Work

Intervention effects of physical exercise on negative emotional symptoms in college students: a three-level meta-analysis

Year: 2026

Type: review

Study designs: Systematic Review, Meta-Analysis

Abstract: Objective To systematically evaluate the effects of physical exercise on negative emotional symptoms in college students and examine potential moderators and dose–response relationships. Methods Five English-language databases were searched from inception to March 22, 2026. Risk of bias was assessed using the Risk of Bias 2.0 (RoB 2.0) tool, and certainty of evidence using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach. Hedges’ g values with 95% confidence intervals (CIs) and prediction intervals (PIs) were pooled using three-level random-effects models. Influence diagnostics, sensitivity analyses, subgroup analyses, publication bias tests, and dose–response analyses were conducted. Results Forty-three studies were included. The full-dataset three-level model, retained as the primary analysis, showed a non-significant overall effect (Hedges’ g = −1.48, 95% CI −3.21 to 0.26, p = 0.094; 95% PI −12.95 to 9.99). Two highly influential effect sizes from one study were identified. Excluding these effects, the influence-adjusted analysis showed a significant effect (Hedges’ g = −0.56, 95% CI −0.73 to −0.39, p < 0.001; 95% PI −1.45 to 0.33). In influence-adjusted exploratory analyses, baseline population characteristics, geographic region, and outcome measurement instrument were significant moderators, whereas symptom category, exercise modality, frequency, session duration, intervention duration, and publication year were not. No significant linear or nonlinear dose–response relationships were found for weekly or total exercise duration. Overall certainty of evidence was very low. Conclusion The full-dataset analysis did not provide statistically significant evidence of an overall effect. Although the influence-adjusted analysis suggested a potential benefit, the result was sensitive to influential observations and certainty of evidence was very low. Accordingly, no definitive conclusion regarding efficacy can be drawn. Potential moderator findings require confirmation in future high-quality randomized controlled trials.

Source: Frontiers in Public Health

DOI: 10.3389/fpubh.2026.1911631

Authors: Zhenyao Wang, Lingling Dong, Yan Rao, Haiyan Wen, Cong Liu, Min Huang