Evaluating Certainty of Evidence Assessment in Systematic Reviews of Yoga for Health: A Meta-Research Study.

Background: The assessment of the trustworthiness and relevance of available evidence is a key element of evidence synthesis. For example, Cochrane systematic reviews (SRs) mandate the assessment of certainty of evidence (CoE) using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) method. However, most SRs are published outside Cochrane, and the use of CoE outside Cochrane is not universal. The use of CoE in SRs of yoga for maintaining or improving health is not known. Objective: To: (a) estimate the proportion of SRs of yoga for health that assessed CoE; (b) identify which CoE systems and domains were used; and (c) examine whether CoE assessment was associated with SR publication and methodological characteristics. Methods: The Cochrane Database of Systematic Reviews (CDSR), MEDLINE (Ovid), PubMed, EMBASE (Ovid), AMED (Ovid), CINAHL (EBSCO), and PsycINFO (EBSCO) were searched to April 25th 2024 for SRs of randomized controlled trials (RCTs) of yoga for health. Two reviewers independently screened studies against predefined eligibility criteria for SRs of yoga: the study (1) sought to comprehensively identify relevant RCTs on a yoga-focused research question, (2) stated inclusion criteria for the studies, (3) searched at least one electronic database, and (4) reported the critical appraisal of the included studies. Data on SR characteristics and CoE were independently extracted from included SRs by two authors. Data analysis included frequencies and univariate and multivariable logistic regressions, conducted using R version 4.5.2. Statistical significance was set at Results: A total of 250 SRs published between 2005 and 2025 were included. More than a third of the SRs (88/250; 35.2%) were published by authors based in Asia, America and Europe. Among the 246 reviews with included RCTs, 39 (16%, 95% CI = 11.8%-20.9%) assessed CoE for at least one effect estimate. Among reviews that assessed CoE, GRADE was the most frequently used framework (35/39, 89.7%). In multivariable analysis, CoE assessment was independently associated with presence of a meta-analysis (OR = 3.01, 95% CI = 1.08-10.21), systematic-review journal type (OR = 56.47, 95% CI = 5.33-1151.95), and use of a Cochrane risk-of-bias tool (OR = 12.17, 95% CI = 2.71-124.93). Region, journal impact factor, and study types included were not independently associated with CoE assessment after adjustment. Conclusion: Approximately 16% of SRs of yoga for health assessed CoE. Since this study was limited to English-language studies identified from major databases through 2024, we cannot be certain whether this result reflects true levels of CoE assessment across all SRs of yoga for health conducted to date. Strengthening reviewer training, instituting journal requirements, and encouraging adherence to PRISMA and GRADE guidelines might increase the future percentage of yoga SRs assessing CoE.