Alesia Sadosky, PhD

Prescription carriers (also known as prescription drug plans) commonly implement priorauthorization (PA) policies with the intent to promote safe and appropriate use of selected medications. PA policies often require the involvement of a physician or other authorized individual to verify that a patient meets carrier-specified eligibility criteria for coverage of a medication that is subject to PA. While PA and other similar medication-access policies, such as “step therapy,” can be effective at achieving their intended goals, they have also been associated with a variety of unintended economic and clinical consequences within a wide variety of illnesses and diseases, including mental health, diabetes, and chronic pain.

Fibromyalgia (FM) and painful diabetic peripheral neuropathy (DPN) are chronic pain conditions that can share similar pharmacologic treatments.

 Two pharmacologic treatments that are FDA approved for both FM and painful DPN are pregabalin and duloxetine. These agents differ in the way that they are processed by the body; consequently, they carry different risks of interacting with other medications or conditions. For example, pregabalin’s pharmacokinetics are unlikely to be affected by other agents through metabolic interactions because it undergoes negligible metabolism in humans and is mainly excreted unchanged in the urine.

 Duloxetine, however, carries the risk for potential interaction with a variety of other medications because it is an inhibitor of cytochrome P450 enzyme (CYP) 2D6 and CYP1A2, and undergoes hepatic metabolism via CYP2D6 and CYP1A2.

 Furthermore, duloxetine has a greater potential than pregabalin to interact with several more medical conditions, as indicated by the number of different conditions listed in these agents’ “Warnings and Precautions” package insert sections.

Prior research involving FM and painful DPN patients has yielded findings that are consistent with the aforementioned differences between pregabalin and duloxetine, demonstrating that the potential for medication-medication and medication-condition interactions (MMI/MCI) is substantially higher for patients when treated with duloxetine than when treated with pregabalin.

 In the same studies, healthcare cost analyses suggested that exposure to duloxetine MMIs/MCIs was associated with increases in healthcare costs.

If PA for pregabalin leads to preferential prescribing of duloxetine, patients who are subject to pregabalin PA may be placed at increased risk for MMIs/MCIs if they are prescribed duloxetine instead of pregabalin and have an existing condition or need for medication that could interact with duloxetine. Thus, the purpose of this study was to examine the association between pregabalin PA and exposure to potential MMI/MCI among FM and painful DPN patients treated with either pregabalin or duloxetine.

This was a retrospective cohort study based on US insurance claims data for healthcare provided in the setting of a routine clinical practice. The data were extracted from the Truven Health MarketScan Commercial Claims and Encounters (Commercial) and Medicare Supplemental and Coordination of Benefits (Medicare Supplemental) databases, which comprise inpatient and outpatient medical and outpatient prescription drug claims for more than 30 million (annually) employees, dependents, and retirees with employer-sponsored primary or Medicare Supplemental insurance.

The Commercial and Medicare Supplemental databases are derived from large self-insured employers and health plans, resulting in a diverse mix of prescription drug coverage policies and medical insurance arrangements, including fee-for-service, managed care, and preferred provider organizations, among others. Information on payments from both Medicare and Medicare Supplemental health insurance plans are included within the Medicare Supplemental database. At the time that this study was conducted, the most recent data in the Commercial and Medicare databases extended through March 31, 2011.

The data contained in these databases are statistically de-identified and have been certified to satisfy the conditions set forth in Sections 164.514 (a)-(b)1ii of the Health Insurance Portability and Accountability Act. As such, Institutional Review Board approval and written informed consent were not sought for this study.

Study patients were required to meet all of the following inclusion criteria: filled 1 or more outpatient prescription for pregabalin or duloxetine between July 1, 2008, and October 1, 2010 (the index date was set as the date corresponding to the first of such prescription fills, and the index medication was set as the medication filled on the index date); filled only 1 of the medications of interest (pregabalin or duloxetine) on the index date; were 18 years or older on the index date; were continuously enrolled in medical and prescription insurance benefits for the 12 months before (pre-period) and 6 months after (post period) the index date; and had 1 or more inpatient or 2 or more outpatient medical claims with an 

International Classification of Diseases, Ninth Revision, Clinical Modification

) diagnosis code for FM (729.1x) or DPN (250.6x, 357.2x) in the 18 months corresponding to the pre- through post periods. At least 1 of the medical claims with an 

diagnosis code for FM or DPN was required to occur within 60 days before or on the index date, which was a requirement imposed to increase the likelihood that the index medication was intended to treat 1 of these medical conditions.

