Jiarong Ge1†
Qiyi Wang
Shimeng Wang
Xuan Xiong
Yifei Zhai- 1Sports Science Institution, Nanjing University, Nanjing, Jiangsu, China
- 2College of Physical Education, Yangzhou University, Yangzhou, Jiangsu, China
- 3Institute of Sports Science, Nantong University, Nantong, Jiangsu, China
- 4Department of Physical Education, Nanjing University, Nanjing, Jiangsu, China
Introduction: Cognitive impairment is a core feature of major depressive disorder (MDD) that often persists during remission, significantly affecting psychosocial functioning. While exercise is known to alleviate depressive symptoms, its specific impact on cognitive domains remains variable. This study aims to evaluate the dose–response relationship and comparative effectiveness of different types of exercise interventions on cognitive function in adults with MDD.
Methods: Five electronic databases (Ovid Medline, PsycINFO, Web of Science, Cochrane Central, and Embase) were systematically searched on September 20, 2024.
Study selection: Randomized controlled trials (RCTs) involving adults aged 18 years or older with MDD were included if they examined the effects of exercise interventions and reported at least one cognitive outcome. Two independent reviewers conducted data extraction and quality assessment according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses for Network Meta-Analyses (PRISMA-NMA) guidelines.
Results: A three-level meta-analysis and a Bayesian model-based network meta-analysis were performed. Fifteen RCTs with a total of 1,196 participants were included. A non-linear dose–response relationship was observed, with a minimum effective dose of 967 METs-min/week—within the WHO-recommended exercise range. Among exercise types, Tai Chi/Qigong demonstrated the greatest cognitive benefits. Moderation analysis indicated that younger age and a higher proportion of female participants were associated with stronger cognitive effects.
Conclusion: Exercise significantly improves cognitive function in adults with MDD, with aerobic and mind–body exercises offering particularly strong benefits. These results support the integration of tailored exercise programs into treatment strategies for cognitive dysfunction in MDD.
Systematic review registration: The systematic review was registered in PROSPERO (CRD42024546289): https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42024546289.
1 Introduction
Major depressive disorder (MDD) is among the most common and debilitating neuropsychiatric disorders (1). Low mood, diminished interest or enjoyment in previous activities (2), and frequent suicidal or death-related thoughts (3) are all signs of MDD. It significantly increases health burdens, disability, and suicidality worldwide, affecting approximately 5% of adults globally (4, 5). Notably, cognitive function (CF) plays a critical role in depression, individuals with MDD often report experiencing cognitive impairments (6). This condition often persists even after symptom remission (7), regardless of the circumstances, improving cognitive impairments will be a key focus in the treatment of depression (8, 9). Cognitive impairment is highly prevalent among individuals with MDD, affecting approximately 40–60% of patients during acute episodes and often persisting into remission phases. These deficits primarily involve attention, executive function, and memory, which are closely related to daily functioning and relapse risk. Increasing evidence suggests that such cognitive disturbances are not merely secondary symptoms but represent a core and enduring feature of MDD, contributing to poor treatment outcomes and psychosocial dysfunction (10).
Some studies suggest cognitive deficits recover after remission (10), while others indicate they persist, potentially contributing to recurrent depression (11). Depressed individuals with cognitive impairment experience reduced quality of life, interpersonal dysfunction, and limited academic productivity (12), and antidepressants offer limited cognitive benefits with significant side effects (13). This underscores the need for effective non-pharmacological interventions to address cognitive impairments in depression.
Exercise, as a low-cost and low-side-effect intervention, has been effective in alleviating depression for over 30 years (14). Recent studies have emphasized the cognitive advantages of physical activity, particularly aerobic exercise, which has been demonstrated to enhance memory and executive functioning in individuals with depression (15). However, the impact of exercise on cognition may depend on the type of exercise (16). For example, Tai Chi combined with escitalopram has demonstrated significant cognitive improvements, as has resistance training, which is particularly beneficial for slowing cognitive decline in older adults (17).
