Drugs for relief of pain in patients with sciatica: systematic review and meta-analysisBMJ 2012; 344 doi: http://dx.doi.org/10.1136/bmj.e497 (Published 13 February 2012) Cite this as: BMJ 2012;344:e497
- Rafael Zambelli Pinto, PhD student1,
- Chris G Maher, director1,
- Manuela L Ferreira, research fellow1,
- Paulo H Ferreira, senior lecturer2,
- Mark Hancock, senior lecturer23,
- Vinicius C Oliveira, PhD student4,
- Andrew J McLachlan, professor5,
- Bart Koes, professor6
- 1George Institute for Global Health, University of Sydney, PO Box M201, Camperdown, Sydney, NSW, 2050, Australia
- 2Discipline of Physiotherapy, Faculty of Health Sciences, University of Sydney, Sydney
- 3Faculty of Human Sciences, Macquaire University, Sydney
- 4Faculty of Health Sciences, University of Sydney, Sydney
- 5Faculty of Pharmacy, University of Sydney, Centre for Education and Research on Ageing, Concord Hospital, Sydney
- 6Department of General Practice, Erasmus MC, University Medical Centre, Rotterdam, Netherlands
- Correspondence to: R Z Pinto
- Accepted 21 November 2011
Objective To investigate the efficacy and tolerability of analgesic and adjuvant pain drugs typically administered in primary care for the management of patients with sciatica.
Design Systematic review.
Data source International Pharmaceutical Abstracts, PsycINFO, Medline, Embase, Cochrane Central Register of Clinical Trials (CENTRAL), CINAHL, and LILACS.
Study selection Randomised controlled trials assessing the efficacy and tolerability of drugs versus placebo or other treatment for sciatica.
Data extraction Two independent reviewers extracted data and assessed methodological quality using the PEDro scale. Pain and disability outcomes were converted to a common 0 to 100 scale. Data were pooled with a random effects model, and the GRADE approach was used in summary conclusions.
Results Twenty three published reports met the inclusion criteria. The evidence to judge the efficacy of non-steroidal anti-inflammatory drugs (NSAIDs), corticosteroids, antidepressants, anticonvulsants, muscle relaxants, and opioid analgesics ranged from moderate to low quality. Most of the pooled estimates did not favour the active treatment over placebo. The pooled results of two trials of corticosteroids (mean difference in overall and leg pain −12.2, 95% confidence interval −20.9 to −3.4) and a single trial of the anticonvulsant gabapentin for chronic sciatica (mean difference in overall pain relief −26.6, −38.3 to −14.9) showed some benefits but only in the short term. The median rate of adverse events was 17% (interquartile range 10-30%) for the active drugs and 11% (3-23%) for placebo. Trial limitations included failure to use validated outcome measures, lack of long term follow-up, and small sample size.
Conclusions As the existing evidence from clinical trials is of low quality, the efficacy and tolerability of drugs commonly prescribed for the management of sciatica in primary care is unclear.
The clinical syndrome of sciatica has been recognised since ancient times. Currently believed to arise from a disorder of the nerve root, the syndrome is known by a range of synonyms such as lumbosacral radicular syndrome, nerve root compromise, nerve root pain, and nerve root entrapment or irritation.1 2 Sciatica is considered to be a prognostic indicator of poor outcome among patients with low back pain3 4 with a substantial proportion continuing to have persistent pain for two years or longer.5 The annual prevalence of sciatica is estimated to be between 14%6 and 2%.7 While there are a range of definitions of sciatica,2 the key clinical features that can help clinicians to distinguish it from non-specific low back pain include unilateral leg pain that is worse than the low back pain, pain radiating below the knee, presence of numbness or pins and needles in a dermatomal distribution, positive results on a straight leg raise test, and weakness or reflex changes, or both, in a myotomal distribution.1
Analgesic and adjuvant pain drugs are often prescribed for patients with sciatica.8 Patients with a clinical diagnosis of sciatica are about five times more likely to take drugs than those with low back pain only.4 Drugs commonly prescribed for the management of sciatica include non-steroidal anti-inflammatory drugs (NSAIDs), skeletal muscle relaxants, opioid analgesics, benzodiazepines, systemic corticosteroids, antidepressants, and anticonvulsants.9 10 While guidelines provide clear and generally consistent recommendations for the prescription of drugs for non-specific low back pain,11 12 this is not the case for sciatica.
