Abstract
Skeletal muscle relaxants are a heterogeneous group of medications used to treat two different types of underlying conditions: spasticity from upper motor neuron syndromes and muscular pain or spasms from peripheral musculoskeletal conditions. Although widely used for these indications, there appear to be gaps in our understanding of the comparative efficacy and safety of different skeletal muscle relaxants. This systematic review summarizes and assesses the evidence for the comparative efficacy and safety of skeletal muscle relaxants for spasticity and musculoskeletal conditions. Randomized trials (for comparative efficacy and adverse events) and observational studies (for adverse events only) that included oral medications classified as skeletal muscle relaxants by the FDA were sought using electronic databases, reference lists, and pharmaceutical company submissions. Searches were performed through January 2003. The validity of each included study was assessed using a data abstraction form and predefined criteria. An overall grade was allocated for the body of evidence for each key question. A total of 101 randomized trials were included in this review. No randomized trial was rated good quality, and there was little evidence of rigorous adverse event assessment in included trials or observational studies. There is fair evidence that baclofen, tizanidine, and dantrolene are effective compared to placebo in patients with spasticity (primarily multiple sclerosis). There is fair evidence that baclofen and tizanidine are roughly equivalent for efficacy in patients with spasticity, but insufficient evidence to determine the efficacy of dantrolene compared to baclofen or tizanidine. There is fair evidence that although the overall rate of adverse effects between tizanidine and baclofen is similar, tizanidine is associated with more dry mouth and baclofen with more weakness. There is fair evidence that cyclobenzaprine, carisoprodol, orphenadrine, and tizanidine are effective compared to placebo in patients with musculoskeletal conditions (primarily acute back or neck pain). Cyclobenzaprine has been evaluated in the most clinical trials and has consistently been found to be effective. There is very limited or inconsistent data regarding the effectiveness of metaxalone, methocarbamol, chlorzoxazone, baclofen, or dantrolene compared to placebo in patients with musculoskeletal conditions. There is insufficient evidence to determine the relative efficacy or safety of cyclobenzaprine, carisoprodol, orphenadrine, tizanidine, metaxalone, methocarbamol, and chlorzoxazone. Dantrolene, and to a lesser degree chlorzoxazone, have been associated with rare serious hepatotoxicity.
Keywords
1. Introduction
Skeletal muscle relaxants are a heterogeneous group of medications commonly used to treat two different types of underlying conditions: spasticity from upper motor neuron syndromes and muscular pain or spasms from peripheral musculoskeletal conditions.
Spasticity from the upper motor neuron syndrome (a complex of signs and symptoms that can be associated with exaggerated reflexes, autonomic hyperreflexia, dystonia, contractures, paresis, lack of dexterity, and fatigability, in addition to spasticity) can result from a variety of conditions affecting the cortex or spinal cord. Some of the more common conditions associated with spasticity include multiple sclerosis,
2.
spinal cord injury,3.
traumatic brain injury, cerebral palsy, and post-stroke syndrome.4.
In many patients with these conditions, spasticity can be disabling and painful, with a marked effect on functional ability and quality of life.Common musculoskeletal conditions causing tenderness and muscle spasms include fibromyalgia,
6.
tension headaches,7.
myofascial pain syndrome, and mechanical low back or neck pain. If muscle spasm is present in these conditions, it is related to local factors involving affected muscle groups. These conditions are commonly encountered in clinical practice and can cause significant disability and pain in some patients. Skeletal muscle relaxants are one of several classes of medications frequently used to treat these conditions.8.
, 9.
, 10.
Drugs classified as skeletal muscle relaxants include baclofen, carisoprodol, chlorzoxazone, cyclobenzaprine, dantrolene, metaxalone, methocarbamol, orphenadrine, and tizanidine. Only baclofen, dantrolene, and tizanidine are approved for the treatment of spasticity. These three medications act by different mechanisms: baclofen blocks pre- and post-synaptic GABAB receptors,
11.
, 12.
tizanidine is a centrally-acting agonist of α2 receptors,13.
, 14.
and dantrolene directly inhibits muscle contraction by decreasing the release of calcium from skeletal muscle sarcoplasmic reticulum.15.
Other medications used to treat spasticity but not formally approved for this indication include benzodiazepines, clonidine, gabapentin, and botulinum toxin.15.
, 16.
, The skeletal muscle relaxants carisoprodol, chlorzoxazone, cyclobenzaprine, metaxalone, methocarbamol, and orphenadrine have been approved for the treatment of musculoskeletal disorders. Cyclobenzaprine is closely related to the tricyclic antidepressants,
18.
carisoprodol is metabolized to meprobamate,19.
methocarbamol is structurally related to mephenesin,18.
chlorzoxazone is a benzoxazolone derivative,20.
and orphenadrine is derived from diphenhydramine.21.
The mechanism of action for most of these agents is unclear, but may be related in part to sedative effects. These drugs are often used for treatment of musculoskeletal conditions, whether muscle spasm is present or not.10.
Although there is some overlap between clinical usage (tizanidine in particular has been studied in patients with musculoskeletal conditions),22.
in clinical practice each skeletal muscle relaxant is used primarily for either spasticity or for musculoskeletal conditions.There is little data regarding the comparative efficacy and safety of different skeletal muscle relaxants. In 2001, Senate Bill 819 was passed by the Oregon Legislature and signed into law by the Governor. The law mandates development of a Practitioner-Managed Prescription Drug Plan (PMPDP) for the Oregon Health Plan (OHP) and evidence-based reviews of the state's most expensive drug classes. The Oregon Health Resources Commission (OHRC) requested such a review of the skeletal muscle relaxant drug class. In consultation with a multidisciplinary committee of experts, we selected the following key questions to guide the review:
- What is the comparative efficacy of different muscle relaxants?
- What is the comparative safety of different muscle relaxants?
- Are there subpopulations of patients for which one muscle relaxant is more effective or associated with fewer adverse effects?
2. Methods
2.1 Literature search
To identify articles relevant to each key question, we searched (in this order): the Evidence-Based Medicine Library (2003, Issue 1) (from the Cochrane Collaboration), MEDLINE (1966–January 2003), EMBASE (1980–January 2003), and reference lists of review articles. In electronic searches we combined terms for spasticity, conditions associated with spasticity, and musculoskeletal disorders with included skeletal muscle relaxants (see Appendix A on the Web site for complete search strategy). In addition, the State of Oregon created and disseminated a protocol to pharmaceutical manufacturers for submitting data. All citations were imported into an electronic database (EndNote 6.0).
2.2 Study selection
All English-language titles and abstracts and suggested additional citations that met the following eligibility criteria were included:
2.2.1 Population
The population included in this review is adult or pediatric patients with spasticity or a musculoskeletal condition. We defined spasticity as muscle spasms associated with an upper motor neuron syndrome. Musculoskeletal conditions were defined as peripheral conditions resulting in muscle or soft tissue pain or spasms. We excluded obstetric and dialysis patients, and patients with restless legs syndrome or nocturnal myoclonus. Senate Bill 819 specifically excludes patients with HIV and patients with cancer.
2.2.2 Drugs
We included the following oral drugs classified as skeletal muscle relaxants: baclofen, carisoprodol, chlorzoxazone, cyclobenzaprine, dantrolene, metaxalone, methocarbamol, orphenadrine, and tizanidine. Other medications used for spasticity but considered to be in another drug class, such as benzodiazepines, quinine, tricyclic antidepressants, gabapentin, and clonidine, were not considered primary drugs in this report, but were reviewed when they were directly compared to an included skeletal muscle relaxant. We excluded trials
20.
, 23.
, 24.
, 25.
, 26.
, 27.
in which an included skeletal muscle relaxant was combined with an analgesic medication unless the comparison arm included the same analgesic medication and dose, trials28.
which evaluated skeletal muscle relaxants not approved in the United States, and trials29.
which only compared one dose of an included skeletal muscle relaxant with another dose.2.2.3 Outcomes
The main efficacy measures were relief of muscle spasms or pain, functional status, quality of life, withdrawal rates, and adverse effects (including sedation, weakness, addiction, and abuse). We excluded non-clinical outcomes such as electromyogram measurements or spring tension measurements.
There is no single accepted standard on how to measure the included outcomes. Spasticity is an especially difficult outcome to measure objectively. The most widely used standardized scales to measure spasticity are the Ashworth
30.
and modified Ashworth31.
scales. In these scales, the assessor tests the resistance to passive movement around a joint and grades it on a scale of 0 (no increase in tone) to 4 (limb rigid in flexion or extension). The modified Ashworth scale adds a “1+” rating between the 1 and 2 ratings of the Ashworth scale. For both of these scales, the scores are usually added for four lower and four upper limb joints, for a total possible score of 0–32, though scoring methods can vary. Other measures of spasticity include the pendulum test, muscle spasm counts, and patient assessment of spasticity severity on a variety of numerical (e.g., 1–3, 1–4, 0–4) or categorical (e.g., none, mild, moderate, severe) scales. Many of these scales have not been validated.Muscle strength is usually assessed with the British Medical Research Council (BMRC) scale, which is based on the observation of resistance provided by voluntary muscle activity.
14.
An assessor grades each muscle or muscle group independently on a scale of 0 (no observed muscle activation) to 5 (full strength).Most studies measure pain using either visual analogue or categorical pain scales. Visual analogue scales (VAS) consist of a line on a piece of paper labeled 0 at one end, indicating no pain, and a maximum number (commonly 100) at the other, indicating excruciating pain. Patients designate their current pain level on the line. Categorical pain scales, on the other hand, consist of several pain category options from which a patient must choose (e.g., no pain, mild, moderate, or severe). Pain control (improvement in pain) and pain relief (resolution of pain) are also measured using visual analogue and categorical scales.
