Limited evidence was defined as a finding in one low-quality randomised trial. Conflicting evidence was defined as inconsistent findings among multiple randomised trials. Definitions of short, intermediate and long term were as per a previous review.18 Short term was defined as less than three months after commencement of treatments. The time point closest to six weeks was used when there were multiple eligible follow-up points. Intermediate term was defined as greater than three months and less than one year after the commencement of treatments. The time point closest to six months was chosen when there were multiple eligible follow-up points. Long term was defined

as greater than or equal to one year after the commencement of treatments. The time point closest to one year this website was chosen if there were check details multiple eligible time points. Figure 1 presents the flow of study

selection. One PhD thesis33 was identified from manual searching and cross-referencing. However, data in the thesis were duplicate and therefore excluded from the review. Five randomised trials34, 35, 36, 37 and 38 were included in this review. Table 1 summarises the five studies. A more detailed description of the studies is available in Table 2, which is available in the eAddenda. Table 3 presents the quality scores. All of the included trials had high quality. No included trials blinded subjects or therapists, although this is not feasible in most rehabilitation trials. Not all studies used therapists who had achieved the highest certification in MDT (diploma). Two trials34 and 35 included a control condition that could be considered as ‘wait and

see’. As pain and disability were reported for the short, intermediate and long term in both trials, meta-analyses were performed. The corresponding author of one study35 provided means and SDs. Based on pooled data from the two trials, MDT STK38 did not significantly improve neck pain intensity in comparison to a wait-and-see control in the short, intermediate or long term, as presented in Figure 2. See Figure 3 in the eAddenda for a more detailed forest plot. Heterogeneity was low (0%) among the short-term and intermediate-term effects, and low to moderate among the long-term effects. The pooled estimates all had 95% CI that were below the threshold of clinical importance. Based on pooled data from the two trials, MDT did not significantly improve disability in comparison to the wait-and-see control in the short, intermediate or long term, as presented in Figure 4. See Figure 5 in the eAddenda for a more detailed forest plot. Heterogeneity was low (0%) at all time points. The pooled estimates all had 95% CI that were below the threshold of clinical importance.