The Effects of Aerobic and Resistance Exercise on Depression and Anxiety: Systematic Review With Meta-Analysis

The paper addresses whether prescribed aerobic exercise, resistance exercise, or a combination of both improves depressive and/or anxiety symptoms in adults diagnosed with depression or anxiety. It situates the question within the global burden of mental illness, reduced life expectancy, and high comorbidity with chronic physical conditions. The WHO notes unique barriers to exercise for people with mental health conditions. Pharmacologic treatments can be costly and have adverse effects, prompting interest in exercise as a non-pharmacological adjunct. Prior evidence suggests physiological (e.g., mitochondrial function, stress hormone regulation, sleep/mood improvements) and psychosocial benefits (social interaction, distraction) of exercise. However, findings across studies vary due to differences in exercise mode, frequency, intensity, and volume, highlighting the need to evaluate the effectiveness of aerobic and resistance modalities relative to depressive and anxiety symptoms and in light of WHO physical activity guidelines.

Prior meta-analyses and trials indicate exercise benefits for mood disorders, but the magnitude varies. Gordon et al. (2018) reported a moderate effect of resistance training on depressive symptoms, while Cooney et al. (2013) found mixed-mode exercise yielded small benefits comparable to psychological treatments. Mechanistic literature notes neurobiological pathways (e.g., improved mitochondrial function, stress hormone regulation) and psychosocial mechanisms (social engagement, distraction). Conflicting findings may reflect heterogeneity in exercise prescription (mode, intensity, volume) and study design. WHO guidelines recommend weekly aerobic and strengthening activities to achieve substantial health benefits, providing a benchmark for evaluating interventions.

The systematic review was prospectively registered on PROSPERO (CRD42019119341) and followed PRISMA 2020 guidelines. A comprehensive search of EBSCOhost, Scopus, Web of Science, PsycNET, and PubMed (plus manual reference checks) was conducted from inception to February 24, 2024, using a PICO-informed strategy targeting exercise AND depression/anxiety, limited to English-language randomized controlled trials (RCTs). Inclusion criteria: adults aged 18–64 with a clinical diagnosis of depression and/or anxiety; interventions comprising prescribed aerobic, resistance, or combined (mixed) exercise, alone or with the same usual care as controls; primary outcomes of depressive and/or anxiety symptoms. Exclusions included non-prescribed exercise, non-aerobic/resistance modalities (e.g., yoga), non-English, non-RCTs, populations outside age range, or no relevant outcomes. Usual care encompassed medication, CBT, counseling, or health education. Study selection involved independent screening by multiple authors, with disagreements resolved via discussion. Data extraction captured study ID, participant characteristics, country, depression/anxiety outcomes, intervention duration, exercise mode, program variables (including intensity per Norton et al. 2010), setting, and additional outcomes (fitness, self-efficacy, adherence, attendance). Risk of bias was assessed using the Cochrane tool; evidence quality via GRADE; reporting quality via CONSORT adherence. Statistical analysis used DerSimonian–Laird random-effects models. For continuous outcomes, standardized mean differences (SMDs) with 95% CIs were computed from post-test summaries, accommodating varied instruments. Means/SDs were estimated when necessary from medians, SEs, CIs, t-values, p-values, ranges, or IQRs. Binary attendance outcomes used risk ratios (RRs) with 95% CIs. Multi-arm exercise groups were combined to yield single pairwise intervention–control comparisons. Heterogeneity was assessed by Cochran’s Q, I², and τ². Sensitivity analyses omitted each study in turn to detect excessive influence. Funnel plots evaluated small-study effects. Secondary meta-analyses summarized physical fitness (6-min walk, VO₂max), self-efficacy, adherence, and attendance. Subgroup analyses compared aerobic vs resistance/mixed modes for depression and anxiety outcomes.