Studies can evaluate functional status using either disease-specific or non-specific scales. Disease-specific scales tend to be more sensitive to changes in status for that particular condition, but non-specific scales allow for some comparisons of functional status between conditions. The most commonly used disease-specific measure of functional and disability status in patients with multiple sclerosis, for example, is the Kurtzke Extended Disability Status Scale (EDSS).
32.
The EDSS measures both disability and impairment, combining the results of a neurological examination and functional assessments of eight domains into an overall score of 0–10 (in increments of 0.5). Disease-specific scales are also available for other musculoskeletal and spastic conditions.33.
, 34.
Scales that are not disease-specific include the Medical Outcomes Study Short Form-36 (SF-36), Short Form-12 (SF-12), and other multi-question assessments. Another approach to measuring function is to focus on how well the medication helps resolve problems in daily living that patients with spasticity or musculoskeletal conditions commonly face, such as getting enough sleep or staying focused on the job. Some studies also report effects on mood and the preference for one medication over another.We focused on the following common adverse events: somnolence or fatigue, dizziness, dry mouth, and weakness. We also paid special attention to reports of serious hepatic injury, abuse, and addiction.
35.
In some studies, only “serious” adverse events or adverse events “thought related to treatment medication” are reported. Many studies do not define these terms. We included information on hospitalizations and deaths when available.Because of inconsistent reporting of outcomes, withdrawal rates may be a more reliable surrogate measure for either clinical efficacy or adverse events in studies of skeletal muscle relaxants. High withdrawal rates probably indicate some combination of poor tolerability and ineffectiveness. An important subset is withdrawal due to any adverse event (those who discontinue specifically because of adverse effects), which may indicate an intolerable adverse event.
2.2.4 Study types
We included the following study types:
- Systematic reviews of the clinical efficacy or adverse event rates of skeletal muscle relaxants for spasticity or musculoskeletal conditions, OR
- Randomized controlled trials that compared one of the included skeletal muscle relaxants listed to another included skeletal muscle relaxant, an antispasticity medication from a different drug class, or placebo in adult patients with spasticity or musculoskeletal conditions, OR
- Randomized controlled trials and large, high quality observational studies that reported adverse event rates for an included skeletal muscle relaxant.
- We did not systematically review case reports and case series in which the proportion of patients suffering an adverse event could not be calculated. We excluded “single-dose” studies, abstracts and unpublished trials unless a pharmaceutical company submitted the full data.
2.3 Data abstraction
One reviewer abstracted the following data from included trials: study design, setting, population characteristics (including sex, age, race, diagnosis), eligibility and exclusion criteria, interventions (dose and duration), comparisons, numbers screened, eligible, enrolled, and lost to follow-up, method of outcome ascertainment (e.g., scales used), and results for each outcome. We recorded intention-to-treat results if available and the trial did not report high overall loss to follow-up. In crossover trials, outcomes for the first intervention were recorded if available to minimize potential bias in results due to differential withdrawal prior to crossover. We also wanted to screen out the possibility of a “carryover” effect from the first treatment in studies without a washout period or “rebound” spasticity from withdrawal of the first intervention.
36.
A second reviewer checked all data.2.4 Quality assessment
We assessed the quality of included trials using predefined criteria (detailed methods available on the Web
37.
or from the authors). Randomized, properly blinded clinical trials are considered the highest level of evidence for assessing efficacy.Anonymous. 2003. Methods for drug class reviews for Oregon Health Plan Practitioner-Managed Prescription Drug Plan. Oregon Evidence-based Practice Center, Portland, OR. Available at: http://www.oregonrx.org/OrgrxPDF/Skeletal%20Muscle% 20Relaxants/Revised%20EPC%20Reprt%204-9-03/ App%20C%20OHP%20Methods%202003.pdf.
38.
, 39.
, 40.
Clinical trials that are not randomized or blinded or that have other methodologic flaws are less reliable. These are discussed in our report with references to specific flaws in study design and data analysis.We rated the internal validity of each trial based on methods used for randomization; allocation concealment and blinding; the similarity of compared groups at baseline; maintenance of comparable groups; adequate reporting of dropouts, attrition, crossover, adherence, and contamination; loss to follow-up; and the use of intention-to-treat analysis. External validity of trials was assessed based on: adequate description of the study population, similarity of patients to other populations to whom the intervention would be applied, control group receiving comparable treatment, funding source, and the role of the funder.
Overall quality was assigned based on criteria developed by the US Preventive Services Task Force and the National Health Service Centre for Reviews and Dissemination (UK).
39.
, 40.
Trials with a fatal flaw in one or more categories were rated poor-quality. Trials that met all criteria were rated “good quality.” The remainder was rated fair quality. As the “fair-quality” category is broad, studies with this rating vary in their strengths and weaknesses. The results of some fair-quality studies are unlikely to be valid, while others are probably or likely to be valid. A “poor-quality” trial is not valid. The results are at least as likely to reflect flaws in the study design as they are true differences between the compared drugs.Many of the studies we reviewed were conducted in the 1970s and early 1980s when standards for reporting clinical trial methodology were generally less stringent. Authors of these trials often did not discuss their methods in what would today be considered adequate detail.
41.
In general, not reporting specific areas of methodology (such as randomization, allocation concealment, or blinding technique) was not considered a “fatal flaw,” but did prevent a trial from achieving a “good” rating for that particular criterion.A particular randomized trial might receive two different ratings: one for efficacy and one for adverse events. Appendix D on the Web site shows the criteria we used to rate studies reporting adverse events. These criteria reflect aspects of the study design that are particularly important for assessing adverse event rates. We rated studies as good-quality for adverse event assessment if they adequately met six or more of the seven pre-defined criteria, fair if they met three to five criteria, and poor if they met two or fewer criteria.
After assignment of quality ratings by the initial reviewer, a second reviewer independently assigned a quality rating. Overall quality rating and quality rating scores (for studies on adverse event assessment) were compared between reviewers. If overall quality ratings differed, the two reviewers came to consensus prior to assigning a final quality rating.
2.5 Data synthesis
We constructed evidence tables showing study characteristics, quality ratings, and results for all included studies. To assess the overall strength of evidence for a body of literature about a particular key question, we examined the consistency of study designs, patient populations, interventions, and results. Consistent results from good-quality studies across a broad range of populations suggest a high degree of certainty that the results of the studies were true (that is, the entire body of evidence would be considered “good quality.”) For a body of fair-quality studies, however, consistent results may indicate that similar biases are operating in all the studies. Unvalidated assessment techniques or heterogeneous reporting methods for important outcomes may weaken the overall body of evidence for that particular outcome or make it difficult to accurately estimate the true magnitude of benefit or harm.
3. Results
Searches identified 3,847 citations: 335 from the Evidence-Based Medicine (Cochrane) Library, 1,155 from MEDLINE, 2,314 from EMBASE, and 43 from reference lists. We received no pharmaceutical company submissions. We identified 377 reports of clinical trials and excluded 227 of these (see Appendix B on the Web site for detailed search results). Sixty-seven were excluded because they did not evaluate an included population, 148 were excluded because they did not evaluate an included intervention (skeletal muscle relaxant), seven were excluded because they did not evaluate an included outcome (spasms, pain, strength, functional ability, or adverse events), one was excluded because it was a single-dose study, and four were excluded because they were not English-language. We retrieved 150 reports on clinical trials for more detailed evaluation. After this second review, we excluded 52: 39 because they did not evaluate an included intervention, one because it did not evaluate an included population, one because it did not contain original data, two because they did not evaluate an included outcome, six because of study design (results published in another reviewed trial, not a controlled trial, or no data), and three because they were not in the English language.
Ninety-eight reports presenting data for 101 randomized controlled trials of patients with spasticity (55 trials reported in 54 publications) or musculoskeletal conditions (46 trials reported in 44 publications) provided usable data and were included. We also identified four relevant systematic reviews
41.
, 42.
, 43.
, 44.
and three relevant meta-analyses (not systematic).45.
, 46.
, 47.
In all trials, external validity was difficult to assess. Numbers screened and enrolled were usually not reported, eligibility and exclusion criteria were often poorly specified, and funding sources were often not stated. When exclusion criteria were reported, numbers of patients excluded for each criterion were not reported.3.1 Comparative efficacy: spasticity
3.1.1 Systematic reviews and meta-analyses
Three systematic reviews evaluated skeletal muscle relaxants used to treat patients with spasticity (Table 1). One was a good-quality systematic review
41.
