Botulinum toxin injections for low‐back pain and sciatica

Abstract

Background

Adequate relief from low‐back pain (LBP) is not always possible. Emerging evidence suggests a role for botulinum neurotoxin (BoNT) injections in treating pain disorders. Proponents of BoNT suggest its properties can decrease muscle spasms, ischemia and inflammatory markers, thereby reducing pain.

Objectives

To determine the effects of botulinum toxin injections in adults with LBP.

Search methods

We searched CENTRAL (The Cochrane Library 2009, issue 3) and MEDLINE, EMBASE, and CINAHL to August 2009; screened references from included studies; consulted with content experts and Allergan. We included published and unpublished randomised controlled trials without language restrictions

Selection criteria

We included randomised trials that evaluated BoNT serotypes versus other treatments in patients with non‐specific LBP of any duration.

Data collection and analysis

Two review authors selected the studies, assessed the risk of bias using the Cochrane Back Review Group criteria, and extracted the data using standardized forms. We performed a qualitative analysis due to lack of data.

Main results

We excluded evidence from nineteen studies due to non‐randomisation, incomplete or unpublished data. We included three randomised trials (N =123 patients). Only one study included patients with chronic non‐specific LBP; the other two examined unique subpopulations. Only one of the three trials had a low risk of bias and demonstrated that BoNT injections reduced pain at three and eight weeks and improved function at eight weeks better than saline injections. The second trial showed that BoNT injections were better than injections of corticosteroid plus lidocaine or placebo in patients with sciatica attributed to piriformis syndrome. The third trial concluded that BoNT injections were better than traditional acupuncture in patients with third lumbar transverse process syndrome. Both studies with high risk of bias had several key limitations. Heterogeneity of the studies prevented meta‐analysis. There is low quality evidence that BoNT injections improved pain, function, or both better than saline injections and very low quality evidence that they were better than acupuncture or steroid injections.

Authors' conclusions

We identified three studies that investigated the merits of BoNT for LBP, but only one had a low risk of bias and evaluated patients with non‐specific LBP (N = 31). Further research is very likely to have an important impact on the estimate of effect and our confidence in it. Future trials should standardize patient populations, treatment protocols and comparison groups, enlist more participants and include long‐term outcomes, cost‐benefit analysis and clinical relevance of findings.

Plain language summary

Botulinum toxin injections as a treatment for low‐back pain and sciatica

Back pain is a common symptom affecting roughly 50% of the population every year. For the majority of people, back pain goes away gradually ‐ usually within several weeks.

However, many people will experience another bout of back pain in the future. About  5% to 10% of the population will develop back pain that never goes away. As a result, there is a need for treatments that can provide safe and predictable pain relief.

Botulinum toxin injections ‐ i.e., one or more injections of a drug to temporarily numb or weaken nerves and muscles that might contribute to low‐back pain ‐ are an increasingly popular treatment. Some of the commercial names of botulinum toxin include "Botox", "Lantox", "Myobloc" and "Neurobloc".

Government regulatory agencies, such as the Food and Drug Administration (FDA) in the United States or Health Canada in Canada, have never approved the use of botulinum toxin for low‐back pain.  So the safety and effectiveness of these injections are still open to question.  Rare reports of potentially life‐threatening side effects have prompted Health Canada and the FDA to require warnings on BoNT products.

This review looked at botulinum toxin injections for patients with non‐specific low‐back pain ‐ i.e., back pain without an obvious underlying cause, with or without sciatica ‐ i.e., pain that shoots down the back into the buttocks, leg and often into the foot. It included three randomised controlled clinical trials involving 123 individuals with long‐term back pain, sciatica or both.

Because of the way these trials were designed and carried out, the review concluded that the evidence in favour of botulinum toxin injections is only of low or very low quality. This means that at best, further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.

Background

Description of the condition

Low‐back pain (LBP) is common. Estimates derived from two U.S. National surveys conducted in 2002 reported that 26.4% of adult respondents had at least one day of LBP within the last three months, and about 15% of adults reported LBP lasting longer than two weeks annually (Deyo 2006). Individuals with LBP consumed $90.7 billion in U.S. health care expenditures based on 1998 estimates (Luo 2004). In the United Kingdom, there were similar health care and economic ramifications with LBP being the single largest contributor to absenteeism in the workplace from 1988 to 1989, and costing more than £11 billion (Dagenais 2008).

The vast majority of cases of LBP are self‐limiting and resolve within a few weeks. However, up to 5% to10% of individuals can have persistent symptoms associated with psychological illness and disability (Manchikanti 2009). Several treatment options are available, yet despite this, many individuals are unable to attain adequate relief and seek different therapeutic modalities.

One such option is intramuscular botulinum neurotoxin (BoNT) injections. Freeze‐dried BoNT is reconstituted with saline solution and injected multiple times intramuscularly into the posterior lumbar region. The erector spinae muscle is the most superficially accessible muscle in this area and is located roughly 2 cm from midline of the lumbar spine on either side. Multiple injections on the affected side into the muscle at L1, L2, L3, L4, L5, or a combination of sites, are administered to encompass all or part of the entire mass of muscle. If pain involves the sacral region, injections can also be administered into the paraspinal sacral muscles.

Description of the intervention

BoNT injections block nerve impulses rendering muscles unable to contract. Muscles then enter a state of relaxation, paralysis or both. It can take up to two to four days for BoNT injections to take effect and the effects tend to last from three to six months, at which time injections can be repeated.

Commercially, botulinum toxin is made available in minute doses and distributed as two different serotypes: A and B. Botulinum toxin serotype A (BoNT‐A) is marketed as Botox® (Allergan Inc., Irvine, CA), Dysport® (Ipsen Limited, Wrexham, UK), Lantox® (Lanzhou Institute of Biological Products, Bogota, Columbia) and Xeomin® (Merz Pharma, Frankfurt, Germany), while Botulinum toxin serotype B (BoNT‐B), is marketed as Myobloc® in the U.S. (Solstice Neurosciences, Inc., CA), and Neurobloc® in Europe (Solstice Neurosciences Inc., Malvern, PA).

The Food and Drug Administration (FDA) has approved Botox® for treating cervical dystonia (abnormal involuntary neck posture), axillary hyperhidrosis (extreme armpit sweating), blepharospasm (involuntary eyelid contraction) and glabellar lines (facial wrinkles). Myobloc® is approved for use in cervical dystonia. The 2008 American Academy of Neurology Guidelines have advocated offering botulinum toxin for the treatment of spasticity (increased muscle tone) (Simpson 2008a), select movement disorders (Simpson 2008b) and axillary hyperhidrosis and detrusor overactivity (hyperactive bladder) (Naumann 2008). Emerging evidence may suggest a role for botulinum toxin in treating pain disorders, such as myofascial, neck and back pain (Wheeler 1998; Lang 2000).

