Article Text
PDF
Abstract
Background/importance Fibromyalgia is a complex chronic pain disorder that significantly impairs patient well-being. Evaluating the efficacy of muscle relaxants for treating fibromyalgia is crucial for improving patient care.
Objective This study aimed to evaluate the analgesic efficacy of muscle relaxants in patients with fibromyalgia.
Evidence review A comprehensive literature search was conducted using PubMed, EMBASE, Web of Science, ClinicalTrials.gov, and the Cochrane Library. The search included randomized controlled trials (RCTs) comparing skeletal muscle relaxants with placebo/active analgesics for fibromyalgia. The primary outcome was pain intensity, measured by standardized mean difference (SMD) in pain scores. The risk of bias of included RCTs was assessed using the Cochrane Risk of Bias Assessment Instrument for Randomized Controlled Trials.
Findings 14 RCTs (1851 participants) were included. Muscle relaxants were associated with a small but statistically significant reduction in pain scores compared with placebo or active treatment (SMD=–0.24, 95% CI=–0.32 to –0.15, p<0.001, 95% prediction interval=–0.40 to –0.08), with no significant inconsistency (I2=0, 95% CI=0% to 50.79%) and a moderate Grading of Recommendation, Assessment, Development and Evaluation rating. Secondary outcomes showed small, but statistically significant improvements in depression, fatigue and sleep quality. Muscle relaxants were associated with increased incidence of overall adverse effects, fatigue, abnormal taste, and drug withdrawal due to adverse effects.
Conclusions Moderate quality evidence showed that muscle relaxants were associated with a small reduction in pain intensity for patients with fibromyalgia.
- CHRONIC PAIN
- Fibromyalgia
- Analgesia
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, an indication of whether changes were made, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
Introduction
Chronic pain, which lasts for over 3 months, is a prevalent global medical issue, impacting approximately 20% of the global population.1 2 Fibromyalgia is a kind of chronic pain condition characterized by persistent widespread musculoskeletal pain, sleep issues, fatigue, muscle stiffness and depression.3 This condition greatly impairs patients’ quality of life, psychosocial well-being and daily functioning.4–6 Current standard treatment includes a combination of pharmacological drugs, exercise therapy and cognitive behavioral therapy.7 8 Oral medications are often used first and are therefore an important component of the management of fibromyalgia. Commonly used medication includes the Food and Drug Administration (FDA)-approved drugs serotonin-norepinephrine reuptake inhibitors duloxetine, milnacipran and pregabalin; as well as amitriptyline and tramadol.9 10 However, the overall efficacy of these medications are modest and clinical use is often limited by adverse effects.11 This may have contributed to the use of strong opioids for managing fibromyalgia, which is generally not appropriate due to a lack of long-term analgesic effectiveness and issues with dependence and abuse.11 12 Alternative pharmacological treatment options are needed.
Skeletal muscle relaxants, including benzodiazepines (BZDs) and non-BZD anti-spasmodic, have been used to treat chronic pain in patients with fibromyalgia.13–15 They may decrease musculoskeletal pain in fibromyalgia by reducing muscle tone.16 Nociplastic pain is commonly found in patients with fibromyalgia.17 18 Muscle relaxants could potentially reduce nociplastic pain by enhancing the activity of neurotransmitters including serotonin, norepinephrine and gamma-aminobutyric acid (GABA) within the central nervous system.19–22 Although skeletal muscle relaxants are used in up to 20% of patients with fibromyalgia in the USA and Europe,23–25 the evidence available regarding their analgesic efficacy and adverse effect profile is limited. Two previous meta-analyses suggested that skeletal muscle relaxant may reduce pain, improve sleep quality and enhance physical functioning in fibromyalgia.26 27 However, both meta-analyses were conducted over 20 years ago and did not directly compare the analgesic efficacy of skeletal muscle relaxants against a control group. Moreover, adverse effects were not evaluated, and one of the meta-analyses exclusively focused on cyclobenzaprine.26 27
We conducted an updated and comprehensive systematic review and meta-analysis of randomized controlled trials (RCTs) to examine the use of muscle relaxants in the treatment of fibromyalgia. Our primary outcome was to determine its efficacy on chronic pain intensity. We also studied their adverse effect profile as well as other pain-related functional outcomes such as depression, physical functioning, and sleep quality.
Methods
This systematic review and meta-analysis was conducted and registered in PROSPERO (CRD42024495520) and reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis 2020 statement.28 Post hoc changes to the protocol are shown in online supplemental file 1.
