Abstract
Objective. To assess the effect of intramuscular botulinum toxin type A (BoNT-A) injections on pain, function/disability, global perceived effect, and quality of life (QOL) in adults with neck pain (NP).
Methods. We searched Central, Medline, and Embase databases up to June 2010. A minimum of 2 authors independently selected articles, abstracted data, and assessed risk of bias and clinical applicability. We estimated standard mean differences (SMD) with 95% CI, relative risks (RR), and performed metaanalyses (SMDp) using a random-effects model for nonheterogeneous data. The approach of the Grading of Recommendations Assessment, Development, and Evaluation working group summarizes the quality of evidence.
Results. We selected 14 trials. High-quality evidence suggested BoNT-A was no better than saline at 4 weeks [4 trials/183 participants; SMDp −0.21 (95% CI −0.50 to 0.07)] and 6 months for chronic NP. Moderate-quality evidence showed a similar effect for subacute/chronic whiplash-associated disorder (WAD) on pain [4 trials/122 participants; SMDp −0.21 (95% CI −0.57 to 0.15)], disability, and QOL. Very low-quality evidence indicated BoNT-A combined with exercise and analgesics was not significant for chronic NP reduction at 4 weeks [3 trials/114 participants; SMDp −0.08 (95% CI −0.45 to 0.29)] but was at 6 months [2 trials/43 participants; SMDp −0.66 (95% CI −1.29 to −0.04)].
Conclusion. Current evidence does not confirm a clinically or statistically significant benefit of BoNT-A used alone on chronic NP in the short term or on subacute/chronic WAD pain, disability, and QOL. Larger trials, subgroups, and predictors of responses defined a priori (to facilitate selection of patients most likely to benefit) and factorial designs to explore BoNT as an adjunct treatment to physiotherapeutic exercise and analgesics are needed.
- BOTULINUM TOXIN
- NECK PAIN
- WHIPLASH-ASSOCIATED DISORDER
Neck disorders are common, disabling, and costly1,2. The 12-month prevalence of neck pain (NP) in adults varies from 30% to 50%2. Among Saskatchewan adults, 66% reported NP during their lifetime and 5% reported significant disability from NP in the previous 6 months1. Conceptually, botulinum toxin (BoNT) should decrease neck pain by reducing excessive muscle spindle activity, inhibiting retrograde neuronal flow to the central nervous system, inhibiting release of neuropeptides by nociceptors3, and blocking release of acetylcholine by nerve endings, without interfering with neural conduction4. A single course of treatment could be expected to last for 3 to 4 months5. Our objective was to assess the effect of BoNT intramuscular injections, used alone or with an adjunctive treatment (e.g., physiotherapy, exercise, or additional medication), on pain, function, patient global perceived effect (GPE), and quality of life (QOL) in subacute and chronic NP, including NP accompanied by cervicogenic headache (CGH) and whiplash-associated disorders (WAD), over about 4 weeks (short term) to 6 months (intermediate term).
MATERIALS AND METHODS
Characteristics of included studies
Included studies were randomized controlled trials (RCT) or quasi-RCT of BoNT injections for adults age > 18 years, with NP of any duration (acute, < 30 days; subacute, 30–90 days; chronic, > 90 days) including NP associated with myofascial pain, degenerative changes6, headache (CGH)7, WAD grades I–III8,9, and with10 or without radiculopathy (lower motor neuron signs)9,11,12.
Excluded were studies of NP with long tract (upper motor neuron) signs, infection, or inflammation6, WAD grade IV9, NP grade IV8, and noncervical or “mixed” headache types. BoNT could be compared to placebo or another treatment (e.g., ultrasound), or combined with an additional treatment and compared to placebo plus the additional treatment. Outcomes of interest included pain, function/disability, GPE, and QOL measured by patient self-report or performance tests13,14. Given the expectation that BoNT efficacy would last 3 to 4 months, the analysis was limited to periods of < 6 months.
