ORIGINAL INVESTIGATION Cyclobenzaprine and Back Pain A Meta-analysis Robert Browning, MD; Jeffrey L. Jackson, MD, MPH; Patrick G. O’Malley, MD, MPH Background: Back pain is a common problem for which 14 as were those treated with placebo. Slightly fewer than cyclobenzaprine hydrochloride is frequently pre- 3 individuals (2.7; 95% confidence interval, 2.0-4.2) scribed. needed treatment for 1 to improve. The magnitude of this improvement was modest, with an effect size of 0.38 to Objective: To perform a systematic review of cyclo- 0.58 in all 5 outcomes (local pain, muscle spasm, ten- benzaprine’s effectiveness in the treatment of back pain. derness to palpation, range of motion, and activities of daily living). Treatment efficacy for these 5 outcomes was Methods: We searched MEDLINE, PsycLIT, CINAHL, greatest early, in the first few days of treatment, declin- EMBASE, AIDSLINE, HEALTHSTAR, CANCERLIT, the ing after the first week. Patients receiving cyclobenzap- Cochrane Library, Micromedex, Federal Research in rine also experienced more adverse effects, the most com- Progress, and the references of reviewed articles, and con- mon being drowsiness. tacted Merck, Sharpe and Dohme for English-language, randomized, placebo-controlled trials of cyclobenzap- Conclusions: Cyclobenzaprine is more effective than rine in adults with back pain. Outcomes included global placebo in the management of back pain; the effect is improvement and 5 specific domains of back pain (local modest and comes at the price of greater adverse effects. pain, muscle spasm, range of motion, tenderness to pal- The effect is greatest in the first 4 days of treatment, sug- pation, and activities of daily living). Study quality was gesting that shorter courses may be better. Studies com- assessed using the methods of Jadad. Summary out- paring the relative value of acetaminophen, nonsteroidal comes were obtained using a random-effects model. anti-inflammatory drugs, and cyclobenzaprine individu- ally and in combination in the treatment of back pain Results: Patients treated with cyclobenzaprine were are needed. nearly 5 times (odds ratio, 4.7; 95% confidence interval, 2.7-8.1) as likely to report symptom improvement by day Arch Intern Med. 2001;161:1613-1620 ACK AND NECK pain are com- Agency for Health Care Policy and Re- mon, accounting for 1.8% search (AHCPR) back pain guidelines,11 of all office visits in the released in 1994, suggest that acetamino- United States, approxi- phen is the safest and most effective medi- mately 200000 visits per cation for acute low back problems and year.B1 Among adults presenting with a that nonsteroidal anti-inflammatory drugs physical complaint, these problems are sec- (NSAIDs), although effective, should be ond only to upper respiratory tract infec- used sparingly since they can cause gas- tions as a reason for seeking medical care.2-4 trointestinal irritation, ulceration, and, less Since only about a fourth of patients with commonly, renal or allergic problems. The a physical symptom seek medical atten- panel recommended that muscle relax- tion,5 back and neck pain are even more ants not be used, because they were be- prevalent in the general population. It is lieved to be no more effective than NSAIDs; estimated that up to half of working adults their use in combination with acetamino- experience back pain each year,6 and 70% phen or NSAIDs had no additional dem- From the Departments of of all adults experience it at some time in onstrated benefit; and adverse effects, in- Medicine, National Naval their life.7 Most episodes are self-limited, cluding drowsiness, are common. Medical Center (Dr Browning), with 90% of patients recovering to full ac- Despite these guidelines, cyclobenza- and Uniformed Services tivity within 1 month.8 Even among the prine hydrochloride is still commonly used University of the Health 2% with sciatica, more than half improve in the management of acute back pain. Na- Sciences (Drs Jackson and substantially within 6 weeks and only 5% tional Ambulatory Medical Care Survey data O’Malley), Bethesda, Md; and 6 the Department of Medicine, to 10% eventually require surgery. from 1995 to 1997 indicate that among 7 Walter Reed Army Medical Guidelines for treating back pain rec- million ambulatory visits to primary care Center, Washington, DC ommend empirical therapy, with no ra- physicians for back pain, 13% of the pa- (Drs Jackson and O’Malley). diologic evaluation in most patients.7-10 The tients received cyclobenzaprine.12 Be- (REPRINTED) ARCH INTERN MED/ VOL 161, JULY 9, 2001 WWW.ARCHINTERNMED.COM 1613 ©2001 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 10/04/2021 MATERIALS AND METHODS graphics, number of participants enrolled, follow-up losses, adverse effects, and