ORIGINAL ARTICLE
Evaluating Guideline-recommended Pain
Medication Use Among Patients with Newly
Diagnosed Fibromyalgia
Rac...
disturbances, often resulting in functional disability and
lost productivity.5,6
U.S. studies show that employees
with FM ...
patients treated with “core” guideline-recommended
drug classes (AEDs, TCAs, SSRIs, and SNRIs) and
opioids.
METHODS
Data S...
To evaluate treatment patterns, patients were cate-
gorized into 1 of 4 outcomes based on the earliest change
in their ind...
from the database (Figure 1). Of these, 96,175 (10.3%)
were included in the final sample.
Most patients were assigned to th...
(35%); the prevalence of switching was 10% (Table 3).
The high rate of discontinuation was driven by the SAO
cohort, which...
LAO cohort were to SAOs. Most switches that occurred
in the SAO cohort were to SSRIs (48%).
Pain medication treatment patt...
1.1 times as likely to receive guideline-recommended
pain medications after the index date (P  0.001), and
pre-index SAO u...
Table4.CoxProportionalHazardsRegression
Covariate
Model1DependentVariable:Guideline-recommended
PainMedicationafterIndexDa...
pre-index depression and gender also were consistent.
The hazard ratio for the LAO cohort remained signif-
icant for some ...
as a monotherapy or in combination with opioids. These
data suggest that many patients were not receiving
guideline-recomm...
rately, although the selection criteria of at least 2 FM
diagnostic claims mitigated the effect of the latter
limitation. ...
9. Wolfe F, Clauw DJ, Fitzcharles MA, et al. The Amer-
ican College of Rheumatology preliminary diagnostic criteria
for fib...
Próximos SlideShares
Carregando em…5
×

FMS & Opioid Guidelines, Honored More In the Breach.

128 visualizações

Publicada em

60% receive against recommendations.

Publicada em: Saúde
  • Seja o primeiro a comentar

  • Seja a primeira pessoa a gostar disto

FMS & Opioid Guidelines, Honored More In the Breach.

  1. 1. ORIGINAL ARTICLE Evaluating Guideline-recommended Pain Medication Use Among Patients with Newly Diagnosed Fibromyalgia Rachel Halpern, PhD, MPH*; Sonali N. Shah, RPh, MBA, MPH† ; Joseph C. Cappelleri, PhD, MPH, MS‡ ; Elizabeth T. Masters, MS, MPH† ; Andrew Clair, PhD† *Optum, Eden Prairie, Minnesota; † Pfizer Inc, New York City, New York; ‡ Pfizer Inc, Groton, Connecticut, U.S.A. & Abstract Objectives: To compare pain medication treatment changes across cohorts of newly diagnosed patients with fibromyalgia (FM) treated with guideline-recommended medications or opioids. Methods and Design: Retrospective claims data analysis examined adult commercial health plan members newly diagnosed with FM (initial diagnosis = index date) from January 2008 to February 2012. Patients had 6-month pre- index and 12-month postindex periods and received pain medication within 6 months postindex; cohorts were based on the first postindex medication. Guideline-recommended medication cohorts were anti-epileptic drug (AED), sero- tonin–norepinephrine reuptake inhibitor (SNRI), selective serotonin reuptake inhibitor (SSRI), and tricyclic antidepres- sant (TCA). Short-acting and long-acting opioid (SAO, LAO) cohorts were also identified. Pairwise comparisons with the SAO cohort were conducted. Cox proportional hazards regressions modeled the likelihood of receiving guideline- recommended therapy. Results: The final sample was 96,175 patients (mean age 47.3 years; 72.5% female), distributed into SAO (57%), SSRI (22%), AED (10%), SNRI (6%), TCA (3%), and LAO (2%) cohorts. The SAO cohort had the most discontinuation (49% vs. 6% to 22%, P < 0.01) and the least augmentation (29% vs. 35% to 50%, P < 0.01). Regression analyses indicated that patients with (vs. without) pre-index guideline-recom- mended medications were 2 to 4 times more likely to receive them postindex. Patients in the opioid cohorts were about half as likely to receive subsequent guideline-recommended medications. Conclusions: Opioid use was widespread among patients with FM. Once patients received opioids postdiagnosis, the likelihood of receiving guideline-recommended medications was small. These real-world results indicate an opportunity may exist for improved FM management using recommended therapies in clinical practice. & Key Words: fibromyalgia, pain, therapeutics, guidelines, adherence, opioid use INTRODUCTION Fibromyalgia (FM) is a chronic pain condition in approximately 2% to 6% of the population in the United States (U.S.), affecting about 7 to 9 times more women than men.1–4 FM is typically diagnosed between the ages of 20 and 50 years, and its prevalence rises most notably in middle age.1,2 FM is characterized by widespread pain and tenderness, with symptoms of fatigue and sleep Address correspondence and reprint requests to: Rachel Halpern, PhD, Optum, 12125 Technology Dr., Eden Prairie, MN 55344, U.S.A. E-mail: rachel.halpern@optum.com. Submitted: April 2, 2015; Revision accepted: July 20, 2015 DOI. 10.1111/papr.12364 © 2015 World Institute of Pain, 1530-7085/16/$15.00 Pain Practice, Volume 16, Issue 8, 2016 1027–1039
  2. 2. disturbances, often resulting in functional disability and lost productivity.5,6 U.S. studies show that employees with FM incur 86% more medical, drug, and indirect costs vs. employees without FM and miss an average of 3 times more work days per year.7 Furthermore, patients with severe FM have 4 times greater direct and indirect healthcare costs, exhibit a lower health-related quality of life, have more comorbid conditions and more disability, and report severe pain more often compared with patients with mild FM severity.6,8 The American College of Rheumatology (ACR) first introduced its classification criteria for FM in 1990 based on a combination of criteria for widespread pain occur- ring for at least 3 months and pain in at least 11 of 18 tender points on digital examination. In 2010, the ACR added a composite symptom severity scale that incorpo- rated cognitive problems, unrefreshed sleep, fatigue, and somatic symptoms as part of its diagnostic criteria.9,10 Although the symptoms for FM can mimic other disor- ders including chronic fatigue syndrome, hypothy- roidism, and inflammatory rheumatic disease, the diagnosis of FM should not be based solely on a process of exclusion, which might delay diagnosis and treatment and prolong the burden of disease for patients.2,5,11,12 Assessment of multiple evidence-based FM treatment guidelines reveals common therapeutic approaches for the management of FM pain. Most recently in 2012, the Canadian guidelines for the diagnosis and management of FM recommended all classes of antidepressants, including tricyclic antidepressants (TCAs; such as amitriptyline), selective serotonin reuptake inhibitors (SSRIs; such as paroxetine, fluoxetine, citalopram, sertraline), and serotonin–norepinephrine reuptake inhi- bitors (SNRIs; such as duloxetine, milnacipran) for the treatment of pain and other symptoms of FM. Anticon- vulsants/anti-epileptic drugs (AEDs; gabapentin, prega- balin) and cyclobenzaprine were also recommended. Opioids, beginning