Abstract
Objective:
To assess comorbidities, pain-related pharmacotherapy, and healthcare resource use among patients with fibromyalgia (FM) newly prescribed pregabalin or duloxetine (index event) in usual care settings.
Methods:
Using the LifeLink™ Health Plan Claims Database, patients with FM (International Classification of Diseases, Ninth Revision, Clinical Modification code 729.1X) were identified. Patients initiated on duloxetine were propensity score-matched with patients initiated on pregabalin (n = 826; mean age [standard deviation] of 48.3 [9.3] years for both groups). Prevalence of comorbidities, pain-related pharmacotherapy, and healthcare resource use/costs were examined during the 12-month pre-index and follow-up periods.
Results:
Both patient groups had multiple comorbidities and a substantial pain-related and adjuvant medication burden. In the pregabalin group, use of other anticonvulsants decreased significantly (31.6% vs 24.9%), whereas use of serotonin-norepinephrine reuptake inhibitors (SNRIs; 16.5% vs 22.5%) and topical agents (10.1% vs 13.2%) increased in the follow-up period (p < 0.01). In the duloxetine group, there were significant decreases in the use of other SNRIs (13.0% vs 5.7%), selective serotonin reuptake inhibitors (41.3% vs 21.7%), and tricyclic antidepressants (18.8% vs 13.2%), and an increase in the use of anticonvulsants (28.6% vs 40.1%; p < 0.0001). There were significant increases (p < 0.0001) in pharmacy and total healthcare costs in both cohorts, and a significant increase in outpatient costs (p = 0.0084) in the duloxetine cohort from pre-index to follow-up. There were no significant differences in median total healthcare costs between the pregabalin and duloxetine groups in both the pre-index ($10,159 vs $9,556) and follow-up ($11,390 vs $11,746) periods.
Limitations:
Limitations of this study are typical of those associated with retrospective database analyses.
Conclusions:
Patients with FM prescribed pregabalin or duloxetine were characterized by a significant comorbidity and pain/adjuvant medication burden. Although healthcare costs increased in both groups, there were no statistically significant differences in direct healthcare costs between the two groups.
Introduction
Fibromyalgia (FM), a chronic pain condition with an estimated prevalence of 2–4% in the US population, is characterized by widespread musculoskeletal pain, tenderness, and fatigue, and is often accompanied by a constellation of comorbid symptoms including sleep and mood disturbances and cognitive dysfunctionCitation1–5. Fibromyalgia is generally diagnosed symptomatically based on guidelines published by the American College of Rheumatology (ACR)Citation6, which emphasizes the presence of both pain (widespread pain including in the axial, plus upper and lower body segments, and left- and right-sided pain) and tenderness (at ≥11 of the 18 specific tender point sites). However, recent guidelines by the ACR stress pain and symptom severity (on cognition, mood, sleep, fatigue, and other somatic symptoms) and do not rely on the physical or tender point examinationCitation7. Despite the evolving diagnostic criteria, it is often difficult to distinguish FM from other chronic pain syndromes, resulting in patients spending considerable time in the healthcare system until appropriately diagnosedCitation1. Consequently, patients with FM incur significant healthcare costs both leading up to and after the diagnosis of FMCitation8,Citation9.
Patients with FM experience myriad comorbid conditions that may or may not be related to FM. Studies suggest an increased prevalence of depression, anxiety, sleep disorders, chronic fatigue, cardiovascular disorders, gastrointestinal disorders, and musculoskeletal and neuropathic pain conditionsCitation5,Citation10–16. Further, the prevalence of several of these conditions has been documented to be higher in patients with FM compared with non-FM controls and patients with other chronic pain conditionsCitation10–12,Citation16. The presence of comorbidities exacerbates the human and economic burden of FMCitation17,Citation18. The estimated direct costs of FM are substantialCitation8,Citation10–12,Citation16,Citation19–24, and patients with FM have significantly higher healthcare resource use in comparison with non-FM reference groupsCitation10–12,Citation16–20,Citation22. Moreover, owing to the relatively younger age of patients diagnosed with FMCitation8,Citation19, disability and work productivity are affected and indirect costs associated with absenteeism and work productivity are substantialCitation2,Citation16,Citation23,Citation25,Citation26.
The pathophysiology of FM is not fully understood. However, evidence indicates that FM symptoms are partially associated with the augmentation of pain and sensory processing pathways within the central nervous systemCitation27,Citation28. Pharmacologic management has been typically targeted toward the major symptom, pain, and comorbid conditions including depression, anxiety, and sleep disorders. The American Pain Society and the European League Against Rheumatism developed evidence-based treatment guidelines for the management of FM and recommend the use of anticonvulsants, low-dose tricyclic antidepressants (TCAs), selective serotonin reuptake inhibitors (SSRIs), serotonin-norepinephrine reuptake inhibitors (SNRIs), benzodiazepines, dopaminergic agonists, tramadol, and weak opioids (when other therapies have failed)Citation29–31.