Supplemental material
Study eligibility
RCTs were selected if they compared skeletal muscle relaxants with placebo/active analgesic control in adult patients with fibromyalgia diagnosed by a physician based on predefined inclusion criteria, having pain measurement as a study outcome. Active analgesic control was defined as treatment recommended by European League Against Rheumatism (EULAR) for the management of fibromyalgia, including duloxetine, milnacipran, pregabalin, amitriptyline and tramadol. The clinical studies were limited to those involving patients with physician-diagnosed fibromyalgia. We included studies in which randomization took place at the individual or group level. For unpublished studies, records of trials registered on ClinicalTrial.gov were considered eligible. Excluded studies were open-label trials, preclinical animal trials, reviews, qualitative studies, observational studies, case reports, incomplete studies, studies without available results, studies not reported in English, studies focusing on pediatric patients, and patients with non-fibromyalgia chronic pain conditions. Studies having a follow-up period of less than 2 weeks were also excluded, as very short follow-up periods may overestimate outcomes in chronic pain trials.29
Data sources
A systematic literature search was performed on PubMed, EMBASE, Web of Science, ClinicalTrial.gov, and the Cochrane Library, including the Cochrane Central Register of Controlled Trials (CENTRAL) and Cochrane reviews, from inception until November 10, 2023. This search included both published and unpublished studies to minimize publication bias. For the unpublished studies, we included only unpublished clinical trials from ClinicalTrials.gov, while excluding abstracts from conference proceedings and master’s theses or dissertations. The search strategy was developed by the research team and the database search was conducted by two authors (CHS and PML) separately. The title or abstract must include the following keywords: “fibromyalgia” or “fibrositis” and the name of skeletal muscle relaxants, for example, “cyclobenzaprine” or “tizanidine”, as well as “RCT” or “randomized”. Medical subject headings were also used to consolidate the search. The complete search strategy for each database is available in online supplemental file 2.
Study selection
Two investigators (CHS and PML) independently screened all included articles by title, abstract and full text. Differences in opinion were addressed through discussion (CHS, FW and PML). In cases where a mutual agreement could not be achieved, another investigator (SSCW) was involved in making the final decision.
Data extraction
Two reviewers (CHS and PML) independently extracted data from included studies using a standardized data form which included information about trial design, region of enrollment, year of publication, diagnostic criteria for fibromyalgia, type of intervention and control medication, route of administration, dosage, sample size, follow-up duration, and study outcomes. If a selected article contained more than two study arms, data were extracted only from the relevant arms. Both intent-to-treat and per-protocol analyses were considered. If a selected article reported both types of analyses, the intent-to-treat results were extracted to ensure reliability, minimize bias, maintain randomization integrity, and provide a realistic, generalizable estimate of the treatment effect.30 For studies with multiple follow-up periods, the outcome of the longest follow-up was extracted. In studies with at least 12 weeks of follow-up, the outcome from another period of follow-up less than but closest to 12 weeks was also extracted for subgroup analysis. CHS and PML were responsible for study selection. Discrepancies were addressed through discussion. Another investigator (SSCW) was involved in making the final decision if mutual agreement could not be achieved. If the mean or SD was not provided in the trial, we estimated them from available summary statistics (eg, t-value, median, and range) based on recommendations from the Cochrane Handbook and Wan et al.31 32 When data were missing from the trial, we contacted the corresponding author twice over a 4-week period.
The following outcome measurement instruments were considered eligible for assessing pain intensity: Visual Analog Scale (VAS), Numerical Rating Scale (NRS), and other relevant measurement scales. Pain responders were defined as patients with a clinically significant pain score reduction after receiving muscle relaxants, indicated by at least a 30% reduction in pain intensity or at least a 2-point reduction in NRS pain score (0–10 scale).33 Patient global impression of change or other arbitrary ratings of overall improvement were considered eligible for assessing patient-rated overall improvement.34 Rating scales such as the VAS, NRS, Patient-Reported Outcome Measurement Information System (PROMIS) Sleep Disturbance Score, Sleep Quality Scale, and other questionnaires were eligible for evaluating sleep quality. Measurement tools considered eligible for assessing fatigue included rating scales such as VAS, NRS, PROMIS Fatigue Score, Brief Fatigue Inventory, and other questionnaires. VAS, NRS, revised Fibromyalgia Impact Questionnaire, Health Assessment Questionnaire Disability Index, and other questionnaires were considered eligible for assessing physical function. The Beck Depression Inventory, Patient Health Questionnaire-9, Hospital Anxiety and Depression Scale, and other relevant questionnaires were eligible for evaluating depression.
Risk of bias assessment
The Revised Cochrane risk-of-bias tool for randomized trials was used to examine the risk of bias of included RCTs separately for each outcome.35 This tool, developed by the Cochrane Collaboration, systematically assesses several domains of potential bias, including the randomization process, deviations from intended interventions, missing outcome data, measurement of outcomes, and selection of reported results. Each domain is rated as “low risk”, “some concerns”, or “high risk” of bias. An overall low risk of bias was assigned if all five domains received low-risk ratings. Conversely, if at least one domain received a high-risk rating, the overall risk of bias was deemed high. In cases where no domains received a high-risk rating but not all domains received low-risk ratings, the overall risk of bias was considered to have some concerns. The assessments were performed separately for each outcome to ensure a thorough evaluation of potential biases affecting the findings, thereby enhancing the reliability and validity of the analyses. CHS and FW separately assessed the risk of bias. Differences in opinion were addressed through discussion. Another investigator (SSCW) was involved in making the final decision if mutual agreement could not be achieved.
