Uncompensated Medical Provider Costs Associated with Prior Authorization for Prescription Medications in an HIV Clinic

Abstract

Although prior authorization (PA) for prescription medications is widely employed for cost-containment for third-party insurers, it is a frustrating aspect of outpatient clinical care that imposes uncompensated costs to medical providers. To characterize these costs, we monitored the PA-associated administrative and operational process at the University of Alabama at Birmingham 1917 HIV Clinic over a 2-year period. A total of 288 PAs were processed with a mean (± standard deviation [SD]) of days 3.1 ± 5.8 delay in the patient’s access to medication. A mean (± SD) of 26.8 ± 18.4 min was spent by the nurse practitioner and 6.5 ± 2.9 min was spent by a clerk per PA. Nearly three-quarters (73%) of PAs were approved, 18% were denied, and 10% were voided. The mean (± SD) pages of paperwork was 5.8 ± 6.5. The overall cost was $41.60 per PA. Although evidence supports that PA reduces third-party expenditures, it significantly delays medication accessibility for patients and imposes high costs that negatively impact operating margins for health care providers.

In 2007, US health expenditures reached $2.2 trillion, and prescription drug expenditures reached $227.5 billion [1]. Formularies or preferred drug lists and prior authorizations (PAs) evolved to control the rapid growth of pharmacy expenditures and to manage pharmaceutical drug-selection practices [2]. Although PA is widely implemented as a cost-containment measure, it is a labor-intensive procedure for health care providers, patients, pharmacists, and pharmacy benefit plans [3]. PA requires that the prescriber receive preapproval for a medication under the terms of the benefit plan [4]. PA programs were designed to change prescribing behavior, contain costs, and approve coverage of medications when specific criteria are met [5]. Increasingly, commercial insurance companies, Medicaid, Medicare Part D plans, and Health Maintenance Organizations use PA to target new, costly, or potentially toxic medications and to encourage the use of less expensive, safer alternatives. In this article, the term PA will represent the associated requisite activities, including the acquisition, completion, and transmission of forms by the medical provider that are used to determine whether a prescription medication will be authorized for payment by the patient’s third-party insurer.

Although formularies and PA are effective ways to reduce health care costs, they are associated with unintended effects. PA has the potential to divert clinician time away from direct patient care, delay patient receipt of and adherence with the new medication, negatively impact clinician and patient satisfaction, and create administrative costs. Canadian physicians report dissatisfaction with the time burden for completing up to 10 PAs per day, and some physicians refused to complete them [6]. In the United States, following the implementation of Medicare Part D in 2006, the administrative chaos that ensued was largely attributed to restrictive formularies and PA [7]. More than 2 years after the implementation of Medicare Part D, several challenges, including PA, remain unresolved. As of 2008, there were 1824 stand-alone Part D benefit plans available [8], and the number of available plans varied by region. The lowest was 27 (in Alaska), and the highest was 63 (in Pennsylvania and West Virginia). A study involving 3247 psychiatrists who treated dually Medicare- and Medicaid-eligible patients found that the psychiatrist or their staff spent 45 min in administrative tasks for every hour of direct patient care [9].

Little is known about PA in other health care settings. Using data collected at the UAB human immunodeficiency virus (HIV) outpatient clinic, we aimed to quantify the administrative burden required to complete PA, the PA-related costs, and delays in medication acquisition for patients to gain a more comprehensive understanding of the impact of the PA process.

METHODS

The study was conducted at the UAB 1917 HIV/AIDS Outpatient Clinic, where among the entire clinic population, ~23% of patients were covered by Medicare, 13% were covered by Medicaid, 39% were covered by a commercial payor, and 25% had no insurance. The UAB 1917 HIV/AIDS Clinic Cohort Database Project [10] is a 100% quality-controlled Institutional Review Board–approved protocol [11]. The UAB 1917 Clinic uses a proprietary, locally programmed electronic medical record. PA requests for the HIV-practice of 11 attending and 8 fellow infectious diseases physicians, 2 nurse practitioners (NPs), and 1 nurse midwife were processed by an experienced NP (J.L.R.) and a clerk. In the absence of the NP, senior registered nurses managed PA. The NP was vested with the authority to act on the prescriber’s behalf. From March 2006 through February 2008, all PA-related activities and associated time for completion were recorded as they were executed. Data were entered by the clerk and NP into a database to record PA-related variables and minutes of NP and clerk processing time. PA procedures are depicted in Figure 1. A cumulative number of minutes was tallied and recorded by the NP and clerk immediately after completion of each task (Figure 1). The statistician (H.Y.L.) and NP worked collaboratively to ensure the quality of the data. All documents were maintained in files accessible to the NP and clerk.

Figure 1.

