The association between risk of bias and effect sizes of spinal manipulative therapy in low back pain trials: meta-epidemiological protocol

Systematic error undermines the internal validity of randomized controlled trials (RCTs). Elucidating how risk of bias (RoB) domains distort effect estimates strengthens evidence trustworthiness. Low back pain (LBP) research provides a representative field for this investigation, with spinal manipulative therapy (SMT) serving as a widely utilized intervention. Objective: To explore the association between risk of bias domains and effect estimates of SMT treatment in LBP trials. Methods: RCTs from the Cochrane systematic reviews which have examined the effect of SMT for acute and chronic LBP will be included. The Cochrane RoB 2 tool will be used. The influence of RoB domains on effect estimates for pain intensity and physical functioning will be explored through univariable and multivariable meta-regression models. Models will be adjusted for confounders identified through a directed acyclic graph, including sample size, trial registration, country income level, and comparator type, while exploring interactions. Significance: This study builds upon research in exercise therapy, investigating whether the association between RoB and effect estimates is consistent across other non-pharmacological LBP interventions. Findings may help refine the application of GRADE on SMT, and improve evidence synthesis and decision-making for researchers and policymakers.