Benefits and harms of spinal manipulative therapy for the treatment of chronic low back pain: systematic review and meta-analysis of randomised controlled trials
BMJ 2019; 364 doi: https://doi.org/10.1136/bmj.l689 (Published 13 March 2019) Cite this as: BMJ 2019;364:l689
All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
Many thanks to Alastair D Hay for the suggestions on improving our visual abstracts.
It is always a tricky decision which information to include in a visual abstract. Our template for RCT visual abstracts does include information on clinical significance (for example see: https://www.bmj.com/content/364/bmj.k5432 and https://www.bmj.com/content/364/bmj.l236). However, when subsequently making the template for meta analyses, space was too limited to include it, as the results section tends to require more room.
The suggestion to include this information as a line on the forest plot however is something I hadn't considered. It's an excellent idea, and I have adjusted my template for meta analysis abstracts in future so that we can try it out.
Competing interests: No competing interests
I wish to congratulate both Rubinstein et al (on the publication of an interesting and useful article regarding the effectiveness of spinal manipulative therapy for the treatment of chronic low back pain) and The BMJ for introducing clear, well-designed 'visual abstracts', including the one presented here. These infographics allow the rapid assimilation and interpretation of complex information and I welcome them.
One weakness of the visual abstract associated with this article is the lack of information regarding the clinical significance of the effects. In its current form, the infographic displays: the timing (1, 6 and 12 months) and type of benefits (reduced pain and improved function); and the likelihood the results could be due to chance (the 95% confidence interval horizontal bar and its relation to the vertical 'null effect' line). My proposal is that future infographics also present the minimum clinically important (or maximum clinically unimportant) difference for superiority (and non-inferiority) trials. For this infographic, this would have meant the addition of one more vertical line to the left of the null effect line, above which could sit a 'clinical difference symbol'. Where the clinical important difference is disputed, the infographic could present a semi-transparent block the width of which presents the dispute boundaries, and where it is unknown, the symbol could have an adjacent "?" .
Competing interests: No competing interests
Response to O’Keeffe & O’Connell
We want to thank O'Keeffe and O’Connell for their interest in our article. They raise three issues: 1) efficacy of SMT; 2) whether we categorized the contrast therapies correctly for the pooling of studies; and 3) our application of GRADE.
From their comments, it would appear that they are of the opinion that studies on SMT can be designed and evaluated in a similar fashion as pharmaceutical trials. SMT is a hands-on therapy, which cannot be adequately blinded in a trial; therefore, pragmatic trials with a convincing comparison are more valuable than when comparing SMT to sham. In addition, SMT is applied by a variety of therapists in different populations, so it is important to understand how SMT works in different settings. Furthermore, it is remarkable that following more than 40 years of research on low-back pain, the ‘ideal’ landmark trial on SMT has not been conducted, nor is it likely to come. Even the large UK BEAM trial (1) which included >1,300 patients left many questions, as is illustrated by the letters to the editor. We have to consider the comments on our publication therefore, in this light. We are of the opinion that studies which investigate the efficacy of SMT (i.e. compare SMT to sham) are not as informative as these contributors suggest, because the prevention of performance bias (i.e. ‘blinding’ of subjects) is often not successful. To illustrate: of the eight studies we included, three were deemed unsuccessful (2-4), and two questionable (5,6).
Secondly, they question our choice of categorization of contrast therapies when pooling studies. We categorized contrast therapies into main clusters using national guidelines as a frame of reference. Categorization is not as straightforward as one may suppose and therefore open to discussion. To address the issue which has been raised, we conducted a sensitivity analysis within the category ‘SMT vs. recommended therapies’, and selected only those studies wherein the effect of SMT was compared with exercise. We chose this particular intervention because exercise is accepted as a first-line intervention in multiple national guidelines for the treatment of low-back pain, as is described in the Lancet Series (7). This additional analysis confirmed our earlier findings, namely that SMT is equally as effective as exercise for short-term pain reduction and improvement in functional status. The results are similar for these outcomes at 6 and 12 months. Furthermore, we carried out an additional sensitivity analysis for ‘SMT vs. non-recommended therapies’, excluding the interventions thought by O'Keeffe and O’Connell to be unsuitably categorized (e.g. no treatment, waiting-list control). We also included Hondras et al. (a medical care control group using paracetamol), which was formerly included in the ‘recommended’ cluster. Also this analysis demonstrated similar results for short-term pain reduction and improvement in functional status, as found in our publication (see below). In other words, our findings are robust and sensitivity analyses confirm our earlier findings, including re-categorization of comparative interventions within the cluster, 'SMT vs. non-recommended therapies'.