Pregabalin (Lyrica®)Citation32 was the first drug to receive approval for the treatment of FM by the US Food and Drug Administration (FDA); duloxetine (Cymbalta®)Citation33 and milnacipran (Savella®)Citation34 have subsequently been approved. (Lyrica is a registered trademark of Pfizer Inc., Cymbalta is a registered trademark of Lilly USA, LLC, and Savella is a registered trademark of Forest Laboratories, Inc.) Although studies have analyzed the comparative efficacy of approved FM treatmentsCitation35,Citation36, and several studies described earlier have evaluated the costs associated with FMCitation2,Citation8,Citation10–12,Citation16–24, there is limited information in the literature on the direct healthcare costs of patients initiating treatment with approved or recommended FM treatments in clinical practice. Moreover, whether any observed differences in direct healthcare costs associated with competing treatments may result from preferentially channeling patients with more severe disease to one treatment vs the other is not known. Anecdotal evidence suggests that the introduction of new medications often results in a selection bias, whereby patients with more severe disease are often channeled to the newer drugs. Indeed, results of a retrospective database evaluation of treatment patterns in patients with FMCitation14 did suggest a trend towards selection bias, where patients with greater pain severity were being prescribed pregabalin, a relatively newer drug compared with gabapentin.
Building on the results of the previous study, the present study was conducted to evaluate treatment patterns and costs among patients newly prescribed pregabalin vs duloxetine for the treatment of FM in usual care settings using propensity-score matching (PSM) at baseline to account for potential patient selection bias.
Materials and methods
Data source
This retrospective study is based on data gathered from the LifeLink™ (formerly PharMetrics) Health Plan Claims Database (IMS, Inc., Watertown, MA). The LifeLink database contains adjudicated medical and pharmaceutical claims data from more than 98 managed care organizations throughout the US (Midwest 34%, Northeast 22%, South 29%, West 15%). It includes more than 61 million patients and ∼16 million covered lives per year. The data from participating managed care organizations undergo rigorous data quality review, all patient identifiers in the database have been fully encrypted (thereby maintaining patient confidentiality), and the database is fully protected under the Health Insurance Portability and Accountability Act of 1996. The database includes information on inpatient and outpatient diagnoses (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] format), procedures (Current Procedure Terminology, 4th Edition and Healthcare Common Procedural Coding System formats), as well as demographic variables, provider specialty, and start and stop dates for plan enrollment. Additional data elements include retail and mail order prescription records, the National Drug Code numbers, days’ supply, and quantity dispensed. Importantly, the data are subject to rigorous data quality review and are standardized to ensure consistency across plans and over time.
Sample selection
All patients with one or more healthcare encounters with an associated diagnosis of FM (ICD-9-CM code 729.1X) during each of the years starting July 1, 2005 through June 30, 2006 and July 1, 2006 through June 30, 2007 were identified. Patients who were newly prescribed pregabalin or duloxetine on or after July 1, 2007 (index date) were then selected. The study period constituted the 12 months before (pre-index) and 12 months after (follow-up) the index date. Patients with incomplete data for the study period, missing data for age or gender, younger than 18 years, not naïve to pregabalin and duloxetine for at least 6 months before the index date, had a prescription for both pregabalin and duloxetine on the index date, or were 65 years and older and not enrolled in a Medicare risk plan (claims histories of these patients may be incomplete) were excluded. In addition, PSMCitation37,Citation38 was used to correct for sample selection bias due to observable differences between the pregabalin and duloxetine cohorts.
To estimate the propensity score of each patient in the pregabalin and duloxetine cohorts, logistic regression was applied given a set of observed predictors (covariates), which included age, gender, pre-index Charlson Comorbidity Index score (CCI)Citation39, pre-index pain comorbidity count, and total pre-index healthcare costs. Patients in the duloxetine cohort were matched 1:1 (within 0.01 units) to patients in the pregabalin cohort based on their respective propensity scores.
Measures and analyses
Demographic and clinical characteristics of patients prescribed pregabalin or duloxetine were assessed and included average age, gender distribution, and coprevalence of selected chronic conditions such as cardiovascularCitation40, gastrointestinal, and sleep disorders, and musculoskeletal and neuropathic pain conditionsCitation5,Citation10–16. The prevalence of selected comorbidities (i.e., those considered as being highly prevalent in patients with rheumatic diseases or FM) was determined based on the presence of one or more healthcare encounters with an associated ICD-9-CM diagnosis code for the specific comorbidity during the pre-index period (see ).
Table 1. Diagnostic codes used to identify relevant comorbidities.