Clinic prior authorization (PA) processing procedures. NP, nurse practitioner.

The number of days between the PA submissions until a final benefit plan decision was calculated as the number of calendar days between events. The number of approved days of medication was calculated as the number of days between the dates of approval until the PA’s expiration.

Direct administrative costs were calculated as the rates of pay per minute plus fringe benefit costs (27% of salary) for the NP ($0.83 per min) and clerk ($0.29 per min) multiplied by the total minutes that they spent processing PAs. The rate of the NP’s annual salary plus benefits ($103,584) is comparable to that of nationally based salaries in 2007 for NPs working 40 h a week in an HIV clinic ($100,589) [12]. Opportunity costs were calculated by dividing the minutes that the NP was involved by 30 min, which represents the time allotted for a return patient visit multiplied by $55.19, the clinic’s Medicare reimbursement for an “incident to” level 3 established patient visit. The total number of pages of documents (PA request forms and supporting medical record documentation) required for each PA was recorded at the cost of $0.25 per page [13].

Patient demographic data, benefit plan, and PA characteristics were summarized using descriptive statistics. For continuous variables, mean values and standard deviations (SDs) were calculated. For categorical variables, frequency and percentages were listed. Analysis of variance (ANOVA) tests were used to evaluate differences among 3 categories of benefit plans (Commercial, Medicare Part D, and Alabama Medicaid) for processing time, days of approved medication, processing time requirements, and the number of pages of required paperwork. For a significant result in an ANOVA test, pairwise comparisons were performed using the Tukey post hoc test.

RESULTS

From a population of 1683 patients (50% of whom were white and 23% of whom were female) who received care during the 24-month study, there were 288 PA requests for 144 individual patients (8.6%), representing a mean (± SD) of 2 ± 2.33 PAs for each patient for whom a PA was requested. More than one-half of PAs came from 33 Medicare Part D benefit plans (Table 1). Alabama Medicaid required the most PAs (42% of patients).

Table 1.

Prior Authorization (PA) Request Characteristics

Variable	Commercial health insurance	Medicaid	Medicare	Total
PA request, proportion (%) of PAs
Approved	10/14 (71)	101/122 (83)	98/152 (64)	209/288 (73)
Denied	3/14 (21)	15/122 (12)	33/152 (22)	51/288 (18)
Voided	1/14 (7)	6/122 (5)	21/152 (14)	28/288 (10)
Total	14/288 (5)	122/288 (42)	152/288 (53)	288/288 (100)
Pages, mean no. of pages per PA (±SD)a	4.9 ± 6.8	4.3 ± 4.3	7.1 ± 7.7	5.8 ± 6.5
Clerk time, mean min (±SD)	6.9 ± 3.7	5.7 ± 1.7	6.7 ± 3.2	6.5 ± 2.9
NP time, mean min (±SD)	32.9 ± 29.1	24.5 ± 14.4	28.1 ± 19.7	26.8 ± 18.4
Total time, mean min (±SD)	36.8 ± 26.9	24.0 ± 15.6	29.2 ± 20.3	27.4 ± 19.0
Completion of PA, mean days (±SD)b	6.3 ± 8.6	2.1 ± 5.1	3.6 ± 5.9	3.1 ± 5.8
Duration of approved therapy, mean days (±SD)c	275.0 ± 91.4	150.0 ± 93.1	306.0 ± 119.0	223 ± 131

NOTE. NP, nurse practitioner; SD, standard deviation.

^aP< .001 (Medicare more than Medicaid).

^bP=.013 (commercial health insurance more than Medicaid).

^cP< .001 (Medicaid less than commercial health care or Medicare).

PAs for pain management (eg, methadone sustained release or oxycodone), behavioral health (eg, duloxetine hydrochloride or dronabinol), and anti-infective (eg, atovaquone or voriconazole) agents constituted 60% of the PAs. There were 37 PAs (13%) for HIV antiretroviral medications (eg, fixed-dose lopinavir-ritonavir or fixed-dose efavirenz-emtricitabine-tenofovir). The majority of PAs (78%) were first-time requests. Overall, 73% of PAs were approved. Of the 224 initial PAs, 67% were approved, whereas 92% of the 64 renewal PAs were approved. “Alternate Formulary Medication” was the most common reason for PA denial (accounting for 88% of denials), including all 5 denials for antiretroviral medications. The antiretroviral denials were based on the availability of formulary single agents versus fixed-dose combinations, which are prescribed to reduce pill burden and improve adherence. All “voided” PAs requests (10%) were resolved by formulary substitution by the NP, which obviated the need to submit a completed request form. The mean number of pages of paperwork was 5.8 pages per PA. Medicare Part D, Medicaid, and commercial plans required 7.1, 4.3, and 4.9 pages per PA, respectively. Medicare Part D plans required significantly more paperwork than did Medicaid (P = .001, by Tukey test).

