Rheumatology Advance Access originally published online on March 18, 2009
Rheumatology 2009 48(5):464-465; doi:10.1093/rheumatology/kep048
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
EDITORIALS |
Small effects of treatments for non-specific low back pain: how can we improve patients’ outcomes?
1Arthritis Research Campaign National Primary Centre, Keele University, Keele, UK and 2VU University Medical Centre, Amsterdam, The Netherlands
Correspondence to: Daniëlle van der Windt, Arthritis Research Campaign, National Primary Care Centre, Primary Care Sciences, Keele University, Keele, Staffordshire, ST5 5BG. E-mail: d.van.der.windt{at}cphc.keele.ac.uk
Machado et al. [1] in a recent systematic meta-analysis conclude that ‘the analgesic effects of many treatments for non-specific low back pain (NSLBP) are small’. An increasingly large effort is being expended in the research on treatments for NSLBP, represented by the long trail of more than 1000 publications on randomized trials in the past 25 years, and numerous systematic reviews summarizing the evidence for potentially beneficial interventions. For a very common health problem, with major impact on everyday activities and large socioeconomic consequences, this is a desirable situation. However, it is paralleled by an equally growing realization among the back pain research community that more and more trials and reviews are failing to show clear superiority or benefit of any particular treatment, with a large number of widely varying types of treatments seeming to produce similar improvement in back pain symptoms [2–4]. The most recent is the systematic meta-analysis conducted by Machado et al. [1] summarizing available evidence on the analgesic effects of treatments for NSLBP. Beyond the main conclusion the authors arrived at, this review raises some important issues that deserve further discussion.
One impression some might get from reading this review is that the randomized placebo-controlled trial is not only the gold standard design with the least possible bias and therefore the best way to test the efficacy of a treatment, but the suggestion seems to go further in implying that the placebo-controlled trial is the only valid design that should be used to evaluate treatments.
The role of the randomized placebo-controlled trial is certainly very important with rigorously applied methodology controlling for possible bias to clearly identify the specific effect of a particular treatment. However, for a decision to be made by practising healthcare professionals, policy makers and healthcare service providers that a treatment would be beneficial in practice, it subsequently needs to be taken away from the artificial sterile setting of the placebo-controlled RCT to be tested closer to the real environment of clinical practice. It is only when the use of treatment is examined within the interaction and pollution of patients’ and practitioners’ factors that its benefit is understood. Large pragmatic randomized clinical trials, therefore, have an important role in addition to, and not instead of, randomized placebo-controlled trials.
It is obvious that pragmatic trials pose challenges. The most important is related to their less than stringent inclusion criteria and control interventions, which means that the resulting response to treatment might not be specifically related only to the particular ingredient or component of the treatment itself, but may be influenced by various other factors. ‘Non-specific’ is the description or name often given to these factors, which may include patients’ characteristics, symptom presentation, and interaction with the practitioner, practitioner characteristics or setting. The effects of these factors might be large or even just too large that they would obscure the specific effect of the treatment. Combined with the fact that pragmatic trials often use another active treatment as the comparator, the scope to show superiority of the index active treatment can be reduced.
There is a clear and important role for both the randomized placebo-controlled trial, as well as large pragmatic clinical trials to build an evidence-base for the management of back pain. The dilemma remains, namely the inability to clearly show benefits of treatments. To look for a possible solution, it seems that we have to look at three main areas: firstly whether the estimated effect sizes are indeed as small as they seem; secondly, if they are, whether we are targeting the right type of patients in clinical trials; and thirdly, whether the treatments we are testing are addressing the right components of the back pain problem.
Let us first look at the issue of treatment effect sizes. In this review, the authors considered a difference between groups of >10 points on a 0–100 point pain scale to be small, and a difference of 20 points moderate [5]. A 10-point difference between groups may indeed seem small, but given the fact that in most trials the mean back pain score at baseline is 
50–60 points, a 10-points difference between active and placebo will often approximate a difference of 15–20% in reduction of pain from baseline. Most trials are powered to detect a difference of 15–20% and consider this to be a clinically relevant difference.
