Felipe B. Schuch, Brendon Stubbs, Jacob Meyer, Andreas Heissel, Philipp Zech, Davy Vancampfort, Simon Rosenbaum, Jeroen Deenik, Joseph Firth, Philip B. Ward, Andre F. Carvalho, Sarah A. Hiles

• Background Prospective cohorts have suggested that physical activity (PA) can decrease the risk of incident anxiety. However, no meta-analysis has been conducted. Aims To examine the prospective relationship between PA and incident anxiety and explore potential moderators. Methods Searches were conducted on major databases from inception to October 10, 2018 for prospective studies (at least 1 year of follow-up) that calculated the odds ratio (OR) of incident anxiety in people with high PA against people with low PA. Methodological quality was assessed using the Newcastle-Ottawa Scale (NOS). A random-effects meta-analysis was conducted and heterogeneity was explored using subgroup and meta-regression analysis. Results Across 14 cohorts of 13 unique prospective studies (N = 75,831, median males = 50.1%) followed for 357,424 person-years, people with high self-reported PA (versus low PA) were at reduced odds of developing anxiety (adjusted odds ratio [AOR] = 0.74; 95% confidence level [95% CI] = 0.62, 0.88; crude OR = 0.80; 95% CI =Background Prospective cohorts have suggested that physical activity (PA) can decrease the risk of incident anxiety. However, no meta-analysis has been conducted. Aims To examine the prospective relationship between PA and incident anxiety and explore potential moderators. Methods Searches were conducted on major databases from inception to October 10, 2018 for prospective studies (at least 1 year of follow-up) that calculated the odds ratio (OR) of incident anxiety in people with high PA against people with low PA. Methodological quality was assessed using the Newcastle-Ottawa Scale (NOS). A random-effects meta-analysis was conducted and heterogeneity was explored using subgroup and meta-regression analysis. Results Across 14 cohorts of 13 unique prospective studies (N = 75,831, median males = 50.1%) followed for 357,424 person-years, people with high self-reported PA (versus low PA) were at reduced odds of developing anxiety (adjusted odds ratio [AOR] = 0.74; 95% confidence level [95% CI] = 0.62, 0.88; crude OR = 0.80; 95% CI = 0.69, 0.92). High self-reported PA was protective against the emergence of agoraphobia (AOR = 0.42; 95% CI = 0.18, 0.98) and posttraumatic stress disorder (AOR = 0.57; 95% CI = 0.39, 0.85). The protective effects for anxiety were evident in Asia (AOR = 0.31; 95% CI = 0.10, 0.96) and Europe (AOR = 0.82; 95% CI = 0.69, 0.97); for children/adolescents (AOR = 0.52; 95% CI = 0.29, 0.90) and adults (AOR = 0.81; 95% CI = 0.69, 0.95). Results remained robust when adjusting for confounding factors. Overall study quality was moderate to high (mean NOS = 6.7 out of 9). Conclusion Evidence supports the notion that self-reported PA can confer protection against the emergence of anxiety regardless of demographic factors. In particular, higher PA levels protects from agoraphobia and posttraumatic disorder.…