with a weak opioid such as tramadol, were reserved for patients with moderate to severe pain that was unresponsive to all other treatments recom- mended.13 Previous guidelines published in 2008 by the European League Against Rheumatism (EULAR) rec- ommended similar regimens for the treatment of FM. EULAR also recommended alternatives including mono- amine oxidase inhibitors, dopamine agonists, and sero- tonin receptor antagonists, while recommending against the use of corticosteroids and strong opioids.14 In the United States, guidelines for the treatment of FM were published in 2005 by the American Pain Society (APS), prior to U.S. Food and Drug Adminis- tration (FDA) approval of medications for FM. These recommendations included TCAs, SSRIs, SNRIs, amitriptyline, cyclobenzaprine, and tramadol, while the use of strong opioids was recommended only after all other pharmacologic and nonpharmacologic thera- pies were attempted. Treatment with nonsteroidal anti- inflammatory drugs (NSAIDs) was not recommended.11 More recently, the FibroCollaborative (2012), an ini- tiative of professional organizations and advocacy groups that, through expert consensus, furthered the understanding, assessment, and treatment of FM, included TCAs, AEDs, SSRIs, SNRIs, weak opioids (tramadol), and the central nervous system depressant sodium oxybate as a U.S.-based evidence-based frame- work for the pharmacotherapy of FM.15 Therefore, over the past decade, a common core of recommended pharmacotherapies for the symptomatic management of FM has emerged across various treatment guidelines, which span 4 drug classes—AEDs, TCAs, SSRIs, and SNRIs—and uniformly discourage first-line use of strong opioids or anti-inflammatory medications. Only 3 drugs are approved by the FDA for the treatment of FM: the AED pregabalin (Lyricaâ , Pfizer, Inc., New York, NY, U.S.A), the SNRIs duloxetine (Cymbaltaâ , Eli Lilly and Company, Indianapolis, IN, U.S.A) and milnacipran (Savellaâ , Forest Pharmaceuticals, Inc., St. Louis, MO, U.S.A).15,16 Previous studies in patients with FM have evaluated the cost and type of treatments,6 changes in treatment patterns and/or costs relative to specific pharmacologic treatment regimens,17–21 and predictors of pain medi- cation selection following diagnosis.22,23 Their findings suggest that although guideline-recommended therapies such as AEDs, TCAs, SSRIs, and SNRIs are utilized, there remains substantial use of opioids following a diagnosis of FM. However, patient selection for these studies was based on specific medications prescribed for FM (gabapentin, pregabalin, duloxetine, and TCAs), making the assessment of the use of other pain medica- tions such as opioids restricted to circumscribed pre- scription patterns. This study applied broader patient selection criteria to examine pharmacologic treatment patterns for newly diagnosed FM. The objective was to further explore pain medication selection and treatment patterns in patients with newly diagnosed FM before and after their diag- nosis, with a focus on guideline-recommended pharma- cotherapies and opioids. We compared medication treatment patterns (treatment discontinuation, switch- ing, augmentation, and continuation) across cohorts of 1028 HALPERN ET AL.
  3. 3. patients treated with “core” guideline-recommended drug classes (AEDs, TCAs, SSRIs, and SNRIs) and opioids. METHODS Data Source This study analyzed data from the Optum Research Database (ORD), a proprietary research database con- taining medical and pharmacy claims data with linked enrollment information covering the period from 1993 to present. In 2013, data for approximately 12.7 million commercial health plan members were available in the ORD. The ORD commercial population is representa- tive of the privately insured U.S. population based on age, gender, U.S. census region, race/ethnicity, and income. Pharmacy claims data include National Drug Code, dosage form, fill date, days of supply, and quantity supplied. Medical claims data include Interna- tional Classification of Diseases, Ninth Revision, Clin- ical Modification (ICD-9-CM) diagnosis and procedure codes, Current Procedural Terminology and Healthcare Common Procedure Coding System procedure codes, site of care, and date of service. All study data were accessed in compliance with the Health Insurance Portability and Accountability Act of 1996. Sample Selection Patient selection began with the first FM diagnosis, defined as the “index date,” which was the service date of the first “qualifying” medical claim with an ICD-9- CM diagnosis code for FM (729.1), in any position on the claim, during the identification period from January 1, 2008, through February 29, 2012. Qualifying medical claims were affiliated with direct healthcare encounters, such as office visits, outpatient facility (eg, outpatient hospital clinic) visits, and emergency room visits. Nonqualifying medical claims were affiliated with diagnostic radiology, pathology, or laboratory venues. Patients were required to be at least 18 years old during the year of the index date and to have 6 months of continuous health plan enrollment with medical and pharmacy benefits before the index date (pre-index period) and 12 months of continuous enrollment after the index date (postindex period). The entire study period was from July 2007 through February 2013. In addition, patients needed at least 1 additional qualifying medical claim with an FM diagnosis, in any position, during the postindex period, and a prescription fill for their first pain medication on or within 6 months (182 days) after the index date. This first FM treatment medication was defined as the “index pain medication.” Qualifying index pain medications included the FM guideline-recommended AEDs pregabalin and gabapen- tin; SNRIs duloxetine and milnacipran; TCA amitripty- line; and SSRIs fluoxetine, citalopram, escitalopram, fluvoxamine, paroxetine, and sertraline; and also included short-acting opioids (SAOs) and long-acting opioids (LAOs). SAOs comprised butorphanol, codeine, dihydrocodeine, hydrocodone, levorphanol, meperi- dine, pentazocine, and propoxyphene. LAOs included buprenorphine and methadone. In addition, fentanyl, hydromorphone, morphine, oxycodone, oxymorphone, tramadol, and tapentadol each are manufactured in short- and long-acting formulations and were dis- tributed into the appropriate opioid cohort. Accord- ingly, patients were assigned to 1 of 6 mutually exclusive index pain medication cohorts, based on the drug class of the qualifying index pain medications (AED, SNRI, TCA, SSRI, SAO, or LAO). Patients were excluded if they had missing sex or U.S. census region data, any medical claims (qualifying or nonqualifying) with an ICD-9-CM FM diagnosis code during the pre-index period, if they could not be assigned to 1 of the pain medication-based study cohorts, or if they had more than 1 pain medication on the “medication date,” defined as the date for the index pain medication. Time to the index pain medica- tion was the number of days from the index diagnosis date to the first medication date. Study Measures Patient demographics included age, sex, and the U.S. census region where the patient was enrolled in the health plan. Pre-index clinical characteristics were based on pain-related comorbid conditions, including neuro- pathic pain conditions, musculoskeletal pain conditions, sleep disorders, depression and other mental disorders (bipolar disorder, anxiety, generalized anxiety disorder, panic disorder, post-traumatic stress disorder), cardio- vascular disorders, and gastrointestinal disorders. Patients were identified with comorbid conditions based on at least 1 medical claim with ICD-9-CM diagnosis code for the condition, in any position on the claim, during pre-index period. The proportion of patients with at least 1 pre-index claim for a guideline medica- tion or opioid was also captured. Guideline Pain Medications for Fibromyalgia 1029