The proportions of patients who received pain- and non-pain-related medications recommended and/or used for the treatment of FM in clinical practiceCitation8–14,Citation16,Citation29–31 and the average number of prescriptions for the various medication classes during the pre-index and follow-up periods were examined. The medication classes that were examined in this study are presented in and . Because opioid analgesics are often prescribed and used as rescue pain medications, in order to distinguish potentially ‘as needed’ vs more regular use, these data were stratified by the number of opioid prescriptions (only 1, ≥1, and ≥2) in the pre-index and follow-up periods. The number of pregabalin and duloxetine prescriptions was also stratified to determine if persistent use of these medications (≥2 prescriptions) affected opioid use.
Table 2. Number of patients prescribed pregabalin and duloxetine who had ≥1 claims for pain-related medications in the pre-index and follow-up periods.
Table 3. Medication use (among users) for pain-related medications in the pre-index and follow-up periods.
Use of healthcare resources during the pre-index and follow-up periods for each medication cohort were evaluated and included physician office visits, emergency room (ER) visits, hospitalizations, other outpatient services (e.g., labs, radiology, and imaging), and direct healthcare costs associated with the use of various categories of healthcare resources (e.g., pain-related medications, physician office visits, ER visits, hospitalizations, other outpatient services, and total direct healthcare costs).
Statistics
Descriptive statistics were used to summarize demographic and clinical characteristics. Chi-square or Fisher exact tests and analysis of variance models were used to calculate the statistical significance of differences between pregabalin and duloxetine for proportions and means, respectively. Differences in medication use between the pre-index and follow-up periods were assessed within each cohort as well as between the two cohorts. Resource use was examined in terms of proportions of patients using these services as well as the magnitude of use during the study periods. McNemar or Wilcoxon sign-rank tests were used to determine the statistical significance of within-group changes in medication use, resource use, and costs between the pre-index and follow-up periods. Chi-square or Fisher exact tests and logistic regressions were used to assess the statistical significance of between-group differences in percent medication and resource use in the pre-index and follow-up periods, respectively. Wilcoxon rank-sum tests were used to assess the statistical significance of between-group differences in the magnitude of use (number of prescriptions, inpatient and outpatient visits) and costs in both the pre-index and follow-up periods. For magnitude of use and cost variables that were significantly different between the pregabalin and duloxetine groups in the follow-up period based on the bivariate comparisons (Wilcoxon rank-sum tests), generalized linear models (GLM) with gamma distributions and log-links (regression analyses) were used to assess if these differences were still significant after accounting for relevant covariates. For each GLM model, the independent variables (referred to as ‘study-defined covariates’ in the results section) included the pre-index CCI, the pre-index value of the dependent variable, and dichotomous variables indicating the presence of comorbidities that still remained significantly different between the pregabalin and duloxetine cohorts after PSM. These comorbidities included anxiety, depression, rheumatism, generalized anxiety disorder, panic disorder, low back pain, abdominal pain, irritable bowel syndrome and atypical facial pain. All analyses were performed using the SAS software system, PC version 8.0 (SAS Institute, Inc., Cary, NC), and p < 0.05 was considered statistically significant for all analyses.
Results
Prior to PSM, a total of 1395 patients initiated on pregabalin and 853 patients initiated on duloxetine satisfied all the study entry criteria. More than 95% (pregabalin 95.8% and duloxetine 97.5%) of patients in both groups had their index dates between July 1, 2007 and December 31, 2007. Prior to PSM, the mean (standard deviation) age for patients prescribed pregabalin, 50.0 (9.4) years, was higher than for patients prescribed duloxetine, 48.3 (9.4) years; p < 0.0001; the gender distribution was similar with ∼86% women in each group. After PSM, a total of 826 patients initiated on duloxetine could be matched with 826 patients initiated on pregabalin, and were retained in the analyses. The demographic and clinical characteristics of the matched pregabalin and duloxetine cohorts are presented in . The two cohorts were similar in age (mean [SD] of 48.3 [9.3] years in both cohorts), and ∼86% of patients were women in both cohorts. There were significant differences (p < 0.05) between the matched pregabalin and duloxetine cohorts in the prevalence of mental disorders, low back pain, atypical facial pain, and abdominal pain—each of which were higher in the duloxetine cohort; rheumatism and irritable bowel syndrome were higher in the pregabalin cohort.
Table 4. Demographic and clinical characteristics of the study cohorts.
Depression was the most common neuropsychiatric disorder in each cohort (pregabalin 30.6%, duloxetine 52.3%; p < 0.0001), and more than 20% of patients in each cohort had insomnia/sleep disorders (pregabalin 23.4%, duloxetine 26.4%; not significant). Hyperlipidemia was the most frequent cardiovascular comorbidity in each cohort; 38.5% of patients receiving pregabalin and 38.7% of patients receiving duloxetine (not significant). More than 90% of patients in each cohort had a variety of comorbid musculoskeletal pain conditions including arthritis and other arthropathies (pregabalin 57.6%, duloxetine 52.9%; not significant), low back pain (pregabalin 60.2%, duloxetine 65.1%; p = 0.0370), back and neck pain (pregabalin 49.2%, duloxetine 52.1%; not significant), and rheumatism (pregabalin 57.7%, duloxetine 52.8%; p = 0.0425). Back and neck pain with neuropathic involvement was the most frequent neuropathic pain condition in each cohort (pregabalin 30.6%, duloxetine 31.5%; not significant), and the prevalence of neuropathic pain conditions in each cohort was over 40%.
