Sham treatment effects in manual therapy trials on back pain patients: a systematic review and pairwise meta-analysis

To assess the effects and reliability of sham procedures in manual therapy (MT) trials in the treatment of back pain (BP) in order to provide methodological guidance for clinical trial development.

Systematic review and meta-analysis.

Different databases were screened up to 20 August 2020. Randomised controlled trials involving adults affected by BP (cervical and lumbar), acute or chronic, were included.Hand contact sham treatment (ST) was compared with different MT (physiotherapy, chiropractic, osteopathy, massage, kinesiology and reflexology) and to no treatment. Primary outcomes were BP improvement, success of blinding and adverse effect (AE). Secondary outcomes were number of drop-outs. Dichotomous outcomes were analysed using risk ratio (RR), continuous using mean difference (MD), 95% CIs. The minimal clinically important difference was 30 mm changes in pain score.

24 trials were included involving 2019 participants. Very low evidence quality suggests clinically insignificant pain improvement in favour of MT compared with ST (MD 3.86, 95% CI 3.29 to 4.43) and no differences between ST and no treatment (MD -5.84, 95% CI -20.46 to 8.78).ST reliability shows a high percentage of correct detection by participants (ranged from 46.7% to 83.5%), spinal manipulation being the most recognised technique.Low quality of evidence suggests that AE and drop-out rates were similar between ST and MT (RR AE=0.84, 95% CI 0.55 to 1.28, RR drop-outs=0.98, 95% CI 0.77 to 1.25). A similar drop-out rate was reported for no treatment (RR=0.82, 95% 0.43 to 1.55).

MT does not seem to have clinically relevant effect compared with ST. Similar effects were found with no treatment. The heterogeneousness of sham MT studies and the very low quality of evidence render uncertain these review findings.Future trials should develop reliable kinds of ST, similar to active treatment, to ensure participant blinding and to guarantee a proper sample size for the reliable detection of clinically meaningful treatment effects.

CRD42020198301.