outcomes. Outcomes were extracted as either dichotomous or continuous variables (or both), de- For this review, we searched MEDLINE (January 1966- pending on how they were reported in the studies. December 1999), PsycLIT (November 1887-December All analyses were done using Stata 6.0 (Stata Corp, Col- 1999), CINAHL (1982-1999), EMBASE (January 1974 lege Station, Tex). Assessment for publication bias was done December 1999), AIDSLINE, HEALTHSTAR, CANCERLIT, using the methods of Egger et al14; heterogeneity was as- and Micromedex using the text word and keyword (all lan- sessed visually with Galbraith plots15 andQ(x2) statistics guages) cyclobenzaprine. An additional search using the using the methods of Mantel-Haenszel.16 Although most Medical Subject Heading terms dibenzocycloheptenes and pooled results were homogeneous, the more conservative ran- propylamines was combined with the following text words: dom-effects model using the method of DerSimonian and back pain, backache, low back pain, lumbago, acute muscle Laird16 was used to calculate summary odds ratios (ORs), spasm,ormuscle spasm. We used the Cochrane Library, risk differences, and standardized mean differences. Analy- searching the clinical trials registry for randomized trials, sis of the continuous outcome involved comparing stan- and the Cochrane Database of Systematic Reviews for sys- dardized differences in means between control and treat- tematic reviews. We searched Federal Research in Progress ment groups. Mean outcome scores for the 5 continuous to identify unpublished literature. We searched the refer- outcomes (local pain, muscle spasm, tenderness to palpa- ences of reviewed articles for additional articles missed by tion, range of motion, and activities of daily living) were stan- the computerized database search. We also contacted Merck, dardized by dividing the means for the placebo and treat- Sharpe and Dohme, West Point, Pa. ment groups in each study by their SD. The differences Studies were screened for inclusion (through review between these standardized outcome scores, also known as of the abstract or the published article, if the abstract was effect sizes, were calculated for each study and analyzed. This unclear) based on the following criteria: randomization, pla- approach is especially appropriate when studies measure the cebo control, at least one group receiving cyclobenzap- same concept but use various continuous scales. By stan- rine, and measurable outcomes reported. Studies evaluat- dardization, the study results are transformed to a common ing cyclobenzaprine in the treatment of muscle spasticity scale (SD units) that facilitates pooling. of central nervous system origin were excluded from this Several measures of the sensitivity of the meta-anal- meta-analysis. Authors of abstracts that appeared promis- ysis results to various assumptions were conducted. The ing were contacted and asked to provide data. sensitivity of the results to the existence of unpublished stud- Included study quality was assessed using a 6-item ies was assessed using a “file drawer” test,17 conducted by instrument developed and validated by Jadad et al.13 The 6 determining how many studies with negative results (OR, items in this scale include description of randomization, 1.0; standardized mean difference, 0), each with variance adequacy of blinding, description of withdrawals and and size equal to the average found among the included dropouts, appropriateness of statistical analysis, descrip- studies, would have to exist to negate our findings. A test tion of inclusion and exclusion criteria, and method for for the relative influence of each individual study on the assessing adverse treatment effects. The study quality was results was determined by sequentially dropping indi- assessed independently in duplicate, with substantial inter- vidual studies and calculating summary measures. We ex- rater agreement (k, 0.79). Disagreements were arbitrated plored several sources of heterogeneity, including year of by consensus. publication, type of syndrome (acute vs chronic back pain), Abstracted data included setting, country of origin, treat- study quality scores, and duration and dose of therapy us- ment characteristics (dose, duration, and follow-up), demo- ing metaregression. cause cyclobenzaprine is frequently used for the manage- all studies were randomized and placebo controlled, there ment of back pain, we conducted a meta-analysis of the were several specific quality problems (Table 1), includ- English-language literature to assess the efficacy of this ing inadequate description of the randomization method treatment option. in 9,18,21,22,25,30-32,35,37 failure to describe the placebo as iden- tical to the active medicine