The proportions of patients who received at least one prescription for the various study medication classes are presented in , and the average numbers of prescriptions are presented in . A high burden of pain-related medications, including traditional analgesics such as non-steroidal anti-inflammatory drugs and cyclooxygenase 2 inhibitors, short- and long-acting opioids (SAOs and LAOs), antidepressants, muscle relaxants, and anticonvulsants characterized both the pregabalin and duloxetine cohorts in the pre-index period. Additionally, in the pre-index period, many of the patients (19–44%) were prescribed ‘adjunctive’ medications often used to treat conditions associated with pain, such as depression, anxiety, and insomnia. Between-group comparisons indicated that significantly higher proportions (p < 0.05) of patients in the pregabalin cohort were prescribed opioids (LAOs, SAOs, or any opioids), anticonvulsants, muscle relaxants, triptans, and corticosteroids compared with the duloxetine group in the pre-index period, whereas significantly higher proportions (p < 0.05) of patients in the duloxetine cohort were prescribed attention-deficit/hyperactivity disorder (ADHD) drugs, and sedative/hypnotics in the pre-index period.
The medication burden remained high in the follow-up period in both the pregabalin and duloxetine cohorts. Between-group comparisons indicated that significantly higher proportions (p < 0.01) of patients in the pregabalin cohort were prescribed SAOs, any opioids, SSRIs, TCAs, muscle relaxants, and corticosteroids compared with the duloxetine group in the follow-up period.
Within-group comparisons indicated that in the pregabalin group there were significant increases in the proportions of patients who received SNRIs (16.5% vs 22.5%; p < 0.0001) and topical agents (10.1% vs 13.2%; p = 0.0093), whereas there was a significant decrease in the use of other anticonvulsants (31.6% vs 24.9%; p = 0.0001) during the follow-up relative to the pre-index period. There was no change in opioid use in the pregabalin cohort from the pre-index to the follow-up periods, both on an overall basis as well as when opioid use was stratified by the number of opioid and pregabalin prescriptions.
Pre-index vs follow-up comparisons within the duloxetine cohort indicated that the proportions of duloxetine-treated patients who received weak opioids (55.1% vs 49.3%; p = 0.0027), SSRIs (41.3% vs 21.7%; p < 0.0001), other SNRIs (13.0% vs 5.7%; p < 0.0001), and TCAs (18.8% vs 13.2%; p < 0.0001) decreased significantly from the pre-index to the follow-up periods, whereas the proportion of patients who received anticonvulsants (28.6% vs 40.1%; p < 0.0001) increased. When opioid use was stratified by the number of duloxetine prescriptions, among those patients who received two or more duloxetine prescriptions in the follow-up period, there was an increase in the proportion of patients who received two or more prescriptions for LAOs (14.1% vs 16.0%; p = 0.0326).
Between-group comparisons indicated that among patients who received the various study medications, the number of prescriptions for SAOs, weak opioids, any opioids, and muscle relaxants were significantly higher (p < 0.05) in the pregabalin group compared with the duloxetine group in the pre-index period. In the follow-up period, the number of prescriptions for SSRIs, muscle relaxants, and salicylates were significantly higher (p < 0.05) in patients prescribed pregabalin compared with duloxetine among users of these medications.
These significant differences in the number of prescriptions for SSRIs, muscle relaxants, and salicylates between the pregabalin and duloxetine groups in the follow-up period were further examined using multivariate regression analyses. Results indicated that when study-defined covariates were included in the models, only the difference in the number of prescriptions for SSRIs between the pregabalin and duloxetine groups remained significant (treatment β = −0.1733, p = 0.0099), with the number of these prescriptions remaining higher in the pregabalin group in the follow-up period. When holding other covariates constant, the pre-index CCI and the number of prescriptions for SSRIs in the pre-index period were also significant predictors of the number of SSRI prescriptions in the follow-up period.
Within-group comparisons indicated that, in the pregabalin cohort, there were significant increases (all p < 0.05) in the number of prescriptions for strong opioids, SNRIs, and ADHD drugs, and a significant decrease in the number of prescriptions for weak opioids (p = 0.0381) from the pre-index to the follow-up periods among users of these medications.
Pre-index to follow-up comparisons within the duloxetine cohort indicated that there were increases (all p < 0.05) in the number of prescriptions for strong opioids, anticonvulsants, and benzodiazepines, and decreases in the number of prescriptions for SSRIs and TCAs among users of these medications (both p < 0.01).