  4. 4. To evaluate treatment patterns, patients were cate- gorized into 1 of 4 outcomes based on the earliest change in their index pain medication: discontinuation, switch, augmentation, and continuation (for those who did not discontinue, switch, or augment). Discontinuation was defined as a gap of 30 days or more between the run-out of the last observed index pain medication prescription fill (prescription fill date + days of supply) and the end of the postindex period. The discontinuation date was the day before the gap. A switch between index pain medication classes occurred if a patient had 1 or more claim(s) for a nonindex pain medication and no subse- quent claims for the index pain medication. Augmenta- tion of the index pain medication occurred if a patient had 1 or more claim(s) for a nonindex pain medication and 1 or more subsequent claim(s) for the index pain medication. The pain medication switched to or used to augment the index pain medication was also captured. Patients’ use of guideline medications and opioids, defined as at least 1 pharmacy claim, during the 12- month postindex period was measured each quarter (3 months). For this evaluation, patients were assigned to 1 of 4 categories based on their treatment status quarterly: no FM guideline medication or opioid; FM guideline medication only; opioid plus a FM guideline medication; and opioid only. Analysis Descriptive statistics summarized patient demographic and clinical characteristics. All measures were stratified across the 6 pain medication cohorts. Pairwise compar- isons were performed between the SAO cohort and each of the other medication cohorts. This approach was chosen as it was more conducive to interpreting treat- ment patterns vs. comparisons across all pairs of cohorts. The SAO cohort was selected as the reference because preliminary data showed a clear majority of patients assigned to this group. Differences in continu- ous variables were evaluated with t-test with adjust- ments for unequal variance as necessary.24 Differences in binary and categorical variables between cohorts were tested with chi-square statistics.24 Two Cox proportional hazards regressions were used to model the likelihood of receiving postindex guideline- recommended pain medications for FM.25,26 Hazards measured the likelihood of receiving a guideline-recom- mended mediation (the dependent variable) at any day of the postindex period conditional on not having already received one. Model 1 assessed the likelihood of receiving a guideline-recommended therapy after the index date among all patients. Model 2 assessed the likelihood of receiving a guideline-recommended ther- apy after the medication date for patients in the opioid cohorts (ie, after an opioid index pain medication). Covariates included age, sex, U.S. census region, pre- index pain medication use indicators for index pain medications and other pain therapies, pre-index pain- related comorbid conditions, pre-index inpatient visits, and pre-index emergency department visits. Model 1 controlled for the index pain medication and time- dependent previous opioid use between the index date and medication date. In addition, Model 2 also con- trolled for time from the index date to medication date. The coefficients of the models are reported as hazard ratios; the hazard ratio for each covariate is evaluated holding all other covariates constant. The assumption of proportional hazards was tested with Schoenfeld resid- uals.26,27 Sensitivity testing was conducted to evaluate the impact of pre-index pain medication use and the impact of covariates that did not meet the assumption of proportional hazards. Sensitivity testing for Model 1 included limiting the sample to patients who did not have pre-index claims for their index pain medication, the interaction of time-dependent opioid flag with the time from index date to medication date, and inclusion of interactions between all independent variables and time. Sensitivity testing for Model 2 included limiting the sample to patients whose medication dates were no more than 30 days after their index dates, patients with no pre-index opioid use, and inclusion of interactions between time and the independent variables that did not meet the assumption of proportional hazards in the initial regressions. To account for a series of 5 pairwise multiple comparisons for the 6 cohorts, where each of 5 cohorts was compared against the SAO cohort, the threshold for statistical significance for all tests was P 0.01. The threshold for significance in the regression analyses was P 0.05. All analyses were performed using SAS version 9.2.28 RESULTS Patient Selection and Baseline Characteristics A total of 934,694 commercial enrollees with 1 or more medical claims with a qualifying ICD-9-CM diagnosis code for FM during identification period were identified 1030 HALPERN ET AL.
  5. 5. from the database (Figure 1). Of these, 96,175 (10.3%) were included in the final sample. Most patients were assigned to the SAO cohort (57%), followed by the SSRI (22%), AED (10%), SNRI (6%), TCA (3%), and LAO (2%) cohorts (Table 1). Of the 56,995 patients in the opioid cohorts, 9,538 (17%), or 1.0% of the total sample, received a fill of tramadol on their medication dates (data not shown in table); tramadol was the index pain medica- tion for 9,121 of 54,867 patients in the SAO cohort (16.6%) and for 417 of 2,128 patients in the LAO cohort (19.6%). The most prevalent SAOs were hydrocodone (28,976 patients, 53% of the SAO cohort), oxycodone (9,700 patients, 18% of SAO cohort), tramadol, and propoxyphene (3,982 patients, 7% of SAO cohort). Long-acting oxycodone was the most common LAO (532 patients, 25% of the LAO cohort), followed by fentanyl (428 patients, 20%), tramadol, and morphine (360 patients, 17%: data for individual SAOs and LAOs are not shown in Table S1 [LINK here]). The SAO cohort was slightly younger on average vs. the other cohorts. Most patients (73%) were female, with a higher prevalence of females across the guideline-recommended cohorts. About two-thirds of the patients (66%) had pre- index musculoskeletal pain conditions, and 23% had pre-index