Answer to the letter to the editor of J. Hebert et al. concerning “Hancock MJ, Maher CG, Latimer J, Herbert RD, McAuley JH (2008) Independent evaluation of a clinical prediction rule for spinal manipulative therapy: a randomised controlled trial. Epub ahead of publication DOI:10.1007/s00586-008-0679-9”

Abstract

The primary concern of the letter to editor from Drs Hebert and Perle revolves around the fact that in our trial [4] the spinal manipulative therapy was delivered in a manner that was at the discretion of the therapists. In our trial, most patients received low velocity techniques while in the original trial by Childs et al. [2] all therapists were required to use the same single high velocity technique. We see this as strength of the analysis, not a weakness, because the aim of our analysis was to see if the clinical prediction rule developed by Childs and colleagues [2] would generalize to other forms of spinal manipulative therapy selected by highly skilled and experienced clinicians. The consensus view from the literature on clinical prediction rules is that it is important to test the generalizability of these rules [1, 7–9]. It is recommended that clinical prediction rules should be tested in different settings on diverse patients, with different therapists providing the intervention in different ways [1, 7–9]. Childs and colleagues clearly appreciate the importance of testing the generalizability of their clinical prediction rule because they have published the protocol for a trial to examine if their rule generalizes to other forms of spinal manipulative therapy (a different high velocity technique or non-thrust mobilization technique) in diverse clinical settings [3]. We went to great lengths in our manuscript to point out the differences between our trial and the trial by Childs and colleagues that may explain why the clinical prediction rule did not generalize. Drs Hebert and Perle imply that the clinical prediction rule did not generalize to our study because of differences in the way the treatment was provided in our study. It is important to appreciate that this is only one of several explanations for why the clinical prediction rule did not generalize. The lack of generalization could equally be due to differences in patients, settings or co-interventions in the two studies. Currently the clinical prediction rule, which has been tested in only two randomised controlled trials [2, 4], has not been shown to generalize to a setting different to that of the original trial. Further investigation is required to determine if the clinical prediction rule does generalize. Drs Herbert and Perle raise concerns about the terminology used to describe the treatment in our trial. While we agree that the terminology in this area can be confusing we designed and conducted a trial where therapists could use a mix of high or low velocity techniques and believe that the term “spinal manipulative therapy” is the most appropriate and widely used terminology for such treatment. Our use of terminology is consistent with that agreed to by United Kingdom chiropractors, physiotherapists and osteopaths prior to the BEAM trial [5]. We clearly described both the treatments available to therapists and the proportion of high velocity and low velocity techniques used [4], so that interested readers can make up their own mind about the importance of different types of spinal manipulative therapy used. We wish to dispute the assertion of Drs Perle and Herbert that allowing therapists choice as to the type of technique has been shown to lead to smaller treatment effects. The paper they cite [6] in support of this assertion is a non-randomized comparison of a small number of trials with no adjustment for other factors known to influence the magnitude of treatment effects (such as methodological quality). The effect was only apparent in some comparisons. The authors of the paper concluded that “there is insufficient data available to definitely quantify the effect of clinician treatment choice on the results of RCTs of manual therapy for LBP” [6].