Send letter to journal:
Re: Time to Rethink RCTs for Back Pain

Dear editor,

We read with interest 'A randomised controlled trial of spinal manipulative therapy in acute back pain' by Peter Jüni, Markus Battaglia, Eveline Nüesch et al. in Annals of the Rheumatic Diseases (online 5 Sept, 2008). We congratulate the authors on their attempt to investigate the effectiveness of spinal manipulative therapy (SMT) in the routine clinical practice setting. We also congratulate the authors for their meticulous attention to detail in various parts of their methodology, particularly to reduce bias through randomization, in-tention-to-treat analysis and control and monitoring of co-interventions.

However, we wish to raise some methodological and theoretical concerns regarding the ability of this study to answer the research question as to whether SMT in addition to standard medical treatment is associated with clinically relevant early reductions in pain and analgesic consumption.

Sufficient power is a critical part of the randomised controlled trial (RCT). Jüni et al. calculated that 51 patients per group were needed to achieve more than 80% power in their multilevel model to answer their research question. The research question clearly states that SMT is the treatment being investigated. Unfortunately as described in the methods, spinal mobilization and muscle energy (ME) techniques were also employed at the discre-tion of the treating clinician. The result was that only 38 out of the 52 experimental group participants (73%) actually received SMT in an 'estimated' 80% of the treatment sessions. It is not made clear what the other 37% of SMT group participants received. The unmistakeable bottom line is that this study is underpowered and therefore unable to answer its primary research question. We respectfully ask how the authors can justify their conclusions that SMT is unlikely to result in early pain reductions in acute low back pain patients using an underpowered study?

We have two major concerns regarding recruitment. Firstly, we are unclear as to the reason for using an Emergency Department in combination with a private medical care prac-tice for patient recruitment. The types of patients encountered in an Emergency Ward are unlikely to be typical of acute back pain patients found in routine clinical practice. Therefore we question how applicable are these participants for this study? We are unsure of the methodological rationale for using two such diverse clinical settings.

Secondly, and following on from the issue of recruitment sites, we are also concerned that it took more than 3 years to recruit the 104 acute low back pain patients. We assume that the purpose of this study was to look at the benefit of SMT in the typical clinical setting. As recruitment time lengthens to achieve the sample size estimate, so also do the questions about how representative are the participants to routine clinical practice? Both of these recruitment issues raise further concerns about the ability of this study to answer its fundamental research question.

Another concern is the comparability of the experimental and control groups at baseline as identified in Table 1. An assessment of these groups shows that the experimental (Standard Care & SMT) showed considerably greater duration of pain complaint compared to the Standard Care control group (46% versus 25% >/= 7 days); greater pain intensity (BS 11 score >/= 7: 58% versus 48%) yet were more fully able to work (53% versus 37%), and took less medication (116 versus 131 Diclofenac equivalence dose).

Taken together the experimental group typically had more pain for longer, were likely still working and were less inclined to take medication. It is well known that psychosocial variables are important factors for back pain. These differences at baseline are likely to have influenced the final results of this study. It is unfortunate that no psychosocial variables were investigated as some of these may have been important in terms of baseline characteristics to help explain some of the observed differences. As such, the differences in these baseline characteristics are likely to have impacted on the overall results of this study in meaningful ways that cannot be explained by the present study.

Finally, we suggest that most clinicians and researchers in the field of back pain would agree that the current categorization of mechanical/simple low back pain is insufficient. There are likely a number of subgroups that make up what we currently call low back pain of mechanical origin. We respectfully suggest that the current focus of research for low back and neck pain should be two-fold. First, the identification of subgroups and their defining characteristics is paramount. Second, we suggest that back and neck pain are functional conditions that need appropriate functional outcome measures for their as-sessment. Therefore we suggest that the second focus should be on the development and psychometric testing of new functional outcome measures that are valid, reliable and sensitive for use in mechanical low back and neck pain subgroups.