of various anti-spasticity agents, including skeletal muscle relaxants, for treating symptoms of multiple sclerosis (Table 1 and Evidence Table 1). It identified 11 head-to-head and 12 placebo-controlled trials (five trials of baclofen, four dantrolene, and three tizanidine) of included skeletal muscle relaxants. Seven of the head-to-head trials compared tizanidine to baclofen (including one German-language trial, one unpublished trial and one abstract that were not included in our search). Four other trials compared baclofen, dantrolene, or tizanidine to diazepam. No evaluated trial was rated good quality, and many trials used unvalidated measures of spasticity or muscle strength and inconsistent reporting methods. The authors found no pattern to suggest that one included skeletal muscle relaxant was any better than the others. Meta-analysis was not possible because of marked heterogeneity in study designs, interventions used, and outcomes measured.Table 1Overview of Included Systematic Reviews on Skeletal Muscle Relaxants
| Author/Year | Purpose of Study | Skeletal Muscle Relaxants Evaluated | Number of Included Studies and Patients | Quality | Main Findings |
|---|---|---|---|---|---|
| Systematic Reviews | |||||
| Browning 2001 42. | Assess the effectiveness of cyclobenzaprine in low back pain. | Cyclobenzaprine | 14 trials 3315 patients on cyclobenzaprine | Good. | Included studies of generally fair quality. Cyclobenzaprine moderately effective in improving symptoms compared to placebo. No information on comparative efficacy and safety. |
| Shakespeare 2001 41. | Assess the comparative effectiveness and tolerability of anti-spasticity agents in multiple sclerosis patients. | Tizanidine Baclofen Dantrolene Diazepam | 36 trials (7 tizanidine vs. baclofen, 2 tizanidine vs. diazepam, 1 baclofen vs. diazepam, 1 dantrolene vs. diazepam) 1359 patients overall | Good. | Included studies of fair or poor quality. Tizanidine more effective than baclofen for muscle strength in 2 out of 7 head-to-head trials, otherwise no significant differences in efficacy. No differences in efficacy between tizanidine, baclofen, and dantrolene compared to diazepam; diazepam associated with more sedation and less preferred. |
| Taricco 2000 44. | Assess the effectiveness and safety of drugs for spasticity in spinal cord injury patients. | Tizanidine Baclofen | 9 trials (2 baclofen vs. placebo, 1 tizanidine vs. placebo) 218 patients overall | Fair. Some identified studies not assessed. | Included studies of fair or poor quality. Tizanidine more effective than placebo for Ashworth score but not for functional status. No difference between baclofen and placebo. |
| Lataste 1994 43. | Assess the comparative efficacy of tizanidine compared to other anti-spastic agents. | Tizanidine Baclofen Diazepam | 20 trials (14 vs. baclofen, 6 vs. diazepam) 385 patients on tizanidine, 392 on baclofen or diazepam | Poor. Methods of search not reported, study quality not assessed, insufficient detail of included studies. | Unable to assess quality of included studies. No significant differences between tizanidine and baclofen or diazepam for muscle tone, muscle spasms, clonus, muscle strength, functional status, or overall antispastic effect. Tizanidine slightly better tolerated than diazepam and baclofen. Withdrawals due to adverse events 4% on tizanidine vs. 9% on baclofen or diazepam. |
| Meta-analyses | |||||
| Groves 1998 46. | Assess the efficacy and tolerability of tizanidine using unpublished trials held by the manufacturer. | Tizanidine Baclofen Diazepam | 10 trials (7 vs. baclofen, 3 vs. diazepam) 270 patients overall | Fair. Insufficient detail of included studies and not clear if data combined appropriately. | No significant differences between tizanidine and baclofen or diazepam for spasticity by Ashworth score or mean change in muscle strength. “Global tolerability to treatment” favored tizanidine compared to baclofen (P=0.008) and diazepam (P=0.001). |
| Wallace 1994 45. | Assess the efficacy and tolerability of tizanidine using unpublished trials held by the manufacturer. | Tizanidine Baclofen Diazepam | 3 placebo-controlled trials with 525 patients 11 head-to-head studies (8 vs. baclofen, 3 vs. diazepam) with 270 patients | Fair. Insufficient detail of included studies and not clear if data combined appropriately. | See results for Groves 1998 for results of head-to-head studies. In placebo-controlled studies, there were increased withdrawals due to adverse events (44/284 vs. 15/277) on tizanidine. Frequent adverse events on tizanidine were dry mouth (49%), somnolence (48%), asthenia (41%), dizziness (16%), headache (12%). |
| Nibbelink 1978 47. | Assess the efficacy of cyclobenzaprine using unpublished trials. | Cyclobenzaprine Diazepam Placebo | 20 randomized trials 434 patients on cyclobenzaprine, 280 on diazepam, 439 on placebo | Fair. Insufficient detail of included studies and not clear if data combined appropriately. | ‘Global response’ equivalent for cyclobenzaprine and diazepam and significantly better than placebo. Muscle spasms, tenderness on palpation, limitation of motion, and limitation of daily living (but not local pain) significantly better in patients on cyclobenzaprine compared to diazepam at Week 2 using unvalidated methods. |
a Comparator
One systematic review evaluated pharmacologic interventions for spasticity following spinal cord injury.
44.
It was rated fair quality because the authors had not yet assessed 15 identified potentially relevant studies. Of the nine studies included, two were placebo-controlled trials evaluating baclofen or tizanidine. There were no head-to-head trials, and no study was rated good quality. There was insufficient evidence to judge the comparative efficacy of tizanidine versus baclofen.One systematic review
43.
evaluated 20 studies of tizanidine versus baclofen (14 studies) or diazepam (6 studies) in patients with spasticity. This systematic review included both published and unpublished trials and was rated poor quality (see Table 1). Although this systematic review found some evidence of increased effectiveness of tizanidine compared to baclofen and diazepam, it is not possible to determine whether these conclusions are valid.Two fair-quality meta-analyses (not systematic reviews) evaluated unpublished trials on tizanidine versus baclofen or diazepam.
45.
, 46.
Authors of these trials were employed by Athena Neurosciences (San Francisco, CA), a pharmaceutical company marketing tizanidine in the U.S., and analyzed the same trials (ten trials in one meta-analysis46.
and eleven in the other45.
). Both studies found no significant differences between tizanidine compared to diazepam orbaclofen for outcomes of tone (Ashworth scale) or muscle strength (summed BMRC strength scores).3.1.2 Head-to-head trials
Of 55 trials evaluating included skeletal muscle relaxants in patients with spasticity, 17 (total enrolled = 654) were head-to-head trials of two skeletal muscle relaxants or a skeletal muscle relaxant versus another medication used to treat spasticity (Table 2). The majority (10) of the trials focused on patients with multiple sclerosis, but other clinical conditions (children with cerebral palsy,
48.
post-stroke or head trauma,49.
spinal cord injury,50.
and spasticity from various causes51.
, 52.
, 53.
, 54.
) were also evaluated. Except for one study lasting one year,51.
all of the head-to-head trials were of relatively short duration, ranging from 2 to 8 weeks per intervention. All of the trials except one50.
were published before 1990. Although elderly patients were included in most trials, no trial specifically evaluated only elderly patients. One trial included only children.48.
Table 2Overview of Head-to-Head Trials of Skeletal Muscle Relaxants for Spasticity
| Interventions/Dose | Study/Year/Quality | Population/Number Enrolled | Main Outcomes Assessed | Main Results | Withdrawals (overall) |
|---|---|---|---|---|---|
| Tizanidine versus Baclofen | |||||
| Tizanidine mean 17 mg/day Baclofen mean 35 mg/day | Bass 1988 60. Fair | Multiple sclerosis 66 | Spasticity: 6-point scale Strength: 6-point scale Functional status: Kurtzke functional scale Disability: Pedersen functional disability scale Preference: patient assessment | No significant differences between interventions for main outcomes. | 11% (5/46) 28% (13/46) |
| Tizanidine titrated to 24 mg/day Baclofen titrated to 60 mg/day | Eyssette 1988 58. Fair | Multiple sclerosis 100 | Spasticity: 5-point scale Stretch reflex: 1–5 scale Functional status: Unspecified methods Efficacy and tolerability: Unspecified methods | No significant differences between interventions. | 16% (8/50) 12% (6/50) |
| Tizanidine 12–24 mg/day Baclofen 15-60 mg/day | Hoogstraten 1988 57. Fair | Multiple sclerosis 16 | Spasticity: Ashworth scale and patient self-report (5-point scale) Disability: Kurtzke Expanded Disability Status Scale Functional status: Kurtzke Functional Systems Incapacity status: Minimal record of disability for multiple sclerosis Ambulation: Ambulation index Clonus and reflexes: Unspecified methods Muscle strength and pain: 5-point scales Efficacy and tolerance: −3 to +3 scales | No significant differences between interventions (Ashworth scale scores not reported). | 6% (1/16) 25% (4/16) |
| Tizanidine mean 20 mg/day Baclofen mean 50 mg/day | Medici 1989 51. Fair | Spasticity due to various causes 30 | Spasticity: Ashworth scale and patient self-report (4-point scale) Muscle strength: 5-point scale Clonus: 3-point scale Functional status: Kurtzke Expanded Disability Status Scale Global assessments: Unspecified methods | No significant differences between interventions (Ashworth scale scores not reported). | 7% (1/15) 27% (4/15) |
| Tizanidine titrated to 16 mg/day Baclofen titrated to 40 mg/day | Newman 1982 59. Fair | Multiple sclerosis (32) or syringomyelia (4) 36 | Spasticity: Ashworth scale Functional status: Kurtzke and Pedersen scales | No significant differences between interventions (Ashworth scale scores not reported). | 11% (4/36) 17% (6/36) |
| Tizanidine mean 11 mg/day Baclofen mean 51 mg/day | Rinne 1980 (2) 53. Fair | Multiple sclerosis (24) or cervical myelopathy (8) 32 | Spasticity: Ashworth scale | No significant differences between interventions (Ashworth scale scores not reported). | 6% (1/16) 6% (1/16) |
| Tizanidine 8 mg tid Baclofen 20 mg tid | Smolenski 1981 56. Fair | Multiple sclerosis 21 | Tone: Ashworth scale Spasticity: 5-point scale Muscle strength: 6-point scale Global assessment of change in condition: Unspecified methods Tolerance to medication: Unspecified methods | No significant differences between interventions (Ashworth scale scores not reported). | None reported |
| Tizanidine mean 23 mg/day Baclofen mean 59 mg/day | Stien 1987 36. Fair | Multiple sclerosis 40 | Tone/spasticity: Ashworth scale Functional status: Kurtzke Expanded Disability Status Scale Functional assessment: Pederson scale | No significant differences between interventions (Ashworth scale scores not reported). | 6% (1/18) 5% (1/20) |
| Tizanidine, Baclofen, or Dantrolene versus Diazepam | |||||
| Tizanidine mean 17 mg/day Diazepam mean 20 mg/day | Bes 1988 49. Fair | Post-stroke or head-trauma 105 | Spasticity: 5-point scale Functional status: walking distance Severity of spasms: 5-point scale Muscle strength: Unspecified methods Clonus: Unspecified methods | No significant differences between interventions. | 12% (6/51) 31% (17/54) |
| Tizanidine mean 14 mg/day Diazepam mean 15 mg/day | Rinne 1980 (1) 53. Fair | Multiple sclerosis 30 | Spasticity: Ashworth scale | No significant differences between interventions (Ashworth scale scores not reported). | 0% (0/15) 27% (4/15) |
| Baclofen 30 mg/day and 60 mg/day Diazepam 15 mg/day and 30 mg/day | Cartlidge 1974 63. Fair | Multiple sclerosis 40 | Spasticity: Ashworth scale | No significant differences between interventions (mean Ashworth score improvement 0.227 vs. 0.202 on high-doses). | Not clear |
| Baclofen mean 61 mg/day Diazepam mean 27 mg/day | From 1975 61. Fair | Multiple sclerosis inpatients 16 | Spasticity: Ashworth scale, clinical exam (unspecified methods) Clinical assesments of spasms, clonus, bladder function, walking: Unspecified methods Patient preference | No significant differences between interventions (Ashworth scale scores not reported). | 6% (1/16) 0% (0/16) |
| Baclofen mean 47 mg/day Diazepam 28 mg/day | Roussan 1985 52. Fair | Spasticity due to various causes 13 | Global response to treatment: 0 (no improvement) to 3+ (marked improvement) | No significant differences between interventions. | None reported |
| Dantrolene 100 mg qid Diazepam 5 mg qid | Glass 1974 54. Fair | Spasticity due to various causes 16 | Spasticity/tone: 6-point scale Reflexes: 6-point scale Clonus: 6-point scale Strength: 6-point scale | No significant differences between interventions. | 19% (3/16) 6% (1/16) |
| Dantrolene titrated to 75 mg qid Diazepam titrated to 12 mg/day | Nogen 1976 48. Fair | Children with cerebral palsy 22 | Tone: Unspecified method Tendon jerk: Unspecified method Clonus: Unspecified method Strength: Unspecified method Overall evaluation: Unspecified method | No significant differences between interventions. | None reported |
| Dantrolene titrated to 75 mg qid Diazepam titrated to 5 mg qid | Schmidt 1976 62. Fair | Multiple sclerosis 46 | Spasticity: 6-point scale Clonus: 6-point scale Reflexes: 6-point scale Functional status: Methods not specified, derived from ACTH cooperative study | No significant differences between interventions for spasticity or clonus. Reflexes, station stability, and hand coordination favor dantrolene. | Not clear |
None of the 17 head-to-head trials was rated good quality. All studies had at least two of the following methodological flaws: randomization technique not described, eligibility criteria not described, blinding technique not described, allocation concealment technique not described, or high loss to follow-up (Evidence Table 3). Adequate blinding is an especially important factor in studies using subjective outcomes, such as patient preference, global assessments, spasm severity, or pain. One trial comparing baclofen to clonidine that found no differences for spasticity was rated poor quality because it was not randomized and did not perform blinding, and was excluded from the tables.