Use of healthcare resources including proportions of patients using services and magnitude of use of physician office visits, ER visits, hospitalizations, and use of other outpatient services are presented in . A majority of patients in both medication cohorts had physician office and other outpatient visits, over 30% had ED visits, and between 14–18% had hospitalizations during the pre-index and follow-up periods. There were no differences in the proportions of patients using healthcare resources between the two cohorts in the pre-index or follow-up periods.
Table 5. Healthcare resource use among patients prescribed pregabalin and duloxetine in the pre-index and follow-up periods.
Within-group comparisons indicated that, among patients prescribed pregabalin, there was an increase in the proportions of patients with hospitalizations from the pre-index period to follow-up (14.5% vs 18.4%; p = 0.0141).
Between-group comparisons suggested that the number of physician office visits were higher (both p < 0.05) among patients prescribed duloxetine compared with pregabalin in both the pre-index and follow-up periods. However, after factoring in the potential influence of the study-defined covariates in regression analysis, the difference in physician office visits in the follow-up period was not significant.
Within the pregabalin cohort, there were significant decreases (p < 0.05) in the number of patients who had visits to a chiropractor, the number of visits to chiropractors, and the number of visits to general practitioners (GPs), and increases (p < 0.05) in the number of visits to neurologists and psychiatrists in the follow-up period relative to pre-index. Within the duloxetine cohort, there were significant decreases (p < 0.05) in the number of patients with visits to chiropractors and neurologists and in the number of visits to chiropractors and GPs, and increases in both the proportion of patients receiving and the magnitude of cognitive-behavioral therapy in the follow-up relative to the pre-index period (both p < 0.05).
Direct healthcare costs associated with the use of various categories of healthcare resources, including pharmacy, physician office visits, ED visits, hospitalizations, other outpatient services, and total costs, are presented in . Between-group comparisons indicated no differences in the costs of physician office visits, outpatient visits, hospitalizations, and total costs in the pre-index period between the pregabalin and duloxetine groups. Total medication costs were higher in the pregabalin group compared to the duloxetine group in the pre-index period (median $3090.1 [interquartile range (IQR) $1375.3–$6157.1] vs median $2672.8 [IQR $1133.4–$5589.8]; p = 0.0317). In the follow-up period, costs of physician office visits were higher (median $1508.4 [IQR $783.1–$2613.0] vs median $1315.7 [IQR $801.7–$2228.6]; p = 0.0285) in the duloxetine group compared to the pregabalin group, whereas costs of hospitalizations were higher (p = 0.0430) in the pregabalin group relative to the duloxetine group. However, these differences in costs between the two groups were not significant when the study-defined covariates were accounted for in regression analyses.
Table 6. Direct healthcare costs among patients prescribed pregabalin and duloxetine in the pre-index and follow-up periods.
Within-group evaluations suggested that there were significant increases in total medication and total costs from the pre-index to the follow-up periods in both the pregabalin and duloxetine groups (p < 0.05). There were also significant increases in the costs of other and total outpatient visits in the duloxetine group from the pre-index to the follow-up periods, whereas within the pregabalin group increases in outpatient visit costs from the pre-index to the follow-up period barely attained significance.
Discussion
Results of this study demonstrate that patients with FM who are prescribed pregabalin or duloxetine are characterized by a substantial comorbidity and medication burden including both pain-related and adjuvant medications. The prevalence of comorbidities observed in this study is not only higher than what was previously described in a cohort of 33,176 patients with FM relative to a comparison group without FMCitation10, but also higher than a previous descriptive study of patients with FM initiated on pregabalin or gabapentinCitation14. The comparatively higher burden observed in this study may be indicative of a potential selection bias, where FM patients with more severe disease are being initiated on the two recently approved treatments for FM. The high prevalence of comorbid illnesses is noteworthy, because, although these conditions may or may not be etiologically related to FM, they complicate and confound the management of patients with FM.
As previously demonstrated, patients in this study had a variety of cardiovascular, gastrointestinal, and neuropsychiatric comorbiditiesCitation5,Citation10–16. A majority of patients had one or more musculoskeletal pain conditions, and the prevalence of neuropathic pain conditions, present in over 40% of patients in each cohort, was higher than the ∼23% reported in two previous studiesCitation10,Citation14. In both cohorts, hypertension and hyperlipidemia were the two most common cardiovascular comorbidities, rheumatism and low back pain were the most frequent musculoskeletal pain conditions, and neuropathic low back pain was the most prevalent neuropathic pain condition. The prevalence of neuropsychiatric disorders, including anxiety, depression, and panic disorder, were higher in the duloxetine cohort. This finding is not surprising and points to the fact that duloxetine was potentially being used to treat comorbid depression in these patients.
