Appropriate translational designs for determining causal force-based (manual therapy) treatment mechanisms.

BACKGROUND: Treatment mechanisms of force-based manipulation (manual therapies) aim to explain not just "whether" the treatment works but "why" and "how" a treatment works. To date, a majority of studies have failed to address whether treatment mechanisms are causal, signifying that they are the steps or processes through which a manual therapy treatment leads to measurable changes in clinical outcomes. The objectives of this commentary were: 1) to outline various research designs capable of investigating causal treatment mechanisms, and 2) to illustrate each design type with examples drawn from existing literature. MAIN TEXT: This commentary identified five causal designs that bridge the gap between basic scientific research and practical applications, allowing causal inference of mechanisms to outcomes. The five designs outlined were mediation analysis, longitudinal structural equation modeling, joint latent class models for longitudinal and time-to-event data, time-lagged analysis, and cross-lagged panel designs. Each design is described and when available, an example from the literature is provided for context. CONCLUSION: Each of the five designs have strengths and weaknesses, involve challenging research requirements such as larger sample sizes, concurrent mechanisms and clinical outcomes capture, and a sophisticated biostatistician to run and interpret research. Nonetheless, despite the challenges in performing this research, these studies are needed to reduce the potential for redundancy in research findings and waste of resources.

Duke Scholars

Author Chad E. Cook Orthopaedic Surgery, Physical Therapy

Author Cynthia Lea Green Biostatistics & Bioinformatics, Division of Biostatistics