55.
The remainder were rated fair quality. Possible confounding factors in these trials included different methods of medication titration or target doses, differential withdrawals during the first intervention period in crossover trials, and previous use of an intervention or other muscle relaxant, which was inconsistently reported. In crossover trials, results of the first intervention were usually not reported.Table 3Overview of Placebo-Controlled Trials of Included Skeletal Muscle Relaxants for Spasticity
| Medication | Trial/Quality | Population/Number Enrolled | Main Outcomes for Spasticity/Tone |
|---|---|---|---|
| Baclofen | Basmajian 1974 4. Fair | Various spasticity 15 | Favors baclofen based on “EMG and force recordings” (P not reported). |
| Baclofen | Basmajian 1975 65. Fair | Various spasticity 14 | Favors baclofen using unspecified method (P not reported). |
| Baclofen | Brar 1991 66. Fair | Multiple sclerosis 38 | Favors baclofen using Ashworth scale (P not reported). |
| Baclofen | Duncan 1976 67. Poor | M.S. or spinal cord lesions 25 | Baclofen superior using 5-point scale (P<0.01). |
| Baclofen | Feldman 1978 68. Fair | Multiple sclerosis 33 | Baclofen superior using unspecified method (P not reported). |
| Baclofen | Hinderer 1990 69. Poor | Spinal cord lesions 5 | No improvement on baclofen using unspecified method. |
| Baclofen | Hulme 1985 70. Fair | Post-stroke (elderly patients) 12 | Not assessed; study stopped due to excess adverse events (somnolence). |
| Baclofen | Jones 1970 71. Fair | Spinal cord injury 6 | Favors baclofen using 5-point scale for spasm and spasm counts (P not reported). |
| Baclofen | McKinlay 1980 72. Fair | Children with spasticity (criteria not specified) 20 | No significant difference using Ashworth scale. |
| Baclofen | Medaer 1991 73. Fair | Post-stroke 20 | Baclofen superior using Ashworth scale (P<0.001). |
| Baclofen | Milla 1977 74. Fair | Various spasticity (children) 20 | Baclofen superior using Ashworth scale (P<0.001). |
| Baclofen | Orsnes 2000 75. Fair | Multiple sclerosis 14 | No significant difference using Ashworth scale. |
| Baclofen | Sachais 1977 76. Fair | Multiple sclerosis 166 | Baclofen superior using unspecified method (P<0.01). |
| Baclofen | Sawa 1979 77. Fair | Multiple sclerosis 21 | Baclofen superior using 6-point scale (P<0.001). |
| Dantrolene | Basmajian 1973 78. Poor | Upper motor neuron disease 25 | Spasticity not assessed. |
| Dantrolene | Chyatte 1973 79. Fair | Athetoid cerebral palsy (children) 18 | No measurable difference using 4-point scale. |
| Dantrolene | Denhoff 1975 80. Fair | Various spasticity (children) 18 | Dantrolene superior for “neurologic measurements” using unspecified methods (P<0.04). |
| Dantrolene | Gambi 1983 81. Fair | Multiple sclerosis or myelopathy 24 | Dantrolene superior using 6-point scale (P<0.05, raw data not reported). |
| Dantrolene | Gelenberg 1973 82. Poor | Multiple sclerosis 20 | Spasticity assessed using unspecified method; outcomes not reported. |
| Dantrolene | Glass 1974 54. Fair | Various spasticity 16 | Favors dantrolene for resistance to active stretch and tendon jerk using 6-point scales (P not reported). |
| Dantrolene | Haslam 1974 83. Fair | Perinatal brain injury (children) 26 | No statistical difference using 5-point scale. |
| Dantrolene | Joynt 1980 84. Fair | Cerebral palsy (children) 21 | No statistical difference using 4-point scale. |
| Dantrolene | Katrak 1992 85. Fair | Post-stroke 38 | No measurable difference using 0–6 motor assessment scale. |
| Dantrolene | Ketel 1984 86. Poor | Post-stroke 18 | Favors dantrolene, assessment method not reported. |
| Dantrolene | Luisto 1982 87. Fair | Various spasticity 17 | Dantrolene superior using Ashworth scale (P=0.05). |
| Dantrolene | Monster 1974 88. Fair | Various spasticity 200 | Outcomes not clear, results for placebo not reported. |
| Dantrolene | Nogen 1979 89. Fair | Children with spasticity and epilepsy 21 | No increased seizures on dantrolene; other outcomes not reported. |
| Dantrolene | Sheplan 1975 90. Fair | Various spasticity (all men) 18 | Outcomes not clear (unspecified methods), results for placebo not reported. |
| Dantrolene | Tolosa 1975 91. Fair | Multiple sclerosis 23 | Favors dantrolene using 7-point scale (P not reported). |
| Dantrolene | Weiser 1978 92. Fair | Spinal cord disease 35 | Dantrolene superior for spasms using unspecified scale (P<0.002); no differences for walking/staircase time. |
| Tizanidine | Knutsson 1982 93. Fair | Various spasticity 13 | No significant difference using Ashworth scale. |
| Tizanidine | Lapierre 1987 94. Fair | Multiple sclerosis 66 | No significant difference using unspecified method. |
| Tizanidine | Meythaler 2001 95. Fair | Various spasticity 17 | No significant difference using Penn Spasm Frequency Scale, favors tizanidine using Ashworth scale (P=0.006). |
| Tizanidine | Nance 1994 50. Fair | Spinal cord injury 124 | Tizanidine superior using Ashworth scale (P<0.0001) and pendulum test (P=0.004); no difference in daily spasm frequency. |
| Tizanidine | Smith 1994 96. Fair | Multiple sclerosis 220 | No significant difference using Ashworth scale, 4-point scale, or daily counts. |
| Tizanidine | UK Tizanidine Trial Group 1994 97. Fair | Multiple sclerosis 187 | Tizanidine superior using Ashworth scale (P=0.004). |
| Chlorzoxazone | Losin 1966 98. Poor | Various spasticity (children) 30 | Outcomes not clear using 5-point scale. |
| Cyclobenzaprine | Ashby 1972 100. Fair | Various spasticity 15 | No significant difference using 5-point scale. |
| Methocarbamol | Bjerre 1971 99. Poor | Cerebral palsy (children) 44 | No significant difference for overall condition using 3-point scale, methocarbamol superior for motor function (P<0.01) using Johnson scale for lower extremities but no significant difference for upper extremities. |
In eight trials of tizanidine vs. baclofen, the average dose of tizanidine ranged from 11 mg/day
53.
to 24 mg/day56.
, 57.
, 58.
and the dose of baclofen ranged from 15 mg/day57.
to 90 mg/day.58.
Most of these trials evaluated patients with multiple sclerosis. In each of these eight trials, tizanidine and baclofen appeared to have roughly equivalent efficacy (Table 2). Outcomes measured included muscle tone, muscle spasm, clonus, functional assessments, patient or physician global assessments, and patient or physician preference. These outcomes were assessed using a variety of methods, including unvalidated or unspecified scales. Six trials36.
, 51.
, 53.
, 56.
, 57.
, 59.
used the Ashworth scale to measure spasticity or tone, but methods of reporting these results were inconsistent and raw scores were usually not presented. In most trials, regardless of the method used to assess outcomes, patients receiving either baclofen or tizanidine reported significant improvements compared to baseline. The longest trial (52 weeks compared to 8 weeks or less for the other trials) reported results similar to shorter trials.51.
