A systematic review and meta-analysis on the safety of yoga

Purpose: To systematically assess and meta-analyze the frequency of adverse events in randomized controlled trials of yoga. Methods: Medline/Pubmed, Scopus, Cochrane Library, IndMED and tables of content of yoga specialty journals not listed inmedical databaseswere screened from their inception until February 2014. Randomized controlled trials comparing yoga to other interventions, and reporting non-serious adverse events, serious adverse events, intervention-related adverse events, and/or drop-outs due to adverse events were included. Study characteristics and risk of bias (Cochrane risk of bias tool) were assessed by 2 reviewers independently. Risk differences (RD), odds ratios (OR), and their respective 95% confidence intervals (CI) were calculated and meta-analyzed using a random effects model. Results: Out of 2,520 initially identified records, 301 were randomized controlled trials of yoga, of which 94 (total of 8,430 participants) reported on adverse events. No differences in the frequency of non-serious, serious, or interventionrelated adverse events and of drop-outs due to adverse events were found when comparing yoga to usual care or exercise. Compared to psychological or educational interventions, more non-serious adverse events (RD=0.11; 95% CI=0.02, 0.19; p=0.01; OR=7.29; 95% CI=1.91, 27.89; p<0.01) and more intervention-related adverse events (RD=0.05; 95%CI=-0.02, 0.12; p=0.10; OR=4.72; 95% CI=1.01, 21.99; p=0.05) occurred in the yoga group; serious adverse events and drop-outs due to adverse events were comparable between groups. Conclusion:Only 31% of the identified randomized trials of yoga reported on adverse events. While non-serious adverse events are more frequent with yoga than with psychological interventions, yoga appears as a generally safe intervention, comparable to exercise and usual care. Despite the limitations of the available evidence, recommending yoga to healthy people and those with underlying illnesses should not be discouraged based on safety and can be considered if sufficient evidence of effectiveness is available. Contact:HolgerCramer, h.cramer@kliniken-essen-mitte.de