Side effects include injection site pain, irritation, or both, muscle weakness and rash. Serious reactions are rare but include anaphylaxis, paralysis, fainting and arrhythmias. There have been recent concerns about the distant spread of the botulinum neurotoxin beyond the treatment site to produce botulism‐like symptoms, such as, severe muscle weakness, trouble swallowing, breathing difficulties and speech disorders. This has prompted Health Canada and the FDA to require black box labelling on BoNT products, warning of this uncommon but potentially life‐threatening side effect.

How the intervention might work

Increased muscle contraction, contractures or spasms are suspected to be at least partially responsible for pain from local ischemia. Muscle injury can stimulate the release of inflammatory mediators and nociceptive substances that contribute to acute focal pain and possibly, through influences at the central nervous system level, contribute to chronic, diffusely distributed pain.

Although the precise mechanisms whereby BoNT injections result in pain relief is presently unknown, BoNT injections lead to muscle relaxation. Furthermore, BoNT injections have been demonstrated to reduce levels of inflammatory markers following injury in animal models (Aoki 2003; Cui 2004).

It is also presently unclear which population with LBP benefits from BoNT injections. It is possible that patients with muscular pain, spasms, or both may experience pain improvement from the muscle relaxation effects of BoNT injections.

Why it is important to do this review

The effectiveness of BoNT in the treatment of myofascial pain syndrome involving different muscle groups throughout the body has been systematically reviewed (Pereda 2006); there is a Cochrane review on cervical dystonia (Costa 2004), and there is a Cochrane review currently underway for the treatment of subacute and chronic neck pain (Peloso 2010). Although there are narrative reviews of the efficacy of BoNT toxin for LBP (Jabbari 2008), no systematic review exists. The current systematic review summarizes the available scientific evidence on the effectiveness of BoNT for LBP.

Objectives

The objective of this review was to determine the effectiveness of botulinum toxin on pain, function, disability (including return‐to‐work) and patient satisfaction in adults with LBP.

Methods

Criteria for considering studies for this review

Types of studies

We considered published and unpublished completed randomised controlled trials (RCTs) with no language restrictions for inclusion.

Types of participants

• Adult subjects (18 years of age or older), male or female with non‐specific LBP and/or sciatica were included.

• LBP was defined as pain occurring between the costal margin or 12^th rib to the inferior gluteal fold. Sciatica was defined as pain in the buttock radiating to the leg and/or feet.

• Non‐specific was defined as LBP where no specific pathological cause, such as infection, metastasis, neoplasm, osteoporosis, fracture, lumbar/sacral radiculopathy, inflammatory arthropathy or disease was identified.

• Patients with acute (less than six weeks), subacute (six to 12 weeks) or chronic (longer than 12 weeks) LBP were included.

Types of interventions

• All BoNT serotypes that were injected intramuscularly after reconstitution were considered for the intervention.

• Injection sites included the posterior lumbar, posterior sacral regions, or both. Unilateral, bilateral injections, or both were evaluated.

• No restrictions were placed on the number of BoNT units administered, the number of sites injected, or the number of injections.

• Comparison groups included sham or placebo intramuscular injections, other therapeutic injections (e.g., epidural corticosteroid injection, acupuncture), other medical treatments, or no treatments.

Types of outcome measures

Primary outcomes

RCTs were included if they reported on at least one of the following patient‐centred outcomes:

• Symptoms (e.g. pain intensity measured by visual analogue scale, subjective improvement of symptoms)

• Disability, including return‐to‐work or work status

• Overall improvement or proportion of patients recovered

• Back‐specific functional status (as measured by Roland Morris Questionnaire, Oswestry Disability Index, etc.)

• Well‐being (as measured by SF‐36, Nottingham Health Profile, Sickness Impact Profile, etc.).

Secondary outcomes

• Secondary outcomes such as physiological outcomes of physical examination (e.g. back range of motion, muscle strength) and general health status were included if the trial did not report on any primary outcome measure

• Satisfaction with care

• Adverse events (intentional and unintentional) were extracted

• Outcomes were reported for different follow‐up periods; short‐ (four weeks), intermediate‐ (four weeks to one year) and long‐term (longer than one year)

Search methods for identification of studies

The Trials Search Co‐ordinator for the Cochrane Back Review Group (CBRG) assisted with the development of the search strategy and conducted the searches using the highly sensitive search strategy described in the 2009 updated method guidelines for systematic reviews in the Cochrane Back Review Group (Furlan 2009).

Electronic searches

Relevant RCTs were identified by computer‐aided searches of the following databases: CENTRAL (The Cochrane Library 2009, Issue 3); MEDLINE (January 1966 to August 2009), EMBASE (January 1980 to August 2009) and CINAHL (January 1982 to August 2009). See Appendix 3 for search strategies.

Searching other resources

Additional search strategies included:

• Contacting editorial base of the CBRG to search the CBRG Trials Register.

• Screening references from identified RCTs and relevant narrative and systematic reviews.

• Searching the International Standard Randomized Controlled Trial Number Register and Current Controlled Trials.

• Personal communication with content experts and authors of identified RCTs.

• Contacting the pharmaceutical company Allergan (maker of Botox). At the time of the study, Botox was the commercially available botulinum toxin with widespread use in Canada, as such, for ease of contact, other manufacturers were not contacted. In future updates, other manufacturers will be contacted.

• Searching the grey literature via Internet search engines Google and Yahoo.

Data collection and analysis

Selection of studies

The CBRG Trials Search Co‐ordinator conducted searches using the search strategies in Appendix 3. One reviewer (ZW) conducted the remaining search using the additional search strategies described in the previous section. All studies identified from the searches were entered into RefWorks with duplicate studies eliminated. Three independent review authors (ZW, FI and CB) piloted the inclusion criteria, applying them to five ineligible abstracts to identify and clarify any misinterpretation in the application of inclusion criteria. Two of the reviewer authors (ZW, FI) then independently applied the inclusion criteria to all the titles and abstracts of the identified papers to select studies for inclusion. If eligibility of the study was unclear from the abstract, the full text of the article was obtained and assessed. Disagreements were resolved by discussion, otherwise, a third reviewer (AF) settled the dispute. Reasons for excluding trials that were retrieved in full text but subsequently excluded are listed in the Characteristics of Excluded Studies table.