Certainty of evidence
The Grade of Recommendations Assessment, Development, and Evaluation (GRADE) method was used to examine the evidence level for each meta-analysis outcome.36 The level of evidence was assessed based on the following five domains: risk of bias, inconsistency, imprecision, indirectness, and publication bias. The evidence level was classified into high, moderate, low, or very low certainty. If a limitation was found in any of the domains, the level of evidence would be downgraded according to the instructions specified in online supplemental file 3. Two reviewers (CHS and FW) separately performed a GRADE assessment for each outcome. Differences in opinion were addressed through discussion. If mutual agreement could not be achieved, a third investigator (SSCW) was involved in making the final decision.
Data synthesis and analysis
The primary outcome of this systematic review and meta-analysis was to explore the effect of muscle relaxants on pain scores. Secondary outcomes encompassed patient-rated overall improvement, depression, fatigue, sleep quality, and adverse effects. Due to the heterogeneity of scales used to assess different outcomes, the effect size was evaluated using the standardized mean difference (SMD) for various pain scales or other secondary scores. SMD effect size was categorized as follows according to Cohen’s classification:37 SMD<0.2 (insignificant), 0.2≤SMD<0.5 (small), 0.5≤SMD<0.8 (moderate), and SMD≥0.8 (large). SMD was adjusted for small sample sizes and expressed as Hedges’s g by applying a correction factor J to Cohen’s d.38
We have used SMD≥0.2, the cut-off for a small effect size, to represent minimally clinically important difference for study outcomes.39 Negative values indicated smaller measurements in the treatment group and reflected lower pain intensity. The random effects model, using the restricted maximum likelihood (REML) method combined with the Knapp-Hartung adjustment, was used for estimating the SMD. Additionally, a 95% prediction interval was reported, providing a range within which the true effect size of a future study is expected to fall. This interval accounts for the estimated overall effect and the between-study heterogeneity.40
To evaluate the robustness of the meta-analysis results, leave-one-out (LOO) sensitivity analyses were performed. This method involved systematically removing one study at a time from the data set and recalculating the overall effect size and its 95% CI for each iteration to determine whether the exclusion of any single study significantly impacted the overall findings. The stability of the effect sizes and heterogeneity measures across iterations were assessed to ensure that the meta-analysis results were not unduly influenced by any individual study. In addition, influence analysis was conducted to identify studies that had a disproportionately large impact on the meta-analysis results. Several influence diagnostics were computed, including Cook’s distance, Difference in Beta Statistics (DFBETAS), hat values, and covariance ratios. These metrics helped identify studies with high influence or leverage that could affect the overall findings. Influence plots were generated to visualize and identify studies exerting excessive influence on the overall results.
The OR was used to assess the association between an exposure (eg, treatment or intervention) and an outcome (eg, a clinical event or response) for the following outcomes: pain responder, patient-rated overall improvement, and adverse effects. Inconsistency was assessed using Q, I2, and tau2 statistical measures. Meta-regression was conducted to investigate the relationship between covariates and effect size, using a mixed-effects model with the REML method and the Knapp-Hartung adjustment. Small-study effects were assessed using a funnel plot, Egger’s regression test, and Begg and Mazumdar’s test. Egger’s test assesses the asymmetry of the funnel plot by regressing the standard normal deviation (effect size divided by its SE) against the SE. A significant intercept suggests the presence of publication bias.41 Begg and Mazumdar’s test assesses the correlation between the ranks of effect sizes and their variances. A significant correlation indicates potential publication bias.42 Using both methods can provide a more robust assessment of potential small-study effects, as each method has its strengths and limitations. Egger’s test is sensitive and more powerful for detecting bias but can be affected by the number of studies.43 Begg and Mazumdar’s test is less sensitive but more robust for outliers and small sample sizes.44 When all three criteria were met, it can be concluded that there were no small-study effects. All the analyses were conducted using the “metafor” package from R V.4.4.0 (2024, URL https://www.r-project.org). Statistical significance was determined by a two-sided p value of <0.05. Multiple adjustments were not performed to avoid the risk of being overly conservative and potentially missing true effects, given that the data being analyzed were actual observations derived from real-world studies and not random numbers.45
Results
Search results
Overall, 503 articles were identified for assessment (figure 1). After removing the duplicates, 388 studies were screened based on the title and abstract. 34 articles were assessed for eligibility through a detailed review of the full text, leading to the exclusion of 20 studies that did not meet the eligibility criteria (online supplemental file 4). A total of 14 RCTs were selected for inclusion in the systematic review and meta-analysis.46–59 Some articles presented multiple comparisons. The relevant comparisons were considered in accordance with the study objectives.