Several recent systematic reviews and meta-analyses have explored the cognitive effects of exercise in depression (18, 19). However, most of these studies primarily compared different exercise interventions without exploring dose–response patterns or accounting for dependencies among multiple cognitive outcomes within trials. Moreover, few adopted a Bayesian analytical framework, which enables probabilistic estimation of optimal exercise doses and comparative rankings across modalities. Our study extends the existing literature by integrating three-level meta-analysis and Bayesian model-based dose–response network meta-analysis to comprehensively quantify how exercise dosage and type influence cognitive function in adults with MDD.
Nevertheless, the aforementioned studies overlooked two key priorities when using exercise interventions as treatment methods. First, the type of exercise may moderate the exercise-cognition relationship. Resistance exercise has also been found to enhance cognitive outcomes (20). Second, exercise prescriptions based on the FITT principle (i.e., frequency, intensity, time, and type) are frequently employed to evaluate how each exercise component affects cognitive development. However, considering these exercise characteristics in isolation may overlook the complex interactions between these factors. Additionally, the World Health Organization (WHO) 2020 Guidelines Development Group for Physical Activity and Sedentary Behavior also emphasizes that mood dysregulation in depression may hinder exercise adherence (21, 22), thus the need to determine the optimal and safe exercise dose for cognitive benefits in depressed individuals is necessary.
Despite growing recognition of depression as a leading contributor to the global disease burden, research investigating the cognitive dimension of depression remains fragmented and inconclusive (7). Existing meta-analyses have largely focused on symptom reduction rather than cognitive restoration, leaving a critical knowledge gap in understanding how exercise interventions may optimize cognitive outcomes in individuals with MDD. Furthermore, most prior studies have relied on simplified comparisons between exercise and control groups without addressing the nuanced dose–response patterns that could inform personalized exercise prescriptions (23). Given the World Health Organization’s call for scalable and sustainable strategies to improve mental and cognitive health worldwide, a comprehensive synthesis that quantifies the relationship between exercise parameters and cognitive improvement is urgently needed. This study therefore provides a timely contribution by integrating advanced meta-analytic approaches to clarify how different exercise modalities and doses can be optimized to support cognitive recovery in depression.
Overall, in order to address multiple effect sizes and retrieve pertinent randomized controlled trials, this study employs two innovative meta-analysis techniques: three-level meta-analysis and model-based dose–response network meta-analysis in a Bayesian framework. It aims to explore the dose–response relationship between exercise and CF in depressed patients and identify the optimal exercise dosage and type.
2 Methods
This study protocol was preregistered (PROS-PERO reference number #CRD42024546289) and all study procedures have followed guidelines recommended by the Preferred Reporting Items for Systematic Reviews and Network Meta-Analyses (PRISMA-NMA checklist) (24).
2.1 Search strategy and selection criteria
Five electronic databases were searched for relevant literature (e.i., Ovid Medline, PSYCINFO, Web of Science, Cochrane Central, and Embase) up to September 2024 (see Supplementary File 1). Two investigators (JRG and QYW) independently selected researches, retrieved pertinent information, and evaluated the possibility of bias. Discrepancies were resolved by consensus with the review team (JRG, QYW, and XX).
Inclusion criteria for this meta-analysis were pre-determined: (1) only chronic randomized controlled trials with at least one type of exercise for depression; (2) individuals (18 years of age or older) diagnosed with MDD based on accepted or established diagnostic criteria (e.g., Beck Depression Inventory-II ≥13) (25); (3) a control group receiving usual care or placebo (e.g., stretching); (4) studies reporting global cognition or domain-specific measures (e.g., executive function, attention); (5) articles in English language journals. We excluded studies combining multiple treatments (e.g., exercise plus cognitive therapy), those with participants having severe health conditions (e.g., Parkinson’s), and studies with intervention durations <4 weeks.
2.2 Data extraction and coding strategy
General study characteristics was extracted by three reviewers (JRG, QYW, XX), including participant characteristic, exercise protocol, control group, and main results (i.e., mean and standard deviation (SD) for calculating effect sizes). The diagnostic criteria for MDD are determined by a physician, while cognitive function is assessed by psychologists using standardized instruments and tasks. (see Supplementary Table 5).
