Application of variance ratios to assess heterogeneity of spinal manipulative therapy for chronic low back pain: A meta-epidemiological study.

Abstract

To compare the pooled estimates and variability in pain and function outcomes for spinal manipulative therapy (SMT) versus sham SMT in randomized clinical trials (RCTs), which informs the potential for heterogeneity of treatment effect. Meta-epidemiological study. We used results from the previous Cochrane review and meta-analysis and performed updated electronic searches from May 2018 to June 2025. RCTs comparing SMT to sham SMT for chronic low back pain. For the meta-epidemiological analysis, we estimated the variance ratios of pain and function outcomes for both pooled results and individual trials. Variance ratio statistically different than 0 indicate greater variability in one group, providing evidence that heterogeneity of treatment effect may be present. Fifteen trials (n = 1141) were included in the analysis. No statistically significant or meaningful variance ratios were found in 1-, 3-, 6-, and 12-month pooled results for pain and function. Among individual RCTs, greater variability was found in two trials for the SMT group and four trials in the sham SMT for pain or function at the 1-month analysis. There were no significant differences in the variability of pain and function outcomes for SMT groups compared to sham SMT groups for chronic low back pain. This suggests there is no evidence the distribution around the mean estimate is different between interventions at all time points and does not support the potential for heterogeneity of treatment effect in this population. Nevertheless, we cannot conclusively rule out heterogeneity of treatment effect using variance ratios alone. This study provides an example of how using variance ratios in meta-analyses and RCTs can add other statistically and clinically meaningful findings.