Flow chart of study selection process. RCT, randomized controlled trial.
Study characteristics
The 14 included RCTs (19 comparisons) consisted of 1,851 patients (table 1, online supplemental file 5). There were 12 parallel trials46 47 49–53 55–59 and two crossover trials.48 54 All trials randomized participants at the individual level. Intent-to-treat analysis was performed in seven trials,46 47 50 51 57–59 and per-protocol analysis was performed in the other seven trials.48 49 52–56 Most of the clinical trials diagnosed fibromyalgia using established diagnostic criteria, such as Yunus et al (N=4),49 52 53 56 American College of Rheumatology (ACR) 1990 criteria (N=3),47 51 57 and ACR 2010 criteria (N=3).50 58–62 The remaining studies (N=4)46 48 54 55 used arbitrary criteria consistent with widespread pain, multiple tender points and other somatic symptoms. Non-BZD anti-spasmodic were the most frequently investigated muscle relaxants (N=12),46–48 50–54 56–59 with cyclobenzaprine being the most widely used drug (N=10).46–48 50 51 53 54 57–59 Carisoprodol and chlormezanone were the other two non-BZD anti-spasmodic studied.52 56 The remaining studies focused on alprazolam (N=2).49 55 All studies compared muscle relaxants against a placebo, and one three-arm study compared cyclobenzaprine with amitriptyline in addition to a placebo.47 The mean duration of the studies was 9.8 weeks, with five studies lasting 12 weeks or longer.46 47 50 58 59 10 studies reported receiving industry funding.47–51 53–55 58 59
Characteristic of included studies
Meta-analysis of overall analgesic effect and meta-regression
Meta-analysis of all 19 comparisons revealed a statistically significant and small reduction in pain scores with muscle relaxants compared with placebo or active treatment (SMD=–0.24, 95% CI=–0.32 to –0.15, p<0.001, 95% prediction interval=–0.40 to –0.08) (figure 2A). The prediction interval not including zero indicates that the effect is expected to be negative in future studies as well. This provides additional confidence that the observed effect is not only statistically significant but also likely to be observed in new studies. Two comparisons had a low risk of bias, 10 had some concerns, and another seven had a high risk of bias. The pooled result was associated with a GRADE rating of moderate level (table 2 and online supplemental file 6) and no significant inconsistency was found (Q=21.00, p>0.05; I2=0, 95% CI=0% to 50.79%; tau2=0, 95% CI=0 to 0.10). The majority of the controls were placebos, with only one clinical study (consisting of two comparisons) comparing against active drug. A small effect size in pain reduction was found for muscle relaxants when compared with placebo only (SMD=–0.26, 95% CI=–0.35 to –0.18, p<0.001, 95% prediction interval=–0.43 to –0.10) (figure 2B). The pooled result was associated with a moderate GRADE rating (table 2 and online supplemental file 6) and there was no significant inconsistency (Q=16.77, p>0.05; I2=0.25%, 95% CI=0 to 47.86%; tau2=0.0001, 95% CI=0 to 0.09). Two comparisons had a low risk of bias, eight had some concerns, and another seven had a high risk of bias. In the clinical study involving active drugs, cyclobenzaprine was compared against amitriptyline. There were no substantial differences in pain scores between the two groups at 1 month (mean NRS pain score 2.23±0.88 vs 2.22±0.99) and 6 months (2.11±0.93 vs 2.17±1.02) follow-up.
Comparison of analgesic effects between muscle relaxants and placebo/active treatment. (A) All clinical trials included; (B) muscle relaxants versus placebo; (C) pain responder; (D) cyclobenzaprine as a muscle relaxant; (E) other types of muscle relaxants (non-cyclobenzaprine); (F) trials with a follow-up period shorter than 3 months; (G) trials with a follow-up period of 3 months or longer; (H) trials involving oral administration; and (I) trials involving sublingual administration. RE, random effects; SMD, standardized mean difference.
Assessment of quality of evidence using Grade of Recommendations Assessment, Development, and Evaluation (GRADE)
Leave-one-out sensitivity analysis and influence analysis
The LOO sensitivity analysis demonstrated that the overall effect size estimates remained stable after the exclusion of each individual study. The effect sizes ranged from –0.26 to –0.29, with corresponding 95% CIs consistently excluding zero. This indicates that the exclusion of any single study did not significantly alter the overall results (online supplemental file 7 and 8).
Influence analysis showed that no single study had a disproportionately large impact on the overall results. These diagnostics help identify studies with high influence or leverage that could affect the overall findings. Influence plots further confirmed that no studies exerted excessive leverage on the overall results. While certain studies (such as studies 9, 10, and 14) exhibited higher influence metrics, their exclusion did not substantially alter the overall results of the meta-analyses (online supplemental file 9 and 10).