Data extraction and management

Two review authors (ZW, CB) pilot tested data extraction from an included study from the Cochrane Review of BoNT for cervical dystonia (Costa 2004) and compared the data extracted to results found in the actual review. This helped ensure agreement on the use and interpretation of the data extraction form. Two review authors (ZW, CB) independently performed the data extraction. Data describing the characteristics of (i) the study population (e.g., total number of patients, number of patients in each arm, age, gender, location and duration of LBP) and (ii) of the BoNT injection and reference intervention (e.g., type of reference intervention, sites injected, unilateral or bilateral sites targeted, number of injections, number of BoNT units, BoNT serotype) were extracted. Also, data on the previously described primary and secondary outcomes including adverse effects were extracted. We attempted to retrieve missing data by contacting the study authors directly.

Assessment of risk of bias in included studies

Two review authors (ZW, FI) pilot tested the risk of bias assessment on an included study from the Cochrane Review of BoNT for cervical dystonia (Costa 2004) and compared the data extracted to results found in the actual review. This helped ensure agreement on the use and interpretation of the risk of bias assessment criteria. The risk of bias of the selected studies was assessed independently by two review authors (ZW, FI) using the 12 criteria presented in the 2009 Updated method guidelines for systematic reviews in the Cochrane Back Review Group (Appendix 1; Furlan 2009). Each criterion was scored as yes, unclear or no and reported in the Risk of Bias table. Studies were rated as having a “low risk of bias” when at least six of 12 criteria were met and the study contained no serious flaws, such as large imbalances in drop‐out rates, or severe imbalances in baseline prognostic factors, not accounted for in the analysis. Studies with serious flaws or those meeting fewer than six of the 12 criteria were rated as having a “high risk of bias”. Disagreements in the risk of bias assessment were resolved with discussion, with persistent disputes subject to assessment by another review author (AF). Blinding of authors, institution or journal source was not done.

Measures of treatment effect

All quantitative results were entered into RevMan 5.0. The plan was to combine the outcome measures from the individual trials through meta‐analysis where possible, depending upon homogeneity between studies and sufficient presentation of data.

The following questions were utilized to assess the clinical relevance of each included study: 
 1. Are the patients described in detail so that you can decide whether they are comparable to those that you see in your practice? 
 2. Are the interventions and treatment settings described well enough so that you can provide the same for your patients? 
 3. Were all clinically relevant outcomes measured and reported? 
 4. Is the size of the effect clinically important? (30% on Visual Analogue Scale/Numerical Rating Scale for pain, or two to three points on the Roland‐Morris Disability Questionnaire for function, will be considered clinically significant for the respective outcomes) 
 5. Are the likely treatment benefits worth the potential harms?

Unit of analysis issues

Group data reported in the included studies were used as the unit of analysis

Dealing with missing data

In the event that studies lacked data or information, the authors were contacted directly for clarification. Therefore, the authors of Foster 2001 and Fishman 2002 were contacted for clarification (see results).

Assessment of heterogeneity

We assessed the homogeneity of the trials in terms of population, intervention, outcomes, and timing of measurements. Had these been similar in at least two studies, we had intended to pool the results statistically.

Assessment of reporting biases

An assessment of reporting biases was performed using the item in the updated method guidelines (Furlan 2009): Are reports of the study free of suggestion of selective outcome reporting? Defintions from the updated method guidelines (Appendix 1; Furlan 2009) were used to determine if a "yes" could be assigned. We were unable to perform a more general reporting bias analysis, such as funnel plots, given the lack of included studies.

Data synthesis

The quality of the evidence for each outcome across studies was assessed using the GRADE approach, which is based on performance against five domains: limitations of study design (risk of bias), consistency of results, directness (generalizability), precision of data and reporting bias (Atkins 2004; Furlan 2009).

• High quality evidence = at least 75% of the RCTs with no limitations of study design have consistent findings, are generalizable to the population in question, with precise data and no known or suspected publication biases. Further research is very unlikely to change the level of evidence or the estimate of effect.

• Moderate quality evidence = one of the domains is not met. Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.

• Low quality evidence = two of the domains are not met. Further research is very likely to have an important impact on the estimate of effect and is likely to change our confidence in it.

• Very low quality evidence = three of the domains are not met. There is great uncertainty about the estimate.

• No evidence = no RCTs identified that addressed the outcome

Subgroup analysis and investigation of heterogeneity

We had planned to perform subgroup analysis based on the dose of the BoNT injection. However, since we only had one trial for each comparison, these analyses were not completed.

Sensitivity analysis

We had planned to see if the conclusions of the review were effected by modifying the definition of "low risk of bias", however,since we only had one trial for each comparison, these analyses were not completed.

Results

Description of studies

See Characteristics of included studies and Characteristics of excluded studies tables.

Three trials (123 participants) were selected (Foster 2001; Fishman 2002; Liu 2008). One study was in Chinese (Liu 2008) and was assessed and extracted by a review author recommended by the CBRG (KT) who was able to translate. Unlike the other two studies Liu 2008 was not dual‐reviewed as enlisting another reviewer who was able to translate was not possible. Another study (Herskowitz 2004), was considered for inclusion, however, was only available as an unpublished abstract. Despite the author forwarding additional material for review, a complete assessment of the study could not be made from the available data.

Although the subjects included in the trials all had chronic LBP, they differed regarding age distribution and diagnosis; Fishman 2002 enrolled subjects with electrophysiologic evidence of piriformis syndrome, Foster 2001 included those with lateralizing low‐back pain and Liu 2008 considered patients who met the criteria for diagnosis of third lumbar transverse process syndrome.

Although all three trials used Botulinum toxin serotype A (BoNT‐A), there were differences with respect to BoNT preparation, dose, number of injections, muscles targeted, injectant volume and needle used to administer the intervention. Fishman 2002 injected 300 units of BoNT‐A, Foster 2001 used 200 units and Liu 2008 injected 100 units. Fishman 2002 targeted an electromyographically identified motor point in the piriformis muscle with four injection sites using two millilitres of solution with a 3.5 inch teflonized needle. Foster 2001 injected five lumbosacral paraspinal muscles with a total of two millilitres using a 27 gauge needle. Liu 2008 infiltrated an unspecified number of trigger points in the back and did not specify the volume of injectant or type of needle used.

In one study (Fishman 2002), subjects were given BoNT injections plus a standard physical therapy protocol twice weekly for twelve weeks, while the control group only received standard physical therapy.