To clarify the posterior data synthesis process, we defined data coding in four hierarchical levels. First, interventions were coded as “Exercise” or “Control”; second, as “Aerobic Training,” “Resistance Training,” or “Control.” At the third level, interventions were categorized by exercise type: “Cycling,” “Mixed aerobic exercises,” “Strength,” “Tai-chi/Qigong,” “Walking/Jogging,” and “Placebo” (control). Finally, interventions were classified based on their energy expenditure (Metabolic Equivalent of Task, MET), expressed as METs-min per week, determined by the product of exercise duration, frequency, and intensity (26), We then grouped interventions into four categories based on METs-min per week: 0 (control), 500, 750, and 1,000 METs-min. This classification was made to improve network connectivity, necessary for the network meta-analysis (27).
2.3 Three-level meta-analysis
This method was applied to address multiple effect sizes and their non-independence. It accounts for sampling variance (Level 1), within-study variance (Level 2), and between-study variance (Level 3), using restricted maximum likelihood (REML) estimation to calculate standardized mean difference (SMD) (e.g., Hedges’ g) and 95% confidence intervals (CI) for the effect of exercise on CF. The meta-analysis synthesized CF assessments, conducted subgroup analyses on CF subdomains and study/sample characteristics, and explored significant factors through regression analysis.
2.4 Bayesian model-based network meta-analysis
A Bayesian model-based network meta-analysis (MBNMA) (28) was employed to provide an overview of the dose–response relationship between cognitive performance and exercise. No violations of key assumptions for network meta-analysis (connectivity, consistency, and transitivity) were found (see Supplementary File 2). Effect sizes were standardized as Hedges’ g (29), with posterior medians and 95% credible intervals (CrI) to assess certainty (30).
We used non-linear models to describe the data and illustrated how various forms and dosages of physical activity affected CF, including Emax, restricted cubic splines and non-parametric monotonically up functions (31). Fit indices (32) and deviance plots were used to compare models (see Supplementary Figures 6, 7), with restricted cubic splines providing the best fit for dose–response associations (see Supplementary Figure 5). Based on the optimal fit model (see Supplementary Table 2) and biological plausibility (31), three knots were positioned at the 10th, 50th, and 90th percentiles of exercise dose (33). To ensure interpretability and adequate network connectivity, exercise doses were grouped into four METs-min categories (0, 500, 750, and 1,000 METs-min per week). These thresholds were derived based on WHO recommendations for physical activity (500–1,000 METs-min/week as the optimal range for health benefits) and the dose distribution across included studies. This categorization allowed for both clinical relevance and statistical stability in the MBNMA. The “optimal dose” of physical activity for the maximal possible cognitive benefit was estimated using beta coefficients from restricted cubic splines. This data was then used to order the examined treatments from worst to best according to their likelihood of improving CF.
The Minimum Clinically Important Difference (MCID) (34) was defined as 0.5 SD, following established conventions in neuropsychological and clinical trials research. A change of this magnitude has been shown to correspond to a perceptible and clinically meaningful improvement in cognitive functioning among individuals with MDD. MCID was used to determine the exercise dose for clinically significant cognitive improvements in MDD, assessed via Montreal Cognitive Assessment (MoCA) (35) and Mini-Mental State Examination (MMSE) (36), with effect sizes compared to a standardized MCID of 0.5 SD (37).
2.5 Statistical analysis
Both three-level meta-analysis and Bayesian meta-analyses were conducted in R 4.4.1 (38). We used the ‘MBNMA dose’ package (39) to perform Bayesian dose–response MBNMA models; the ‘metaphor’ package to conduct a three-level meta-analysis (40); the ‘brms’ package to perform regression with moderator factors (41) and the ‘ggplot2’ package for plotting and visualization (42).
2.6 Risk of bias and quality assessment
The studies were evaluated separately by three reviewers (JRG, XX, and YFZ) using the PEDro scale (43) for study quality and the Cochrane Risk of Bias Tool (44) for randomized trials. Study quality was classified as high, moderate and poor. Disagreements were resolved by discussion. In order to assess their influence on the total dose–response estimates, a sensitivity analysis was also carried out, eliminating studies with a high risk of bias.