The high medication burden observed in both cohorts in this study has been consistently noted in patients with FM in clinical practiceCitation8,Citation10–14,Citation16,Citation24. Comparatively, patients in the pregabalin group were characterized by a higher pain medication burden (both with respect to the number of patients who received these medications as well as the magnitude of use) in both the pre-index and follow-up periods relative to patients treated with duloxetine. However, the differences in medication use between the two cohorts were not confirmed in multivariate analyses that controlled for pre-index use. The higher pain medication burden found in the pregabalin group could be ascribed to a potential bias of channeling patients with more severe FM to pregabalin. Although we used PSM to address potential selection bias in this study, the PSM model could not directly account for pain severity owing to the limitation of our data source (a retrospective medical and pharmacy claims database). Therefore, it is possible that patients with more severe FM pain were initiated on pregabalin, whereas similarly, patients with mood/neuropsychiatric disorders may have been channeled to duloxetine, an antidepressant, since this cohort had significantly higher proportions of patients with these disorders relative to pregabalin. Of note, among the patients initiated on duloxetine, significant reductions were observed for all classes of antidepressants (SSRIs, SNRIs, and TCAs) in the follow-up relative to the pre-index period. The channeling of patients with more severe disease is common after the introduction of a new medication, and the launch of the first approved treatment for any indication can often serve to evolve prescribing patterns for that condition. This pattern is also consistent with what was seen in the previous descriptive study of patients initiated on pregabalin or gabapentin, where patients with more severe disease were treated with pregabalinCitation10. Since pregabalin was the first FDA-approved treatment for FM and duloxetine was not specifically approved for FM during our study period, it is possible that, in the future, different trends in the use of pain-related medications will be evident as physicians become more familiar with other FM treatments.
There was an increase in the use of SNRIs in the pregabalin group in the follow-up period and a corresponding increase in the use of anticonvulsants in the duloxetine group in the follow-up period, which could suggest switching and/or augmenting between these two medications. In the pregabalin group, there was no change in opioid use on an overall basis as well as when opioid use was stratified by the number of prescriptions. In the duloxetine group, among those patients who received at least two duloxetine prescriptions in the follow-up period, there was a statistically significant increase in the proportion of patients who received at least two prescriptions for LAOs, albeit this increase was small (1.9%). Opioid analgesics are often prescribed and used as rescue pain medications or on an as-needed basis. Treatment guidelines for FM recommended the use of opioids when other therapies have failed. While there was no increase in opioid use after initiation of pregabalin, the increase in opioid use, particularly in the number of patients who received more than two prescriptions for LAOs among persistent duloxetine users (those who received at least two duloxetine prescription), is noteworthy.
There were no remarkable differences in physician office visits, ED visits, hospitalizations, or outpatient visits between the pregabalin and duloxetine groups in the pre-index and follow-up periods, or within groups from pre-index to follow-up. There was a decrease in the number of visits to chiropractors and GPs in both groups. Whether or not this decrease was related, in part, to improved pain control due to the initiation of pregabalin or duloxetine cannot be ascertained owing to database limitations. Within the pregabalin and duloxetine cohorts, there were modest increases in medication costs and total costs in the follow-up period; however, there were no differences in total costs between the two groups in either the pre-index or follow-up periods.
Our findings of no differences in costs between patients initiated on pregabalin or duloxetine are consistent with another recent studyCitation41 of employees with FM, which indicated cost neutrality between pregabalin and duloxetine. Contrary to our findings, cost differences in favor of duloxetine were reported in a studyCitation42 of patients with FM in usual care. However, patients initiated on pregabalin in that study were inherently different and potentially sicker than patients initiated on duloxetine due to disparate inclusion criteria applied to the two groups. Our results, and those of the aforementioned study in FM employeesCitation41, confirm no overall cost differences between these two medications when similar patient populations are evaluated. There were modest cost increases in both the pregabalin and duloxetine groups in the follow-up period in our study. Similar findings were noted in another study where costs increased with each subsequent stage of FM (pre-diagnosis, post-diagnosis, and established FM)Citation9.
When interpreting this study, as with all studies that rely on retrospective database analyses, some limitations should be noted. These limitations include potential errors in coding and recording, which might result in misdiagnosis in a proportion of patients, particularly due to the difficulty in diagnosing FM in the absence of an established diagnostic test. Moreover, in studies using retrospective claims data, it is not possible to link the condition of interest, FM, with the prescribing of a particular pain medication, since this population is characterized by multiple comorbidities that may be associated with intermittent or chronic pain. Therefore, the prescribing of pregabalin, duloxetine, or any of the adjunctive medications often prescribed for pain-related sequelae (i.e., depression/anxiety and sleep disorders) may have been for indications other than pain related to FM. A similar limitation of retrospective database studies is that, since patient compliance cannot be confirmed, a prescription claim merely implies that the patient filled the prescription. However, whether or not the patient used the medication as directed is unknown.