Two studies (Foster 2001; Fishman 2002) used pain intensity as a primary outcome measure. Foster 2001 also administered the Oswestry Low‐back Pain Questionnaire to subjects. Liu 2008 assessed categorical data with four levels of improvement defined by pain and function. Two studies (Foster 2001; Fishman 2002) reported short and intermediate follow‐up, while Liu 2008 reported intermediate follow‐up only. The timing of outcome measurement was different; Fishman 2002 assessed subjects every two weeks after the intervention for a twelve‐week period, Foster 2001 followed patients at the three‐ and eight‐week marks and Liu 2008 evaluated patients two months after the injections. Two studies (Foster 2001; Liu 2008) commented on adverse effects, while none of the trials made reference to the cost of intervention or work‐related outcomes.

Results of the search

The initial search yielded one hundred and thirty‐four potential references, from which twenty‐two studies were selected. After application of the inclusion criteria, three studies were chosen as the included studies for the review, and nineteen were excluded.

Included studies

See Characteristics of included studies table.

Three studies were included (Foster 2001; Fishman 2002; Liu 2008).

Excluded studies

See Characteristics of excluded studies table.

Nineteen studies were considered but excluded after further review. Fourteen of these studies were excluded because they were not randomised, four did not report results of back pain subjects separately and one was unpublished, with incomplete data.

Risk of bias in included studies

Twelve criteria, recommended by the CBRG (Appendix 1) were applied to each of the included studies. The authors of Foster 2001 were contacted directly for clarity on their blinding method and the authors of Fishman 2002 were contacted for clarity on blinding, identity of outcome assessors and follow‐up. One study (Foster 2001) had a low risk of bias, while the other two (Fishman 2002; Liu 2008) had a high risk of bias. Foster 2001 fulfilled ten of the twelve criteria. Intention‐to‐treat analysis was incomplete as drop‐outs were unaccounted for in the final results. There were serious concerns with Fishman 2002, given the method of randomisation, lack of blinding of providers and assessors, unequal baseline characteristics, drop‐out rate, incomplete outcome data and different timing of outcome assessments. Only two criteria were satisfied as the remaining items were unclear from information provided in the study. Liu 2008 satisfied five criteria as the treatment allocation was not concealed, patients and providers were not blinded. There was lack of clarity regarding the randomisation process, blinding of assessors, selective outcome reporting and co‐interventions. See Figure 1.

1.

A summary table of review authors' judgements for each risk of bias item for each study.

Effects of interventions

All three studies evaluated patients with chronic LBP. Therefore, there is no evidence for BoNT injections in the acute and sub‐acute periods.

Two studies compared BoNT‐A to control injections (Foster 2001; Fishman 2002). Fishman 2002 also included a third group, lidocaine/corticosteroid, that received a 2 ml injection containing 1.5 ml 2% lidocaine and 20 mg triamcinolone acetonide in a 0.5 ml solution. Liu 2008 compared BoNT‐A to acupuncture.

Patients were not followed long‐term. In the lone study with a low‐risk of bias (Foster 2001), outcomes were assessed at eight weeks. Fishman 2002 followed patients for twelve weeks and Liu 2008 followed patients for eight weeks. Therefore, there are no data on the long‐term benefits and harms of BoNT injections for LBP.

Planned subgroup analysis based on dose of therapy could not be performed due to heterogeneity among the populations evaluated in the three studies. Only one study (Foster 2001) included patients with 'standard' chronic LBP, while the other two examined unique subpopulations; one study (Fishman 2002) only included patients with sciatica due to piriformis syndrome and another study (Liu 2008) only involved patients with chronic LBP from third lumbar transverse process syndrome.

1) BoNT‐A injections compared to control injections

One study with high risk of bias (67 subjects; Fishman 2002) showed that BoNT‐A injections in conjunction with physical therapy were significantly superior to saline injections in conjunction with physical therapy on measurements of pain intensity for patients with chronic LBP from piriformis syndrome. Pain was assessed on a visual analogue scale two weeks after the intervention and every two weeks thereafter, for a twelve‐week period. The percentage of subjects with 50% or more improvement on VAS at each of the last two visits was 65% (13/21) in the BoNT‐A group compared to 6% (1/15) in the saline group.

One study with low risk of bias (31 subjects; Foster 2001) showed that BoNT‐A injections were significantly better than control injections on measurements of pain intensity and improved function for patients with chronic LBP. Pain was measured on a visual analogue scale. At three weeks follow‐up, the percentage of participants with pain relief exceeding 50% on VAS was 73.3% (11/15) of subjects injected with BoNT‐A versus 25% (4/16) receiving placebo (P = 0.012). At eight weeks, 60% (9/15) of subjects in the BoNT‐A group versus 12.5% (2/16) in the control group experienced pain relief exceeding 50% on VAS (P = 0.009). Function was measured using the Oswestry Low‐back Pain Questionnaire (OLBPQ). At eight weeks, improvement in scores was seen in 66.7% (10/15) of those injected with BoNT‐A compared to 18.8% (3/14) in the control group (P = 0.011).

It was not possible to combine the two studies in a meta‐analysis as Fishman 2002 evaluated a very specific subset of patients with LBP, while Foster 2001's inclusion criteria were broader.

There was low quality evidence in the short term, and very low quality in the intermediate term, that BoNT‐A injections reduced pain intensity better than saline injections in participants with chronic LBP. There was no evidence for long‐term improvement in pain intensity.

There was also low quality evidence that BoNT‐A injections improved function better than saline injections in the intermediate term. Again, there was no evidence for long‐term functional improvement.

2) BoNT‐A injections compared to other treatments

2a) BoNT‐A injections compared to corticosteroid injection

One study with high risk of bias (67 subjects; Fishman 2002) showed that BoNT‐A injections in conjunction with physical therapy were significantly better than triamcinolone injections with physiotherapy and lidocaine injections with physiotherapy on measurements of pain intensity for patients with chronic LBP from piriformis syndrome. The percentage of subjects with 50% or more improvement on VAS at each of the last two visits was 65% (13/21) in the BoNT‐A group compared to 32% (10/31) in the triamcinolone/lidocaine group.

There was very low quality evidence that BoNT‐A injections were better than corticosteroid injections for reducing pain intensity in chronic LBP in the short term. There was no evidence on intermediate or long‐term improvement in pain intensity.

There was no evidence on the effects of BoNT‐A injections compared to corticosteroid injections for functional improvement in patients with LBP.

2b) BoNT injections compared to acupuncture

One study with high risk of bias (25 subjects; Liu 2008) showed that BoNT‐A injections were significantly better than traditional acupuncture in improving pain and function among subjects with chronic LBP from third lumbar transverse process syndrome. Pain and function were evaluated using categorical data with four levels of improvement (non‐validated tool). Between‐group comparisons showed the BoNT‐A group with greater improvement in pain and function than the acupuncture group (P < 0.05).

