The effects of exercise on generalized anxiety symptoms in adults: a systematic review and meta-analysis

This study aimed to systematically evaluate the interventional effect of exercise on symptoms of generalized anxiety.

This systematic review and meta analysis was prospectively registered on PROSPERO (Registration Number: No. CRD420251146169). A comprehensive literature search was conducted across seven electronic databases (Embase, Web of Science, PubMed, The Cochrane Library, Wanfang, VIP, and the China National Knowledge Infrastructure) from their respective inceptions to June 2025. This search specifically targeted randomized controlled trials (RCTs) examining the effects of exercise interventions on generalized anxiety symptoms. The inclusion criteria were defined as follows: adult participants (aged 18 years or older) presenting with generalized anxiety symptoms, without any major comorbidities. The PEDro scale was utilized to assess the methodological quality of the included trials. All statistical analyses, including the primary meta analysis, subgroup evaluations, sensitivity analyses, and publication bias assessments, were performed using Stata 17.0 software. The standardized mean difference (SMD) with a 95% confidence interval (CI) was utilized as the primary effect measure. Finally, the GRADE framework was employed to evaluate the overall certainty of the evidence.

A total of 10 studies (involving 2408 participants) were included. The meta analysis revealed that exercise interventions provided suggestive evidence for a potential reduction in generalized anxiety symptoms (SMD=-0.502,95%CI[-0.830,-0.174], P = 0.003). Preliminary subgroup analyses observed a potential positive trend in middle aged and older adults (SMD=-1.063,95%CI[-1.831,-0.295], P = 0.007) compared with younger adults (SMD=-0.388,95%CI[-0.698,-0.057], P = 0.022). Mind body exercise (SMD=-0.680,95%CI[-0.914,-0.445], P < 0.001) and aerobic exercise (SMD=-0.543,95%CI[-1.064,-0.022], P = 0.041) demonstrated potentially beneficial trends, whereas current evidence remains insufficient to determine the efficacy of resistance exercise (SMD=-0.302,95%CI[-0.964,0.361], P = 0.372). Regarding exercise parameters, potential positive trends were observed for frequencies of ≥ 4 sessions per week (SMD=-1.063,95%CI[-1.831,-0.295], P = 0.007), single session durations of 21 to 40 min (SMD=-1.079, 95%CI[-1.575,-0.582], P < 0.001), and interventions lasting ≥ 12 weeks (SMD=-0.868,95%CI[-1.255,-0.481], P < 0.001) or 8 weeks (SMD=-0.780,95%CI[-1.315,-0.246], P = 0.004). Evidence was insufficient to support 6 week programs (SMD=-0.061,95%CI[-0.517,0.395], P = 0.794) or ≤ 20 min sessions (SMD=-0.022, 95%CI[-0.618,0.574], P = 0.943). What was subjectively classified as light intensity exercise demonstrated a potential positive trend (SMD=-0.915,95%CI[-1.402,-0.427], P < 0.001). While leave one out sensitivity analyses confirmed a consistent beneficial direction (SMD ranging from - 0.43 to -0.61, all P < 0.05), the overall interpretability is severely limited by exceptionally high heterogeneity (I²=90.37%).

The included studies demonstrated acceptable methodological quality (PEDro scores 6 to 8); however, limitations in blinding and subjective intensity classifications introduce substantial bias. In conclusion, while current findings provide suggestive evidence that exercise may assist in ameliorating generalized anxiety symptoms, the exceptionally high heterogeneity dictates that these results must be interpreted with low confidence rather than as definitive efficacy. Furthermore, while analyses across the literature indicate that specific exercise parameters (such as higher frequencies, longer intervention durations, and mind body modalities) exhibit potential positive trends, the specific strata demonstrating the largest effects are often derived from severely underpowered subgroups. Consequently, these observations must be viewed strictly as hypothesis generating rather than optimal clinical prescriptions, highlighting the critical need for robust future trials.