Additionally, we were not able to quantify over-the-counter medications use for FM because the database is limited to prescription medications. Alternative treatments including acupuncture and chiropractic care are both recommended and used in the management of FM in clinical practice. However, since these services are often not reimbursed by insurance companies, their use may also be under-represented in our study. Finally, the database does not include information on pain severity levels and, hence, it is not possible to know what effects, if any, the prescribing of pregabalin or duloxetine may have had on pain-related outcomes.
Conclusions
In conclusion, our results demonstrate the prevalence of a myriad of comorbidities and extensive pain-related/adjuvant medication prescribing in patients with FM who initiated treatment with pregabalin or duloxetine in clinical practice. The use of adjuvant medications was also high in both cohorts. Patients in the pregabalin cohort were characterized by a higher pain medication burden in both the pre-index and follow-up periods relative to duloxetine, although medication burden was also high in the duloxetine group in both periods. An increase in the use of SNRIs after initiating treatment with pregabalin and a corresponding increase in the use of anticonvulsants after initiating treatment with duloxetine may potentially be indicative of switching and/or add-on treatment patterns between these two FDA-approved treatments for FM. Although healthcare costs increased in both groups, there were no statistically significant differences in direct healthcare costs between the two groups. Given the human and economic burden of FM, future research may benefit from a focus on efficacy and safety parameters to further differentiate treatment options and to investigate the causal relationships among medication use and clinical outcomes in patients with FM.
Transparency
Declaration of funding
This research was funded by Pfizer Inc. All authors contributed to the concept design of the study and to data preparation and analysis. All authors drafted, reviewed, and revised the manuscript.
Declaration of financial/other relationships
Dr Gore is Principal Consultant and Kei-Sing Tai is Principal Statistician at Avalon Health Solutions, Inc., and they were paid consultants to Pfizer Inc in connection with the development and execution of both this article and the research it describes. Dr Gore also owns stock in Pfizer. Dr Zlateva and Ms Chandran are employees and stockholders in Pfizer. Dr Leslie was paid an honorarium by Avalon Health Solutions, Inc. for his participation in this research and for his review of and input for this article. Dr Leslie also has performed consulting for Kurron Bermuda Ltd.
Acknowledgments
The authors would like to thank Christopher Rice for editorial assistance in the preparation of this article. Christopher Rice is an employee of Avalon Health Solutions, Inc. Editorial support to prepare this manuscript for journal submission was provided by UBC Scientific Solutions and funded by Pfizer Inc.
References
- Arnold LM, Crofford LJ, Mease PJ, et al. Patient perspectives on the impact of fibromyalgia. Patient Educ Couns 2008;73:114-20
- Bennett RM, Jones J, Turk DC, et al. An internet survey of 2,596 people with fibromyalgia. BMC Musculoskelet Disord 2007;8:27
- Lawrence RC, Felson DT, Helmick CG, et al. National Arthritis Data Workgroup. Estimates of the prevalence of arthritis and other rheumatic conditions in the United States. Part II. Arthritis Rheum 2008;58:26-35
- Wolfe F. The epidemiology of fibromyalgia. J Musculoskelet Pain 1993;1:137-48
- Wolfe F, Ross K, Anderson J, et al. The prevalence and characteristics of fibromyalgia in the general population. Arthritis Rheum 1995;38:19-28
- Wolfe F, Smythe HA, Yunus MB, et al. The American College of Rheumatology 1990 criteria for the classification of fibromyalgia. Report of the Multicenter Criteria Committee. Arthritis Rheum 1990;33:160-72
- Wolfe F, Clauw DJ, Fitzcharles MA, et al. The American College of Rheumatology preliminary diagnostic criteria for fibromyalgia and measurement of symptom severity. Arthritis Care Res 2010;62:600-10
- Berger A, Sadosky A, Dukes EM, et al. Patterns of healthcare utilization and cost in patients with newly diagnosed fibromyalgia. Am J Manag Care 2010;16(5 Suppl):S126-37
- White L, Birnbaum H, Kaltenboeck A, et al. Medical care and APS-guideline-related prescription drug use in newly diagnosed and established fibromyalgia patients. J Pain 2008;9:16
- Berger A, Dukes E, Martin S, et al. Characteristics and healthcare costs of patients with fibromyalgia syndrome. Int J Clin Pract 2007;61:1498-508
- Berger A, Martin S, Dukes E, et al. Patterns of pain-related pharmacotherapy in patients with fibromyalgia. Manag Care Interface 2008;20:30-8