There was very low quality evidence that BoNT‐A injections were better than acupuncture for reducing pain intensity in chronic LBP in the intermediate‐term. There was no evidence on the short or long‐term improvement in pain intensity with the use of BoNT‐A injections compared to acupuncture for LBP.

There was also very low quality evidence that BoNT‐A injections improved function better than acupuncture in the intermediate‐term. There was no evidence on the use of BoNT‐A injections compared to acupuncture for short or long‐term functional improvement.

Adverse Events

Two studies (Foster 2001; Fishman 2002) reported pain immediately after injection but no other side effects. In Foster 2001, two subjects in the control group had worsening pain after saline injections; no patients in the BoNT‐A group reported any problems. In Liu 2008, an unspecified number of patients in the BoNT‐A group reported injection site pain after the intervention, but no other adverse events.

Costs and Work‐Related Outcomes

None of the trials made reference to the cost of intervention or work‐related outcomes.

Discussion

BoNT injection is increasingly being used as a therapeutic modality for the treatment of LBP although there is at best little or no evidence that it benefits patients. Despite an extensive search, only three trials met our inclusion criteria and were selected for this review.

Summary of main results

Only one of the three studies had a low risk of bias. Foster 2001 (low‐risk of bias) demonstrated that BoNT injections were superior to saline for pain at three and eight weeks and improved function at eight weeks, but only had 31 participants. Fishman 2002 (high risk of bias) showed that BoNT injections were better than lidocaine or placebo for pain in piriformis syndrome. Liu 2008 (high risk of bias) concluded that BoNT injections were better than traditional acupuncture.

Overall completeness and applicability of evidence

Meta‐analysis was not possible given the heterogeneity of the trials; patient populations were different, differences existed with respect to BoNT injection preparation, dose, site injected, needle used to inject, number of injections, volume of injectate and comparison groups. None of the trials attempted to provide a cost‐benefit analysis which, given the expenses associated with BoNT injection, would have been useful. Only one study (Foster 2001) considered functional outcomes. None of the studies commented on the clinical relevance of the findings. Both Foster 2001 and Fishman 2002 showed statistically significant differences in pain relief exceeding 50% on the VAS, but how much the pain improved was uncertain. Since the absolute changes in VAS were unavailable, an extrapolation of the clinical significance of the findings could not be made.

It should be noted that evidence from nineteen studies was not included due to non‐randomisation, lack of reported data and incomplete or unpublished data.

Quality of the evidence

Although Foster 2001 was a study with a low risk of bias, findings from this study should be balanced against its small patient numbers (N = 31) and lack of long term follow‐up.

There were serious concerns with Fishman 2002, given the method of randomisation, lack of blinding, unequal baseline characteristics of control and treatment groups, high drop‐out rate, incomplete outcome data and inconsistency of patient follow‐up for reporting of pain scores in the final two weeks.

Liu 2008 also had key methodological limitations. There was lack of clarity regarding the randomisation process, blinding of assessors, selective outcome reporting and co‐interventions.

Potential biases in the review process

Trials in all languages were considered and one non‐English trial (Liu 2008) met the inclusion criteria. Unpublished studies were also considered and although one was closely examined (Herskowitz 2004), the lack of complete data led to its subsequent exclusion.

Agreements and disagreements with other studies or reviews

This is the first systematic review that attempts to determine the efficacy of BoNT injections for LBP, as such, no comparisons can be made to previous studies.

Clinical Practice Guidelines from the American Pain Society (Chou 2009) state that there was insufficient evidence to properly evaluate the merits of BoNT injections for LBP.

The European guidelines for the management of chronic nonspecific low‐back pain (Airaksinen 2006) indicated limited evidence for BoNT injections in the treatment of chronic LBP.

Authors' conclusions

Implications for practice.

There is a lack of high quality studies evaluating BoNT injections for patients with standard LBP. Among the studies that exist, there is significant heterogeneity in trial design and outcome parameters. The current body of evidence does not support the use of BoNT injections to improve pain or function in patients with LBP. There is only low quality evidence that BoNT injections are more effective than saline or corticosteroid injections or acupuncture for reducing low‐back pain. Therefore, future research is very likely to change the results and our confidence in them. The present literature has yet to address the long term benefits of BoNT injections or the cost‐benefits of this therapy. Finally, published studies have not addressed how pain relief from BoNT injections translates into clinically relevant outcomes for patients with LBP.

Implications for research.

There is a need for more high quality studies evaluating BoNT injections for LBP. Future trials should attempt to standardize patient populations, treatment protocols and comparison groups while including more participants. Future trials should also attempt long‐term patient follow‐up to assess the long‐term implications of BoNT injections. In addition, cost‐benefit analysis and commentary on the clinical relevance of findings is needed in future studies.

Appendices

Appendix 1. Criteria and operationalisation for Risk of Bias Assessment ‐ RCTs and CCTs

1. Was the method of randomisation adequate? A random (unpredictable) assignment sequence. Examples of adequate methods are coin toss (for studies with two groups), rolling a dice (for studies with two or more groups), drawing of balls of different colours, drawing of ballots with the study group labels from a dark bag, computer‐generated random sequence, pre‐ordered sealed envelops, sequentially‐ordered vials, telephone call to a central office, and pre‐ordered list of treatment assignments

Examples of inadequate methods are: alternation, birth date, social insurance/security number, date in which they are invited to participate in the study, and hospital registration number

2. Was the treatment allocation concealed? Assignment generated by an independent person not responsible for determining the eligibility of the patients. This person has no information about the persons included in the trial and has no influence on the assignment sequence or on the decision about eligibility of the patient.

Was knowledge of the allocated interventions adequately prevented during the study? 
 3. Was the patient blinded to the intervention? 
 This item should be scored “yes” if the index and control groups are indistinguishable for the patients or if the success of blinding was tested among the patients and it was successful. 