Search methods
Search databases included Central (The Cochrane Library 2010, issue 6), Medline, and Embase, from beginning to June 2010, without language restriction. Subject headings (MeSH) and key words included anatomical, disorder or syndrome, treatment, and methodological terms. Additional searches included review of article references, personal files, contacts with identified experts, and meeting abstract searches. Authors were contacted for additional unpublished data.
Data collection
At least 2 authors with differing clinical backgrounds independently selected studies, abstracted data, assessed study quality, and evaluated clinical applicability. Agreement was assessed using the quadratic weighted κ statistic, Cicchetti weights15. A third author was consulted in cases of persisting disagreement for all components of data collection. Prepiloted forms were used for all elements of data abstraction, except for the clinical applicability criteria, which were developed for this review based on Cochrane standards16,17.
Data analysis
Descriptive statistics summarized treatment groups, interventions, outcomes, adverse effects, and costs, and used intention-to-treat (ITT) principles. Standard mean differences (SMD) with 95% CI were calculated for continuous data, to accommodate the different outcome measures used. Effect sizes were calculated for continuous outcomes reporting medians18. The minimum clinically important difference (MCID) was assumed to be 10 on a 100-point pain intensity scale19, and 7/50 neck disability index units20. For dichotomous outcomes, relative risks (RR) were calculated, where RR < 1 represents treatment benefit. When data were not extractable and contacted authors did not respond, we used the statistical significance reported in the original study. Data imputation may have been performed. The number needed to treat and the treatment advantage were calculated to indicate the magnitude of treatment effect21 (Table 1).
Assessment of heterogeneity and subgroup analysis
Studies were assessed for heterogeneity prior to combination in metaanalysis, first by consideration of clinical features (symptom duration, NP subtype, intervention and treatment application, and outcome measures) and then by statistical methods (chi-squared test for trend, p > 0.10, I2 < 40%). Results were calculated as pooled SMD (SMDp) or RR using a random-effects model.
Methodological quality assessment
The inherent trial validity was assessed through risk of bias (ROB) evaluations using the updated Cochrane criteria22, and an interprofessional team. The results reported in Table 2 represent group consensus. Studies scoring ≥ 6/12 were deemed to have high validity as assessed by a low ROB. The influences of ROB, duration of pain, and subtypes of NP (WAD, non-WAD, headache, myofascial pain) were assessed by subgroup analysis.
The overall quality of the summarized evidence was evaluated using the Grading of Recommendations Assessment, Development, and Evaluation working group approach, as recommended by Cochrane16,22 (Table 1).
RESULTS
Description of studies
Table 3 represents the 14 trials selected of 147 eligible trials (estimated κ = 0.84, 95% CI 0.75 to 0.94). Only botulinum toxin type A (BoNT-A) was used in the included studies. No trial included neck pain subjects with radicular findings. Of the 8 trials for chronic myofascial NP23,24,25,26,27,28,29,30, 1 did not have extractable data30. There was 1 study of chronic CGH31, 3 studies of subacute and chronic WAD grade I or II32,33,34, 1 study of WAD and chronic CGH35, and 1 study combining chronic WAD and non-WAD36.
The following comparisons were made to BoNT-A: saline in 11 trials23,25,26,28,29,31,32,33,34,35,36, dry needling and lidocaine in 1 trial27, lidocaine in 2 trials24,27, and ultrasound, stretching and lidocaine in 1 trial24.
Cointerventions included exercise24,27, physiotherapy and medications25,28,32, hot packs and massage31, and additional prescription medications25.
Two studies used a crossover design with BoNT-A/saline23,29.
The timeframe of interest meant that data were abstracted from the first period in the Ojala (2006)29 trial and from both periods in Cheshire (1994)23.