- Berger A, Sadosky A, Dukes E, et al. Characteristics and patterns of healthcare utilization of patients with fibromyalgia in general practitioner settings in Germany. Curr Med Res Opin 2008;24:2489-99
- Gore M, Sadosky A, Zlateva G, et al. Initial use of pregabalin, patterns of pain-related pharmacotherapy, and healthcare resource use among older patients with fibromyalgia. Am J Manag Care 2010;16:144-53
- Gore M, Sadosky AB, Zlateva G, et al. Clinical characteristics, pharmacotherapy and healthcare resource use among patients with fibromyalgia newly prescribed gabapentin or pregabalin. Pain Pract 2009;9:363-74
- Nampiaparampil DE, Shmerling RH. A review of fibromyalgia. Am J Manag Care 2004;10:794-800
- White LA, Birnbaum HG, Kaltenboeck A, et al. Employees with fibromyalgia: medical comorbidity, healthcare costs, and work loss. J Occup Environ Med 2008;50:13-24
- Robinson RL, Birnbaum HG, Morley MA, et al. Depression and fibromyalgia: treatment and cost when diagnosed separately or concurrently. J Rheumatol 2004;31:1621-9
- Weir PT, Harlan GA, Nkoy FL, et al. The incidence of fibromyalgia and its associated comorbidities: a population-based retrospective cohort study based on International Classification of Diseases, 9th revision codes. J Clin Rheumatol 2006;12:124-8
- Hughes G, Martinez C, Myon E, et al. The impact of a diagnosis of fibromyalgia on health care resource use by primary care patients in the UK: an observational study based on clinical practice. Arthritis Rheum 2006;54:177-83
- Lachaine J, Beauchemin C, Landry PA. Clinical and economic characteristics of patients with fibromyalgia syndrome. Clin J Pain 2010;26:284-90
- Robinson RL, Birnbaum HG, Morley MA, et al. Economic cost and epidemiological characteristics of patients with fibromyalgia claims. J Rheumatol 2003;30:1318-25
- Spaeth M. Epidemiology, costs, and the economic burden of fibromyalgia. Arthritis Res Ther 2009;11:117
- Wassem R, Hendrix TJ. Direct and indirect costs of fibromyalgia to patients and their families. J Orthop Nurs 2003;7:26-32
- Wolfe F, Anderson J, Harkness D, et al. A prospective, longitudinal, multicenter study of service utilization and costs in fibromyalgia. Arthritis Rheum 1997;40:1560-70
- Kleinman N, Harnett J, Melkonian A, et al. Burden of fibromyalgia and comparisons with osteoarthritis in the workforce. J Occup Environ Med 2009;51:1384-93
- White KP, Speechley M, Harth M, et al. Comparing self-reported function and work disability in 100 community cases of fibromyalgia syndrome versus controls in London, Ontario: the London Fibromyalgia Epidemiology Study. Arthritis Rheum 1999;42:76-83
- Bennett RM. Emerging concepts in the neurobiology of chronic pain: evidence of abnormal sensory processing in fibromyalgia. Mayo Clin Proc 1999;74:385-98
- Staud R. Evidence of involvement of central neural mechanisms in generating fibromyalgia pain. Curr Rheumatol Rep 2002;4:299-305
- Burckhardt C, Goldenberg D, Crofford L, et al. Guideline for the management of fibromyalgia syndrome pain in adults and children (Clinical practice guideline, no. 4). Glenview, IL: American Pain Society, 2005
- Carville SF, Arendt-Nielsen S, Bliddal H, et al. EULAR evidence-based recommendations for the management of fibromyalgia syndrome. Ann Rheum Dis 2008;67:536-41
- Goldenberg DL, Burckhardt C, Crofford L. Management of fibromyalgia syndrome. JAMA 2004;292:2388-95
- Lyrica [package insert]. New York, NY: Pfizer Inc, 2007
- Cymbalta [package insert]. Indianapolis, IN: Eli Lilly & Co., 2008
- Savella [package insert]. New York, NY: Forest Pharmaceuticals, Inc., 2009
- Häuser W, Petzke F, Sommer C. Comparative efficacy and harms of duloxetine, milnacipran, and pregabalin in fibromyalgia syndrome. J Pain 2010;11:505-21
- Staud R, Price DD. Long-term trials of pregabalin and duloxetine for fibromyalgia symptoms: how study designs can affect placebo factors. Pain 2008;136:232-4
- D'Agostino RB, Jr. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med 1998;17:2265-81
- Rosenbaum P, Rubin D. The central role of the propensity score in observational studies for causal effects. Biometrika 1983;70:41-55
- Charlson ME, Pompei P, Ales KL, et al. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40:373-83
- Turesson C, Jacobsson LT, Matteson EL. Cardiovascular co-morbidity in rheumatic diseases. Vasc Health Risk Manag 2008;4:605-14
- Harnett J, Margolis J, Cao Z, et al. Real-world evaluation of health-care resource utilization and costs in employees with fibromyalgia treated with pregabalin or duloxetine. Pain Pract 2011;11:217-29
- Zhao Y, Sun P, Sun S, et al. Comparison of dosing patterns between fibromyalgia patients with high versus low duloxetine or pregabalin compliance. Value Health 2010;13:311