4. Was the care provider blinded to the intervention? This item should be scored “yes” if the index and control groups are indistinguishable for the care providers or if the success of blinding was tested among the care providers and it was successful 

5. Was the outcome assessor blinded to the intervention? Adequacy of blinding should be assessed for the primary outcomes.  This item should be scored  “yes” if the success of blinding was tested among the outcome assessors and it was successful or:

• for patient‐reported outcomes in which the patient is the outcome assessor (e.g., pain, disability): the blinding procedure is adequate for outcome assessors if participant blinding is scored “yes”

• for outcome criteria assessed during scheduled visit and that supposes a contact between participants and outcome assessors (e.g., clinical examination): the blinding procedure is adequate if patients are blinded, and the treatment or adverse effects of the treatment cannot be noticed during clinical examination

• for outcome criteria that do not suppose a contact with participants (e.g., radiography, magnetic resonance imaging): the blinding procedure is adequate if the treatment or adverse effects of the treatment cannot be noticed when assessing the main outcome

• for outcome criteria that are clinical or therapeutic events that will be determined by the interaction between patients and care providers (e.g., co‐interventions, hospitalisation length, treatment failure), in which the care provider is the outcome assessor: the blinding procedure is adequate for outcome assessors if the item for care provider is scored “yes”

• for outcome criteria that are assessed from data of the medical forms: the blinding procedure is adequate if the treatment or adverse effects of the treatment cannot be noticed on the extracted data

Were incomplete outcome data adequately addressed? 
 6. Was the drop‐out rate described and acceptable? The number of participants who were included in the study but did not complete the observation period or were not included in the analysis must be described and reasons given. If the percentage of withdrawals and drop‐outs does not exceed 20% for short‐term follow‐up and 30% for long‐term follow‐up and does not lead to substantial bias a 'yes' is scored. (N.B. these percentages are arbitrary, not supported by literature).

7. Were all randomised participants analysed in the group to which they were allocated? All randomised patients are reported/analysed in the group they were allocated to by randomizations for the most important moments of effect measurement (minus missing values) irrespective of non‐compliance and co‐interventions.   

8. Are reports of the study free of suggestion of selective outcome reporting? In order to receive a ‘yes’, the review author determines if all the results from all pre‐specified outcomes have been adequately reported in the published report of the trial.  This information is either obtained by comparing the protocol and the report, or in the absence of the protocol, assessing that the published report includes enough information to make this judgment.

Other sources of potential bias: 
 9.  Were the groups similar at baseline regarding the most important prognostic indicators? In order to receive a “yes”, groups have to be similar at baseline regarding demographic factors, duration and severity of complaints, percentage of patients with neurological symptoms, and value of main outcome measure(s).

10. Were co‐interventions avoided or similar? This item should be scored “yes” if there were no co‐interventions or they were similar between the index and control groups.

11. Was the compliance acceptable in all groups? The reviewer determines if the compliance with the interventions is acceptable, based on the reported intensity, duration, number and frequency of sessions for both the index intervention and control intervention(s). For example, physiotherapy treatment is usually administered over several sessions; therefore it is necessary to assess how many sessions each patient attended. For single‐session interventions (for ex: surgery), this item is irrelevant.

12. Was the timing of the outcome assessment similar in all groups? Timing of outcome assessment should be identical for all intervention groups and for all important outcome assessments.

Appendix 2. Assessment of Clinial Relevance

1.    Are the patients described in detail so that you can decide whether they are comparable to those that you see in your practice?

2.    Are the interventions and treatment settings described well enough so that you can provide the same for your patients?

3.    Were all clinically relevant outcomes measured and reported?

4.    Is the size of the effect clinically important?

5.    Are the likely treatment benefits worth the potential harms?

Appendix 3. Search Strategies

CENTRAL 
 #1        MeSH descriptor Back explode all trees 
 #2        MeSH descriptor Buttocks, this term only 
 #3        MeSH descriptor Leg, this term only 
 #4        MeSH descriptor Back Pain explode tree 1 
 #5        MeSH descriptor Back Injuries explode all trees 
 #6        MeSH descriptor Low Back Pain, this term only 
 #7        MeSH descriptor Sciatica, this term only 
 #8        (low next back next pain) 
 #9        (lbp) 
 #10      (#1 OR #2 OR #3 OR #4 OR #5 OR #6 OR #7 OR #8 OR #9) 
 #11      MeSH descriptor Botulinum Toxins explode all trees 
 #12      MeSH descriptor Botulinum Toxin Type A explode all trees 
 #13      MeSH descriptor Clostridium botulinum explode all trees 
 #14      botulin* toxin* 
 #15      clostridium botulin* 
 #16      Botox* 
 #17      Dysport* 
 #18      oculinum* 
 #19      (#11 OR #12 OR #13 OR #14 OR #15 OR #16 OR #17 OR #18) 
 #20      (#10 AND #19)

MEDLINE

1    randomised controlled trial.pt. 
 2    controlled clinical trial.pt. 
 3    randomized.ab. 
 4    placebo.ab,ti. 
 5    drug therapy.fs. 
 6    randomly.ab,ti. 
 7    trial.ab,ti. 
 8    groups.ab,ti. 
 9    or/1‐8 
 10  (animals not (humans and animals)).sh. 
 11  9 not 10 
 12  dorsalgia.ti,ab. 
 13  exp Back Pain/ 
 14  backache.ti,ab. 
 15  (lumbar adj pain).ti,ab. 
 16  coccyx.ti,ab. 
 17  coccydynia.ti,ab. 
 18  sciatica.ti,ab. 
 19  sciatica/ 
 20  spondylosis.ti,ab. 
 21  lumbago.ti,ab. 
 22  exp low back pain/ 
 23  exp Back Injuries/ 
 24  or/12‐22 
 25  11 and 24 
 26  exp Botulinum Toxins/ 
 27  exp Clostridium botulinum/ 
 28  botulin* toxin*.mp. 
 29  botox*.mp. 
 30  dysport*.mp. 
 31  oculinum*.mp. 
 32  or/26‐31 
 33  25 and 32