Methodological quality
The ROB assessments showed variation in reviewer agreement for study quality (estimated κ = 0.23 to 1.00) with disagreement usually secondary to poor trial reporting. Ten high-quality studies scored ≥ 623,25,26,28,30,32,33,34,35,36, with 4 studies having a high ROB24,27,29,31 (Table 2). Methodological weaknesses in multiple trials included failure to describe or use appropriate randomization (64%, 9/14), improper allocation concealment (57%, 8/14), and ineffective blinding procedures for patients (29%, 4/14), care providers (36% 5/14) and outcome assessors (36% 5/14). Other concerns included dropout rates, inadequate ITT analyses, unexplained baseline differences, and a lack of cointervention standardization across treatment arms. Table 1 summarizes the findings by population, quality of the evidence, and comparison type.
Subacute/chronic NP; BoNT-A vs placebo
High-quality evidence from 4 trials, 183 participants23,26,28,29, showed no short-term statistically significant difference [SMDp −0.21 (95% CI −0.50 to 0.07)] for chronic NP (Figure 1).
Low-quality evidence from 1 trial, 24 participants28, showed no difference up to 6 months for chronic NP.
Very low-quality evidence from 1 trial, 31 participants29, showed a short-term difference in GPE favoring BoNT-A in chronic NP [SMD −1.12 (95% CI −1.89 to −0.36)].
Low-quality evidence from 1 trial, 45 participants36, showed no difference at 6 months in disability or GPE for chronic NP.
WAD; BoNT-A vs placebo
Moderate-quality evidence from 4 trials, 122 participants32,33,34,35, showed no difference up to 4 weeks [SMDp −0.21 (95% CI −0.57 to 0.15)] and low-quality evidence from 1 trial, 19 participants32, demonstrated no difference up to 6 months in pain (Figure 1) or GPE. Low-quality evidence from 2 trials, 63 participants33,35, showed no difference at 4 weeks for disability associated with WAD grade II. Additionally, low-quality evidence from 1 trial, 39 participants34, noted no short-term difference for GPE for chronic WAD grades I or II.
Cervicogenic headache, BoNT-A vs placebo
Very low-quality evidence from 2 trials, 58 participants35,31, showed no difference for short-term pain and one of the trials, 32 participants31 also showed no benefit over 6 months of BoNT-A for CGH. The same trial31 provided low-quality evidence of no difference for CGH-associated disability in the short term.
Combination with exercise and medication
Very low-quality evidence from 3 trials, 114 participants25,28,32, suggested no difference in the short term for chronic NP or WAD [SMD −0.08, (95% CI −0.45 to 0.29)] when BoNT-A was combined with exercise/medication versus exercise/medication alone (Figure 1). However, very low-quality evidence from 2 trials, 43 participants28,32, indicated a difference of 6 months, favoring BoNT-A plus exercise/medication for pain [SMDp −0.66 (95% CI −1.29 to −0.04)] in subacute or chronic NP or subacute WAD (Figure 1).
Combination with exercise
Very low-quality evidence from 1 trial, 36 participants24, demonstrated no short-term difference for chronic NP with BoNT-A plus exercise versus exercise alone.
BoNT-A plus exercise versus dry needling/lidocaine plus exercise
Very low-quality evidence from 1 trial, 19 participants27, showed a short-term difference for pain [SMD −1.03 (95% CI −2.01 to −0.06)] but not for disability or QOL, comparing BoNT-A plus exercise versus dry needling plus exercise, in chronic NP. Very low-quality evidence from 2 trials, 55 participants24,27, showed no short-term difference in pain [SMDp 0.35 (95% CI −0.18 to 0.89)] and 1 trial, 19 participants27, showed no short-term difference in disability or QOL comparing BoNT-A plus exercise with lidocaine plus exercise for chronic NP.
BoNT-A plus exercise versus ultrasound plus exercise
Very low-quality evidence from 1 trial, 36 participants24, demonstrated no short-term difference for BoNT-A plus exercise versus conventional ultrasound plus exercise for chronic NP. This trial showed a difference between BoNT-A plus exercise compared to pain-threshold ultrasound plus exercise [SMD −1.41 (95% CI −2.15 to −0.67)].