The proportion of pain responders was also analyzed. Meta-analysis of four comparisons showed that the odds of a positive pain response were 1.72 times higher in the muscle relaxant group than the control group (OR=1.72, 95% CI=1.58 to 1.88, p=0.001, 95% prediction interval=1.04 to 2.86) (figure 2C). One comparison had a low risk of bias, one had some concerns, and another two had a high risk of bias. This result had a high-level GRADE rating (table 2 and online supplemental file 6) and no significant inconsistency (Q=0.34, p>0.05; I2=0, 95% CI=0 to 29.96%; tau2=0, 95% CI=0 to 0.04).
Meta-regression was further performed to compare the analgesic efficacy (in terms of reduction in pain intensity) of varying follow-up durations and different routes of drug administration methods. Since cyclobenzaprine was the most commonly studied muscle relaxant, its efficacy was separately compared with other muscle relaxants. No significant differences were found for all these variables (online supplemental file 11). Subgroup meta-analysis was conducted for duration of treatment, route of administration and types of muscle relaxants.
Meta-analysis of analgesic effect with specific muscle relaxants
There were 15 comparisons specifically for cyclobenzaprine and four on other muscle relaxants (carisoprodol, chlormezanone and alprazolam). Subgroup meta-analysis with cyclobenzaprine studies only showed a statistically significant and small reduction in pain scores compared with the control group in favor of cyclobenzaprine (SMD=–0.23, 95% CI=–0.31 to –0.15, p<0.001, 95% prediction interval=–0.39 to –0.07) (figure 2D). Two comparisons had low risk of bias, eight had some concerns, and another five had a high risk of bias. The GRADE rating was moderate (table 2 and online supplemental file 6), and no significant inconsistency was found (Q=12.65, p>0.05; I2=0, 95% CI=0 to 35.28%; tau2=0, 95% CI=0 to 0.04). Patients who received other types of muscle relaxants (non-cyclobenzaprine) did not have a statistically significant reduction in pain scores compared with the control group (SMD=–0.42, 95% CI=–1.34 to 0.50, p>0.05, 95% prediction interval=–1.77 to 0.93) (figure 2E). Two comparisons had some concerns, and another two had a high risk of bias. The GRADE rating was very low (table 2 and online supplemental file 6) and significant inconsistency was found (Q=7.62, p>0.05; I2=60.32%, 95% CI=0 to 97.05%; tau2=0.20, 95% CI=0 to 4.62).
Meta-analysis of analgesic effect: duration of treatment
There were 13 comparisons with a follow-up duration of shorter than 3 months and six comparisons with a duration of at least 3 months or more. For those with a duration of less than 3 months, patients who received muscle relaxants experienced a statistically significant but small reduction in pain scores compared with control (SMD=–0.29, 95% CI=–0.48 to –0.11, p=0.005) (figure 2F). Future studies might observe a wider range of effects, including the possibility of no effect (95% prediction interval=–0.77 to 0.18). One comparison had a low risk of bias, six had some concerns, and another six had a high risk of bias. This pooled result was linked to a GRADE rating of moderate level (table 2 and online supplemental file 6) and had no notable inconsistency (Q=16.50, p>0.05; I2=30.12%, 95% CI=0 to 64.77%; tau2=0.02, 95% CI=0 to 0.23). Similarly, for those with a duration of at least 3 months, patients who received muscle relaxants exhibited a statistically significant and small reduction in pain scores in the fibromyalgia group compared with control (SMD=–0.23, 95% CI=–0.35 to –0.11, p=0.005, 95% prediction interval=–0.41 to –0.04) (figure 2G). One comparison had a low risk of bias, four had some concerns, and another one had a high risk of bias. The pooled result, which had a GRADE rating of moderate level (table 2 and online supplemental file 6), showed no significant inconsistency (Q=4.43, p>0.05; I2=0, 95% CI=0 to 81.87%; tau2=0, 95% CI=0 to 0.10).
Meta-analysis of analgesic effect with different routes of administration
There were 15 comparisons involving oral administration and four comparisons involving sublingual administration. Subgroup analysis for muscle relaxants given orally demonstrated a statistically significant and small reduction in pain scores compared with the control group (SMD=–0.31, 95% CI=–0.47 to –0.14, p=0.001) (figure 2H). Eight comparisons had some concerns, and another seven had a high risk of bias. Future studies might observe a wider range of effects, including the possibility of no effect (95% prediction interval=–0.76 to 0.14). The result was associated with a GRADE rating of moderate level (table 2 and online supplemental file 6) and no significant inconsistency was found (Q=18.37, p>0.05; I2=25.99%, 95% CI=0 to 59.74%; tau2=0.02, 95% CI=0 to 0.17). For the sublingual administration group, patients who received muscle relaxants experienced a statistically significant and small reduction in pain scores compared with control (SMD=–0.21, 95% CI=–0.32 to –0.09, p=0.01, 95% prediction interval=–0.38 to –0.04) (figure 2I). Two comparisons had a low risk of bias, and another two comparisons had some concerns. The GRADE rating was moderate (table 2 and online supplemental file 6), and there was no significant inconsistency (Q=1.63, p>0.05; I2=0, 95% CI=0 to 90.52%; tau2=0, 95% CI=0 to 0.10).