3 Results
3.1 Study selection
The study selection procedure is depicted in the PRISMA flow diagram (Figure 1). We retrieved 8,896 studies, and after removing duplicates, 6,128 remained. After two reviewers independently vetted abstracts and titles, 105 pertinent studies were found. Following the application of inclusion/exclusion criteria, the meta-analysis contained 15 studies (43–59), with 36 arms and 1,196 participants. Supplementary Table 5 details the references and characteristics of these studies. Of the participants, 438 (36.6%) were female, with an average age of 42.3 years (SD = 14.5). A total of 337 (28.2%) were overweight, and 630 (52.7%) had mild to moderate depression. The treatment-level and agent-level network diagram is shown in Supplementary Figures 1, 2.
Figure 1. Flow of studies through review.
3.2 Global cognitive function
This study demonstrated that different types of exercise (135 effect sizes, k = 15) had a small but significant effect on overall cognitive function [(SMD, 0.329; 95% CI, 0.25 to 0.41); standard error (SE) = 0.04]. The likelihood ratio tests (LRTs) showed nonsignificant variance in effects within studies at level 2 (LRT = 0, p > 0.05), while showing significant variance between studies at level 3 (LRT = 6.75, p = 0.0094). The statistical heterogeneity in the aggregated effect magnitude was minimal (I2 = 17.30%). A p-value below 0.05 indicates significant heterogeneity, necessitating a moderator analysis to explore its origins.
3.3 Moderator analyses
Table 1 presents subgroup analyses based on gender, age, BMI, depression severity, and CF subdomains, identifying key moderators of CF. Exercise had a significant impact, especially in females (SMD, 0.41; 95% CI, 0.22–0.60) and young adults (SMD, 0.32; 95% CI, 0.14–0.51), with minimal effects in middle-aged (SMD, 0.09; 95% CI, −0.21–0.38) and older adults (SMD, −0.13; 95% CI, −0.68–0.41). BMI showed no significant role, whether normal (SMD, 0.09; 95% CI, −0.28–0.45) or overweight (SMD, 0.36; 95% CI, −0.06–0.79). Exercise had the largest effect in individuals with mild to moderate depression (SMD, 0.26; 95% CI, 0.08–0.4). However, it did not significantly improve cognitive subdomains, including executive function (SMD, −0.07; 95% CI, −0.53–0.40), memory (SMD, 0.05; 95% CI, −0.38–0.48; p = 0.81), attention (SMD, −0.01; 95% CI, −0.45–0.44), and processing speed (SMD, −0.10; 95% CI, −0.55–0.37).
Table 1. Moderator analyses of exercise on cognition.
Figure 2 shows the linear regression between cognitive function and the two key moderators: gender and age. Depression severity was excluded to maintain objectivity. The results indicate that younger participants and those with a higher proportion of females experience more significant cognitive improvements from exercise. In studies with participants aged 18–44 and a female proportion >50%, the fitted dose–response relationship shows MCID values of 501 METs-min/week and 761 METs-min/week, respectively (see Supplementary Figures 8, 9).
Figure 2. Regression analysis between effects of cognitive function and moderators (sex and age). (A) Regression between proportion of females and cognitive effect size; (B) Regression between mean age and cognitive effect size.
3.4 Dose–response relationships
Figure 3 illustrates the estimated dose–response relationship between physical activity dose and cognitive improvement in adults with MDD.
Figure 3. Dose–response relationship between overall physical activity and different types of physical activity on cognitive function in adults with MDD. MBNMA, Model-based network meta-analysis; MCID, Minimum clinically important difference; ECT, Effective critical threshold. (A) Overall dose–response relationship between physical activity and cognitive function; (B) Dose–response relationships by exercise type.
Panel A shows the nonlinear dose–response relationship between exercise and CF. CF improves significantly beyond 600 METs-min (WHO’s lower bound), with a slight plateau after 1,200 METs-min (WHO’s upper bound) (57). The combined effect size, based on the minimal clinically important difference (MCID), is moderate (SMD = 0.39), with values of 0.29 [95% CrI, 0.17–0.43] at 600 METs-min and 0.42 [95% CrI, 0.32–0.53] at 1200 METs-min. The minimum exercise dose to achieve this effect is approximately 967 METs-min/week (SMD, 0.39; 95% CI, 0.27–0.53). Predicted effects for the lower and upper WHO activity levels are shown in Supplementary Table 4. Panel A indicates that cognitive benefits increase rapidly up to approximately 967 METs-min/week, after which the curve plateaus. This threshold corresponds to the minimum effective dose reaching the minimum clinically important difference (MCID, 0.5 SD), which lies within the WHO-recommended range (500–1,000 METs-min/week).