Finally, they take issue with our application of GRADE, which is used to assess the evidence and strength of recommendations. We agree that selection bias and performance bias are important issues, and those studies which demonstrate these types of bias may exaggerate treatment effects. However, while the GRADE Working Group offers some examples concerning application, there is of yet, no consensus as to how to apply these (8). Consequently, the choice when to downgrade for each of the determinants is a question of good judgement, including a combination of content and methodological knowledge.
In summary, there are large degrees of variation between trials, and the interventions are sometimes complex. Summarizing them is never straightforward. However, we are convinced that our conclusion that ‘SMT produces similar effects as recommended therapies for the treatment of chronic low-back pain, and better effects than non-recommended interventions for short-term improvement in function’, stands.
On behalf of the project team,
Sidney Rubinstein
Additional sensitivity analyses conducted for the purpose of this response:
SMT vs. exercise at 1 month for pain and functional status
Pain (0-100 points): MD -0.34, 95% CI -6.86 to 6.18; participants = 1569; studies = 9; I2 = 91%;
Note: removal of the one extreme outlier (Sarker 2017) reduces the statistical heterogeneity to an acceptable level: MD 0.46, 95% CI -1.88 to 2.79; participants = 1499; studies = 8; I2 = 12%.
Functional status: SMD -0.01, 95% CI -0.11 to 0.09; participants = 1499; studies = 8; I2 = 0%.
Data for 6 and 12 months are available upon request.
SMT vs. non-recommended treatments at 1 month (with the changes stated above)
Pain (0-100 points): MD -6.04, 95% CI -8.79 to -3.30; participants = 970; studies = 7; I2 = 5%.
Functional status: SMD -0.37, 95% CI -0.63 to -0.11; participants = 835; studies = 6; I2 = 64%
References
1. UK BEAM Trial Team. United Kingdom back pain exercise and manipulation (UK BEAM) randomised trial: effectiveness of physical treatments for back pain in primary care. BMJ. 2004;329(7479):1377-1381.
2. Bialosky JE, Bishop MD, Price DD, Robinson ME, George SZ. Spinal manipulative therapy has an immediate effect on thermal pain sensitivity in people with low back pain: a randomized controlled trial. Physical therapy. 2009;89:1292-1303.
3. Hidalgo B, Pitance L, Hall T, Detrembleur C, Nielens H. Short-term effects of Mulligan mobilization with movement on pain, disability, and kinematic spinal movements in patients with nonspecific low back pain: a randomized placebo-controlled trial. J Manipulative Physiol Ther 2015 Jul-Aug;38(6):365-74. 2015;38(6):365-374.
4. Krekoukias G, Gelalisa ID, Xenakisa T, Gioftsosb G, Dimitriadisc Z, Sakellarib V. Spinal mobilization vs conventional physiotherapy in the management of chronic low back pain due to spinal disk degeneration: a randomized controlled trial.". Journal of Manual and Manipulative Therapy. 2017;25(2):66-73.
5. Ghroubi S, Elleuch H, Baklouti S, Elleuch MH. Chronic low back pain and vertebral manipulation. [French]. Annales de Readaptation et de Medecine Physique. 2007;50(7):570-576.
6. Licciardone JC, Stoll ST, Fulda KG, et al. Osteopathic manipulative treatment for chronic low back pain: a randomized controlled trial. Spine. 2003;28(13):1355-1362.