Adverse events
Pooled data from the 14 trials reported an adverse event rate estimated at 30% (109/360 participants treated with BoNT-A). Adverse event reports included transient effects of injection site soreness, shoulder or arm weakness, fatigue, heaviness, numbness, flu-like symptoms, systemic fever, shivering, generalized muscle soreness, vertigo, and headache. For comparison intervention participants, mild adverse events were reported and estimated at 21% (71/343 subjects). Adverse event reporting was poorly done in general across trials. Cost of care was not reported in most studies. Strategies recommended to reduce costs included single low-dose injections27 or very restricted patient numbers29. A benchmark cost of $335 US per 100 units was noted37.
Clinical applicability
All studies were assessed to determine whether readers could implement the findings into clinical practice. Six criteria (participant description, interventions, outcomes, relevance, benefits vs harms, timing of evaluation) were developed for this review16,17. Considering this, the study population was adequately described in 13/14 studies. One did not detail exclusion criteria, 8 provided sufficient detail for protocol replication, and 6 omitted necessary details of treatment administration or provider training. Twelve studies evaluated outcomes at clinically sensible times, while 2 studies used a 4-week evaluation period, limiting understanding of intermediate-term treatment effects. In 85% of the studies, the treatment effects were rated as not clinically important based on the MCID standards, and in 100% of the studies, the treatment benefits as reported were not considered greater than the potential harms.
DISCUSSION
In a previous review37 we found moderate evidence of no benefit for BoNT-A over saline for chronic neck pain (7 trials; 270 participants). Since the 2007 publication, data on 7 additional trials, 396 participants, with BoNT-A have been identified. This review found high-quality (4 trials/183 participants)23,26,28,29 and moderate-quality evidence (4 trials/122 participants)32,33,34,35 demonstrating a lack of benefit for BoNT-A over saline injections for subacute or chronic NP in the short term. Very low-quality evidence from 2 studies, 143 participants28,32, showed benefit of BoNT-A plus exercise/medication over placebo plus exercise/medication at 6 months. These results extend the findings of our prior review and suggest that BoNT-A is not effective as a standalone agent in subacute or chronic neck pain. While Colhado, et al4 advocated use of BoNT-A in chronic pain disorders, our data do not support its use as a standalone treatment for chronic neck pain. Our analysis differs from that of Jeynes and Gauci38, who noted that the Gobel study26 reported superior pain benefits for BoNT-A versus placebo at 5–8 weeks. However, analysis of earlier and later time-points shows no significant differences. Given these mixed results, accompanied with a 17% dropout rate, we disagree that “there is level 2A evidence in support of using of BoNT-A in the treatment of myofascial pain”38. Our conclusions are in agreement with the qualitative systematic review of Ho and Tan39, who concluded that the evidence did not support the use of BoNT-A in myofascial pain syndrome. Interestingly, in low back pain, Chou, et al5 noted positive short-term results (3 weeks) for low back pain and disability, with cessation of benefit after 3–4 months. Additionally, we suggest that whether BoNT-A has utility as an adjunctive agent to exercise requires further study. A review by Lang40 suggested that BoNT-A, used as part of a multifaceted approach, may improve the results in chronic pain associated with muscle disorders.
Further, our study estimated transient adverse events at a rate of 30%, consistent with reports by Mejia, et al41 (36%), Kessler, et al42 (22%), and Naumann and Jankovic43 (25%). While BoNT-A appeared generally safe to administer, case reports of allergic reactions, including fatal anaphylaxis, have been reported5.
Whether there is value in transient effects at 5 to 8 weeks is a point of conjecture26. Given the lack of superiority to lidocaine injections, lack of effects earlier and later, as well as costs and transient adverse events, we suggest that BoNT-A is not recommended in the treatment of chronic neck pain, CGH, or WAD. Also, the pharmacological action of BoNT-A is limited to muscle tissue, and does not directly influence the commonly affected articular or neuro-meningeal tissues. Given the varied etiologies in the development of “myofascial” pain, it is possible that poor diagnostics contributed to the limited results of the injections.