The overall withdrawal rate was higher with baclofen than with tizanidine in three out of seven trials51.
, 57.
, 60.
and roughly equivalent in the other four. Of the three trials with differential withdrawal rates, two had low numbers of overall withdrawals (five in each trial), making the significance of these differential rates difficult to assess. In two of the trials,51.
, 60.
withdrawals due to adverse events accounted for most of the observed differences in overall withdrawal rates (see section on adverse events).There were no trials directly comparing dantrolene to baclofen or tizanidine. In the eight trials
48.
, 49.
, 52.
, 53.
, 54.
, 61.
, 62.
, 63.
of tizanidine, baclofen, or dantrolene versus diazepam, there was no pattern to suggest that any of these skeletal muscle relaxants was superior to the others for assessed clinical outcomes including spasm, strength, functional status, or patient preference (Table 2 and Evidence Table 3). Differences in study design, patient populations, outcomes evaluated, and roughly similar efficacy of each skeletal muscle relaxant compared to diazepam in individual trials made it impossible to make accurate judgments about the comparative efficacy of tizanidine, baclofen, and dantrolene from these trials as a whole.3.1.3 Placebo-controlled trials
In addition to one head-to-head trial
54.
that also included a placebo arm, we identified an additional 38 additional placebo-controlled trials (Table 3). Fourteen evaluated baclofen,64.
, 65.
, 66.
, 67.
, 68.
, 69.
, 70.
, 71.
, 72.
, 73.
, 74.
, 75.
, 76.
, 77.
15 dantrolene,78.
, 79.
, 80.
, 81.
, 82.
, 83.
, 84.
, 85.
, 86.
, 87.
, 88.
, 89.
, 90.
, 91.
, 92.
six tizanidine,55.
, 93.
, 94.
, 95.
, 96.
, 97.
one chlorzoxazone,98.
one methocarbamol,99.
and one cyclobenzaprine.100.
Conditions evaluated in these studies were multiple sclerosis, cervical myelopathy, cerebral palsy, post-stroke, traumatic brain injury, spinal cord injury, and spasticity from various causes. Nine placebo-controlled trials evaluated children72.
, 74.
, 79.
, 80.
, 83.
, 84.
, 89.
, 98.
, 99.
and one specifically evaluated elderly post-stroke patients.70.
We identified no placebo-controlled trials of carisoprodol, metaxalone, or orphenadrine in patients with spasticity.None of the placebo-controlled trials was rated good quality (Evidence Table 4). Main results from placebo-controlled trials for spasticity are summarized in Table 3. Most of the placebo-controlled trials found either significant benefits or trends towards benefit from baclofen, dantrolene, and tizanidine compared to placebo for spasticity, functional ability, and strength. However, because of the use of unvalidated outcomes scales and inconsistent methods for reporting outcomes, the relative magnitude of benefit for each of these medications could not be compared across studies. There was inadequate evidence from one trial
98.
of chlorzoxazone (rated poor quality), one trial100.
of cyclobenzaprine (no significant differences), and one trial99.
of methocarbamol in children with cerebral palsy (rated poor quality) to show that these skeletal muscle relaxants are effective for treatment of spasticity. These three medications are not approved for this indication.Table 4Overview of Head-to-Head Trials of Skeletal Muscle Relaxants for Musculoskeletal Conditions
| Interventions/Dose | Study/Year | Population/Number Enrolled | Main Outcomes Assessed | Main Results | Overall Withdrawals |
|---|---|---|---|---|---|
| Tizanidine versus Chlorzoxazone | |||||
| Tizanidine 2 mg tid Chlorzoxazone 500 mg tid | Bragstad 1979 103. Fair | Back spasms 120 | Muscle tension: 4-point scale Pain intensity: 4-point scale Tenderness: 4-point scale Interference with normal activities: 4-point scale | No significant differences between interventions. | 0% (0/14) 8% (1/13) |
| Cyclobenzaprine versus Methocarbamol | |||||
| Cyclobenzaprine 10 mg tid Methocarbamol 1500 mg qid | Preston 1984 18. Fair | Localized acute muscle spasm 227 | Muscle spasm: 9-point scale Local pain and tenderness: 9-point scale Limitation of normal motion: 9-point scale Interference with normal activities: 9-point scale | No significant differences between interventions except slightly greater proportion of patients with improvement in local pain with cyclobenzaprine (48% vs. 40%). | 14% (12/87) 13% (12/94) |
| Cyclobenzaprine versus Carisprodol | |||||
| Cyclobenzaprine 10 mg qid Carisoprodol 350 mg qid | Rollings 1983 104. Fair | Back spasms 78 | Pain severity: 1–5 verbal rating scale and 0–100 visual analogue scale Muscle stiffness: VRS and VAS Activity impairment: VRS and VAS Sleep impairment: VRS and VAS Muscle tension: VRS and VAS | No significant differences between interventions. | 24% (9/37) 28% (11/39) |
| Carisoprodol, Cyclobenzaprine or Tizandine versus Diazepam | |||||
| Carisoprodol 350 mg qid Diazepam 5 mg qid | Boyles 1983 105. Fair | Acute back sprain or strain with spasms 80 | Muscle spasm: 5-point scale Tenderness: 5-point scale Mobility restriction: 5-point scale Pain, stiffness, activity, sleep impairment, tension: 5-point scales | Carisoprodol superior to diazpeam for muscle stiffness (P<0.05), tension (P<0.05), and relief (P<0.05) using 5-point scales; trend towards better overall relief (68% vs. 45%) with carisoprodol. | 10% (4/40) 12% (5/40) |
| Cyclobenzaprine 10–20 mg tid Diazepam 5–10 mg tid | Aiken 1978a 107. Fair | Acute back or neck spasms 117 | Muscle spasm: 5-point scale Limitation of motion: 5-point scale Daily activities: 5-point scale Pain: 5-point scale Tenderness: 5-point scale Global response: 5-point scale (worse to marked improvement) | Cyclobenzaprine more effective than diazepam for muscle spasm, tenderness, limitation of motion at Week 1 (P<0.05) and for pain, tenderness, limitation of motion, and global response at Week 2 (P<0.05). | 13% (5/38) 15% (6/40) |
| Cyclobenzaprine 10–20 mg tid Diazepam 5 mg tid | Basmajian 1978 102. Poor | Back or neck spasms 120 | Muscle spasm: 5-point scale | No significant differences between interventions. | Not reported |
| Cyclobenzaprine 10 mg tid Diazepam 5 mg tid | Brown 1978 101. Fair | Back or neck spasms 49 | Global evaluation: 5-point scale | No significant differences between interventions. | None reported |
| Cyclobenzaprine 30–40 mg tid Diazepam 15–20 mg/day | Scheiner 1978 (1) 106. Fair | Acute back or neck spasms 96 | Muscle spasm: 5-point scale Pain: 5-point scale Tenderness: 5-point scale Limitation of motion: 5-point scale Daily activities: 5-point scale Global evaluation: 5-point scale (worse to marked improvement) | No significant differences between interventions except cyclobenzaprine more effective for tenderness at Week 2 (P<0.05), limitation of motion at Weeks 1 and 2 (P<0.01), and global evaluation (marked improvement) (P<0.01). | 35% (12/34) 9% (3/32) |
| Cyclobenzaprine 30–40 mg tid Diazepam 15–20 mg/day | Scheiner 1978 (2) 106. Fair | Acute back or neck spasms 75 | Muscle spasm: 5-point scale Pain: 5-point scale Tenderness: 5-point scale Limitation of motion: 5-point scale Daily activities: 5-point scale Global evaluation: 5-point scale (worse to marked improvement) | Cyclobenzaprine more effective than diazepam (P<0.05) for all outcomes at Weeks 1 and 2 except for muscle spasm and limitation of motion at Week 1. | 8% (2/26) 21% (5/24) |
| Tizanidine 4–8 mg tid Diazepam 5–10 mg tid | Fryda-Kaurimsky 1981 108. Fair | Degenerative spinal disease with acute muscle spasm (inpatients) 20 | Pain: 4-point scale Tenderness: 4-point scale Muscle spasm: 3-point scale Abnormal posture: 3-point scale Daily activities: 4-point scale | No significant differences between interventions. | None reported |
| Tizanidine 4 mg tid Diazepam 5 mg tid | Hennies 1981 109. Fair | Back or neck spasms 30 | Pain: 4-point scale Muscle tension: Unspecified method Daily living activity: Unspecified method | No significant differences between interventions. | 7% (1/15) 0% (1/15) |
Meta-analysis could not be performed on the placebo-controlled trials because of marked differences in interventions (doses used and methods of titration), trial designs, populations studied, outcomes scales, and methods for reporting outcomes. No reliable conclusions about comparative efficacy can be drawn from these placebo-controlled trials.
3.2 Comparative efficacy: musculoskeletal conditions
3.2.1 Systematic reviews and meta-analyses
We identified no systematic reviews comparing different skeletal muscle relaxants in patients with musculoskeletal conditions.
One good-quality systematic review evaluated the efficacy of cyclobenzaprine versus placebo for treatment of back pain (Table 1 and Evidence Table 2).
42.
This systematic review examined 14 trials of fair overall quality and found that cyclobenzaprine was associated with better ‘global improvement’ scores at Day 14 (odds ratio 4.7; 95% confidence interval (CI), 2.7–8.1). For individual symptoms, the systematic review found a modest magnitude of improvement (effect size 0.38–0.58) compared to placebo by Day 14 for five outcomes: local pain, muscle spasm, tenderness to palpation, range of motion, and activities of daily living. Information regarding other skeletal muscle relaxants evaluated in included trials was specifically excluded from analysis in this systematic review.One fair-quality meta-analysis evaluated the comparative efficacy of cyclobenzaprine, diazepam and placebo.
47.