7. Foster NE, Anema JR, Cherkin D, et al. Prevention and treatment of low back pain: evidence, challenges, and promising directions. Lancet. 2018;391(10137):2368-2383.
8. GRADE Working Group. https://gdt.gradepro.org/app/handbook/handbook.html#h.m9385o5z3li7.
Competing interests: No competing interests
In their systematic review on the benefits and harms of spinal manipulative therapy (SMT) for chronic low back pain (CLBP), Rubinstein et al. conclude that SMT has similar effects to ‘recommended’ therapies for pain intensity, and a clinically better effect for function over recommended therapies, ‘non-recommended’ therapies and sham therapy at short-term follow-up.1 However, we believe there are a number of important reasons why these conclusions may not be valid.
The only reliable way of assessing the efficacy of SMT is to compare it with sham treatment. In their analyses, there was low to very low certainty evidence based upon up to eight trials and 831 participants that SMT did not improve pain more than sham treatment at any time point and provided a small benefit for function at one month only. Examination of the forest plot for function at one month (Fig D in the appendix) indicates considerable statistical heterogeneity (I2=91%) explained by a single extreme outlier which reported a likely implausible lack of improvement over time in the sham SMT group.2 Removal of that outlier results in an effect substantially lower than their threshold for clinically important benefit (SMD -0.27 (95% CI -0.52 to -0.02), I2=39%). In fact, five of the remaining six trials failed to report a benefit for function at this time point. The lack of an important benefit in trials with a host of biases likely to favour SMT indicates that higher quality evidence would be extremely unlikely to overturn a conclusion that SMT confers no benefits over sham therapy.
It is also difficult to interpret the comparisons of SMT with ‘recommended’ and ‘non-recommended’ therapies as these comparisons combine a variety of control groups that may not yield similar outcomes. For example, multimodal packages of care, exercise interventions, back school, minimal medical care, and a pain clinic were all included in the ‘recommended’ therapies comparison. The categorisation of therapies as ‘recommended’ or ‘non-recommended’ therapy is also sometimes questionable. For example, inert (e.g. detuned ultrasound) and no treatment controls were both categorised as ‘non-recommended’ therapies. While these are clearly not recommended it seems unreasonable to pool them with non-recommended therapies offered in good faith. On the other hand, a minimal medical care control comprised mainly of provision of paracetamol,3 a treatment that is no longer recommended by the clinical guidelines cited, was categorised as a ‘recommended’ therapy.
Some included trials also provide no reliable estimation of the effectiveness of SMT. For example, one study compared a package of chiropractic care which included stretching, dry needling, ischemic compression, soft tissue massage and home exercise in addition to SMT, and this was compared to a variety of approaches which included gabapentin, facet joint injection, soft-tissue injections and transcutaneous electrical nerve stimulation (TENS).4
A further issue is that the authors do not seem to interpret their results in line with their own thresholds for determining clinical effectiveness. The point estimates for almost all comparisons failed to meet their pre-defined thresholds for clinical importance. In the single comparison where the point estimate exceeded this threshold the lower limit of the 95% confidence intervals fell well outside it. That SMT fails to cross this low bar in open-label trials at risk of various other biases, might be interpreted as evidence of ineffectiveness.
Finally, the assessment of the certainty in evidence and strength of recommendations for SMT overstates the degree of certainty. A body of evidence for a given comparison was only downgraded for the criterion “limitations of studies” where >50% of participants came from studies at risk of both allocation and performance bias. This is lenient since either one of these sources of bias alone represents a meaningful limitation.
In conclusion, these results demonstrate no convincing evidence for the superiority of SMT over sham SMT and a lack of clinically important benefit of SMT when compared with any other treatment. The lack of a benefit of SMT over sham therapy indicates that SMT is unlikely to have any direct benefits and observed improvements are the result of contextual and other effects. It is likely that the apparent equivalence with both ‘recommended’ and ‘non-recommended’ therapies tells us more about the disappointing effectiveness of those approaches than it does about the benefit of SMT.
