How much of the effect of exercise and advice for subacute low back pain is mediated by depressive symptoms?

Background: Exercise and advice can reduce pain and disability in patients with subacute low back pain. However, the mechanisms by which this combined intervention works are unclear. Our objective was to estimate how much of the effect of a physiotherapist-directed exercise and advice intervention on pain and disability is mediated via changes in depressive symptoms. Methods: Causal mediation analysis of a randomized controlled trial. We measured our hypothesized mediator - depressive symptoms (Depression, Anxiety, Stress Scale-21) at 6 weeks, and the outcomes - pain (numerical rating scale) and disability (Roland Morris Disability Questionnaire) at 3 months, estimating the average causal mediation effect (ACME), average direct effect (ADE), and total effect. We pre-specified a causal model to identify potential confounders of the mediator-outcome effect and conducted sensitivity analyses to assess the robustness of the ACME under varying levels of unknown confounding. Results: Data from 240 patients were analyzed (average age 50.5 (SD 15.6) years, 52% male, median depressive symptoms score 4). The effect of exercise combined with advice was not mediated via depressive symptoms: ACME on pain (0.05, 95%CI -0.24 to 0.15), ACME on disability (-0.10, 95%CI -0.59 to 0.38). However depressive symptoms were associated with pain (regression coefficient 0.06; SE 0.03)) and disability (regression coefficient 0.17; SE 0.05).) Conclusions: Depressive symptoms did not mediate the effect of exercise and/or advice in this sample. However, depressive symptoms were associated with pain and disability.