Figure 3 illustrates the dose–response relationship between exercise types and CF in adults with MDD. Cycling, Mixed aerobic exercises, and Walking/Jogging show inverted U-shaped patterns, with peak effect sizes at 914 METs-min (SMD, 0.63; 95% CrI, 0.31–0.93), 864 METs-min (SMD, 0.89; 95% CrI, 0.52–1.23), and 810 METs-min (SMD, 0.64; 95% CrI, 0.26–1.12), respectively. The MCID for these exercises ranges from 376 to 1,582 METs-min/week, 243–1,652 METs-min/week, and 312–1,445 METs-min/week. At doses of 1713 METs-min/week (SMD, 0.33; 95% CrI, 0–0.61) and 1724 METs-min/week (SMD, 0.17; 95% CrI, 0–0.31), Cycling and Walking/Jogging no longer show significant effects, roughly three times the lower limit of the WHO’s recommended exercise dose (58). The Strength exercise dose–response curve follows a whale-like shape, with significant effects only within 861–1,403 METs-min/week and no MCID or maximum dose. Tai-chi/Qigong exhibits a U-shaped pattern, but its wide CI limits significant effects to 1,237 METs-min/week (SMD, 0.47; 95% CrI, 0–0.96).
Panel B presents the dose–response patterns by exercise type. Tai Chi/Qigong demonstrated a nearly linear improvement across the exercise range, whereas aerobic modalities (cycling, walking/jogging, and mixed aerobic exercises) exhibited an inverted-U pattern, with optimal effects observed between 750 and 1,200 METs-min/week. Resistance training showed smaller but consistent improvements. These findings highlight both a global optimal dose and exercise-specific variations in cognitive response.
Supplementary Table 3 compares the effects of different exercises at various doses. Tai-chi/Qigong (1,000 METs-min/week) shows the most favorable effect, followed by Mixed aerobic exercise (500 METs-min/week). Table 2 offers practical exercise recommendations based on these findings.
Table 2. Exercise recommendations to improve cognitive function for adults with major depressive disorder.
Across included studies, cognitive domains most frequently assessed were executive function, attention, and memory. Subgroup analyses suggested slightly stronger effects on executive function and attention, consistent with previous evidence linking exercise to fronto-executive network enhancement in MDD.
3.5 Risk of bias and quality assessment
Of the studies, six were rated low risk, five moderate risk, and four high risk (see Supplementary Figure 11). The average quality score was 6.25 (see Supplementary Table 6). Sensitivity analyses, excluding high-risk and unclear studies, confirmed an inverted U-shaped dose–response relationship, with an optimal dose of 923 METs-min/week (see Supplementary Figure 10).
4 Discussion
4.1 Nonlinear associations between exercise and cognitive function
This study reveals several clinically significant findings. First, physical activity interventions improve cognitive function (CF) in adults with MDD. Second, our results demonstrate a distinct non-linear connection between exercise and CF, with impact values ranging from minor to moderate. The minimum clinically effective dose is 967 METs-min per week, equivalent to 250 min of moderate-intensity exercise or 125 min of vigorous exercise, consistent with WHO guidelines (19). This suggests that maintaining a tolerable level of weekly physical activity is beneficial for CF, offering a reference for experts in developing exercise plans for MDD patients, many of whom may lack motivation for physical activity. However, cognitive improvements plateau once the dose exceeds the WHO-recommended 1,200 METs-min/week. Notably, the dose–response relationship between exercise and CF mirrors that observed for alleviating depressive symptoms (59). Specifically, before reaching the WHO-recommended dose, higher exercise levels correlate with better outcomes, which could guide research on increasing exercise doses for patients whose CF remains impaired despite improvements in depression (60).