neuropathic pain conditions. The prevalence of all pre-index pain conditions in the LAO and AED cohorts were significantly higher vs. the SAO cohort (P 0.01). The LAO and AED cohorts also had notably higher percentages of patients with muscu- loskeletal and neuropathic pain conditions and cardio- vascular disorders compared with the other cohorts. Still, the SAO, SNRI, and TCA cohorts had a high prevalence of musculoskeletal pain conditions, occur- ring in about two-thirds of patients in each cohort (Table 1). The SSRI and SNRI cohorts had the highest prevalence of depression. The SAO cohort had the lowest prevalence of sleep disorders compared with the other cohorts. Pre-index Pain Medication Use Forty-seven percent of patients (44,930 of 96,175) received SAOs or LAOs, or both, prior to diagnosis. Over half of patients (53%) in the SAO cohort received a SAO in the pre-index period (Table 2). Compared with patients in the SAO cohort, significantly lower percentages of patients in the guideline-recommended cohorts received SAOs in the pre-index period (ranging from 28% for SSRIs to 46% for AEDs). Most patients (79%) in the SSRI cohort received a SSRI in the pre- index period. Likewise, the highest percentages of patients in each remaining guideline-recommended cohort received the cohort medication class in the pre-index period (AEDs, 43%; SNRIs, 52%; TCAs, 46%). Postindex Treatment Patterns Across all cohorts, 81% of patients had a postindex treatment change; discontinuation was the most preva- lent change (36%) followed closely by augmentation -index date N = 228,111 medical claim with ICD-9-CM diagnosis code for N = 934,694 Newly diagnosed with FM at index date N = 448,739 8 - - N = 522,270 index date, not missing sex and US ce N = 486,857 First treatment medication after index date N = 96,175 medication date N = 102,409 Figure 1. Overview of sample selection and reasons for attrition. FM, fibromyalgia; ICD-9-CM, International Classification of Dis- eases, Ninth Revision, Clinical Modification. Guideline Pain Medications for Fibromyalgia 1031
  6. 6. (35%); the prevalence of switching was 10% (Table 3). The high rate of discontinuation was driven by the SAO cohort, which had the highest prevalence of discontin- uation vs. all other cohorts. The guideline-recommended medication cohorts (accounting for 48% of all augmen- tations) all had significantly higher (P 0.01) rates of augmentation vs. the SAO cohort. The LAO cohort had the highest prevalence of augmentation, but represented only 4.7% of all augmentations. Pain medications with which patients augmented are shown in Figure 2. The guideline-recommended medi- cation cohorts primarily augmented therapy with SAOs, ranging from 57% of augmenting patients (“aug- menters”) in the TCA cohort to 78% of augmenters in the SSRI cohort. Next, patients in the serotonin-based cohorts (SSRI, SNRI) augmented with AEDs, and the AED cohort augmented with serotonin-based therapies (31% of augmenters with SSRIs and SNRIs). Fifty-six percent of augmenters in the LAO cohort augmented with SAOs, and 37% augmented with guideline-recom- mended medications. Switching was the second most common pain med- ication change in the LAO, AED, and TCA cohorts (Table 3). All cohorts had statistically significantly higher percentages of patients switching vs. the SAO cohort, although the difference was smallest for the SSRI cohort. Pain medications that patients were switched to are summarized in Figure 3. Across the AED, SNRI, SSRI, and TCA cohorts, the most common switch was to SAOs, ranging from 51% of patients who switched (“switchers”) in the SNRI cohort to 66% of switchers in the SSRI cohort. Almost two-thirds of switches in the Table 1. Demographic and Clinical Characteristics All (N = 96,175) Index Pain Medication Cohort SAO (N = 54,867) LAO (N = 2,128) AED (N = 9,105) SNRI (N = 6,180) SSRI (N = 20,922) TCA (N = 2,973) Age, mean (SD), year 47.3 (12.6) 46.8 (12.8) 50.3 (12.1)* 50.2 (12.3)* 48.2 (11.3)* 46.6 (12.5) 48.3 (12.8)* Age distribution, n (%) * * * * * 18 to 29 year 8,478 (8.8) 5,349 (9.7) 98 (4.6) 499 (5.5) 379 (6.1) 1,903 (9.1) 250 (8.4) 30 to 39 year 17,974 (18.7) 10,710 (19.5) 289 (13.6) 1,255 (13.8) 969 (15.7) 4,272 (20.4) 479 (16.1) 40 to 49 year 26,291 (27.3) 14,885 (27.1) 589 (27.7) 2,388 (26.2) 1,822 (29.5) 5,816 (27.8) 791 (26.6) 50 to 59 year 27,605 (28.7) 15,168 (27.7) 704 (33.1) 3,050 (33.5) 2,044 (33.1) 5,763 (27.6) 876 (29.5) ≥ 60 year 15,827 (16.5) 8,755 (16.0) 448 (21.0) 1,913 (21.0) 966 (15.6) 3,168 (15.1) 577 (19.4) Sex, n (%) Female 69,740 (72.5) 36,624 (66.8) 1,439 (67.6) 7,214 (79.2)* 5,429 (87.8)* 16,543 (79.1)* 2,491 (83.8)* Male 26,435 (27.5) 18,243 (33.2) 689 (32.4) 1,891 (20.8) 751 (12.2) 4,379 (20.9) 482 (16.2) U.S. Census region, n (%) * * * * * Northeast 8,018 (8.3) 4,352 (7.9) 172 (8.1) 736 (8.1) 472 (7.6) 2,024 (9.7) 262 (8.8) Midwest 25,209 (26.2) 13,557 (24.7) 461 (21.7) 2,496 (27.4) 1,572 (25.4) 6,162 (29.5) 961 (32.3) South 47,303 (49.2) 27,804 (50.7) 1,031 (48.4) 4,546 (49.9) 3,277 (53.0) 9,358 (44.7) 1,287 (43.3) West 15,636 (16.3) 9,148 (16.7) 464 (21.8) 1,327 (14.6) 859 (13.9) 3,376 (16.1) 462 (15.5) Other 9 ( 0.1) 6 ( 0.1) 0 0 0 2 ( 0.1) 1 ( 0.1) Time to index pain medication, mean (SD), day N/A 45.4 (52.5) 15.7 (29.7) 29.5 (41.0) 30.9 (36.9) 37.9 (40.9) 31.1 (42.6) Pre-index pain conditions, n (%)** Neuropathic pain conditions 22,111 (23.0) 12,890 (23.5) 956 (44.9)* 2,949 (32.4)* 1,375 (22.3) 3,313 (15.8)* 628 (21.1)* Musculoskeletal pain conditions 63,146 (65.7) 36,538 (66.6) 1,842 (86.6)* 6,764 (74.3)* 4,154 (67.2) 11,897 (56.9)* 1,951 (65.6) Sleep disorders 9,826 (10.2) 5,132 (9.4) 279 (13.1)* 1,045 (11.5)* 816 (13.2)* 2,221 (10.6)* 333 (11.2)* Depression 15,996 (16.6) 6,232 (11.4) 452 (21.2)* 1,419 (15.6)* 1,669 (27.0)* 5,854 (28.0)* 370 (12.5) Other mental disorders 13,162 (13.7) 5,836 (10.6) 367 (17.3)* 1,184 (13.0)* 1,041 (16.8)* 4,413 (21.1)* 321 (10.8) Cardiovascular disorders 37,112 (38.6) 20,972 (38.2) 1,043 (49.0)* 4,056 (44.6)* 2,500 (40.5)* 7,444 (35.6)* 1,097 (36.9) Gastrointestinal disorders 23,015 (23.9) 12,847 (23.4) 621 (29.2)* 2,526 (27.7)* 1,618 (26.2)* 4,607 (22.0)* 796 (26.8)* AED, anti-epileptic drug; LAO, long-acting opioid; SAO, short-acting opioid; SNRI, serotonin and norepinephrine reuptake inhibitor; SSRI, selective serotonin reuptake inhibitor; TCA, tricyclic antidepressant; SD, standard deviation. Other U.S. census region includes Armed Forces Americas (except Canada), Armed Forces (Europe, Canada, Middle East, Africa), Armed Forces Pacific, American Samoa, Federated State of Micronesia, Guam, Marshall Islands, Commonwealth of the Northern Mariana Islands, Puerto Rico, Palau, Virgin Islands. *P 0.01 vs. SAO cohort; comparisons over all age groups, U.S. census regions. **See Gore and colleagues (2012)18 for further details on the comorbidities used. We added the diagnosis codes for osteoarthritis (715.XX) to their definition of musculoskeletal pain and for diabetic peripheral neuropathy (250.6X, 357.2) to their codes for neuropathic pain. 1032 HALPERN ET AL.