This study summarized results of 20 unpublished short-term (2-week) trials performed in the U.S. in 1153 patients with muscle spasm; the authors were employed by Merck Laboratories. It included patients with post-traumatic injury, musculoskeletal strain, radiculopathy, and osteoarthritis. This study found that the unvalidated outcome measure ‘global response’ was equivalent for cyclobenzaprine and diazepam (66% marked or moderate improvement) and significantly better than placebo (40%).3.2.2 Head-to-head trials
Of 46 trials of included skeletal muscle relaxants in patients with musculoskeletal conditions, 11 (total enrolled = 724) were head-to-head trials (Table 4). All of the head-to-head trials focused on patients with back or neck pain and spasms. One trial
101.
focused on patients with chronic symptoms and the remainder evaluated patients with acute symptoms. The duration of all head-to-head trials ranged from seven18.
to 18102.
days. All of the trials were published before 1985. Although elderly patients were included in most trials, no trial specifically evaluated only elderly patients and none included children.None of the 11 head-to-head trials was rated good-quality; all had at least two important methodological flaws (Evidence Table 5). All trials were rated fair except one trial of cyclobenzaprine versus diazepam that was rated poor because in addition to other flaws, it only reported results for 52 of the 105 enrollees and did not account for the other patients.
102.
A variety of methods was used for measuring outcomes, including various scales for pain (4-, 5-, or 9- point scales and visual analogue scales), tenderness, and functional status. Most assessment scales were unvalidated, and methods of reporting these outcomes were inconsistent. Functional status was either not measured or assessed using unstandardized and unvalidated methods. Doses of medications varied between trials.Table 5Overview of Placebo-Controlled Trials of Skeletal Muscle Relaxants for Musculoskeletal Conditions
| Medication | Trials | Population/Number Enrolled | Main Outcomes (Included Skeletal Muscle Relaxant versus Placebo) |
|---|---|---|---|
| Carisoprodol | Baratta 1976 121. Fair | Low back syndrome 105 | No significant difference for pain using 4-point scale, carisoprodol superior to placebo for various functional measurements and for sleep. |
| Carisoprodol | Cullen 1976 122. Fair | Acute back or neck syndrome 65 | Carisoprodol superior for pain, spasm, and limitation of movement using unspecified methods (all P<0.01). |
| Carisoprodol | Hindle 1972 123. Fair | Low back syndrome (Mexican migrant workers) 48 | Carisoprodol superior for pain, spasm, functional assessments using 4-point scales (all P<0.01) and pain intensity using 0–100 visual analogue scale (P<0.01). |
| Carisoprodol | Soyka 1979 124. Fair | Acute neck or low back syndrome 414 | Favors carisoprodol for muscle spasm (P=0.015) and functional assessment (P=0.04) using 5-point scales, no significant difference for sleep impairment using 4-point scale or pain using 5-point scale. |
| Cyclobenzaprine | Aiken 1978a 125. Fair | Acute neck or low back syndrome 117 (including diazepam arm) | Cyclobenzaprine superior to placebo for pain, tenderness, limitation of motion, daily activities, and global evaluation (all P<0.05) at end of Week 2 using 5-point scales. |
| Cyclobenzaprine | Aiken 1978b 125. Fair | Acute neck or low back syndrome 50 | Cyclobenzaprine superior to placebo for spasm, limitation of motion, daily activities (all P<0.01); pain/tenderness (P<0.05); and global evaluation (P not reported) using 5-point scales. |
| Cyclobenzaprine | Baratta 1982 126. Fair | Various acute muscle spasm 120 | Cyclobenzaprine superior for local muscle spasm (P<0.01) and pain (P<0.01) using 5-point scale. |
| Cyclobenzaprine | Basmajian 1978 102. Fair | Various acute muscle spasm 120 (including diazepam arm) | No significant differences for task performance time or muscle spasms using 5-point scale. |
| Cyclobenzaprine | Basmajian 1989 127. Fair | Various acute muscle spasm 175 | No significant differences for pain, muscle spasm, or functional measurements using unspecified methods. |
| Cyclobenzaprine | Bennett 1988 114. Fair | Fibromyalgia 120 | Cyclobenzaprine superior for pain (P<0.02) using 1–10 visual analogue scale and sleep quality and fatigue using 5-point scale (P<0.02). |
| Cyclobenzaprine | Bercel 1977 128. Fair | Neck or back pain >30 days 54 | Favors cyclobenzaprine for spasm duration using 5-point scale (P not reported). |
| Cyclobenzaprine | Bianchi 1978 129. Fair | Acute neck or low back syndrome 48 | No significant differences at Day 14; cyclobenzaprine superior to placebo for muscle consistency, tenderness, limitation of motion, and global evaluation (all P<0.01) and daily activities (P<0.05) at Day 7. |
| Cyclobenzaprine (+naprosyn in both arms) | Borenstein 1990 110. Poor | Acute low back syndrome 40 | Cyclobenzaprine + naprosyn superior to naprosyn alone for functional capacity using 4-point scale (P<0.05) and muscle spasm using 4 point scale (P<0.05), no difference for resolution of pain (using 0–20 and 4-point scales). |
| Cyclobenzaprine | Brown 1978 101. Fair | Chronic (>12 months) neck or low back pain | Cyclobenzaprine superior to placebo for global evaluation using 5-point scale (P not reported). |
| Cyclobenzaprine | Carette 1994 115. Fair | Fibromyalgia 208 | No significant difference for 6-month improvement using 0–10 visual analogue scale, pain using McGill Pain Questionnaire, functional disability, or psychological status. |
| Cyclobenzaprine | Lance 1972 117. Poor | Chronic tension headache 20 | Favors cyclobenzaprine using 3-point scale (P not reported). |
| Cyclobenzaprine | Preston 1984 18. Fair | Acute local muscle spasm 227 (includes methocarbamol arm) | No differences for muscle spasm or limitation of motion; favors cyclobenzaprine for local pain and daily activities (P not reported) using 9-point scales. |
| Cyclobenzaprine | Quimby 1989 130. Fair | Fibromyalgia 40 | Favors cyclobenzaprine using 5-point scale for patient-rated stiffness and aching, patient-rated poor sleep, and overall patient rating (P<0.05), no difference using 5-point scale for patient rated fatigue or muscle pain. |
| Cyclobenzaprine | Reynolds 1991 113. Fair | Fibromyalgia 12 | No differences for tender point severity count using 5-point scale, pain using 7-point scale, fatigue using 7-point scale, sleepiness using Stanford Sleepiness Rating Scale. |
| Cyclobenzaprine | Scheiner 1978 (1) 106. Fair | Acute back or neck spasm 96 | Cyclobenzaprine superior to placebo for muscle spasm, local pain, tenderness, limitation of motion, daily activities, and global evaluation (P<0.01) using 5-point scales. |
| Cyclobenzaprine | Scheiner 1978 (2) 106. Fair | Acute back or neck spasm 75 | Cyclobenzaprine superior to placebo for muscle spasm, local pain, tenderness, limitation of motion, daily activities, and global evaluation (P<0.01) using 5-point scales. |
| Cyclobenzaprine | Steingard 1980 131. Fair | Back or neck spasm 121 | No significant differences for global evaluation, pain, muscle spasm, or functional measurements using unspecified methods. |
| Metaxalone | Dent 1975 133. Poor | Acute skeletal muscle disorders (not specified) 228 | Metaxolone superior for muscle spasm, local pain, limitation of normal motion, and interference with daily activities using unspecified scales. |
| Metaxalone | Diamond 1966 135. Fair | Muscle pain and spasm, unspecified locations 100 | No significant difference using 5-point scale for muscle spasm or 4-point scale for pain. |
| Metaxalone | Fathie 1964 (1) 134. Fair | Low back pain 100 | Metaxolone superior for global therapeutic response using 4-point scale, range of motion using 5-point scale, and palpable spasm using 5-point scale. |
| Metaxalone | Fathie 1964 (2) 134. Fair | Low back pain 100 | Metaxolone superior for global therapeutic response using 4-point scale, range of motion using 5-point scale, and palpable spasm using 5-point scale. |
| Methocarbamol | Preston 1984 18. Fair | Acute local muscle spasm 227 (includes cyclobenzaprine arm) | No differences for muscle spasm; favors cyclobenzaprine for local pain, limitation of motion, and daily activities (P not reported) using 9-point scales. |
| Methocarbamol | Tisdale 1975 141. Fair | Acute local muscle spasm 180 | Methocarbamol superior for muscle spasm and local pain at 48 hours using 5-point scales; methocarbamol superior for limitation of motion and daily activities at 1 Week (P<0.05) but not for local pain (P<0.10) or muscle spasm (NS) using 5-point scales. |
| Orphenadrine | Gold 1978 21. Poor | Acute low back syndrome 60 | Orphenadrine superior for pain intensity (P<0.01) and pain relief (P<0.01)using unspecified methods. |
| Orphenadrine | Latta 1989 120. Fair | Nocturnal leg cramps (elderly) 59 | Orphenadrine superior for number of nocturnal leg cramps in one-month period. |
| Orphenadrine (+paracetamol in both arms) | McGuinness 1983 111. Fair | Various musculoskeletal conditions 32 | Favors orphenadrine for pain, stiffness and function using 4-point scales (P not reported). |
| Orphenadrine | Valtonen 1975 132. Fair | Low back or neck pain 200 | No significant difference using 3-point scale for ‘overall effect’. |
| Baclofen | Dapas 1985 140. Fair | Acute back syndrome 200 | Baclofen superior for lumbar pain, tenderness, spasm, functional assessments using unspecifie methods (P<0.05). |
| Dantrolene | Casale 1988 142. Fair | Chronic low back syndrome 20 | Dantrolene superior for muscle spasm using “manual semiotic maneuvers” (P<0.001) and pain behavior using visual analogue scale (P<0.001). |
| Dantrolene (+ ibuprofen in both arms) | Salvini 1986 12. Fair | Neck or low back syndromes 60 | Dantrolene superior for muscle contracture using 4-point scale (P=0.04), strength using 5-point scale (P=0.05), no difference for pain on movement using 4-point scale. |
| Tizanidine | Berry 1988a 137. Poor | Acute low back syndrome 105 | Cyclobenzaprine superior for pain on movement (P=0.029), and pain at night (P=0.025) using 4-point scales, no differences for pain at rest or restriction of movement using 4-point scales. |
| Tizanidine | Berry 1988b 136. Fair | Acute low back syndrome 112 | No significant differences for pain at night, pain at rest, or restriction of movement using 4-point scales. |
| Tizanidine | Fogelholm 1992 116. Fair | Tension headache (all women) 45 | Tizanidine superior for headache severity using 0–100 visual analogue (P=0.018) scale and 5-point verbal rating scale (P=0.012) and for analgesic use using pill counts (P=0.001). |
| Tizanidine | Lepisto 1979 138. Fair | Low back syndrome 30 | Tizanidine superior for pain, muscle tension, tenderness using 4-point scales (P <0.05), no differences for limitation on movement using 4-point scale. |
| Tizanidine | Murros 2000 118. Fair | Tension headache 201 | No statistical differences for headache severity using 100 mm visual analogue scale, days free of headache, daily duration of headache, or use of paracetamol. |
| Tizanidine | Saper 2002 119. Fair | Daily headaches 136 randomized | Tizanidine superior for headache index (headache days x average intensity x duration), mean headache days/week, average headache duration, average headache intensity using 5-point scale, pain using 100 mm visual analogue scale, no difference for functional status using Migraine Disability Assessment questionnaire. |
| Tizanidine | Sirdalud Ternelin Asia- Pacific Study Group 1988 139. Fair | Acute neck or low back syndromes 405 | Tizanidine superior for pain using 4-point scale (P<0.05), spasm using 4-point scale (P<0.001), restriction of body movement using 4-point scale (P<0.001), no difference for sleep quality using 4-point scale. |
There was no clear evidence from head-to-head trials that one skeletal muscle relaxant was superior to any other. Three trials evaluated one included skeletal muscle relaxant versus another, but each evaluated a different comparison. In a trial comparing tizanidine and chlorzoxazone in patients with back pain,
103.