Meta-analysis of effects on secondary outcomes
Meta-analyses were conducted on secondary outcomes for patient-rated overall improvement, depression, fatigue, sleep quality, and physical function. For the patient-rated overall improvement, the odds were 1.538 times higher in the muscle relaxant group than control group (OR=1.50, 95% CI=1.18 to 1.90, p=0.006, 95% prediction interval=1.04 to 2.15) (figure 3A). Two comparisons had a low risk of bias, seven had some concerns, and another four had a high risk of bias. This finding had a high GRADE rating (table 2 and online supplemental file 12) and no significant inconsistency (Q=20.73, p>0.05; I2=0, 95% CI=0 to 70.12%; tau2=0, 95% CI=0 to 1.34). Patients who received muscle relaxants experienced a statistically significant and small reduction in depression compared with those in the control group (SMD=–0.20, 95% CI=–0.32 to –0.07, p=0.009) (figure 3B). Future studies might observe a wider range of effects, including the possibility of no effect (95% prediction interval=–0.43 to 0.03). Five comparisons had a low risk of bias, and two had some concerns. The GRADE rating was moderate (table 2 and online supplemental file 12), and no significant inconsistency was found (Q=4.38, p>0.05; I2=0.02%, 95% CI=0 to 82.88%; tau2=0, 95% CI=0 to 0.28). Patients who received muscle relaxants had a statistically significant and small reduction in fatigue compared with control (SMD=–0.20, 95% CI=–0.28 to –0.12, p<0.001, 95% prediction interval=–0.38 to –0.03) (figure 3C). Two comparisons had a low risk of bias, three had some concerns, and another three had a high risk of bias. The GRADE rating was low (table 2 and online supplemental file 12), and no notable inconsistency was found (Q=3.63, p>0.05; I2=0, 95% CI=0 to 32.08%; tau2=0, 95% CI=0 to 0.04). Patients who received muscle relaxants had statistically significant and small enhancement in sleep quality compared with those in the control group (SMD=–0.31, 95% CI=–0.40 to –0.21, p<0.001, 95% prediction interval=–0.47 to –0.14) (figure 3D). Two comparisons had a low risk of bias, five had some concerns, and another five had a high risk of bias. The GRADE rating was high (table 2 and online supplemental file 12), and no notable inconsistency was identified (Q=11.41, p>0.05; I2=0, 95% CI=0 to 60.59%; tau2=0, 95% CI=0 to 0.15). Patients receiving muscle relaxants had statistical improvement in physical functioning scores, but the standard mean difference was negligible (SMD=–0.16, 95% CI=–0.27 to –0.05, p=0.01, 95% prediction interval=–0.35 to 0.04) (figure 3E). Six comparisons had a low risk of bias, three had some concerns, and another two had a high risk of bias. The GRADE rating was moderate (table 2 and online supplemental file 12), and no notable inconsistency was reported (Q=13.37, p>0.05; I2=7.59%, 95% CI=0 to 79.53%; tau2=0, 95% CI=0 to 0.24).
Secondary outcomes comparing muscle relaxants to placebo/active treatment. (A) Patient-rated overall improvement; (B) impact on depression levels; (C) effect on fatigue; (D) improvement in sleep quality; and (E) change in physical function. RE, random effects; SMD, standardized mean difference.
Meta-analysis of adverse effects
The reported adverse effects were predominantly of mild-to-moderate intensity. In comparison to placebo, no statistically significant differences were observed in the incidence of severe adverse effects (OR=0.64, 95% CI=0.21 to 1.95, p=0.46, 95% prediction interval=0.05 to 8.61). However, fatigue (OR=2.99, 95% CI=1.84 to 4.85, p=0.02, 95% prediction interval=0.66 to 13.46), abnormal taste (OR=14.12, 95% CI=11.66 to 17.11, p=0.001, 95% prediction interval=1.95 to 102.34), drug withdrawal due to adverse effects (OR=2.12, 95% CI=1.46 to 3.09, p=0.002, 95% prediction interval=0.98 to 4.58) and overall adverse effect (OR=4.76, 95% CI=2.00 to 11.29, p=0.01, 95% prediction interval=0.30 to 75.28) were statistically more likely to occur in patients treated with muscle relaxants compared with placebo. There were no statistically significant differences observed in the occurrence of dry mouth (OR=4.54, 95% CI=1.39 to 14.82, p=0.07, 95% prediction interval=0.16 to 126.44), drowsiness (OR=2.45, 95% CI=0.74 to 8.04, p=0.21, 95% prediction interval=0.12 to 49.42), constipation (OR=1.79, 95% CI=0.33 to 9.59, p=0.57, 95% prediction interval=0.06 to 57.55), dizziness (OR=1.35, 95% CI=0.56 to 3.27, p=0.55, 95% prediction interval=0.19 to 9.44), headache (OR=1.23, 95% CI=0.26 to 5.74, p=0.82, 95% prediction interval=0.04 to 41.69) and nausea (OR=0.72, 95% CI=0.24 to 2.19, p=0.61, 95% prediction interval=0.06 to 8.76) (online supplemental file 13).