The mean number of days required for the clinic to process a PA was 3.1 days. Because of the use of an on-line system, Medicaid PAs required significantly less time (mean time required, 2.1 days) than did PAs involving commercial plans (mean time required, 6.3 days; P = .034, by Tukey test). On average, approved PAs were authorized for 223 days. The average number of days of approved medication was significantly shorter in duration for patients with Medicaid (Mean duration, 150 days; P < .001, by Tukey test).

Direct costs to the provider included NP and clerk expenses (salary plus benefits) and paperwork (Table 2). A mean time of 26.8 min per PA was spent by the NP, and 6.5 min per PA was spent by the clerk. The costs of NP and clerk time were calculated at $0.83 and $0.29 per min, respectively, for a direct cost of $3,684.53 ($12.79 per PA). The mean number of pages of paperwork was 5.8 pages per PA. Using a mean of 5.8 pages per PA at $0.25 per page, the paperwork charge was $1.45 per PA. For all 288 PAs, paperwork costs equaled $418.00. The total direct costs equaled $4,102.53 ($14.24 per PA).

Table 2.

Costs to Medical Provider

Variable	Cost for 288 prior authorizations	Cost per prior authorization
Direct costs
Personnel cost
Nurse practitioner	$3568.33 (=$0.83/min × 4282 min)	$12.39
Clerk	$116.20 (=$0.29/min × 395 min)	$0.40
Paperwork	$418.00 (=$0.25/page × 1672 pages)	$1.45
Opportunity costs	$7877.45 (=$55.19/30-min visit × 143 visits)	$27.35
Total	$11,979.98	$41.60

Opportunity costs included the loss of revenue related to the time the NP was diverted to the non–revenue-generating PA activity instead of spending the same time involved in the revenue-generating activity of proving direct medical services. Overall, 7182 min were spent by the NP processing the 288 PAs (26.8 min per PA). If the NP had not been involved in PA, a total of 143 thirty-minute return visits could have been completed. Thus, the financial opportunity costs equaled the revenue lost for 143 visits. Using an estimate of Medicare reimbursement of $55.19 for auxiliary personnel “incident to” level 3 established patient office visit (CPT code 99213), the lost opportunity financial costs equaled $7,877.45. The combined direct and opportunity costs associated with the 288 PAs cost the medical provider $11,979.98 or $41.60 per PA.

DISCUSSION

PA is commonly used to limit financial exposure of third-party payors. Pharmacy benefit plans assert that PA also decreases inappropriate medication use through its sentinel or “spillover” effects on prescribers [14]. Yet, although saving money for the benefit plan, PA increases administrative burden for pharmacists, benefit plans, providers, and patients. The results of this study reveal a direct cost of $14.24 per PA to the health care provider, which is consistent with the $10–$25 cost reported more than a decade ago [14]. However, earlier studies do not account for the medical provider’s lost opportunity costs. When the opportunity costs are combined with the direct cost, the overall cost per PA increases to $41.60, which represents 75% of the amount Medicare reimburses for a level 3 established patient visit ($55.19).

Many medications that require PA were on the benefit plan’s formulary but had protocols to reinforce prescription guidelines. Documentation of the provider’s adherence to these standards was required. Providing evidence requires that valuable time be spent combing through medical records searching for prior medication use; failure of “first-line” therapies; antiretroviral resistance tests (in the case of HIV treatment); and laboratory, pathology, and consultant reports. At our clinic, the use of an electronic medical record greatly enhanced the NP’s ability to perform these tasks efficiently, which provided a conservative estimate when compared with practices using paper charts.

“Formulary alternative” was the reason for denial for 45 (88%) of the 51 PAs that were denied. Therefore, prescribers must have immediate access to benefit plan formularies to avoid denials. However, providers often lack the awareness of key data, including a patient’s specific benefit plan, requirements for step-therapies approach, maximum quantity limits, and the variability and dynamic nature of benefit plans’ formulary inclusion of therapeutic alternatives. We referenced >34 benefit plans formularies. No single provider could be familiar with this number of formularies. Personal digital assistant handheld devices are useful but are often not sufficiently comprehensive to provide listings for all benefit plan formularies, and many providers are not technologically versed in their use. Locating the benefit plan formularies required significant time searching Web sites. Most formularies were presented in nonsearchable formats that required time-consuming manual review. Ideally, if all practices had access to electronic medical records interfaced with benefit plan systems and used electronic prescribing, providers could potentially be informed about formulary alternatives at the time that the prescription was written. However, at the present time, this technology is not readily available and is beyond the financial scope of many small to mid-sized primary care practices. Additional recommendations to improve PA are provided in Table 3 [15].

