Do Physical Therapy and Yoga Improve Pain and Disability through Psychological Mechanisms? A Causal Mediation Analysis of Adults with Chronic Low Back Pain

Abstract

Objective: To investigate whether indirect effects via psychological mechanisms explain the effects of physical therapy (PT) or yoga versus education on back-related outcomes.

Design: Mediation analyses using data from a randomized controlled trial of PT, yoga, and education interventions for chronic low back pain.

Methods: Primary outcomes were changes in back-related pain on the 11-point numeric rating scale and disability on the modified 23-point Roland-Morris Disability Questionnaire, measured at 52 weeks after randomization. Hypothesized mediators were 12-week changes in pain self-efficacy, fear-avoidance beliefs, depression, anxiety, perceived stress, and sleep quality. We used causal mediation analysis to estimate the total effect, direct effect, indirect effect, and proportion mediated.

Results: We analyzed data from 230 adults (mean age = 46.2 years, 69.6% female, 79.6% non-White). In the PT-versus-education model, when the mediator was perceived stress, the total effect on disability was 2.6 points (95% confidence interval [CI]: 0.3, 4.9) and decomposed into a direct effect of 1.7 points (95% CI: -0.4, 3.8) and an indirect effect 0.9 points (95% CI: 0.1, 2.0; proportion mediated, 34%). No other psychological construct was a significant mediator.

Conclusion: Improvements in perceived stress mediated improvements in disability after PT treatment compared to education. Other psychological outcomes did not mediate the effect of yoga or PT on pain or disability outcomes compared to education. J Orthop Sports Phys Ther 2022;52(7):470-483. Epub: 18 May 2022. doi:10.2519/jospt.2022.10813.

Trial registration: ClinicalTrials.gov NCT01343927.

Keywords: biopsychosocial; chronic pain; mechanism; mediation; physical therapy; yoga.

Conflict of interest statement

The authors certify that they have no affiliations with or financial involvement in any organization or entity with a direct financial interest in the subject matter or materials discussed in the article.

Figures

FIGURE 1.

Hypothesized causal structure of the interventions on outcomes measured at 52 weeks through the psychological score mediators measured at week 12. The straight line from the treatment box to the week 52 box represents the indirect effect. Path a is the effect of the intervention on the mediator. Path b is the effect of the mediator on the outcome. The curved line represents the average direct effect. The lines from the baseline box represent potential confounders of the mediator-outcome effect. Abbreviations: FABQ, Fear-Avoidance Beliefs Questionnaire; GAD, Generalized Anxiety Disorder; PHQ, Patient Health Questionnaire; PSEQ, Pain Self-Efficacy Questionnaire; PSQI, Pittsburgh Sleep Quality Index; PSS, Perceived Stress Scale; PT, physical therapy.

FIGURE 2.

Sensitivity analysis for the confounding assumption. The curve represents the natural indirect effect (ACME) as a function of the sensitivity parameter rho, the correlation of the residuals of the hypothesized mediator and outcome models with confidence bands in gray. Since no unmeasured confounding in the mediator-outcome relationship is assumed, the estimated natural indirect effect occurs at rho equal to zero (dashed line). The value of rho for which the natural indirect effect is equal to zero is 0.3. Abbreviation: ACME, average causal mediation effect.