EMBASE 
 1    Clinical Article/ 
 2    exp Clinical Study/ 
 3    Clinical Trial/ 
 4    Controlled Study/ 
 5    Randomized Controlled Trial/ 
 6    Major Clinical Study/ 
 7    Double Blind Procedure/ 
 8    Multicenter Study/ 
 9    Single Blind Procedure/ 
 10  Phase 3 Clinical Trial/ 
 11  Phase 4 Clinical Trial/ 
 12  crossover procedure/ 
 13  placebo/ 
 14  or/1‐13 
 15  allocat$.mp. 
 16  assign$.mp. 
 17  blind$.mp. 
 18  (clinic$ adj25 (study or trial)).mp. 
 19  compar$.mp. 
 20  control$.mp. 
 21  cross?over.mp. 
 22  factorial$.mp. 
 23  follow?up.mp. 
 24  placebo$.mp. 
 25  prospectiv$.mp. 
 26  random$.mp. 
 27  ((singl$ or doubl$ or trebl$ or tripl$) adj25 (blind$ or mask$)).mp. 
 28  trial.mp. 
 29  (versus or vs).mp. 
 30  or/15‐29 
 31  14 and 30 
 32  human/ 
 33  Nonhuman/ 
 34  exp ANIMAL/ 
 35  Animal Experiment/ 
 36  33 or 34 or 35 
 37  32 not 36 
 38  31 not 36 
 39  37 and 38 
 40  38 or 39 
 41  dorsalgia.mp. 
 42  back pain.mp. 
 43  exp BACKACHE/ 
 44  (lumbar adj pain).mp. 
 45  coccyx.mp. 
 46  coccydynia.mp. 
 47  sciatica.mp. 
 48  exp ISCHIALGIA/ 
 49  spondylosis.mp. 
 50  lumbago.mp. 
 51  exp Low back pain/ 
 52  back injuries.mp. 
 53  or/41‐52 
 54  53 and 40 
 55  exp botulinum toxin/ 
 56  exp clostridium botulinum/ 
 57  exp botulinum toxin A/ 
 58  botulin* tox*.mp. 
 59  botox*.mp. 
 60  dysport*.mp. 
 61  oculinum*.mp. 
 62  or/55‐61 
 63  62 and 54

CINAHL 
 S47 S40 and S46 
 S46 S41 or S42 or S43 or S44 or S45 
 S45 dysport* 
 S44 botox* 
 S43 botulin* toxin* 
 S42 (MH "Clostridium") 
 S41 (MH "Botulinum Toxins") 
 S40 S23 and S39 
 S39 S24 or S25 or S26 or S27 or S28 or S29 or S30 or S31 or S32 or S33 or S34 or S35 or S36 or S37 or S38 
 S38 lumbar W1 pain 
 S37 "lumbago" 
 S36 (MH "Spondylolysis") 
 S35 (MH "Spondylolisthesis") 
 S34 (MH "Thoracic Vertebrae") 
 S33 (MH "Lumbar Vertebrae") 
 S32 coccydynia 
 S31 "sciatica" 
 S30 "coccyx" 
 S29 (MH "Sciatica") 
 S28 (MH "Coccyx") 
 S27 "backache" 
 S26 (MH "Low Back Pain") 
 S25 (MH "Back Pain+") 
 S24 dorsalgia 
 S23 S21 not S22 
 S22 (MH "Animals+") 
 S21 S20 or S19 or S18 or S17 or S16 or S15 or S14 or S13 or S12 or S11 or S10 or S9 or S8 or S7 or S6 or S5 or S4 or S3 or S2 or S1 
 S20 "volunteer*" 
 S19 prospectiv* 
 S18 "control*" 
 S17 "follow‐up stud*" 
 S16 (MH "Prospective Studies+") 
 S15 (MH "Evaluation Research+") 
 S14 (MH "Comparative Studies") 
 S13 "latin square" 
 S12 (MH "Study Design+") 
 S11 (MH "Random Sample+") 
 S10 "random*" 
 S9    "placebo*" 
 S8    (MH "Placebos") 
 S7    (MH "Placebo Effect") 
 S6    "triple‐blind" 
 S5    "single‐blind" 
 S4    "double‐blind" 
 S3    ""clinical W8 trial"" 
 S2    "randomi?ed controlled trial*" 
 S1    (MH "Clinical Trials+")

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Fishman 2002.

Methods	Subjects with buttock tenderness and sciatica considered for study with those having EMG diagnosis of piriformis syndrome being randomised to one of three groups: Botox, Triamcinolone + Lidocaine, Control.
Participants	87 subjects with piriformis syndrome randomised, 67 completed study
Interventions	Each subject was given standard physical therapy protocol twice weekly for 12 weeks. Injection into motor point (spot in muscle giving maximal response with minimal stimulus) of piriformis muscle medial to its musculotendinous junction in the affected buttock with 3.5 inch teflonized needle under electrophysiologic guidance. Botox Group: 2 ml injection containing 300 U Botox in preservative free normal saline Triamcinolone + Lidocaine (T/L) Group: 2 ml injection containing 1.5 ml 2% lidocaine and 20 mg triamcinolone acetonide in a 0.5 ml solution Control Group: 2 ml preservative free normal saline
Outcomes	Visual analogue score measured at baseline then at follow‐up exams at 2 week intervals for 12 weeks. Percentage of subjects with 50% or more improvement on VAS at each of last two visits: T/L 32% (10/31), Botox 65% (13/21), Placebo 6% (1/15) Statistical Significance of study results: T/L vs. placebo X² 2.66, P 0.10 Botox vs placebo X² 10.56, P 0.001 T/L vs Botox X² 4.02, P 0.044
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	High risk	Injections of the three substances were randomised by the nurse loading the syringe, reading the last digit of a random number chart: 1–6 BOTOX, 7 and 8 steroid and lidocaine, 9 and 0 placebo.
Allocation concealment?	Unclear risk	Unclear from text
Blinding? All outcomes ‐ patients	Unclear risk	Unclear from text
Blinding? All outcomes ‐ providers?	High risk	Blinded 75% of the time
Blinding? All outcomes ‐ outcome assessors?	High risk	Blinded 90% of the time
Incomplete outcome data addressed? All outcomes ‐ Withdrawal/drop‐out rate?	High risk	67 of 87 patients completed the study
Incomplete outcome data addressed? All outcomes ‐ intention‐to‐treat analysis?	High risk	All patients were analysed in the groups they were assigned by randomisation. Drop‐out patients unaccounted for.
Free of selective reporting?	Low risk	All pre‐specified outcomes adequately reported, effect size could not be calculated
Groups similar at baseline?	High risk	placebo control group different baseline characteristics (older age)
Co‐interventions avoided or similar?	Low risk	The sole difference between the groups was the injectate received
Compliance acceptable?	Unclear risk	Unclear from text
Timing of outcome assessment similar?	High risk	Although attempted to assess subjects every two weeks for a twelve week period, subjects occasionally failed to show and follow‐up telephone interviews unsuccessful

Foster 2001.