At present the efficacy and tolerability of commonly prescribed analgesic and adjuvant drugs for the management of patients with sciatica has not been established. The problem is that there are no reviews that specifically focus on these drugs that are used in primary care to manage sciatica. Because most guidelines recommend a course of conservative care before surgery is considered, it is imperative to understand what best practice conservative care should entail. In this systematic review we determined the efficacy and tolerability of analgesic and adjuvant pain drugs typically administered in primary care (orally, topically, or parentally) in the management of patients with sciatica when compared with other treatment options or no treatment.
Data sources and searches
We used the review methods advocated by the editorial board of the Cochrane Back Review Group.13 This study searched the following databases from the earliest records to 15 March 2010: International Pharmaceutical Abstracts, PsycINFO, Medline, Embase, Cochrane Central Register of Clinical Trials, CINAHL, and LILACS. Key words related to randomised controlled trial, sciatica, and drugs terms in addition to subject subheadings and word truncations specific for each database were used (see appendix on bmj.com). Electronic searches were supplemented by hand searching reference lists of eligible clinical trials. Our search was restricted to trials published in English, German, Dutch, Portuguese, and Spanish. One reviewer screened all relevant titles and abstracts and excluded clearly irrelevant papers, leaving 173 potentially relevant papers. Two reviewers independently evaluated the full reports for eligibility. Disagreements were resolved by discussion.
Studies were eligible if they were randomised controlled trials evaluating single or any combination of analgesic or adjuvant pain drugs. For the purposes of this review, we focused on pharmacological interventions that could be administered in primary care settings, which include drugs administered via oral, topical, or parental route (that is, intramuscular injection and intravenous bolus). We excluded trials evaluating procedures that require imaging assistance and the monitoring of anaesthesia monitoring capabilities (for example, epidural, extradural, and intradiscal injections) or that require an extended period of time to be administered with monitoring of the patient’s vital signs (for example, intravenous infusion). Eligible comparisons included no treatment, placebo, or other treatment options (for example, other pharmacological treatment, surgery, epidural injections, or other conservative treatment).
To be eligible studies needed to explicitly report that participants had sciatica or a synonym for sciatica. Sciatica synonyms considered included radiculopathy, nerve root compromise, nerve root compression, lumbosacral radicular syndrome, nerve root pain, nerve root entrapment, and pain radiating below the knee. The intensity or duration of symptoms was not restricted, but participants were classified as having acute (less than six weeks), subacute (six to 12 weeks), or chronic (12 weeks or more) symptoms. Trials of mixed groups of patients with low back pain were eligible if it was possible to clearly identify a subgroup with sciatica and data only from these participants were included in the analysis. There was no restriction related to the source of participants. Sciatica definitions across trials were classified as those based on clinical assessment alone and those that also require concordant imaging evidence.
Trial drugs were classified according to the Anatomical Therapeutic Chemical (ATC) classification system advocated by the World Health Organization’s collaborating centre for drugs statistics methodology.14 Table 1 lists the classes of drugs and the ATC codes included in the search strategy of the original protocol⇓.
Trials were included when one of the following outcome measures was reported: overall pain intensity (when not specified as leg or back pain), leg or back pain intensity, disability status, work status, and adverse events. We also included studies that did not report continuous data for these variables but reported percentage of improved patients. As the definition of improved (or not improved) was unclear or varied between studies, however, we did not pool these data. Work status, when identified, was listed as an outcome but pooling of these data was also not possible because of different definitions used across trials.
