A Meta-Analysis of Exercise Intervention for Depression Symptoms

This meta‑analysis estimated the effect of exercise on depression symptoms. Unlike network meta‑analyses that compare exercise modalities, this pairwise meta‑analysis specifically quantifies the effect of structured exercise versus no exercise, using rigorous control-group definitions and updated methods. A literature search up to November 2025 identified 10,123 records. Forty randomized controlled trials met the inclusion criteria. Pooled effect sizes were calculated using random‑effects models; odds ratios (OR) for remission and standardized mean differences (SMD) for depression scores. Heterogeneity was assessed with I². Exercise significantly increased remission rates (OR = 2.40; 95% CI, 1.97-2.93; p < 0.001; I² = 49%) and reduced depression scores (SMD = -0.96; 95% CI, -1.27 to -0.65; p < 0.001; I² = 87%) compared to control groups. The number needed to treat for one additional remission was 2.1. The pooled SMD corresponded to clinically meaningful reductions. Subgroup analyses showed larger effects for supervised exercise and passive controls (e.g., waitlist) versus active comparators (e.g., pharmacotherapy). However, findings have major limitations; only 12/40 studies (30%) had low-risk of bias (RoB 2); most had short follow‑up (≤16 weeks); control conditions varied widely (waitlist to pharmacotherapy); and very high heterogeneity for depression scores (I² = 87%) persisted despite using SMD. The 95% prediction interval for SMD crossed zero, indicating exercise may not benefit all future populations. Grading of Recommendations Assessment, Development and Evaluation (GRADE) was moderate for remission but low for depression scores. Despite these limitations, exercise was associated with significantly higher remission and lower depression scores. These results support exercise as a complementary approach for depression, particularly for achieving remission. However, high‑quality, large‑scale RCTs with standardized outcomes and longer follow‑up are needed before clinical recommendations can be made.