  7. 7. LAO cohort were to SAOs. Most switches that occurred in the SAO cohort were to SSRIs (48%). Pain medication treatment patterns across quartiles during the follow-up period are shown in Figure 4. Over one-third of patients (35%) diagnosed with FM were treated only with opioids in the first quarter (3 months) postindex; 20% or more were treated only with opioids in each remaining quarter. Moreover, 52% of patients were not prescribed guideline-recommended therapies in the first quarter after the index date. The percentage of patients receiving FM guideline medications remained fairly stable (42% to 48%) during the follow-up period. Overall, treatment changes during the first year of therapy were primarily due to decreases in patients receiving “opioid-only” therapy, with pro- portionate increases in patients receiving “no FM guideline medication or opioids.” Cox Proportional Hazard Regression Table 4 summarizes results from the 2 Cox propor- tional hazards regressions modeling the likelihood of receiving a guideline-recommended pain medication after the index date (Model 1) and the likelihood of receiving a guideline-recommended pain medication after an opioid index pain medication (Model 2). The highest hazard ratios in both models were associated with pre-index guideline medication use covariates. Model 1 indicated that patients with pre-index use of guideline-recommended medications were 2.1 to 3.9 times more likely (depending on the medication) to receive guideline-recommended pain medications after the index date relative to patients not using these pre- index medications (holding other covariates constant; all P 0.001). The respective hazard ratios for Model 2 were similar—patients with pre-index use of guide- line-recommended medications were 1.9 to 3.9 times (all P 0.001) more likely to receive guideline-recom- mended medications after an opioid index pain med- ication. Across both regression models, pre-index opioid use was a weaker predictor of guideline-recommended pain medication use than was pre-index guideline medication use. Patients in Model 1 with pre-index LAO use were Table 3. Postindex Treatment Changes Across Cohorts Postindex Treatment Change, n (%) All (N = 96,175) Index Pain Medication Cohort SAO (N = 54,867) LAO (N = 2,128) AED (N = 9,105) SNRI (N = 6,180) SSRI (N = 20,922) TCA (N = 2,973) Discontinuation 34,291 (35.7) 27,108 (49.4) 136 (6.4)* 1,643 (18.0)* 834 (13.5)* 3,921 (18.7)* 649 (21.8)* Switch 9,890 (10.3) 4,055 (7.4) 309 (14.5)* 1,864 (20.5)* 1,057 (17.1)* 1,894 (9.1)* 711 (23.9)* Augmentation 33,201 (34.5) 15,676 (28.6) 1,548 (72.7)* 4,579 (50.3)* 2,896 (46.9)* 7,388 (35.3)* 1,114 (37.5)* Continuers 18,793 (19.5) 8,028 (14.6) 135 (6.4)* 1,019 (11.2)* 1,393 (22.5)* 7,719 (36.9)* 499 (16.8)* AED, anti-epileptic drug; LAO, long-acting opioid; SAO, short-acting opioid; SNRI, serotonin and norepinephrine reuptake inhibitor; SSRI, selective serotonin reuptake inhibitor; TCA, tricyclic antidepressant. Continuers = no discontinuation, augmentation, or switching. n = number with a discontinuation, switch, augmentation, or continuation. *P 0.01 vs. SAOs. Table 2. Pre-index Pain Medication Use: Relevant (Index Pain) Medications Pre-index Pain Medication Use: Patients with ≥ 1 Pharmacy Claim, n (%) All (N = 96,175) Index Pain Medication Cohort SAO (N = 54,867) LAO (N = 2,128) AED (N = 9,105) SNRI (N = 6,180) SSRI (N = 20,922) TCA (N = 2,973) Opioid medications SAO 44,187 (45.9) 29,243 (53.3) 1,454 (68.3)* 4,200 (46.1)* 2,350 (38.0)* 5,852 (28.0)* 1,088 (36.6)* LAO 4,504 (4.7) 1,981 (3.6) 1,492 (70.1)* 421 (4.6)* 233 (3.8) 303 (1.5)* 74 (2.5)* Guideline-recommended medications AED 10,759 (11.2) 4,292 (7.8) 557 (26.2)* 3,917 (43.0)* 713 (11.5)* 1,012 (4.8)* 268 (9.0) SNRI 7,030 (7.3) 2,265 (4.1) 298 (14.0)* 745 (8.2)* 3,223 (52.2)* 388 (1.9)* 111 (3.7) SSRI 26,804 (27.9) 6,929 (12.6) 483 (22.7)* 1,524 (16.7)* 1,015 (16.4)* 16,458 (78.7)* 395 (13.3) TCA 3,740 (3.9) 1,353 (2.5) 112 (5.3)* 408 (4.5)* 189 (3.1)* 322 (1.5)* 1,356 (45.6)* AED, anti-epileptic drug; FM, fibromyalgia; LAO, long-acting opioid; SAO, short-acting opioid; SNRI, serotonin and norepinephrine reuptake inhibitor; SSRI, selective serotonin reuptake inhibitor; TCA, tricyclic antidepressant. Guideline AED, gabapentin, pregabalin; guideline SNRI, duloxetine, milnacipran; guideline SSRI, all medications; guideline TCA, amitriptyline. *P 0.01 vs. SAO cohort. Bold values indicate pre-index use of the particular index pain medication. Guideline Pain Medications for Fibromyalgia 1033
  8. 8. 1.1 times as likely to receive guideline-recommended pain medications after the index date (P 0.001), and pre-index SAO use was not significant. Model 2 showed that patients with pre-index SAO or LAO use were 1.3 and 1.1 times as likely, respectively, to receive a guideline-recommended pain medication after their opioid index pain medication vs. patients who did not receive an SAO or LAO, respectively (both P 0.01). The other notable opioid-related results from Model 1 suggest that patients whose first medication after FM diagnosis was an opioid were less than half as likely (hazard ratio: 0.461, P 0.001) to receive guideline- recommended medications vs. patients whose first medication was not an opioid (Table 4, Time-dependent previous opioid). For Model 2, patients in the LAO cohort were 1.2 times as likely (P 0.001) to receive guideline-recommended medications following an opi- oid index pain medication vs. patients in the SAO cohort (Table 4, Cohort). Age, U.S. census region, pre-index pain conditions, and other pre-index medication use covariates across both regression models did not exhibit any prominent predictors of postindex guideline-recommended medi- cation use. Both models showed that patients with pre-index depression were more likely to receive guide- line-recommended pain medications, whereas males (relative to females) were less likely to receive guide- line-recommended pain medications. The impact of each of the pre-index guideline- recommended medication use covariates was consistent in sensitivity testing for both models. The results for Figure 2. Pain medications with which patients augmented. *Multiple medications mean that patient had fill for more than 1 pain medication on the augmentation date. Figure 3. Pain medications to which patients switched. *Multiple medications mean that patient had fill for more than 1 pain medication on the switch date. Figure 4. Quarterly treatment in first year after index date (N = 96,175). 1034 HALPERN ET AL.