Study patients were required to not meet any of the following exclusion criteria 12 months before to 6 months after the index date: prior use of pregabalin or duloxetine, identified by 1 or more outpatient prescription claims for either during the pre-period; 1 or more medical claims with an 

 diagnosis code for postherpetic neuralgia (053.1x) or epilepsy (345.xx, 780.39); 1 or more medical claims with an 

 or Current Procedural Terminology (CPT) procedure code (available upon request) for transplant surgery; 1 or more inpatient or 2 or more outpatient (on different days, at least 21 days apart) medical claims with an 

 diagnosis code for cancer (140.xx-172.xx, 174.xx-208.xx, 235.xx-239.xx), with the exception of basal cell and squamous cell skin cancers and benign neoplasms; or residence in a long-term care facility for 90 or more total days.

Aside from the non-naïve exclusion criterion, the aforementioned exclusion criteria were imposed to eliminate patients with medical conditions other than FM and DPN in which pregabalin or duloxetine are indicated and to eliminate patients with medical conditions or subject to circumstances that may complicate measurement of variables. Because duloxetine was approved on November 4, 2010, for use for chronic musculoskeletal pain, diagnosis codes for such conditions were unnecessary for use as exclusion criteria. Finally, patients were excluded from the sample if they were a member of a prescription carrier for which formulary data on pregabalin and duloxetine could not be obtained.

Potential MMIs/MCIs were measured using a previously published algorithm.

Potential MMIs were measured using software (Micromedex Medication Management) that is intended to prevent MMIs in clinical practice.

This software was used to identify instances in which prescriptions that carry a potential for MMIs with pregabalin or duloxetine were filled 180 days before to 30 days after index, with days’ supply of the potentially interacting medication extending beyond index. Potential MCI were medical conditions listed in the “Contraindications” and “Warnings and Precautions” sections of the prescribing information for pregabalin and duloxetine.

 With assistance from nosologists (professionals specializing in disease classification and coding), each MCI was assigned an administrative claims-based identification algorithm. Pregnancy was also included as a potential MCI for both pregabalin and duloxetine, as it is listed under “Use in Specific Populations” within these medications’ package inserts. The presence of potential MCIs was measured during the pre-period. The methodology for measurement of potential MCI is described in full detail within the 

As noted in the study eAppendix Table 1, potential MCI were not included for study if they could not be measured from administrative claims data (eg, tumorigenic potential [pregabalin]), if factors that influence heterogeneity in the risk of interaction were unknown (eg, weight gain [pregabalin]), or if the MCI were not thought to affect prescriber decision making in the context of FM or DPN (eg, glycemic control in patients with diabetes [duloxetine]).

While a patient could possess characteristics that would place them at risk for potential MMIs/MCIs with pregabalin or duloxetine, a patient was classified as being exposed to potential MMIs/MCIs only if he or she was actually prescribed the medication with which they could have a potential MMI or MCI. For example, if a patient had pre-period evidence of angioedema (a potential MCI for pregabalin) but was prescribed duloxetine, that patient would have been at risk for a potential MCI with pregabalin but would not have actually been exposed.

Classifying Members Into Pregabalin PA Versus Unrestricted Prescription Carriers

Using prescription formulary data on pregabalin and duloxetine, which were supplied by an outside vendor, patients were classified as being members of either “Pregabalin PA” or “Unrestricted” prescription carrier groups. Patients classified into the Pregabalin PA group were members of prescription carriers that, on the patient’s index date, had a PA policy for pregabalin and no access restrictions for duloxetine. Patients classified into the Unrestricted group were members of prescription carriers that, on the patient’s index date, had access restrictions for neither pregabalin nor duloxetine.

Study covariates included demographics and clinical characteristics listed in the 

, which were used to describe the sample or adjust the statistical comparisons through multivariable models. Demographic characteristics were measured based on enrollment data as of patients’ index dates. Clinical characteristics, which included the Deyo adaptation of the Charlson Comorbidity Index and a count of unique medications, were measured based on claims incurred during the pre-period.