there were no significant differences between treatments for muscle pain, muscle tension, tenderness, and activity. More patients reported ‘excellent’ overall results with tizanidine (57%) compared to chlorzoxazone (23%), but similar proportions of patients reported ‘good or excellent’ results (79% vs. 69%). A trial of cyclobenzaprine versus methocarbamol in patients with localized muscle spasm found that there were no significant differences in the proportion of patients reporting absent or mild muscle spasm, limitation of motion, or limitation of daily activities.18.
In a trial of cyclobenzaprine versus carisoprodol in patients with acute back pain and spasms,104.
there were no significant differences for pain, muscle stiffness, activity impairment, sleep impairment, tension, or relief scores compared to baseline.Eight other head-to-head trials compared an included skeletal muscle relaxant to diazepam. Of these, the trial that appeared to be of best quality compared carisoprodol and diazepam.
105.
This trial was still rated fair quality because the authors did not describe allocation concealment techniques and used unvalidated methods for assessing outcomes. Carisoprodol was significantly superior to diazepam for stiffness, tension, and relief, with average differences about 0.5 on a 1–5 scale.105.
No significant differences were seen for pain, activity impairment, or sleep impairment.Of five trials
101.
, 102.
, 106.
, 107.
comparing cyclobenzaprine to diazepam, two106.
, 107.
found significant differences (using unvalidated measures) for most measurements of pain, muscle spasm, functional status, and ‘global evaluations’ that favored cyclobenzaprine. One other trial106.
reported decreased tenderness, decreased limitation of motion and better ‘global evaluation’ for cyclobenzaprine versus diazepam, but not for other measures (muscle spasm, pain, functional ability). All three of these trials had some support from a manufacturer (Merck Sharp & Dohme, West Point, Pennsylvania, USA) and were published in the same book. For most outcomes that favored cyclobenzaprine, the magnitude of difference between treatments was greater at the end of Week 1 than at the end of Week 2. Two other trials comparing cyclobenzaprine to diazepam101.
, 102.
and two trials108.
, 109.
comparing tizanidine to diazepam found no significant differences for any clinical outcomes including pain, stiffness, or functional ability.The trial
101.
focusing on patients with chronic back or neck symptoms reported results similar to the other trials. The overall withdrawal rates in all head-to-head trials ranged from 0% to 35%. In one trial,106.
the overall withdrawal rate appeared significantly higher on cyclobenzaprine (12/34 [35%]) compared to diazepam (3/32 [9%]), but there was no significant difference in the withdrawal rate between interventions in other trials.We identified no head-to-head trials of orphenadrine, metaxalone, dantrolene, or baclofen in patients with musculoskeletal conditions.
3.2.3 Placebo-controlled trials
In addition to six head-to-head trials (from five publications)
18.
, 101.
, 102.
, 106.
, 107.
with a placebo arm, we identified an additional 35 placebo-controlled trials (Table 5). Three trials evaluated a skeletal muscle relaxant with an equivalent analgesic in each arm.110.
, 111.
, 112.
Most trials evaluated low back or neck syndromes alone or mixed with other musculoskeletal conditions. Other conditions evaluated were fibromyalgia,113.
, 114.
, 115.
tension headaches or mixed headache conditions,116.
, 117.
, 118.
, 119.
and nocturnal leg cramps.120.
No trial included children.In general, placebo-controlled trials were not helpful in assessing comparative efficacy. None of the placebo-controlled trials involving patients with musculoskeletal conditions was rated good quality (Table 5 and Evidence Table 6). The comparative efficacy of each skeletal muscle relaxant was also difficult to assess because of marked heterogeneity in study design, interventions, populations studied, and outcomes assessed.
Table 6Adverse Events, Head-to-Head Trials of Skeletal Muscle Relaxants for Spasticity
| Study | Interventions | Somnolence or Fatigue | Weakness | Dizziness or Lightheadedness | Dry Mouth | Withdrawals Due to Adverse Events |
|---|---|---|---|---|---|---|
| Tizanidine versus Baclofen | ||||||
| Bass 1988 60. | Tizanidine mean 17 mg/day | 29% | 21% | Not reported | 23% | 9% (4/46) |
| Baclofen mean 35 mg/day | 19% | 35% | Not reported | 14% | 26% (12/46) | |
| Eysette 1988 58. | Tizanidine 24 mg/day | 30% | Infrequent (data not reported) | Not reported | 28% | 6% (3/49) |
| Baclofen 60 mg/day | 20% | 20% | Not reported | Infrequent (data not reported) | 6% (3/49) | |
| Hoogstraten 1988 57. | Tizanidine 12-24 mg/day | 57% | 33% | 14% | 36% | 11% (1/9) |
| Baclofen 15-60 mg/day | 29% | 57% | 14% | 14% | 14% (1/7) | |
| Medici 1989 51. | Tizanidine mean 20 mg/day | 33% | 0% | 0% | 7% | 0% (0/15) |
| Baclofen mean 50 mg/day | 29% | 7% | 7% | 0% | 20% (3/15) | |
| Newman 1982 59. | Tizanidine titrated to 16 mg/day | 15% | 8% | 8% | 0% | 6% (2/36) |
| Baclofen titrated to 40 mg/day | 19% | 15% | 15% | 4% | 17% (6/36) | |
| Rinne 1980 (2) 53. | Tizanidine mean 11 mg/day | 62% (6% severe) | 19% (0% severe) | 25% (0% severe) | 50% | 6% (1/16) |
| Baclofen mean 51 mg/day | 80% (20% severe) | 38% (40% severe) | 60% (13% severe) | 27% | 6% (1/16) | |
| Smolenski 1981 56. | Tizanidine 24 mg/day | 45% | 18% | None reported | 9% | 0% (0/11) |
| Baclofen 60 mg/day | 0% | 30% | None reported | 10% | 0% (0/10) | |
| Stien 1987 36. | Tizanidine mean 23/day | 33% (also includes weakness and dry mouth) | Not reported separately | Not reported | Not reported separately | 6% (1/18) |
| Baclofen mean 59 mg/day | 25% (also includes weakness and dry mouth) | Not reported separately | Not reported | Not reported separately | 4% (1/20) | |
| Tizanidine, Baclofen, or Dantrolene versus Diazepam | ||||||
| Bes 1988 49. | Tizanidine mean 17 mg/day | 44% | 2% | None reported | 11% | 12% (6/51) |
| Diazepam mean 20 mg/day | 44% | 18% | None reported | 3% | 28% (15/54) | |
| Rinne 1980 (1) 53. | Tizanidine mean 14 mg/day | 53% (0% severe) | 13% (8% severe) | 7% | 33% | 0% (0/15) |
| Diazepam mean 15 mg/day | 87% (47% severe) | 53% (27% severe) | 13% | 0% | 27% (4/15) | |
| Cartlidge 1974 63. | Baclofen 30 mg/day and 60 mg/day | 14% | 11% | 3% | 3% | 30% (11/37) |
| Diazepam 15 mg/day and 30 mg/day | 11% | 16% | 0% | 0% | 38% (14/37) | |
| From 1975 61. | Baclofen mean 61 mg/day | 31% | 19% | 6% | Not reported | 6% (1/16) |
| Diazepam mean 21 mg/day | 69% | 12% | 6% | Not reported | 0% (0/16) | |
| Roussan 1985 52. | Baclofen mean 47 mg/day | 8% | Not reported | Not reported | Not reported | 0% (0/13) |
| Diazepam mean 28 mg/day | 38% | Not reported | Not reported | Not reported | 0% (0/13) | |
| Glass 1974 54. | Dantrolene 100 mg qid | Not reported | Not reported | Not reported | Not reported | 19% (3/16) |
| Diazepam 5 mg qid | Not reported | Not reported | Not reported | Not reported | 6% (1/16) | |
| Nogen 1976 48. | Dantrolene titrated to 75 mg qid | Not clear | Not reported | Not reported | Not reported | None reported |
| Diazepam titrated to 12 mg/day | Not clear | Not reported | Not reported | Not reported | None reported | |
| Schmidt 1976 62. | Dantrolene 75 mg qid | 31% | 67% | 19% | Not reported | Not clear |
| Diazepam 5 mg qid | 67% | 76% | 19% | Not reported | Not clear |
Carisoprodol (four trials
121.