Assessment of small-study effects
Assessment of small-study effects was conducted using funnel plots, Egger’s regression test, and Begg and Mazumdar’s test. No significant small-study effects were found in the 19 RCTs comparing muscle relaxants to placebo or active treatment, as evidenced by the funnel plot, Egger’s test (p=0.27), and Begg and Mazumdar’s test (p=0.33) (online supplemental file 14 and 15). Subgroup analyses also showed no significant small-study effects except for fatigue. A significant small-study effect was present for fatigues (Egger’s test: p=0.04), possibly due to the small number of studies (n=8). However, Begg and Mazumdar’s test did not confirm this effect (Begg and Mazumdar’s test: p=0.06). Funnel plots displayed also showed a fairly symmetrical pattern (online supplemental file 14 and 15).
Discussion
Overall findings
In this meta-analysis of RCTs, skeletal muscle relaxants were linked to a small but statistically significant reduction in pain scores compared with placebo/active treatment in patients with fibromyalgia. Subgroup analysis showed a small effect in favor of muscle relaxants for both short-term (less than 3 months) and long-term (at least 3 months) treatment duration and for both oral and sublingual administration. Skeletal muscle relaxants were associated with favorable outcomes in patient-rated overall improvement, fatigue, and sleep quality. They were associated with a higher incidence of overall adverse effect, depression, fatigue, abnormal taste, and drug withdrawal due to adverse effects, but no difference in serious adverse effects. The GRADE level of evidence was mostly moderate or high and inconsistency was not significant for most outcomes.
Implications for practice
The results of our study showed a positive analgesic effect with muscle relaxants for the treatment of fibromyalgia. This was consistent regardless of route of administration and duration of follow-up. Skeletal muscle relaxants can increase the levels of serotonin, norepinephrine and GABA in the central nervous system.17 18 This can reduce nociplastic pain, which is an important component of pain in fibromyalgia.17 18 Since pain from muscle tightness is common among patients with fibromyalgia, muscle relaxants may also decrease pain by reducing muscle tone.16 The effect size in pain intensity improvement was small, which was similar to those demonstrated for duloxetine and pregabalin.63 In addition to a reduction in pain intensity, the odds of achieving clinically significant pain reduction were also higher in those given skeletal muscle relaxants.33 64 Improvement in pain-related secondary outcomes, such as patient-rated overall improvement, depression, sleep, and fatigue, further suggests that there is a clinically relevant analgesic benefit. Improvement in secondary outcomes could be attributed to the bidirectional relationship between pain and these outcomes. It is well-established that poorly controlled pain can negatively impact sleep quality and increase fatigue, while poor sleep and fatigue can exacerbate pain.65 66 These findings suggest that muscle relaxants can improve pain and related secondary outcomes in patients with fibromyalgia and can be considered a useful pharmacological treatment option.
