Many different treatments exist for low back pain (LBP). Often they are multidisciplinary, involving physiotherapists, psychologists and medical doctors. However, when these treatments are tested in experimental studies many produce only small treatment effects when they are compared to each other and only small or moderate effects when compared to a placebo or usual care. In other cases we have isolated trials of treatments which have demonstrated large effects (e.g. Albert et al’s trial of antibiotics for LBP [1]; von Korff et al’s [2] trial of activation for LBP) but the mechanism of action for the intervention remains uncertain.
Establishing how our treatments work is one of the key questions we should be asking ourselves. Much of the previous research for LBP has been spent identifying what factors predict which patients will recover and which will not, and it is often assumed (mistakenly) that the same factors that predict outcome will make good targets for treatment. Mediation analysis, which allows us to test which particular aspects of a treatment are associated with patient improvement, can help answer this question.
Our article, recently published in Spine (Mansell et al 2014 [3]) provided an overview of mediation analysis and its clinical importance. It outlined the results of a workshop held at an international LBP conference which brought together experts in the field, to discuss how intervention study designs can be tweaked in order to include mediation analysis.
It is not that our current designs for trials are not fit for purpose; it’s just that current trial design aims to test efficacy – does the new treatment work better than usual care or placebo. In mediation analysis we are able to take this question a step further and ask why the treatment worked or did not work. However, in order to conduct robust mediation analyses, a few small changes to the current design of our LBP clinical trials are required. For me, the most essential element is to take measures of factors thought to be key to changing outcome not only at the beginning and end of the treatment period, but also during the treatment period. Only by taking measures during the course of treatment will we know when specific factors change, and by how much. Current trial methodology tends to take measures at baseline and at follow-up points after treatment has ended. While I understand that additional questionnaires, however short, increases patient and practitioner burden in terms of time, the information we can gain from this may be invaluable in terms of seeing how particular characteristics relate to improvements in pain and disability. Making our interventions more efficient by focusing on particular aspects of treatment at key points is likely to result in stronger treatment effects and better outcomes for patients.
Another key element is to recognise that identifying how and why treatments work is a process; one trial alone cannot adequately answer the question of how treatments work. A programme of research, starting with qualitative research to gain the views and opinions of patients and healthcare professionals as to what they think leads to changes in treatment would be an important first step, followed by observational studies to assess which variables are most likely to be potential mediators of outcome (i.e. which can change, and is change associated with change in outcome) will result in testable hypotheses for randomised controlled trials (RCTs) which can focus on specific elements of treatment in a more controlled way. It goes without saying that RCTs provide the best medium for testing mediators of treatment effect; many mediation studies use observational data to test between different factors, which can be useful for looking at relationships between different factors but do not help us understand how different aspects of treatment can lead to patient improvement.
These changes to trial design to better test for mediating factors have been called upon in other fields (such as Kazdin (2007 [4]) in psychotherapy, and Stanton et al (2013 [5]) in oncology). It seems it is time to make these changes happen in our own field of LBP to ensure we create trials that can answer more than just the efficacy question, in order that we see greater improvements for our patients.
About Gemma
Gemma Mansell is a PhD student and Research Assistant at the Arthritis Research UK Primary Care Centre, Primary Care Sciences, Keele University. Her PhD research involves investigating design and analysis methods for conducting mediation analysis. She has a background in health psychology, and the research studies in her PhD involve secondary analysis of psychological treatment interventions for LBP, but her research interests also include musculoskeletal pain prognosis.
References
[1] Albert HB, Sorensen JS, Schiott Christensen B, Manniche C. Antibiotic treatment in patients with chronic low back pain and vertebral bone edema (Modic type I changes): A double-blind randomized controlled trial of efficacy. European Spine Journal 2013, 22: 697-707.
[2] Von Korff M, Balderson BH, Saunders K, Miglioretti DL, Lin EH, Berry S, Moore JE, & Turner JA (2005). A trial of an activating intervention for chronic back pain in primary care and physical therapy settings. Pain, 113 (3), 323-30 PMID: 15661440
[3] Mansell G, Hill JC, Kamper SJ, Kent P, Main C, & van der Windt DA (2014). How can we design low back pain intervention studies to better explain the effects of treatment? Spine, 39 (5) PMID: 24305571
[4] Kazdin AE. Mediators and mechanisms of change in psychotherapy research. Annual Review of Clinical Psychology 2007, 3: 1-27.
Kazdin AE (2007). Mediators and mechanisms of change in psychotherapy research. Annual review of clinical psychology, 3, 1-27 PMID: 17716046
[5] Stanton AL, Luecken LJ, MacKinnon DP, & Thompson EH (2013). Mechanisms in psychosocial interventions for adults living with cancer: opportunity for integration of theory, research, and practice. Journal of consulting and clinical psychology, 81 (2), 318-35 PMID: 22663900