, 122.
, 123.
, 124.
), cyclobenzaprine (18 trials reported in 17 publications18.
, 101.
, 102.
, 106.
, 107.
, 110.
, 113.
, 114.
, 115.
, 117.
, 125.
, 126.
, 127.
, 128.
, 129.
, 130.
, 131.
including five headto-head trials with a placebo arm), orphenadrine (four trials21.
, 111.
, 120.
, 132.
), metaxalone (four trials in three publications133.
, 134.
, 135.
), and tizanidine (six trials116.
, 118.
, 119.
, 136.
, 137.
, 138.
, 139.
) were evaluated in the highest number of trials. A smaller number of trials evaluated baclofen (1 trial140.
), methocarbamol (2 trials18.
, 141.
), and dantrolene (2 trials112.
, 142.
). Although most of these trials found significant benefits or trends towards benefit on active treatment compared to placebo, cyclobenzaprine has been evaluated and consistently found effective in substantially more trials than the other skeletal muscle relaxants. The data on metaxalone, on the other hand, was mixed. The best fair-quality trial found no differences compared to placebo,135.
but a poor-quality trial133.
and two lesser fair-quality trials134.
reported some benefits compared to placebo using unvalidated outcome measures. We identified no placebo-controlled trials evaluating chlorzoxazone.3.3 Comparative safety: spasticity
3.3.1 Systematic reviews and meta-analyses
We identified no systematic reviews that evaluated comparative adverse event rates from skeletal muscle relaxants in patients with spasticity. One meta-analysis of three placebo-controlled trials was rated poor quality for adverse event assessment because no information about adverse event assessment methods was reported (Table 1).
45.
Adverse events included 49% dry mouth, 48% somnolence, 41% asthenia, 16% dizziness, and 12% headache in patients on tizanidine compared to 10%, 10%, 16%, 4%, and 13% on placebo. Two patients had liver function abnormalities and three had hallucinations. No deaths were reported. Abuse or addiction was not evaluated. Withdrawal rates due to adverse events were 17% for tizanidine and 7% for placebo.3.3.2 Head-to-head trials
No head-to-head trial was rated good quality for adverse event assessment. In general, there was little evidence of rigorous adverse event assessment in these trials (Evidence Table 3). No trial appeared to have significantly better adverse event reporting methods than the others. The most frequently reported adverse event rates were for somnolence, weakness, dizziness, and dry mouth. For the same medication, adverse event rates varied between trials (Table 6). For example, rates of somnolence from baclofen in head-to-head trials of baclofen and tizanidine ranged from 0%
56.
to 80%,53.
and weakness ranged from 7%51.
to 57%.57.
The observed ranges of adverse event rates could reflect differences in populations, dosing of medications in trials, use of a run-in period, the rigor of adverse event assessment, or other factors. No deaths or serious adverse events were reported in these trials. Rates of abuse and addiction were not evaluated.For each skeletal muscle relaxant evaluated in head-to-head trials, rates across trials for common adverse events overlapped with rates found for other skeletal muscle relaxants (Table 6). In individual head-to-head trials of tizanidine and baclofen, however, several patterns emerged. In these eight trials, dry mouth was reported more frequently on tizanidine in five studies (roughly equivalent or not reported in the other three), but weakness was reported more frequently on baclofen in all seven studies in which it was reported. No consistent patterns were seen for somnolence or dizziness. Withdrawal rates due to adverse events, an indicator of intolerable adverse events, were higher on baclofen than tizanidine (12/46 [26%] vs. 4/46 [9%]) in only one trial with significant numbers of withdrawals. Other trials had very low numbers of withdrawals due to adverse events or found no differences.
It was not possible to use trials comparing baclofen, dantrolene, or tizanidine with diazepam to assess comparative adverse event rates between these three medications. Adverse events data were not reported or poorly reported in three trials.
48.
, 52.
, 54.
In the remaining trials, no clear pattern of differential adverse events was apparent for any skeletal muscle relaxant. Withdrawals due to adverse events favored tizanidine over diazepam in one trial49.
(28% [15/54] vs. 12% [6/51]), but in other trials withdrawal rates were equivalent, not reported, or very few in number. The small number (two or three) of trials for each skeletal muscle relaxant, the wide ranges for adverse events (somnolence 11–67%, weakness 12–53%) on diazepam (the common comparator) in different trials, and the limited quality of adverse event assessment limit further interpretation of these data.3.3.3 Placebo-controlled trials
Most placebo-controlled trials showed little evidence of rigorous adverse event assessment. Abuse or addiction was not evaluated. Three trials appeared to have more rigorous adverse event assessment
95.
, 96.
, 97.
and were rated good quality. All three of these trials evaluated tizanidine. Rates of somnolence (41–54%) were similar in these trials but rates for other adverse events (dry mouth, dizziness, weakness, and withdrawal due to adverse events) ranged widely or were not consistently reported (Table 7). In one of the good-quality trials,95.
3 patients (18%) developed elevations of transaminases (highest alanine transaminase 90) that were not thought to be clinically significant.Table 7Adverse Events, Placebo-Controlled Trials of Skeletal Muscle Relaxants for Spasticity
| Intervention | Study and Year | Somnolence or Fatigue | Dizziness or Lightheadedness | Dry Mouth | Withdrawals Due to Adverse Events | Any Adverse Events |
|---|---|---|---|---|---|---|
| Baclofen 5 mg tid | Basmajian 1974 64. | 0% | 0% | 0% | 0% | None reported |
| Baclofen unclear dose | Basmajian 1975 65. | Not reported | Not reported | Not reported | 12% | Not reported |
| Baclofen 5–20 mg/day | Brar 1991 66. | Not reported | Not reported | Not reported | Not reported by intervention | Not reported |
| Baclofen 5 mg tid to 100 mg/day | Duncan 1976 66. | 12% | 24% | 12% | 0% | 60% |
| Baclofen 15–80 mg/day | Feldman 1978 68. | 17% | Not reported | 22% | 0% | Not reported |
| Baclofen 40–80 mg/day | Hinderer 1990 69. | Not reported | Not reported | Not reported | Not reported | Not reported |
| Baclofen 10 mg tid | Hulme 1985 70. | 78% | Not reported | Not reported | 56% | 78% |
| Baclofen 15–60 mg/day | Jones 1970 71. | Not clear | None reported | None reported | None reported | Not reported |
| Baclofen 0.5 mg/kg/day titrated to maximum 60 mg/day | McKinlay 1980 72. | 60% | Not clear | None reported | 0% | 40% |
| Baclofen 30 mg/day | Medaer 1991 73. | 5% | 30% | None reported | None reported | 50% |
| Baclofen 10 mg/day titrated up to 60 mg/day | Milla 1977 74. | 20% | None reported | Not reported | 0% | 25% |
| Baclofen 5 mg tid titrated to 15 mg tid | Orsnes 2000 75. | 36% | 21% | None reported | None reported | 64% |
| Baclofen 5 mg tid titrated to 80 mg/day | Sachais 1977 76. | 71% | 22% | Not reported | Not reported (36% overall) | Not reported |
| Baclofen 5 mg tid titrated to 60 mg/day | Sawa 1979 77. | 29% | 10% | 5% | Not clear | 71% |
| Dantrolene unclear dose | Basmajian 1973 78. | ‘Almost all’ | ‘Several’ | Not reported | Not reported by intervention group | Not reported |
| Dantrolene 25–100 mg qid | Chyatte 1973 79. | Not reported | Not reported | Not reported | 0% | Not reported |
| Dantrolene 1–3 mg/kg qid | Denhoff 1975 80. | Not reported | Not reported | Not reported | None reported | 57% |
| Dantrolene 25 mg bid to 350 mg/day | Gambi 1983 81. | 29% | Not reported | Not reported | 9% | 54% |
| Dantrolene 50–800 mg/day | Gelenberg 1973 82. | 15% | 55% | Not reported | None reported | Not reported |
| Dantrolene 4–12 mg/kg/day | Haslam 1974 83. | Not reported | Not reported | Not reported | 0% | Not reported |
| Dantrolene 4–12 mg/kg/day | Joynt 1980 84. | Not reported | Not reported | Not reported | 9% | 91% |
| Dantrolene 25 mg bid to 50 mg qid | Katrak 1992 85. | 70% | Not reported | Not reported | Not reported by intervention group | Not reported |
| Dantrolene mean 165 mg/day | Ketel 1984 86. | Not reported | Not reported | Not reported | 25% | 75% |
| Dantrolene 75 mg tid to 400 mg qid | Luisto 1982 87. | 88% | 24% | Not reported | Not reported by intervention group | 100% |
| Dantrolene 50–100 mg qid | Monster 1974 88. | Not clear | Not clear | Not clear | Not clear (27% withdrawals overall) | Not reported |
| Dantrolene 6–8 mg/kg/day | Nogen 1979 89. | 82% | Not reported | Not reported | None reported | Not reported |
| Dantrolene titrated to maximum 200 mg qid | Sheplan 1975 90. | Not clear | Not clear | Not clear | Not reported | Not reported |
| Dantrolene 100 mg/day titrated to 800 mg/day | Tolosa 1975 91. | Not clear | Not clear | Not clear | 17% | Not reported |
| Dantrolene titrated to 100 mg qid | Weiser 1978 92. | 23% | Included in somnolence | Not reported | 11% | Not reported |
| Ti |