  9. 9. Table4.CoxProportionalHazardsRegression Covariate Model1DependentVariable:Guideline-recommended PainMedicationafterIndexDate(AllCohorts*) Model2DependentVariable:Guideline-recommendedPainMedication afterOpioidIndexMedication(AllOpioidCohortPatients† ) HazardRatio95%CIPValueHazardRatio95%CIPValue CohortN/AN/AN/A SAON/AN/AN/ARef.–– LAON/AN/AN/A1.1701.095to1.2500.001 Agegroups Age18to290.8620.834to0.8920.0010.9050.853to0.9600.001 Age30to390.9300.908to0.9540.0010.9510.911to0.9930.023 Age40to49Ref.––Ref.–– Age50to591.0200.998to1.0420.0791.0431.004to1.0820.028 Age60+0.9970.971to1.0230.8071.0681.021to1.1170.004 Sex(male)0.7130.698to0.7280.0010.6850.662to0.7090.001 U.S.censusregion NortheastRef.––Ref.–– Midwest1.0631.029to1.0980.0011.1771.107to1.2520.001 South0.9980.968to1.0290.9111.1131.050to1.1800.001 Westor“Other”0.9690.935to1.0030.0741.0961.027to1.1700.006 Pre-indexpainconditions Neuropathic painconditions 1.0080.987to1.0290.4791.0421.008to1.0760.016 Musculoskeletal painconditions 1.0150.995to1.0360.1321.0431.004to1.0840.029 Sleepdisorders1.0110.985to1.0370.4211.0541.010to1.1000.015 Depression1.0861.063to1.1100.0011.1081.065to1.1520.001 Othermental disorders 1.0581.034to1.0830.0011.0140.973to1.0560.521 Cardiovascular disorders 1.0251.006to1.0440.0091.0691.036to1.1030.001 Gastrointestinal disorders 1.0321.013to1.0520.0011.0841.049to1.1190.001 Pre-indexguidelinemedicationuse GuidelineAEDuse2.2862.232to2.3410.0012.2932.206to2.3840.001 GuidelineSNRIuse2.8162.740to2.8930.0012.9172.781to3.0590.001 SSRIuse3.8533.782to3.9250.0013.9353.803to4.0720.001 GuidelineTCAuse2.1102.037to2.1860.0011.9291.813to2.0510.001 Pre-indexopioiduse SAOuse0.9820.964to1.0010.0601.2531.209to1.2980.001 LAOuse1.1371.096to1.1800.0011.0891.032to1.1500.002 Otherpre-indexmedicationuse OtherSNRIuse‡ 0.9900.951to1.0300.6110.9600.903to1.0200.187 OtherTCAuse‡ 1.0340.980to1.0920.2231.0230.939to1.1150.602 OtherAEDuse‡ 1.0661.033to1.1000.0011.0531.000to1.1100.050 Corticosteroid use 0.9970.979to1.0160.7601.0511.019to1.0840.001 Benzodiazepine use 1.0841.063to1.1050.0011.1241.087to1.1620.001 Musclerelaxant use 1.0581.036to1.0790.0011.0741.040to1.1090.001 Guideline Pain Medications for Fibromyalgia 1035
  10. 10. pre-index depression and gender also were consistent. The hazard ratio for the LAO cohort remained signif- icant for some of the sensitivity testing for Model 2; this hazard ratio was not significant when Model 2 was estimated on the subset of the opioid cohorts who did not have any pre-index opioid claims. DISCUSSION Our analysis evaluated pharmacologic treatment pat- terns in adult patients with newly diagnosed FM. Similar to other studies,17–22,29 the patients meeting selection criteria for this analysis were primarily females, and 66% had musculoskeletal pain conditions. Patients in the LAO and AED cohorts appeared to carry the greatest burden of pain-related comorbidities based on a significantly higher percentage of all pre-index pain- related conditions vs. the SAO cohort. Higher percent- ages of the LAO and AED cohorts also had muscu- loskeletal and neuropathic pain conditions and cardiovascular disorders vs. other cohorts. Yet while the LAO cohort had the highest prevalence of pre-index musculoskeletal and neuropathic pain conditions, there was no definitive treatment pattern shift (via treatment augmentation or switching patterns) toward guideline-recommended pain medications after the diagnosis of FM. This is noteworthy given that recent treatment guidelines emphasize that chronic widespread pain is the pivotal symptom to treat in FM, but not with opioids as a frontline therapy.13,30 A recent systematic literature review of prospective survey studies found that physicians who treated chronic noncancer pain with opioids tended to add medications, such as hypnotic agents, to opioids and did not stop opioid therapy when patients did not experience ade- quate pain reduction.31 Consistent with this finding, our results showed that 73% of patients in the LAO cohort had an augmentation. Widespread opioid use also extended to other treat- ment patterns. SAOs were the most prevalent medica- tions to which patients switched, or with which they augmented their index pain medications. Twenty-nine percent of the SAO cohort augmented their SAO therapy. Also, as a pre-index medication and index pain medication, there appeared to be a strong preference toward SAOs. Moreover, even with a decrease in “opioid-only” use in the 12 months following the index date, there was a proportionate increase in “no FM guideline medications or opioids” rather than an increase in guideline-recommended medications either Table4.(Continued) Covariate Model1DependentVariable:Guideline-recommended PainMedicationafterIndexDate(AllCohorts*) Model2DependentVariable:Guideline-recommendedPainMedication afterOpioidIndexMedication(AllOpioidCohortPatients† ) HazardRatio95%CIPValueHazardRatio95%CIPValue Migraine medicationuse 1.0491.020to1.0790.0011.0971.048to1.1480.001 NSAIDuse1.0531.033to1.0730.0011.0801.047to1.1140.001 ≥1pre-index EDvisit 0.9520.934to0.9700.0010.9880.957to1.0200.475 ≥1pre-index inpatientvisit 0.9600.929to0.9930.0180.9620.915to1.0110.129 Time-dependent previousopioid§ 0.4610.451to0.4720.001N/AN/AN/A Daysfromindex dateto medicationdate N/AN/AN/A0.9910.990to0.9910.001 AED,anti-epilepticdrug;ED,emergencydepartment;LAO,long-actingopioid;N/A,notapplicable;SAO,short-actingopioid;SNRI,serotoninandnorepinephrinereuptakeinhibitor;SSRI,selectiveserotoninreuptakeinhibitor;TCA, tricyclicantidepressant;NSAID=nonsteroidalanti-inflammatorydrug. *N=96,175observations. † N=56,995observations. ‡ SNRIs,TCAs,andAEDsnotincludedas“guideline”fibromyalgiatreatment. § Timedependentreferstotimebetweentheindexdateandtheguideline-recommendedmedication. 1036 HALPERN ET AL.
