between institutions and practices, but our data suggest that even with an efficient, dedicated PA group, costs are high.

Previous survey data have highlighted dermatologists’ frustrations with PA administrative burden; many dermatologists report that PAs have negatively affected at least 1 patient daily.⁴ These effects on patient care and physician practice, combined with the data demonstrating substantial nonreimbursable time and cost, highlight the need for specialty-wide PA reform.

The median cost of PAs for biologic medications determined from this study was $15.80 per PA. While only comprising 11% of PAs, biologics required 23% of total PA staff time and up to 31 business days to complete. Given the continual emergence of new biologics and dosing regimens, as well as ever-increasing prices, we worry this may signal worsening PA burden.^13,14 Likewise, our data likely underestimate PA costs for most practices. First, our analysis did not account for clinician time, which can include consideration of alternative therapies, writing appeal letters, and conducting peer-to-peer authorizations. Second, the logistical difficulty and time-consuming nature of tracking each staff member’s time was substantial and thus time was likely underreported. Third, multiple PAs were still pending at the end of the study, thus truncating their attributed time and cost. Together, these estimates likely constitute a conservative floor for time and cost. A recently proposed strategy to reduce PA wait times consists of employing a pharmacy technician, knowledgeable of PAs and insurance formularies, to follow PAs until approval.¹ This, however, would likely be unwieldy for many dermatology offices.

A solution to this problem could be reducing or eliminating unnecessary PAs. One study identified several generic medications (tretinoin, clobetasol, topical tacrolimus) for which PAs could be removed without substantial consequence to payers.¹ We found near-100% approval rates for dermatologic procedures and some nonbiologic medications. In addition, 21.7% of PAs that were initially denied were later approved on appeal. Reducing these PAs would decrease administrative costs for insurers and practices, highlighting the need for collaborative efforts.¹⁵

Additionally, insurers should work with dermatologists to identify therapies where PAs may cause harm and increase costs for little regulatory gain, as was recently noted.^15,16 For example, one of our psoriasis specialists saw a middle-aged patient with severely flaring psoriasis. Cyclosporine was selected as the optimal therapy to rapidly reverse the flare, but the patient experienced a 5-day delay in receiving cyclosporine due to a PA requirement. During this time, the patient developed erythrodermic psoriasis requiring a prolonged hospitalization and infliximab infusion to control the acute flare, costing the patient’s insurance tens of thousands of dollars and substantially affecting the patient’s quality of life. The utility of cyclosporine PAs is questionable. Cyclosporine is a relatively inexpensive treatment for moderate to severe psoriasis, and from our experience, it is unlikely to be overused.

Limitations

This study has limitations. First, it was performed at a single academic center with specialty dermatology clinics. Thus, we see many patients with complex conditions and those who have been referred from other practices who likely require more medication-related PAs than the typical dermatology practice. Second, our study lacked patient insurance information. It is possible our experience differs from other practice types and geographical locations. Third, while weekly trainings helped minimize person-to-person variation and improve reporting accuracy in time-to-completion data, we could not verify the accuracy of each employee’s inputs. Reviewing each employee’s self-reported times, we believe this led to underreporting, as no instances occurred where employees reported more time spent on PAs than was expected (during a standard workday). Fourth, we did not assess indirect PA costs on clinicians’ time or effects on patient quality of life.

Despite these limitations, our findings are similar to prior data.¹ Our department is large, accepts a variety of Medicare, Medicaid, and private insurances, and includes many general dermatologists who vary in clinical styles and practice setups (2 hospital-based and 9 community-based clinics). While insurances and PA requirements vary by state, our aim was to quantify actual administrative burden per PA with less emphasis on the actual PA counts by type, which vary geographically and by prescribing patterns.

Conclusions

Our data should alert many stakeholders to the costly, time-consuming, and nonreimbursable¹⁷ nature of PAs to dermatology practices. In the short term, we need to encourage payers to simplify PAs and reduce PA burden, perhaps first by eliminating unnecessary PAs and appeals.

Supplement.

eTable. List of all medications requiring a prior authorization in September 2016 and their prior authorization status on the last day of the month