Effectiveness and mechanisms of a multimodal treatment for low back pain: a pragmatic matched cohort study

ObjectiveTo investigate the effectiveness and mechanisms of a multimodal treatment including perturbation exercise. MethodsA matched cohort study was conducted. The intervention consists of a 12-week back pain prevention course with perturbation exercise and education embedded in primary health care according to German social law. Participants from the intervention group had chronic or recurrent low back pain with heterogenous but on average rather low pain and chronification. Control groups (usual care) were matched from a multi-center RCT. Outcomes were pain, disability, isokinetic trunk strength and balance. Bayesian regression models were used to estimate the Average Treatment effect on the Treated (ATT). Further, sub-group and mediation analyses within the intervention group using the biopsychological avoidance-endurance model were conducted. Median values with highest posterior density intervals (HPDI) from baseline-adjusted analyses are presented. ResultsOver 12 weeks, intervention and control (n = 128 each) experienced a similar decrease in pain and disability, which led to negligible ATTs for pain (-0.3 (HPDI95% [-4.3, 3.4]) and disability (-0.2 (HPDI95% [-4, 3.7])). Changes in functional parameters (n=18) showed small effects in favor of the intervention group, in particular for monopedal stances (standardized mean difference: -0.5 HPDI95% [-0.79, -0.21]). Depression was higher in drop-outs and decreases in pain/disability were associated with decreases in depression. Distress-endurance subgroups experienced higher baseline pain and disability and showed the highest reductions in both parameters upon completion of the intervention. ConclusionPerturbation exercise with education yielded only small treatment effects in a heterogenous population with rather mild symptoms. Targeting distress-endurance subgroups with a multimodal treatment approach is probably an effective strategy in treatment tailoring.