This study employed bivariate and multivariable statistical approaches. Chi-squared tests compared the proportion of patients treated with duloxetine versus pregabalin and the proportion of patients exposed to potential MMIs/MCIs with their index medication across the Pregabalin PA and Unrestricted groups. Multivariable logistic regressions adjusting for patient demographics (age, sex, geographic region of residence, health plan type, and index year) were used to examine the association between being in the Pregabalin PA versus the Unrestricted group and the following outcomes: 1) treatment with duloxetine versus pregabalin; and 2) exposure to potential MMIs/MCIs with their index medication. These models did not adjust for clinical characteristics, such as comorbidities, because although such characteristics were closely related to the study outcomes, they were unlikely to affect the kind of prescription carrier that was offered by their insurance plans, and therefore would not confound the relationship between study outcome and prescription carrier. Statistical Analysis Software (SAS) 9.2 (SAS Institute Inc, Cary, North Carolina) was used for data management. Stata/MP 10 (StataCorp, College Station, Texas) was used for multivariable analyses. A 

<.05 was set as the a priori threshold for statistical significance on inference related to the study analyses.

 presents the sample attrition associated with the application of each patient inclusion and exclusion criteria. Ultimately, 10,603 patients were selected for study. The Pregabalin PA group comprised 2840 patients from 3 unique prescription carriers. The Unrestricted group represented 7763 patients from 7 unique prescription carriers.

presents patients’ demographics and pre-period clinical characteristics, stratified by the Pregabalin PA group versus the Unrestricted group. Compared with the Unrestricted group, the Pregabalin PA group was slightly younger on average (50.6 years vs 53.8 years; 

<.001), had a marginally lower mean value of the pre-period Deyo- Charlson Comorbidity Index (1.1 vs 1.2; 

= .010), and had slightly lower mean values of the pre-period number of unique medications (13.0 vs. 13.6; 

displays the unadjusted proportions of patients treated with duloxetine or pregabalin, stratified by the Pregabalin PA group versus the Unrestricted group. A greater proportion of patients were treated with duloxetine, as opposed to pregabalin, in the Pregabalin PA group (62.4%) compared with the Unrestricted group (43.8%) (unadjusted odds ratio [OR], 2.13; 

<.001). After multivariable adjustment, the odds of a patient being treated with duloxetine, as opposed to pregabalin, were 95.5% (

 <.001) higher in the Pregabalin PA group compared with the Unrestricted group. Full multivariable analysis results are provided in the

displays the prevalence (unadjusted) of potential pregabalin and duloxetine MMIs/MCIs, irrespective of the index medication, stratified by the Pregabalin PA group versus the Unrestricted group. The data in Figure 3 are presented irrespective of the medication that patients were ultimately treated with, and therefore do not represent actual exposures to MMIs/MCIs. Rather, they represent the risk of potential MMIs/MCIs. This was generally similar across both groups. In the groups, the risk of potential pregabalin MMI/MCI was substantially lower than that of potential duloxetine MMIs/MCIs (

<.001 for both groups). Potential pregabalin MMIs/MCIs were driven entirely by MCI, with no patients having potential pregabalin MMIs; potential duloxetine MMIs/MCIs were primarily driven by MMI, as discussed in greater detail below.

, which takes into account the medication that patients were ultimately treated with (and therefore exposed to), displays the proportions (unadjusted) of patients exposed to potential MMIs/MCIs, stratified by the Pregabalin PA group versus the Unrestricted group. A greater proportion of patients were exposed to potential MMIs/MCIs in the Pregabalin PA group (42.5%) compared with the Unrestricted group (30.8%) (unadjusted OR, 1.66;

 <.001). Of the 1208 patients who were exposed to potential MMIs/MCIs in the Pregabalin PA group, 2% (N = 24) were exposed to potential pregabalin MCIs and 98% (N = 1184) were exposed to potential duloxetine MMIs/MCIs Of the 2394 patients exposed to potential MMIs/MCIs in the Pregabalin PA group, 2.6% (N = 63) were exposed to potential pregabalin MCIs and 97.3% (N = 2331) were exposed to potential duloxetine MMIs/MCIs. After multivariable adjustment, the odds of a patient being exposed to potential MMIs/MCIs were 59.8% (P <.001) higher in the Pregabalin PA group compared with the Unrestricted group. Full multivariable analysis results are provided in the eAppendix.