Table 3.

Recommendations to Improve Prior Authorization (PA) Processes

Challenge	Potential solution
Patients change benefit plans often.	Practice personnel could inquire about pharmacy benefits when patients arrive for visits. Medical providers should maintain a high degree of awareness as to how patients acquire medications.
Office management and electronic medical record software systems do not have the capacity to store and retrieve benefit plan coverage.	Vendors need to develop benefit plan applications that store and retrieve pharmacy benefit related data. Practice personnel could develop work around alternatives to store benefit related information.
Inconsistent inclusion of generic and preferred medications in benefit plan drug formulariesa.	Benefit plans could prominently displayed links to their formularies on the benefit plan’s homepage. Centers for Medicare and Medicaid Services could require coverage of specific medications.
Benefit plan formulary changes generate the possibility of disgruntled prescribers, patients, and pharmacists who have to deal with switching drugs and therapies.	Centers for Medicare and Medicaid Services could restrict formulary changes to once a year.
Providers are not familiar with available formulary alternatives.	Benefit plans could develop integrated systems that provide a list of all available benefit plan specific formulary alternatives when: (1) e-prescriptions are submitted and (2) the prescription is processed for claims adjudication by the pharmacist.
Time consuming process required to complete PA request forms.	With appropriate technical support, the form could be adapted for use with electronic medical records to populate 31 of its required fields.
Nonstandardized approach to submission of complete PA request forms.	Benefit plans could develop on-line PA systems with provider notification for the receipt of PA requests. Centers for Medicare and Medicaid Services could mandate that benefit plans accept the Medicare Part D Coverage Determination Request Formb.
Providers, patients, and pharmacists lack awareness as to the status of PA requests.	Benefit plans could develop viewable systems to assess the status of PA requests in the review process.

^aCommon mechanisms to control costs include pharmacy payments or reimbursements, negotiated prices, generic substitutions, manufacturer price concessions, rebates, and disease management programs.

^bThe form has several restrictions on its use. The most problematic feature of the standardized form is the glaring lack of the specific benefit plan fax numbers.

We were unable to determine whether there were any health-related adverse events directly attributable to PA, but this seems plausible. For HIV-infected patients, delays in access to antiretroviral medications can have significant consequences. Ultimately, 28 (76%) of 37 PAs for antiretroviral medications were approved. However, in every case, patients may have experienced a delay in access to these life-saving medications that require strict adherence to avoid development of drug resistance [16]. In addition, interruptions and delays in receiving anti-infective or pain management medications may impose dramatic short-term implications for patient outcomes while adding to patient frustration and the burden of navigating a dysfunctional health care system.

Although there is a significant financial burden imposed on primary care providers by prescription benefit plans, PA program implementation also comes at a cost to third-party payors. Anecdotal and self-reports have shown that dropping PA requirements for referrals and medical procedures saved UnitedHealth $110 million in annual administrative costs by reassigning those utilization review nurses who denied <2% of treatment requests to other jobs [17].

Provider dissatisfaction with the PA gives cause for long-term concern when coupled with the looming provider shortage in primary care. The uncompensated PA administrative burden, together with the growing piles of other nonclinical paperwork, such as precertification, contributes to job dissatisfaction among primary care providers [18]. An unintended consequence of these frustrations is the decision of some providers to limit their number of Medicare and Medicaid patients [19]. In the case of HIV, the effects of PA on medical providers may result in fewer providers willing to “jump through the administrative hoops,” thereby providing less access to comprehensive care and more episodic medical care in emergency departments.

Although our findings may not be generalizable to other practice settings (because of proportional variations in the number of patients with third-party coverage, Medicaid, and regional private insurers; variations in access to an electronic medical record or high-speed internet; and an expanding library of PA forms). PA is widely used in virtually all primary care practices. The study population (1683 patients) and number of PAs (288) were modest. Delegation of PA to a lesser-paid NP may not be possible for some physician practices. PA centralization may have increased the NP’s efficiency and thereby decreased costs to the provider, which may not be possible in other clinical practice sites.

As observed in other primary care settings, health care workers caring for people living with HIV/AIDS are “burning out” from increased workloads [20, 21]. PA adds to their workload, depletes valuable clinical resources, and results in less time for direct patient care. PA costs are expensive, administratively daunting, and unsustainable for most primary care providers. Left unchecked, PA may further dissuade clinicians from providing care to Medicaid and Medicare beneficiaries. It is reasonable to question whether short-term PA-related cost saving for the benefit plans offset the long-term consequences of shifting administrative expense to medical providers. Health care cannot be managed effectively in a system that does not account for the unintended consequences on all stakeholders.