Randomized trials investigating the effectiveness of treatment for musculoskeletal pain rarely show much larger effect sizes. So, are our expectations from treatments too high? When looking at the effects of treatment from a population or public health perspective, one may end up with a different interpretation of these effect sizes. Given the high prevalence of back pain, and its large impact on disability and work performance, small improvements may be very important. These effects have been acknowledged in other medical fields; think of the management of hypertension or hypercholesteraemia in the prevention of cardiovascular disease for which small effects are considered to be relevant and guide management decisions. Guideline committees and decision makers, therefore, may be interested in effects that can be considered to be small on an individual or group level, as these might result in considerable shifts in the population burden of a disease.
The other question is whether we are targeting the right groups of patients with our interventions. Mean differences in trials may obscure wide individual variation in responses to treatment. A particular subgroups of patients with specific characteristics and risk factor profiles may respond in a different way to treatment than another group with different characteristics. Out of this thought came the idea of sub-grouping the broad, heterogeneous population of NSLBP patients into small groups with similar characteristics with the assumption that individual responses to treatment in these small groups would be more consistent and the overall mean group response would better represent individual responses. The clear empirical evidence for methods of sub-grouping that are feasible in clinical practice and more accurately predict treatment outcome is still developing and more and more studies aim to identify relevant subgroups of spinal pain patients who may respond better to specific interventions [for example, 6–8].
A third area in which an opportunity for improving treatment outcomes might be available is to target important components of NSLBP symptoms, which may reflect physical as well as psychological or social dimensions of health. This may be possible by designing new innovative interventions or by designing intervention programmes that specifically target these components. For example, in Canada, a community-based intervention program [the Pain-Disability Prevention (PDP) Program] has been developed to specifically target psychosocial elements of the pain problem [9]. Also, trials would need to incorporate these different dimensions of symptom presentation when measuring response to treatment. Machado et al. [1], understandably, given the scope of their review, limited their measure of response to pain only, but functional disability is another equally important and widely recognized outcome measure, that reflects another aspect of symptom presentation. Several psychological factors, including depressive symptoms and illness perceptions, are increasingly being recognized as vital aspects of back pain symptomatology. The current understanding is that back pain often has an episodic nature, and is characterized by fluctuations of pain and disability. Physical as well as psychological and social aspects may influence the occurrence, impact and outcome of these episodes of back pain. It seems important to address and target these different dimensions when designing effective treatment and when trying to measure their effectiveness.
Disclosure statement: The authors have declared no conflicts of interest.
References
- Machado LAC, Kamper SJ, Herbert RD, Maher CG, MacAuley JH. Analgesic effects of treatments for non-specific low back pain: a meta-analysis of placebo-controlled randomized trials. Rheumatology (2009) 48:520–7.
[Abstract/Free Full Text] - Keller A, Hayden J, Bombardier C, Van Tulder M. Effect sizes of non-surgical treatments of non-specific low back pain. Eur Spine J (2007) 16:1776–88.[CrossRef][Web of Science][Medline]
- Koes B, Van Tulder MW. Acute low back pain. Am Fam Physician (2006) 74:803–5.[Web of Science][Medline]
- Van Tulder MW, Koes B. Chronic low back pain. Am Fam Physician (2006) 74:1577–9.[Web of Science][Medline]
- Chou R, Qaseem A, Snow V, et al. Diagnosis and treatment of low back pain: a joint clinical practice guideline from the American College of Physicians and the American Pain Society. Ann Intern Med (2007) 147:478–91.
[Abstract/Free Full Text] - Schellingerhout JM, Verhagen AP, Heymans MW, et al. Which subgroups of patients with non-specific neck pain are more likely to benefit form spinal manipulation therapy, physiotherapy or usual care. Pain (2008) 139:670–80.[CrossRef][Web of Science]
- Brennan GP, Fritz JM, Hunter SJ, Thackeray A, Delitto A, Erhard RE. Identifying subgroups of patients with acute/subacute ‘nonspecific’ low back pain: results of a randomized clinical trial. Spine (2006) 31:623–31.[CrossRef][Web of Science][Medline]
- Hay EM, Dunn KM, Hill JC, et al. A randomised clinical trial of subgrouping and targeted treatment for low back pain compared with best current care. The STarT Back Trial Study Protocol. BMC Musculoskelet Disord (2008) 9:58.[CrossRef][Medline]
- Sullivan MJ, Stanish WD. Psychologically based occupational rehabilitation: the Pain-Disability Prevention Program. Clin J Pain (2003) 19:97–104.[CrossRef][Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||