  11. 11. as a monotherapy or in combination with opioids. These data suggest that many patients were not receiving guideline-recommended therapies either before or after their diagnosis of FM. High usage of SAOs (50%) was previously found in patients receiving so-called standard-of-care antide- pressants (amitriptyline, duloxetine, or venlafaxine) prior to their FM diagnosis, with little change in postindex SAO use (57%).21 The prevalence of SAO use also far exceeded that of SSRIs (14%).21 It is suggested that physicians’ preferences toward opioids for FM might stem from meta-analysis data demon- strating the efficacy of opioids for the relief of chronic noncancer pain and for improving functional outcomes and that strong opioids (oxycodone, morphine) are statistically superior to TCAs (nortriptyline) for relief of chronic pain.16,32 However, no trials have directly compared the effects of opioids with those of any other medication class in patients with FM; indeed, the use of opioids for FM is not recommended based on this lack of evidence.13,14 Moreover, a longitudinal study demonstrated that in a multidisciplinary pain clinic setting, improvements in pain, functional, and psychological outcomes were independent of opioid use in patients with FM, and that opioid use might have contributed to poorer outcomes (increased disability and unemployment pay- ments) vs. patients not receiving opioids.33 Chronic use of opioids for FM presents other concerns—patient- perceived benefits of opioids may stem from central reward effects rather than FM symptom relief, reflecting the abuse liability properties of opioids.34 In addition, FM is a syndrome of central pain amplification; treating FM with opioids, which are known to support a state of activation of central neurons involved in pain transmission, could facilitate or aug- ment symptoms (opioid-induced hyperalgesia).34,35 The Canadian guidelines mention that in clinical practice, moving from a weak to a strong opioid may be useful in selecting patients, but should be done with caution given no convincing evidence exists that supports this approach.13 The 41% of patients in our sample who received guideline-recommended pain medications first after their FM diagnoses (ie, the AED, SNRI, SSRI, and TCA cohorts) exhibited some encouraging treatment patterns. Patients who were treated with guideline- recommended therapies before their FM diagnosis tended to continue using them after diagnosis. This was most pronounced with the SSRI cohort, with 79% of patients remaining on their pre-index medication, and less so with AEDs, SNRIs, and TCAs, where only about half of the patients remained on their existing pain medication. The regression analysis showed that pre-index guideline medication use variables were the strongest predictors of receiving a guideline-recom- mended medication after the index date or after an opioid index pain medication. The use of guideline- recommended medications also appeared more consis- tent over time than did opioid use—the AED, SNRI, SSRI, and TCA cohorts had less than half the preva- lence of discontinuation (termination) compared with the SAO cohort, and the SNRI, SSRI, and TCA cohorts had greater percentages of patients who continued therapy. The AED, SNRI, SSRI, and TCA cohorts were more likely to augment or switch their index pain medications compared with the SAO cohort. The high prevalence of augmentation and switches in the AED cohort may be linked to patients’ high burden of pain-related comor- bidities, which may require altered treatment approaches.35 The high prevalence of switches and the low number of patients in the TCA cohort might be due to the limited use of this drug class as a therapy for FM because of purported drug safety, tolerability, and efficacy concerns.13,16,22 Yet only 41% of patients in the TCA cohort were switched to a guideline-recom- mended AED, SNRI, or SSRI. Of note were treatment patterns for SNRIs. Even though duloxetine and milnacipran are 2 of 3 drugs approved for the management of FM in the United States, there were relatively few treatment switches to (6% to 19%), and augmentations with (3% to 13%), SNRIs. The low augmentation of SSRIs and SNRIs with another antidepressant is understandable given that prescription of concurrent antidepressants is ill-ad- vised,19,36 and may explain the higher prevalence of AED as an augmentation medication in our analysis. However, given that the optimal pharmacotherapy for FM may involve a combination of treatment options including multimodal pharmacologic therapies, often at low doses that may benefit treatment adherence,15,30 it is of note that SNRI use did not emerge in some respect as a more prominent therapy. Limitations Claims data have inherent limitations for research. We could not be certain whether medications were used as intended or whether diagnoses were recorded accu- Guideline Pain Medications for Fibromyalgia 1037
  12. 12. rately, although the selection criteria of at least 2 FM diagnostic claims mitigated the effect of the latter limitation. Also, claims data do not include important clinical information, such as severity of FM and outcomes. Previous studies have suggested pregabalin may be prescribed to patients with more severe FM.18 The high burden of disease found in the AED cohort might be indicative of more severe FM, which may have influenced medication choice. In addition, in claims databases it is not possible to directly link the medication with the condition; therefore, it is possible medications were prescribed for other painful condi- tions; and also given the size of the database, that statistically significant differences were likely driven by the large sample rather than by clinical signifi- cance. Finally, because the index pain medication date could occur within 6 months after the index diagnosis date to capture real-world use, the observation period for treatment pattern outcomes was not uniform. More- over, as in all observational studies of this type, the multivariate analysis could not control for unobserved factors that would confound the relationship between the observed covariates and receipt of guideline-recom- mended medications. CONCLUSION FM is a complex condition to manage due to the challenges associated with diagnosis and treatment of numerous comorbid conditions such as cardiovascular disorders, musculoskeletal and neuropathic pain condi- tions, and depression, with no clear lines of therapy for all symptoms. While the premise of this analysis was to retrospectively evaluate pain medications used to treat FM, it should be emphasized that the condition is best managed with the addition of nonpharmacologic inter- ventions.15 This analysis showed considerable variation in the pharmacotherapeutic treatment of FM, although there were some resonant themes: the use of opioids was widespread in this study sample, and a substantial proportion of patients with FM did not receive a guideline-recommended therapy in the year following their diagnosis. Once patients were prescribed opioids after their FM diagnoses, the likelihood of receiving a guideline-recommended medication was small. These real-world utilization results indicate an opportunity may exist for improved FM management using recom- mended therapies in clinical practice. ACKNOWLEDGEMENTS This study was sponsored by Pfizer. Shah, Cappelleri, Masters, and Clair are employees of Pfizer. Halpern is an employee of Optum and was contracted by Pfizer in connection with this study and development of this article. Medical writing support was provided by Rod Gossen, PhD of Optum, and funded by Pfizer. The authors would like to thank Cori Blauer-Peterson, MPH of Optum, for her assistance with data set creation and analysis and Randall Gerdes of Optum for his work in programming the analytic data set. SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article: Table S1. Individual Opioids Among Patients in the Short- and Long-acting Opioid Cohorts. REFERENCES 1. Lawrence RC, Felson DT, Helmick CG, et al. Esti- mates of the prevalence of arthritis and other rheumatic conditions in the United States. Part II. Arthritis Rheum. 2008;58:26–35. 2. Chong YY, Ng BY. Clinical aspects and management of fibromyalgia syndrome. Ann Acad Med Singapore. 2009;38:967–973. 3. Wolfe F, Ross K, Anderson J, Russell IJ, Hebert L. The prevalence and characteristics of fibromyalgia in the general population. Arthritis Rheum. 1995;38:19–28. 4. Vincent A, Lahr BD, Wolfe F, et al. Prevalence of fibromyalgia: a population-based study in Olmsted County, Minnesota, utilizing the Rochester Epidemiology Project. Arthritis Care Res (Hoboken). 2013;65:786–792. 5. Dunne FJ, Dunne CA. Fibromyalgia syndrome and depression: common pathways. Br J Hosp Med (Lond). 2012;73:211–217. 6. Chandran A, Schaefer C, Ryan K, Baik R, McNett M, Zlateva G. The comparative economic burden of mild, moderate, and severe fibromyalgia: results from a retrospective chart review and cross-sectional survey of working-age U.S. adults. J Manag Care Pharm. 2012;18: 415–426. 7. White LA, Birnbaum HG, Kaltenboeck A, Tang J, Mallett D, Robinson RL. Employees with fibromyalgia: medical comorbidity, healthcare costs, and work loss. J Occup Environ Med. 2008;50:13–24. 8. Schaefer C, Chandran A, Hufstader M, et al. The comparative burden of mild, moderate and severe fibromyal- gia: results from a cross-sectional survey in the United States. Health Qual Life Outcomes. 2011;9:71. 1038 HALPERN ET AL.