Methods	Subjects with LBP between L1‐S1 of 6 months or greater and with pain laterality were recruited and randomised to two groups: Botox or control.
Participants	31 subjects with chronic low‐back pain were randomised with a total of 28 completing the study
Interventions	Using 1 CC tuberculin syringe and 27 gauge needle, Injections given at five lumbar levels (L1‐L5), or if pain involved sacral region lumbosacral injections (L2‐S1) given. All patients injected unilaterally, on side of pain or pain predominance. Botox: Each site received 40 Units for a total of 200 Units Saline: Same volume and sites injected in the control group
Outcomes	Visual analogue scores were assessed at baseline, and at 3 and 8 weeks following intervention. Oswestry low‐back pain questionnaire was given at baseline and 8 weeks post intervention. 3 weeks: > 50% pain relief (VAS scores) Botox 73.3% (11/15) vs. Saline 25% (4/16), P = 0.012. Difference between groups 48% (95% CI, 11.7% to 80.1%) 8 weeks: > 50% pain relief (VAS scores) Botox 60% (9/15) vs. Saline 12.5% (2/16), P = 0.009 Difference between groups 47.5% (95% CI, 10.5% to 79.1%) Improvement in OLBPQ at 8 weeks Botox 66.7% (10/15) vs. Saline 18.8% (3/14), P=0.011 Difference between groups 47.9% (95% CI, 10.6%‐79.6%)
Notes	2 patients in saline group reported worsening of pain immediately following injection No other side effects reported
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Low risk	A nurse clinician not familiar with the patients randomly assigned them to Botox (Allergan, Inc.) and placebo groups by drawing a card from a deck of shuffled card that included equal number of Botox and saline cards
Allocation concealment?	Low risk	The allocation was concealed by calling a central trial office that kept the allocation data in sequential numbered and sealed folders
Blinding? All outcomes ‐ patients	Low risk	Patients were blinded to the content of injected solution
Blinding? All outcomes ‐ providers?	Low risk	The physician performing the outcome assessments was blinded to the content of the injection
Blinding? All outcomes ‐ outcome assessors?	Low risk	The physician who examined and rated the patients, and other physicians involved in the study were blinded to the content of the injected solution
Incomplete outcome data addressed? All outcomes ‐ Withdrawal/drop‐out rate?	High risk	28 of 31 patients completed the study: 14 of 15 patients in Botox group, 14/16 patients in placebo group. The drop‐out rate was close to 10% and in this small study represents a potential risk of bias
Incomplete outcome data addressed? All outcomes ‐ intention‐to‐treat analysis?	High risk	All patients were analysed in the groups they were assigned by randomisation. 3 drop‐outs were not analysed, only completers were analysed. An intention‐to‐treat analysis was not performed
Free of selective reporting?	Low risk	All primary outcome variables were specified before un‐blinding and data analysis. All proposed data were reported in the final results. Although an effect size could be calculated, we chose not to because a meta‐analysis was not being performed.
Groups similar at baseline?	Low risk	Demographic data of patients at baseline did not reveal any significant differences
Co‐interventions avoided or similar?	Low risk	The sole difference between the groups was the injectant received
Compliance acceptable?	Low risk	All patients received a one time injection of either saline or Botox
Timing of outcome assessment similar?	Low risk	All patients were assessed at baseline, and the three and eight week mark

Liu 2008.

Methods	Subjects with third lumbar transverse process syndrome, as per definition in National Guideline of Traditional Chinese Medicine Diagnosis 1994, recruited and randomised to two groups: Botox and acupuncture. Categorical data with 4 levels of improvement assessed at baseline and 2 months post intervention. Improvement defined by pain and function (this is outcome)
Participants	25 subjects with third lumbar transverse process syndrome recruited and randomised into Botox and acupuncture groups.
Interventions	2 groups: Group 1: Botox injections to trigger points, maximum 100 IU, given once Group 2: Acupuncture needling over similar trigger points
Outcomes	P < 0.05 Botox group better
Notes	Injection site pain reported in Botox group
Risk of bias
Bias	Authors' judgement	Support for judgement
Adequate sequence generation?	Unclear risk	Unclear from text
Allocation concealment?	High risk	List of subjects assigned to groups were not described as being in a concealed setting
Blinding? All outcomes ‐ patients	High risk	Patients were aware if receiving BoNT or acupuncture in the injection
Blinding? All outcomes ‐ providers?	High risk	Providers could not be blinded while administering the BoNT or acupuncture interventions
Blinding? All outcomes ‐ outcome assessors?	Unclear risk	Unclear from text
Incomplete outcome data addressed? All outcomes ‐ Withdrawal/drop‐out rate?	Low risk	100% follow‐up from each group
Incomplete outcome data addressed? All outcomes ‐ intention‐to‐treat analysis?	Low risk	All patients were analysed in the groups they were assigned
Free of selective reporting?	Unclear risk	Unclear from text
Groups similar at baseline?	Low risk	Demographic data of patients at baseline did not reveal any significant differences
Co‐interventions avoided or similar?	Unclear risk	Unclear from text
Compliance acceptable?	Low risk	All patients received one time treatment with BoNT or acupuncture
Timing of outcome assessment similar?	Low risk	All patients were assessed at baseline, and the two‐month mark

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Alo 1997	Not randomised
Carrasco 2003	Not randomised
Childers 2002	Not randomised
De Andres 2003	low‐back pain group data not reported separately
Fanucci 2001	Not randomised
Fishman 2004	Not randomised
Gobel 2006	low‐back pain group data not reported separately
Hernandez 2003	Not randomised
Herskowitz 2004	Unpublished and incomplete data
Jabbari 2006	Not randomised
Ko 2007	Not randomised, low‐back pain group data not reported separately
Lang 2004a	Not randomised, low‐back pain group data not reported separately
Lang 2004b	Not randomised
Lew 2008	low‐back pain group data not reported separately
Nagarajan 2007	Not randomised
Ney 2006	Not randomised
Porta 1999	low‐back pain group data not reported separately
Subin 2003	Not randomised
Yoon 2007	Not randomised

Differences between protocol and review

There were no differences between protocol and review

Contributions of authors

Dr Zeeshan Waseem was involved with data collection and analysis and wrote the manuscript.

Drs Chris Boulias and Farooq Ismail were involved with data collection and analysis.

Drs Allan Gordon and Geoffrey Sheean assisted in developing the background for the manuscript, and were the contact experts in the utilization of botulinum toxin.

Dr Andrea Furlan served as an expert in the research methodology and systematic reviews.

Sources of support

Internal sources

• Division of Physiatry, Department of Medicine, University of Toronto., Canada.

External sources

• No sources of support supplied

Declarations of interest

Dr Allan Gordon uses Botox for low‐back pain in a small number of patients and has taught the injection technique in the past to prospective injectors. 

Dr Geoffrey Sheean has treated patients with low‐back pain but currently does not do so. He has served on Allergan’s spasticity advisory board in the past. 

Drs Chris Boulias, Farooq Ismail and Andrea Furlan do not use BoNT in their low‐back pain patient populations.

None of the investigators have contracts with manufacturers of BoNT such as Allergan.

New