To our knowledge, this is the first study to examine the association between pregabalin PA policies and exposure to potential MMIs/MCIs among FM and painful DPN patients treated with either pregabalin or duloxetine.

We found that PA for pregabalin was associated with increased treatment with duloxetine and decreased treatment with pregabalin; this finding is consistent with prior studies examining restrictions for pregabalin in a variety of settings, which have demonstrated that PA and step-therapy policies for pregabalin are associated with decreased treatment with the drug pregabalin.

 While the present study did not study pregabalin step-therapy policies, the similarity between such policies and PA, in terms of their mechanism of controlling access to pregabalin, suggests that future research examining the association between these 2 and potential exposure to MMIs/MCIs is warranted.

We also found that the risk of potential pregabalin MMIs/MCIs (1.5% and 2.4% in the Unrestricted group and the Pregabalin PA group, respectively) was substantially lower than that of potential duloxetine MMIs/MCIs (71.4% in the Unrestricted group and 70% in the Pregabalin PA group). This finding is consistent with prior research involving FM and painful DPN patients, which yielded findings that have shown directionally and quantitatively (in some cases) similar results.

 This finding is also consistent with an Institute for Safe Medication Practices study, which ranked duloxetine as the 8th-most often suspected medication in serious adverse drug event reports submitted directly to the FDA by health professionals and patients in 2011.

Finally, we found that patients who had carriers with pregabalin PA were more likely to be exposed to potential MMIs/MCIs, with potential duloxetine MMIs/MCIs accounting for more than 97% of all of such exposures. Exposure to potential MMIs/MCIs is an inherently undesirable outcome. Prior research has confirmed that such exposures may have adverse healthcare cost consequences: among duloxetine-treated patients, exposures to potential duloxetine MMIs/MCIs have been associated with increases in mean healthcare costs of $670 (

 Notably, analogous analyses conducted among pregabalin-treated patients found no statistically significant associations between exposures to potential pregabalin MMIs/MCIs and healthcare costs, the frequency of which was less than 3% in the present study. This study did not attempt to quantify the healthcare cost impact of increased exposure to potential MMI/MCI vis-à-vis pregabalin PA policies. However, it is logical that if such exposures are costly, and if pregabalin PA increases such exposures, pregabalin PA may lead to unintended cost increases. Future research to elucidate this relationship is also warranted.

The chief limitations of this study are related to uncertainty regarding the identification of MMIs/MCIs. First, in cases where patients were exposed to potential MMIs/MCIs, it is unknown whether an MMI or MCI actually occurred. It is also unknown whether potential interacting medications were taken concomitantly with one another because patients could have discontinued 1 of the filled medications. The MicroMedex Medication Management software does not identify the potential for additive adverse effects between medications. The administrative claims—based identification algorithms used to identify potential MCI that may be undercoded, such as “suicide attempt” and “alcohol use,” may have suboptimal sensitivity. These algorithms have been used in previously published studies, but have not been validated against other sources of information, such as medical records. Potential MCIs for which there are no 

 diagnosis codes—including tumorigenic potential (pregabalin), PR interval prolongation (pregabalin), and uncontrolled narrow-angle glaucoma (duloxetine)&mdash;could not be measured using the data source in this study.

 diagnosis code that is specifically intended for painful DPN, relying instead on the combination of an 

 diagnosis code for DPN with what we assume to be medication for pain due to such. In addition, because the databases used for this study are nonprobability samples, results may not be generalizable to the entire US population. Finally, the choice of medication for patients with FM or painful DPN may depend on the presence of a comorbid major depressive disorder, a condition for which duloxetine carries an indication for treatment.

In this retrospective cohort study of FM and painful DPN patients treated with either pregabalin or duloxetine, pregabalin PA was associated with increased treatment with duloxetine. Because duloxetine carries a higher risk for potential MMIs/MCIs, based on patients’ comorbidities and concomitant medications, patients who had carriers with pregabalin PA were more likely to be exposed to potential MMIs/MCIs—more than 9 out of 10 of which were potential duloxetine MMIs/MCIs. Prescription carriers considering implementing pregabalin PA policies should be aware of the unintended potential consequence of patients having increased exposure to potential MMIs/MCIs.

Articles

This study of patients treated with pregabalin or duloxetine found that pregabalin prior authorization policies were associated with increased exposure to potential medication-medication and medication-condition interactions.