21 and 22 A cut-off score of six and above has been used for high-quality studies,21 but reducing the cut-off score from six to five has not affected the overall outcome and a cut-off score of five has been used by some reviews.23, 24, 25 and 26 Hence, in this review, high-quality research was defined as a study with Selleckchem ON-1910 a ≥ 5 PEDro score and was used as a criterion for meta-analysis. The score from the PEDro online database was used, as all studies included in this study were included in the PEDro database. Two assessors (HT and XC) independently extracted data, with no disagreements.

When data reported in a published paper were insufficient to quantitatively analyse the effect of MDT, the corresponding author was contacted and additional data were obtained if possible. Consideration of the quality NVP-BKM120 in vivo of interventions is important27 and therapists’ certification/training levels could

affect outcomes with MDT treatment because treatment strategies are different in each subgroup and reliability of classification of subgroups could vary by certification/training levels. There is a consensus that classification reliability is good in the holders of the highest certification but the reliability level in other therapists is not always good.28, 29 and 30 Thus, the level of MDT certification was also analysed. To enable comparison of outcomes between interventions and trials, data for pain intensity and disability were converted to a point scale of 0 to 100 (0 = no pain or no disability) and then a mean difference with 95% confidence interval (95% CI) was calculated for within-group change scores. A positive mean difference indicates Phosphoprotein phosphatase a favourable effect of MDT in comparison to other therapeutic approaches including wait-and-see control. A value of 20 on the 0-to-100 scale was used as the threshold for clinical importance for both pain and disability. When variability data for within-group change scores were unavailable and when baseline scores were assumed to be comparable,

between-group differences at follow up were used. SD was estimated as one quarter of the mean value when variability data were unavailable.18 When the sample size at a follow-up point was not clear, the sample size before the follow-up point was used to calculate mean differences. When pooling data was appropriate, meta-analysis was undertaken and a weighted mean difference was calculated. I2 was assessed to investigate the degree of between-trial heterogeneity using a random-effects model. I2 values of 25%, 50% and 75% indicate low, moderate and high heterogeneity, respectively.31 When meta-analysis was not undertaken, a quantitative summary was tabulated. Levels of evidence were decided according to a guideline for systematic reviews.32 Strong evidence was defined as consistent findings among multiple high-quality randomised trials.