Limitations of our study
Some limitations are inherent in the primary literature. For instance, we principally considered pain, since there was limited information on disability, GPE, and QOL. Further, it is not clear that an optimal dose, or the dose-response, has been adequately defined for BoNT-A for NP. Finally, there may be some patient subgroups that do respond, although there is not sufficient information on responders and their prediction from the current data. In spite of the increased number of studies since our prior review, the overall sample size is still limited. Our ability to metaanalyze results was restricted by variable trial quality, insufficient subject numbers, and lack of standardization of adjunctive treatments.
Nevertheless, our review has several strengths. Database searches had no language restriction. At least 2 independent reviewers from diverse professional backgrounds selected studies, minimizing both selection and professional bias. Data abstraction and risk of bias assessment were performed independently and final scores represent the group’s consensus. We also searched extensively for unpublished work, and contacted authors and known experts to find further studies.
We calculated that an additional 2 studies, each having active and placebo study arms of size, n = 100, with a similar mean and SD to the 4 high-quality pooled studies23,26,28,29 (6 studies in total) would be needed to show a statistically significant difference from placebo. This result, however, would still fall below established thresholds for clinically important differences from placebo.
Future trials should define responder criteria a priori as well as examine predictors of response to facilitate patient selection. While the current evidentiary basis is not methodologically compelling enough to recommend BoNT-A plus exercise or medications in the clinic, there is weak evidence that these combinations could be effective. Therefore future studies should explore the combination of BoNT-A with a cointervention such as exercise44 and analgesics, where to date limited data suggest a benefit of this combination, to refute or clarify whether BoNT-A might have clinical value. Different designs would be appropriate, such as the use of a 2 × 2 factorial design with double placebo, exercise alone, BoNT-A alone, and the combination. We might even suggest that a BoNT-A-alone arm is not necessary, since it is known to be ineffective versus placebo, based on our findings.
Inclusion criteria should be tightened to ensure that patients have defined trigger points45 and to reduce the possible confounding factor of age-related degenerative changes in the cervical spine. Collection of outcomes, such as disability and GPE, would help place the pain findings in context. Further, trial reporting must improve, with clear descriptions of randomization, allocation, and blinding procedures, as well as the use of ITT analysis and both qualitative and quantitative analysis of baseline differences and treatment outcomes. Authors should provide sufficient details to allow beneficial interventions to be implemented in the clinic.
BoNT-A intramuscular injections produced pain relief similar to saline for chronic neck pain and for whip-lash-associated disorder, as assessed on pain, function, and patient global perceived effect. Consequently, for these populations, we do not recommend use of BoNT-A in the clinic, either alone or combined with any other therapy. Any further investigation of BoNT-A for neck pain should combine BoNT-A with physiotherapy (exercise) and medication in a well constructed design study, as the potential for a positive result would be more likely and could inform clinical practice.
Acknowledgment
We thank the Faculty of the MClSc (Manipulative Therapy) Program, School of Physical Therapy, Faculty of Health Sciences, The University of Western Ontario, London, Ontario, Canada, for their guidance, and Annie Moran and Lina Santaguida for their consultation on this review.
Footnotes
-
Supported by Lifemark Health.
- Accepted for publication September 20, 2010.
REFERENCES
- 1.
- 2.
- 3.
- 4.
- 5.
- 6.
- 7.
- 8.
- 9.
- 10.
- 11.
- 12.
- 13.
- 14.
- 15.
- 16.
- 17.
- 18.
- 19.
- 20.
- 21.
- 22.
- 23.
- 24.
- 25.
- 26.
- 27.
- 28.
- 29.
- 30.
- 31.
- 32.
- 33.
- 34.
- 35.
- 36.
- 37.
- 38.
- 39.
- 40.
- 41.
- 42.
- 43.
- 44.
- 45.