- Included studies: 32 RCTs (n=3243); meta-analyses: 26 studies (n=2681). - Depression outcome: pooled SMD −0.97 (95% CI −1.28 to −0.66), Z=6.10, p<0.001, favouring exercise with a large effect; heterogeneity high (I²=90.1%, τ²=0.531, Q χ²(25)=253.5, p<0.001). Thirteen studies showed significant effects favouring exercise; eleven showed non-significant benefits; none significantly favoured control. Individual SMDs ranged from −6.32 (Merom 2008) to +0.18 (Blumenthal 1999). Sensitivity analyses found no excessive influence; funnel plot indicated some small-study effects. - Anxiety outcome: pooled SMD −0.66 (95% CI −1.09 to −0.23), Z=3.00, p=0.003, favouring exercise with a moderate effect; heterogeneity high (I²=85.8%, τ²=0.468, Q χ²(11)=77.5, p=0.005). Two studies had significant effects favouring exercise; eight had non-significant benefits; two had non-significant effects favouring control. No study excessively influenced results; funnel plot suggested some small-study effects. - Secondary outcomes: - 6-min walking test (n=2): SMD 3.39 (95% CI 2.48 to 4.30), I²=0%, Z=7.31, p<0.001, favouring exercise. - VO₂max (n=5): SMD 0.40 (95% CI −0.35 to 1.10), I²=88.7%, Z=1.02, p=0.306, not significant. - Self-efficacy (n=3): SMD 0.35 (95% CI −0.37 to 1.08), I²=68.4%, Z=0.95, p=0.341, not significant. - Attendance (n=3): RR 1.11 (95% CI 0.93 to 1.32), I²=35.9%, Z=1.19, p=0.235, not significant. - Subgroup analyses: - Depression: Aerobic (n=18) SMD −1.60 (95% CI −2.22 to −0.98), I²=92.2%, p<0.001; Resistance/mixed (n=6) SMD −0.89 (95% CI −1.36 to −0.42), I²=79.6%, p<0.001; no significant between-group difference (Z=0.538, p=0.591). - Anxiety: Aerobic (n=8) SMD −0.56 (95% CI −1.15 to −0.03), I²=89.9%, p=0.090 (not significant); Resistance/mixed (n=4) SMD −0.83 (95% CI −1.42 to −0.25), I²=63.5%, p=0.005; no significant between-group difference (Z=0.408, p=0.683). - Reporting and bias: CONSORT adherence averaged 67% (range 38–84%); 41% high reporting quality (≥75%), 47% adequate (50–75%). Main risks of bias: performance bias (participant/researcher blinding) and outcome assessor blinding (69% high/unclear).

The findings demonstrate that prescribed exercise interventions significantly improve depressive symptoms and moderately improve anxiety symptoms in adults with diagnosed conditions. For depression, benefits were robust across aerobic, resistance, and mixed modalities, indicating flexibility in exercise prescription. For anxiety, resistance or mixed modes showed significant benefits, whereas aerobic alone did not consistently reach significance, suggesting modality may play a more nuanced role. The analyses also suggest that greater adherence to WHO physical activity guidelines might enhance outcomes, as few included anxiety studies met weekly minimums. Despite high heterogeneity, sensitivity analyses indicated stability of pooled effects without undue influence from any single study. Exercise remains underutilized in standard treatment models; integrating exercise as an adjunct or standalone option could address physiological and psychosocial aspects of depression and anxiety and mitigate medication-related drawbacks.

Exercise-based interventions—whether aerobic, resistance, or combined—are effective in reducing depressive symptoms and moderately effective for anxiety symptoms in clinically diagnosed adults. Mode of delivery does not materially alter benefits for depression; resistance or mixed modalities may be preferable for anxiety. Clinicians should consider exercise as an adjunct or standalone treatment within routine care. Future research should improve methodological rigor and reporting (especially detailed exercise prescriptions, adherence, attendance, sample size calculations), adhere closely to CONSORT guidelines, and explore optimal dose, intensity, supervision, and cost-effectiveness to refine exercise recommendations.

- High heterogeneity across studies in exercise mode, intensity, duration, supervision, and delivery; diverse measurement instruments necessitating SMDs. - Many studies had small sample sizes; baseline imbalances in outcomes were observed despite randomization. - Limited reporting quality for key methodological aspects (e.g., blinding, allocation sequence, sample size calculations); frequent high/unclear risk of performance and detection bias. - Some data required estimation from medians/IQRs or extraction from figures, introducing potential imprecision. - Evidence of small-study effects in funnel plots suggests possible publication or related biases. - Limited number of anxiety studies; few met WHO guideline thresholds for exercise volume/intensity. - Restricted age range (18–64) and English-only publications reduce generalizability to youth and older adults and non-English contexts. - Incomplete reporting of adherence, attendance, and medication adherence limits clinical translation.