Data extraction and quality assessment
Two trained independent raters assessed the quality of the trials’ methods using the PEDro scale.15 16 Disagreements were resolved by a third rater. Methodological quality was not an inclusion criterion.
Two independent reviewers extracted means (final scores or change score), standard deviations, and sample sizes from studies using a standardised data extraction form. When there was insufficient information in trial reports, we contacted authors or estimated data using methods recommended in the Cochrane Handbook for Systematic Reviews of Interventions.17 Briefly, when the mean was not reported we used the median; when standard deviations could not be estimated, we adopted the standard deviation from the most similar study.
To evaluate the overall quality of the evidence and the strength of the recommendation, we used the GRADE approach.18 We used an adapted version of the criteria advocated by the Cochrane Back Review Group.19 The quality of the evidence was downgraded by one level for each of four factors we encountered: limitations in the design (for instance, >25% of participants from studies with low quality methods—PEDro score <7 points); inconsistency of results (for instance, ≤75% of the participants report findings in the same direction); and imprecision (for instance, total number of participants <300 for each outcome). We did not assess publication bias with funnel plots as too few studies were included in the meta-analysis. We also did not assess indirectness as this review encompasses a specific population. Two reviewers judged whether these factors were present for each outcome. Single randomised studies (with under 300 participants) were considered inconsistent and imprecise (that is, sparse data) and provided “low quality evidence.” This could be further downgraded to “very low quality evidence” if there were also limitations in design. We applied the following definitions of quality of the evidence20:
High quality—further research is unlikely to change our confidence in the estimate of effect. There are no known or suspected reporting biases; all domains fulfilled
Moderate quality—further research is likely to have an important impact on our confidence in the estimate of effect and might change the estimate; one of the domains was not fulfilled
Low quality—further research is likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate; two of the domains were not fulfilled
Very low quality—we are uncertain about the estimate; three of the domains were not fulfilled.
A GRADE profile was completed for each pooled estimate and for single trials comparing drugs versus placebo.
Data synthesis and analysis
Outcome data were extracted for immediate term (two or fewer weeks after randomisation), short term (more than two weeks but less than three months), intermediate term (more than three months but less than12 months), and long term (12 months or more) follow-up evaluations. When multiple time points fell within the same category, we used the one closest to one week for the immediate term, eight weeks for the short term, six months for the intermediate term, and 12 months for the long term. Scores for pain intensity and disability were converted to a scale from 0 to 100. When more than one outcome measure was used to assess intensity of pain, disability, or work status, we included the outcome measure described as the primary outcome measure for the trial. Descriptive statistics were used to describe adverse events reported in each trial.
Trials that we considered clinically homogeneous were grouped according to class of drugs, comparison group, dose, outcomes (pain, disability, etc), and outcome assessment time points (immediate term, short term, intermediate term, and long term). Pooled estimates were obtained with Comprehensive Meta-Analysis software, version 2.2.04 (Biostat, Englewood, NJ). Pooled effects were calculated with a random effects model. For calculation of effect size (see appendix tables A and B on bmj.com), we pooled leg pain and overall pain, as leg pain is usually worse than pain in the back. When trials were not sufficiently homogeneous, pooling of data via meta-analysis was not performed and outcome measures of the individual studies were reported. We decided a priori that a sensitivity analysis would be conducted to investigate definitions of sciatica as possible sources of heterogeneity in effect size among included studies. The limited number of trials in the meta-analysis, however, prevented further investigation.
Figure 1 outlines the flow of trials through the review⇓. The initial electronic database search identified 2460 potential studies of interest. After screening citations by title and abstract, we considered 197 potentially eligible articles for inclusion and retrieved full articles. We included 23 published reports21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 (reporting 24 different clinical trials, as one28 reported on two trials). Three trials reported in previous systematic reviews12 44 45 46 were not included in this review because of unclear randomisation.47 48 49
The included trials investigated six different classes of drugs: NSAIDs, antidepressants, corticosteroids, opioid analgesics, muscle relaxants, and anticonvulsants. The drugs investigated in the included trials were administered orally in 16 trials21 22 23 24 27 28 32 33 35 36 37 38 41 42 43 and parenterally in five trials (four trials25 30 39 40 investigated intramuscular injections and one trial29 investigated intravenous infusions). Three other trials26 31 34 used mixed administration of a starting intramuscular dose and subsequent oral doses. One study compared oral versus parenteral administration.22 Table 2⇓ shows the characteristics of the included trials, and table 3⇓presents details of methodological quality.