  13. 13. 9. Wolfe F, Clauw DJ, Fitzcharles MA, et al. The Amer- ican College of Rheumatology preliminary diagnostic criteria for fibromyalgia and measurement of symptom severity. Arthritis Care Res (Hoboken). 2010;62:600–610. 10. 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;33:160–172. 11. American Pain Society. Guideline for the Management of Fibromyalgia Syndrome Pain in Adults and Children. Chicago, IL: American Pain Society; 2005. 12. Arnold LM, Clauw DJ, McCarberg BH. FibroCollab- orative. Improving the recognition and diagnosis of fibromyal- gia. Mayo Clin Proc. 2011;86:457–464. 13. Fitzcharles MA, Ste-Marie PA, Goldenberg DL, et al. 2012 Canadian guidelines for the diagnosis and management of fibromyalgia syndrome. 2012. http://www.canadianpainso- ciety.ca/pdf/Fibromyalgia_Guidelines_2012.pdf (accessed October 10, 2014). 14. Carville SF, Arendt-Nielsen S, Bliddal H, et al. EULAR evidence-based recommendations for the management of fibromyalgia syndrome. Ann Rheum Dis. 2008;67:536–541. 15. Arnold LM, Clauw DJ, Dunegan LJ, Turk DC. FibroCollaborative. A framework for fibromyalgia manage- ment for primary care providers. Mayo Clin Proc. 2012;87:488–496. 16. Traynor LM, Thiessen CN, Traynor AP. Pharma- cotherapy of fibromyalgia. Am J Health Syst Pharm. 2011;68:1307–1319. 17. Gore M, Sadosky AB, Zlateva G, Clauw DJ. Clinical characteristics, pharmacotherapy and healthcare resource use among patients with fibromyalgia newly prescribed gabapen- tin or pregabalin. Pain Pract. 2009;9:363–374. 18. Gore M, Tai KS, Chandran A, Zlateva G, Leslie D. Clinical comorbidities, treatment patterns, and healthcare costs among patients with fibromyalgia newly prescribed pregabalin or duloxetine in usual care. J Med Econ. 2012;15:19–31. 19. Gore M, Tai KS, Chandran A, Zlateva G, Leslie D. Clinical characteristics, pharmacotherapy, and healthcare resource use among patients with fibromyalgia newly pre- scribed pregabalin or tricyclic antidepressants. J Med Econ. 2012;15:32–44. 20. 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;11:217–229. 21. Kleinman NL, Sanchez RJ, Lynch WD, Cappelleri JC, Beren IA, Joshi AV. Health outcomes and costs among employees with fibromyalgia treated with pregabalin vs. standard of care. Pain Pract. 2011;11:540–551. 22. Boulanger L, Wu N, Chen SY, et al. Predictors of pain medication selection among patients diagnosed with fibromyalgia. Pain Pract. 2012;12:266–275. 23. White LA, Robinson RL, Yu AP, et al. Comparison of health care use and costs in newly diagnosed and established patients with fibromyalgia. J Pain. 2009;10:976–983. 24. Rosner B. Fundamentals of Biostatistics, 7th ed. Boston, MA: Brooks/Cole; 2010. 25. Cox DR. Regression models and life tables (with discussion). J R Stat Soc Series B. 1972;34:187–220. 26. Kleinbaum DG. Survival Analysis: A Self-Learning Text, 3rd ed.. New York, NY: Springer-Verlag; 2011. 27. Schoenfeld D. Partial residuals for the proportional hazards regression model. Biometrika. 1982;69:239–241. 28. SAS Institute Inc. http://www.sas.com/en_us/ home.html (accessed January 7, 2015). 29. Burke JP, Sanchez RJ, Joshi AV, Cappelleri JC, Kulakodlu M, Halpern R. Health care costs in patients with fibromyalgia on pregabalin vs. duloxetine. Pain Pract. 2012;12:14–22. 30. Fitzcharles MA, Ste-Marie PA, Goldenberg DL, et al. 2012 Canadian Guidelines for the diagnosis and management of fibromyalgia syndrome: executive summary. Pain Res Manag. 2013;18:119–126. 31. Tournebize J, Gibaja V, Muszczak A, Kahn JP. Are physicians safely prescribing opioids for chronic noncancer pain? A systematic review of current evidence Pain Pract. 2016;16:370–383. 32. Furlan AD, Sandoval JA, Mailis-Gagnon A, Tunks E. Opioids for chronic noncancer pain: a meta-analysis of effectiveness and side effects. CMAJ. 2006;174:1589–1594. 33. Fitzcharles MA, Faregh N, Ste-Marie PA, Shir Y. Opioid use in fibromyalgia is associated with negative health related measures in a prospective cohort study. Pain Res Treat. 2013;2013:898493. 34. Painter JT, Crofford LJ. Chronic opioid use in fibromyalgia syndrome: a clinical review. J Clin Rheumatol. 2013;19:72–77. 35. Hawkins RA. Fibromyalgia: a clinical update. J Am Osteopath Assoc. 2013;113:680–689. 36. Ables AZ, Nagubilli R. Prevention, recognition, and management of serotonin syndrome. Am Fam Physician. 2010;81:1139–1142. Guideline Pain Medications for Fibromyalgia 1039

×