References
1. Rubinstein SM, de Zoete A, van Middelkoop M, et al. Benefits and harms of spinal manipulative therapy for the treatment of chronic low back pain: systematic review and meta-analysis of randomised controlled trials. BMJ 2019;364:l689.
2. Krekoukias G, Gelalis ID, Xenakis T, Gioftsos G, Dimitriadis Z, Sakellari V. Spinal mobilization vs conventional physiotherapy in the management of chronic low back pain due to spinal disk degeneration: a randomized controlled trial. J Man Manip Ther 2017;25:66-73.
3. Hondras MA, Long CR, Cao Y, et al. A randomized controlled trial comparing 2 types of spinal manipulation and minimal conservative medical care for adults 55 years and older with subacute or chronic low back pain. J Manipulative Physiol Ther 2009;32(5):330-43.
4. Wilkey A, Gregory M, Byfield D, et al. A comparison between chiropractic management and pain clinic management for chronic low-back pain in a national health service outpatient clinic. J Altern Complement Med 2008;14(5):465-73.
Competing interests: No competing interests
This review confirms previous ones that SMT appears to be both effective and safe, but it wrongly states that SMT is not included in current NICE guidelines as a first-line treatment. It is, and is considered alongside other recommended therapies. SMT qualifies as a treatment which is good enough for the NHS and for its patients.
It's a pity that systematic reviews tend to average results across heterogeneous substrates, but nevertheless the contention that SMT is a real treatment with some effect and few harms, is supported by even this averaging of data. SMT is skill-dependent in both application and execution, and spinal dysfunction differs between affected individuals. There are differences between SMT treatment done by osteopaths, chiropractors or physiotherapists.
The review barely references the landmark UK BEAM trial (attributing a null result to a significant trial pain outcome) - described even by Edzard Ernst as "excellent". UK BEAM was published with a positive cost-effectiveness analysis and had been rigorously designed to avoid criticisms of previous pragmatic trials. The results are significant, with meaningful benefits of SMT over and above 'best care' by GP plus exercise, with benefit still present at twelve months. Although there were further criticisms of the small (but statistically significant) mean differences in RMDQ scores, a further analysis revealed NNTs as low as 3.3 to 5.1 for treatment by SMT with and without exercise, respectively.
Since registering in 2011, I've treated thousands of patients as an NHS GP Osteopath; with longitudinal GP data linked to consultation rates, NSAID/opioid use, referrals, co-morbidities etc., one might reasonably think that's a pretty useful resource for real and linked data. I doubt any data from one source would pass muster under Rubinstein's gaze - an approach which I think would be mistaken.
Using osteopathic SMT in general practice has proved to be very effective. General Practice is a very suitable setting for its delivery, with both clinical and fiscal benefits. GPs are well-placed to spot early and to investigate differential pathologies.
I'm not sure why this systematic review is calling for blocks to funding for further SMT research, on the premise that RCTs have shown us as much as they can. We clearly need more and smarter research to understand what SMT does do and how we might do it better. Any research interest welcomed.
BMJ 2004; 329 doi: https://doi.org/10.1136/bmj.38282.669225.AE
BMJ 2004; 329 doi: https://doi.org/10.1136/bmj.38282.607859.AE
Competing interests: No competing interests
Re: Benefits and harms of spinal manipulative therapy for the treatment of chronic low back pain: systematic review and meta-analysis of randomised controlled trials
The SMT systematic review project team feels that it is appropriate to clarify the following point.
Our BMJ review is an update of our earlier Cochrane review published in 2011; however, is not officially a Cochrane review. The version which recently appeared in The BMJ was not submitted for editorial approval to the Cochrane Back and Neck (CBN) Review Group; did not receive peer-review feedback from them, nor was it signed-off for publication by those co-editors. Readers might have this impression because in the acknowledgements we thank the Editorial Board for feedback and in the review there is a line that says the secondary outcomes will be updated in the Cochrane Library.
We have plans to publish the full review in the Cochrane Library in the near future. We sincerely regret any confusion this may have caused.
On behalf of the project team,
Sidney Rubinstein
Competing interests: No competing interests