Duration of symptoms varied a lot across studies (table 2⇑). Six trials specifically investigated patients with chronic symptoms21 23 37 38 41 43 and eight trials included patients with acute symptoms.22 28 29 30 33 35 42 In six trials, however, patients had mixed duration of symptoms25 27 34 36 39 40; in two trials duration was not specified26 32; and in two trials patients were reported to have acute symptoms but no information regarding duration was provided.24 31 Regarding the definition of sciatica, 15 trials used a definition based on clinical assessment alone,21 23 24 26 28 30 31 32 33 34 35 39 40 42 whereas nine trials also required concordant imaging evidence.22 25 27 29 36 37 38 41 43 Pooling of data via meta-analysis was possible only for comparison of NSAIDs versus placebo in pain intensity at immediate follow-up and for corticosteroids versus placebo in pain intensity at immediate and short term follow-up. Table 2 also presents the number of studies comparing each class of drug and control.⇑
Placebo controlled trials
Fifteen randomised clinical trials compared drugs with placebo with individual effect sizes calculated for nine trials (fig 2⇓). For five trials investigating NSAIDs,31 corticosteroids,32 34 antidepressants,21 and muscle relaxants,24 we did not calculate effect sizes as percentage of improved patients was reported as an outcome.
Four trials compared orally administered NSAIDs with placebo for acute sciatica.28 33 42 Of these, three trials used a three arm design.28 33 Treatment effects for these studies are presented in figure 2⇑. Data for pooling were available from all three studies. Figure 3⇓ shows small and non-significant pooled effect size for pain (overall and leg pain) in the immediate term follow-up (mean difference −4.9, 95% confidence interval −10.2 to 0.4; P<0.07). Because of the three arm design, we calculated effect sizes for three other treatment comparisons (comparison 2 (meloxicam 7.5 mg/day, lornoxicam 8 mg/day, piroxicam 20 mg/day v placebo): mean difference −4.4, −8.9 to 0.1, I2=0.1%, P<0.06; comparison 3 (meloxicam 15 mg/day, lornoxicam 8 mg/day, piroxicam 2 mg/day v placebo): mean difference −3.9, −8.0 to 0.2, I2=0.6%, P<0.07; comparison 4 (meloxicam 15 mg/day, diclofenac 50-150 mg/day, piroxicam 20 mg/day v placebo): mean difference −4.4, −9.4 to 0.6, I2=6.8%, P<0.09) (not shown in fig 3). The quality of evidence (GRADE) for this pooling was rated “low quality” (downgraded for limitation of study design and inconsistency).
Three trials in patients with acute sciatica tested the effect of corticosteroids compared with placebo.29 30 35 The individual studies provided imprecise estimates of treatment effect (fig 2⇑). For the immediate term, pooling showed no effect of steroids on leg pain (mean difference −1.8, −11.1 to 7.5; P=0.71) with a GRADE classification of “moderate quality” of evidence (downgraded for imprecision) (fig 3⇑). For the short term, pooling showed a significant effect of steroids on pain (overall and leg pain) (mean difference −12.2, −20.9 to −3.4; P<0.01) and “moderate quality” of evidence according to the GRADE approach (downgraded for imprecision) (fig 3⇑).