There was an increased incidence of overall adverse effects for muscle relaxants, which were more likely to cause discontinuation of medication. This may be attributed to the inadvertent activation of the central nervous system and muscarinic receptors by muscle relaxants.13 Cyclobenzaprine is known to exhibit a potent anti-cholinergic effect,67 while benzodiazepines can cause central nervous system side effects through the action on BZD receptors.15 These factors may contribute to the fatigue experienced by participants in our study. Nonetheless, muscle relaxants were not associated with an increased risk of serious adverse effects nor with other common anticholinergic adverse effects, such as drowsiness, dizziness, and constipation. On the other hand, these adverse effects are commonly encountered with pregabalin, gabapentin, and duloxetine.68 69 Since these adverse effects limit optimal drug titration and cause treatment withdrawal, this may be a potential advantage of muscle relaxants in clinical practice. Abnormal product taste was a specific adverse effect reported in three clinical trials that studied sublingual cyclobenzaprine, which may be attributed to the new formulation (TNX-102 SL) used in these studies.50 58 59
Implications for research
We only included randomized controlled trials in this systematic review and meta-analysis. This was because randomized controlled trials have a higher validity and are associated with less bias compared with observational studies.70 In addition to pain intensity, other pain-related secondary outcomes including sleep quality, physical functioning, depression, and patient-rated overall improvement were reported. This was essential to provide a holistic evaluation of the overall effect of muscle relaxants for fibromyalgia and enable a more enhanced assessment of clinical benefit. Cyclobenzaprine was the most extensively researched drug. It is a skeletal muscle relaxant that is structurally similar to tricyclic antidepressants.21 Prior to our analysis, there was one other meta-analysis on cyclobenzaprine conducted in 2004 by Tofferi et al, which included five clinical studies.27 They reported a small yet statistically significant short-term benefit for pain relief, sleep improvement, and patient-rated overall improvement compared with baseline in favor of cyclobenzaprine. This is similar to results from our subgroup meta-analysis, which also showed a statistically significant and small reduction in pain scores. In addition, our study showed that there were both short-term (less than 3 months) and longer-term (3 months or more) analgesic effects. While cyclobenzaprine is not FDA-approved for the management of fibromyalgia, it is recommended by the EULAR as one of the pharmacological treatment options.10 Our meta-analysis provides moderate GRADE evidence to support its use for fibromyalgia. Subgroup meta-analysis showed that non-cyclobenzaprine muscle relaxants were also associated with small pain reduction. However, this subgroup meta-analysis only consisted of four clinical studies (carisoprodol, chlormezanone and alprazolam), and was associated with a very low GRADE level of evidence. We were unable to identify randomized controlled trials for other muscle relaxants such as tizanidine and orphenadrine. It is not possible to draw conclusions on the analgesic effect of other muscle relaxants based on the currently available evidence. Clinical trials on other muscle relaxants for fibromyalgia would be worthwhile to explore new pharmacological treatment options.11
Strengths and limitations
In addition to analgesic efficacy, our analysis encompassed a variety of pain-related functional outcomes and adverse effect profile. Two previous systematic review and meta-analyses mainly focused on analgesic efficacy without evaluating adverse effects.26 27 One of the meta-analyses only evaluated cyclobenzaprine.27 In contrast to the two previous meta-analyses,26 27 our study directly compared muscle relaxants to a control group, which is associated with less bias compared with before and after studies.71 To minimize publication bias, we included both published and unpublished studies in our review.72–74 We also used the GRADE approach to systematically rate the quality of evidence in our analysis.36 Most of the quantitative analyses in this study were associated with moderate to high GRADE rating and non-significant inconsistency, suggesting good reliability in the level of clinical evidence.
This systematic review and meta-analysis has several limitations. Although we included clinical studies over different durations, most had a follow-up duration of less than 12 weeks. Therefore, the overall results may be less informative on long-term efficacy and safety. Another limitation was the inconsistency in the definitions of adverse effects used by different clinical trials. To account for this, we reported relative ORs instead of the absolute prevalence of adverse effects. In addition, four trials adopted distinct arbitrary criteria for fibromyalgia as they were conducted before the 1990 ACR criteria was available. This may potentially introduce some inconsistency among patients included. Despite an extensive search of multiple databases, there were only four trials on muscle relaxants that were not cyclobenzaprine. This limits the generalizability of study findings to other muscle relaxants. It also highlights the need for clinical trials on other skeletal muscle relaxants for treating fibromyalgia. Moreover, our review only included clinical trials published in English. Although we did not restrict language during our literature search, we found no relevant studies in other languages. Multiple meta-analyses were performed, but multiple corrections for p values were not applied. Therefore, the results should be interpreted with caution. Finally, since this was an aggregate data meta-analysis there may be potential for ecological fallacy.
Conclusions
Skeletal muscle relaxants resulted in a small reduction in pain intensity for patients with fibromyalgia compared with placebo or active treatment. They were also associated with higher odds of clinically significant pain relief, as well as patient-rated overall improvement, depression, fatigue, and sleep quality. There was most evidence to support the use of cyclobenzaprine. Our findings suggest that muscle relaxants, in particular cyclobenzaprine, may provide clinically relevant analgesic benefit, and can be considered as a potentially useful pharmacological treatment option for treating fibromyalgia.
Supplemental material
Ethics statements
Patient consent for publication
Ethics approval
Not applicable.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
X @stancius2
CHS and FW contributed equally.
Contributors CHS helped with study concept, study design, screening of papers, data extraction, data interpretation, drafting of manuscript and final review. FW helped with study concept, study design, data extraction, data analysis, data interpretation, drafting of manuscript and final review. LNLL helped with data interpretation and final review. PML helped with screening of papers, data extraction and final review. HCH helped with data interpretation and final review. SSCW helped with study concept, study design, data interpretation and final review. SSCW is the guarantor.
Funding This work was supported by the Department of Anaesthesiology, School of Clinical Medicine, Li Ka Shing Faculty of Medicine, The University of Hong Kong.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.