Research waste in randomised trials of exercise treatments for chronic low back pain: trial sequential and cumulative meta-analyses by publication date and size.

Abstract

To determine if there is research waste in controlled trials of exercise therapies compared with usual care/no treatment for adults with chronic non-specific lower back pain. Secondary analysis of previously published systematic review (Cochrane review). Randomised controlled trials comparing exercise treatments for chronic low back pain to usual care/no treatment eligible for inclusion in the 2021 Cochrane review of exercise interventions for chronic low back pain. All data were taken from the 2021 Cochrane review of exercise therapy for chronic lower back pain and the UK BEAM trial. We did trial-sequential meta-analysis and cumulative meta-analyses, exploring changes in effect estimates over time and by trial size. Respective superiority boundaries for pain and disability were crossed in 2004 after four and five trials (n=358/415) were published. A further 43 trials with 2626 participants were included in the Cochrane review. In 2004, the mean effect sizes for pain and disability were -12.85 (95% CI -24.89 to -0.81) and -6.67 (95% CI -11.27 to 3.36), respectively; similar to those reported by Cochrane in 2021. Including small trials substantially affected effect size estimates. When the 33 and 36 trials, respectively, with fewer than 70 participants are excluded, the limits of the 95% CIs for effect size estimates exclude the clinically important differences ((pain; -8.8 (95% CI -11.38 to -5.63): disability -4.27 (95% CI -6.12 to -2.24). It may be difficult to justify any further trials comparing exercise interventions to usual care/no treatment for chronic low back pain. The inclusion of small studies in meta-analyses has produced biased results in previous meta-analyses. Exercise treatments might not have a clinically important effect on people with chronic low back pain.