Antidepressants, anticonvulsants, and opioid analgesics
Three trials investigated the efficacy of antidepressants, anticonvulsants, and opioid analgesics (fig 2⇑).37 38 43 As these were single trials that investigated different classes of drugs or time points of assessment, pooling was not possible. Hence, the quality of evidence for all these medications was considered “low quality” (single studies)37 43 or “very low quality” (single study downgraded for limitation of study design)38 according to the GRADE approach.
Two trials investigated the effect of anticonvulsants in patients with chronic symptoms of sciatica. In one crossover trial,38 topiramate (50-400 mg/day) showed no better immediate effects than placebo for disability and leg or back pain. In contrast, data from another trial43 showed significant overall pain relieving effect with the anticonvulsant gabapentin (900-3600 mg) in the short term (mean difference −26.6, −38.3 to −14.9; P<0.001) compared with placebo.
One crossover study with four periods investigated the relative efficacy of opioid analgesics (sustained release morphine 15 mg/day), antidepressants (nortriptyline 25 mg/day), and a combination of both and placebo in patients with chronic sciatica.37 In this trial, antidepressants, opioid analgesics, and a combination of both had no significant effect compared with placebo in the immediate term for disability or leg or back pain.
NSAIDs v NSAIDs
Five studies compared one type of NSAID with other NSAIDs.25 28 33 39 40 Figure 4⇓ shows all the comparisons involving different types of NSAIDs. None of these studies showed one NSAID to be better than the other. One study found no difference in overall pain between oral and parenteral administration of meloxicam 15 mg a day.22
NSAIDs v other treatments
In three separate trials the NSAID diclofenac (50-75 mg/day) showed no difference in outcomes compared with antidepressant36 or electroacupuncture41 but did worse than caudal epidural injections of corticosteroids27 for pain and disability, specifically in the immediate follow-up (fig 3⇑). In one trial the NSAID ketoprofen (200 mg/day) was no better than a combination of corticosteroids for pain in the immediate follow-up.26
Adverse events were investigated in 75% (18/24) of the included trials (table 4⇓); the median number of adverse events (interquartile range) was 17% (10-30%) for the active drugs and 11% (3-23%) for placebo. The rate and type of adverse events reported varied substantially between drugs and between trials for the same medicine.
There is at best only low quality evidence to judge the efficacy and tolerability of drugs commonly prescribed for the management of sciatica in primary care. The available evidence does not clearly show favourable effects of NSAIDs, corticosteroids, antidepressants, or opioid analgesics in the immediate term, even compared with placebo. Evidence from a single trial and one meta-analysis with two trials provides some limited support (“low quality” and “moderate quality” evidence) for the use of NSAIDs and corticosteroids to relieve pain in the short term in patients with acute sciatica. In one small sample trial there was also limited support (“low quality” evidence) for the short term relief of pain in chronic sciatica with an anticonvulsant drug. At present these data provide a lack of conclusive and high quality evidence to guide the prescription of these drugs for patients with sciatica in primary care.
Strengths and limitations of review
The strengths of this systematic review include the use of a prespecified protocol, inclusion of trials published in languages other than English, and the use of a highly sensitive search strategy to identify trials for sciatica coupled with search terms related to the nine classes of drugs of interest. We assessed methodological quality with the PEDro scale rather than the Cochrane Risk of Bias tool because the PEDro scale has been shown to have acceptable reliability16 and validity,15 50 whereas two studies have reported reliability limitations with the Cochrane tool.51 52 One limitation of our study is the possibility of publication bias as we did not attempt to identify unpublished trials that could be found in clinical trials registries and conference proceedings. Another limitation is that we could accommodate trials in only four languages other than English.
