Mediation Analysis in Manual Therapy Research

The purpose of this study was to illustrate the applicability of mediation analysis in the manual therapy field by assessing whether pain intensity, duration of pain, or the change in systolic blood pressure mediated the heart rate variability (HRV) of patients with musculoskeletal pain who received manual therapy. A secondary data analysis from a 3-arm, parallel, randomized, placebo-controlled, assessor-blinded, superiority trial was performed. Participants were randomized into spinal manipulation, myofascial manipulation, or placebo groups. Cardiovascular autonomic control was inferred from resting HRV variables (low-high frequency power ratio; LF/HF) and blood pressure responsiveness to a sympathoexcitatory stimulus (cold pressor test). Pain intensity and duration were assessed. Mediation models analyzed whether pain intensity, duration, or blood pressure independently affected the improvement of the cardiovascular autonomic control of patients with musculoskeletal pain after intervention. The first assumption of mediation was met for LF/HF with statistical evidence of a total effect of spinal manipulation, as compared with placebo on HRV outcomes (β = 0.77 [0.17-1.30]); second and third assumptions showed no statistical evidence of a relationship between the intervention and pain intensity (β = –5.30 [–39.48 to 28.87]), pain intensity, and LF/HF (β = 0.00 [–0.01 to 0.01]). In this study of causal mediation analysis, the baseline pain intensity, duration of pain, and responsiveness of the systolic blood pressure to a sympathoexcitatory stimulus did not mediate the effects of the spinal manipulation on the cardiovascular autonomic control of patients with musculoskeletal pain. Accordingly, the immediate effect of spinal manipulation on the cardiac vagal modulation of patients with musculoskeletal pain may more likely be related to the intervention rather than the mediators investigated.