A strength of our review over past reviews is that we located 24 randomised clinical trials relevant to our question compared with six,12 three,46 and two44 45 trials included in the four previous reviews. Accordingly we provide more precise estimates of treatment effects, specifically for NSAIDs and corticosteroids, and a better description about tolerability of all classes of drug. Including more trials has allowed us to provide a more in-depth evaluation of the pharmacological treatments that have been tested for sciatica in primary care. For example, we have provided information on the efficacy of antidepressants and opioid analgesics whereas previous reviews did not cover these classes of drugs. Another strength of this review is that we quantitatively pooled trials where possible. This allowed us to explore the size of the treatment effect, which is not possible in a qualitative summary. In addition, we were able to assess the overall quality of the evidence with the GRADE approach. Although conclusions of previous reviews were based on a surprising paucity of evidence, our findings are likely to inform the direction of future research in this topic. In pharmacological trials for sciatica, we consider the priorities to include consideration of duration of symptoms, recruitment of large samples, and collection of outcomes relevant to patients. For instance, regarding duration of symptoms it would be reasonable to test corticosteroids and NSAIDs in trials in patients with acute sciatica, whereas given that chronic sciatica includes a neuropathic pain component, anticonvulsant and antidepressants might need to be tested in patients with a chronic condition.
Although the knowledge of evidence about surgery has changed in the past decade, we still do not know much about a simpler conservative treatment such as pharmacological interventions for sciatica. A recent clinical practice guideline from the American Pain Society53 recommends that surgery should be considered as a treatment option for persistent and disabling radicular pain caused by a herniated lumbar disc. This recommendation relies on moderate short term benefits from large randomised clinical trials such as those by Peul et al54 55 and Weinstein et al,56 with 283 and 501 patients randomised, respectively. At present, there is a limited number of mainly small trials of pharmacological interventions with typically short term follow-up. For example, the evidence to support the use of anticonvulsants in chronic sciatica arises from a single trial with only 50 participants.
Our principal finding is that there is insufficient evidence to confidently guide the use of any analgesic or adjuvant pain medicine for the effective management of pain and disability in patients with sciatica. Until this changes, we would advise clinicians treating such patients who exhibit clinical features of neuropathic pain to consider evidence based guidelines for neuropathic pain57 58 59 and for other patients with sciatica to consider therapeutic recommendations from current guidelines for the management of non-specific low back pain.11
What is already known on this topic
Pain relieving drugs are often prescribed by general practitioners for patients with sciatica
While guidelines provide clear and generally consistent recommendations for the prescription of drugs for non-specific low back pain, this is not the case for sciatica
The efficacy and tolerability of drugs in the management of sciatica administered in primary care has not been established
What this study adds
There is low quality evidence for the efficacy of NSAIDs, corticosteroids, and anticonvulsants
The available evidence does not clearly show favourable effects of NSAIDs, corticosteroids, antidepressants, muscle relaxants, and opioid analgesics in the immediate term and provides limited support for the use of NSAIDs, corticosteroids, and anticonvulsants in the short term
At present these data provide a lack of conclusive and high quality evidence to endorse the prescription of these drugs for patients with sciatica in primary care
Cite this as: BMJ 2012;344:e497
Contributors: RZP, CGM, BK, PHF, MLF, MH, and AJM were involved in the design of the review. RZP, MLF, PHF, and VCO developed the search strategy and performed the study selection. RZP, MH, AJM, and VCO extracted data from included studies. RZP, CGM, BK, AJM, and VCO were involved in data analysis. RZP, CGM, BK, PHF, MLF, and MH were involved in the interpretation and discussion of results. All authors contributed to the writing and review of the various drafts of the report. RZP and CGM are guarantors.
Funding: This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. RZP is supported by Capes Foundation, Ministry of Education of Brazil. CGM is supported by a research fellowship funded by the Australian Research Council.
Competing interests: All authors have completed the ICMJE uniform disclosure form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare: no support from any organisation for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous three years; no other relationships or activities that could appear to have influenced the submitted work.
Ethical approval: Nor required.
Data sharing: No additional data available.
This is an open-access article distributed under the terms of the Creative Commons Attribution Non-commercial License, which permits use, distribution, and reproduction in any medium, provided the original work is properly cited, the use is non commercial and is otherwise in compliance with the license. See: http://creativecommons.org/licenses/by-nc/2.0/ and http://creativecommons.org/licenses/by-nc/2.0/legalcode.