4.2 Exercise types and dose–response relationships in cognitive function
Most exercise types in our study effectively improve CF, with identifiable dose–response patterns. Notably, Tai Chi/Qigong emerged as the most effective for enhancing CF in MDD patients, though a certain intensity is required, and further validation is needed. For other exercise types, dose–response relationships generally follow an inverted U-shape, consistent with a similar research in older adults (61), though our effect sizes may be more pronounced. This relationship likely varies by exercise type, reflecting different mechanisms that induce cognitive changes. Tai Chi/Qigong’s effectiveness may stem from its inherent integration of physical exercise with cognitive therapy. Beyond physiological adaptations, the superior efficacy of Tai Chi and Qigong observed in this study may be mediated by their dual emphasis on physical exertion and mental focus. Previous research indicates that mind–body interventions significantly reduce sympathetic nervous system activity and cortisol levels more effectively than aerobic exercise alone (62). These modalities require sustained attention and motor planning, which may directly engage the prefrontal cortex and hippocampus—regions often compromised in MDD. Consequently, the integration of mindfulness with movement appears to offer a synergistic benefit for executive control and working memory that pure resistance or aerobic training may not fully replicate. However, the minimum threshold dose for this intervention is 1,237 METs-min per week, which requires a considerable time and physical commitment from most patients. Therefore, patients might consider more accessible exercises like cycling, mixed aerobic exercises, and walking, which also produce significant effects. Their minimum clinically important differences are 376, 243, and 312 METs-min per week, respectively.
Our study categorizes exercises into two main types based on training characteristics: aerobic exercise and resistance training. Adults with MDD can improve their CF with aerobic exercise (63), which aligns with our findings. Other meta-analyses also show its significant impact on cognitive function in conditions like schizophrenia and dementia (64), making it the most common and effective intervention. The mechanisms behind its effects on cognitive performance require further exploration from a neuropsychological perspective (65). In contrast, resistance training appears to be a higher-threshold and less effective intervention for MDD patients. We could not identify its optimal dose or MCID, and its effectiveness is limited to a range of 861 to 1,403 METs-min per week. However, some studies suggest that even low doses of resistance training can induce clinically significant cognitive changes and may offer superior overall cognitive improvement (66). This discrepancy could stem from their focus on older populations.
4.3 The influence of moderators’ factors on cognitive function
Gender and age were identified as significant moderators of CF in adults with MDD. Evidence shows that individuals aged 18–44 years are particularly affected, with a weekly beneficial dose of at least 501 METs-min. The moderating effect is more pronounced in studies with over 50% female participants.
The moderation analysis revealed stronger cognitive benefits in younger adults (18–44 years) and females. Biological mechanisms likely underpin the age-related findings; younger adults typically exhibit greater neuroplasticity and higher baseline levels of Brain-Derived Neurotrophic Factor (BDNF) in response to physical stress compared to older populations (67). Regarding gender, the stronger response observed in women may be linked to the interaction between sex hormones and the dopaminergic system. Estrogen has been shown to enhance synaptic plasticity and may potentiate the neurotrophic effects of exercise (68). Furthermore, sociodemographic data suggests that women with MDD may exhibit different baseline physical activity behaviors and adherence rates compared to men, potentially influencing the net effectiveness of the intervention.
4.4 Strengths, limitations and future directions
Mechanistically, the cognitive enhancement observed in MDD through exercise may involve increased neuroplasticity, elevated brain-derived neurotrophic factor (BDNF) levels, reduced systemic inflammation, and improved stress regulation via the hypothalamic–pituitary–adrenal (HPA) axis. Exercise can also enhance prefrontal and hippocampal activity, regions crucial for attention, memory, and emotional regulation. While part of the cognitive improvement may result from alleviation of depressive symptoms, emerging evidence suggests that exercise-induced neurobiological changes independently contribute to cognitive restoration.
Our research has a number of advantages. First, we applied a three-level meta-analysis alongside traditional network meta-analysis to address test result merging and manage heterogeneity within and between studies. Second, we used the latest Bayesian models to estimate exercise doses, providing more precise insights, particularly for developing exercise prescriptions. Finally, our findings align with the WHO’s Global Action Plan on Physical Activity (GAPPA) target for 2030, promoting “more active people for a healthier world.” Precise exercise dosing allows office workers and students to achieve similar benefits to continuous exercise by accruing brief bursts of exercise over the course of the week, known as “exercise snacks” (69). Nevertheless, this study has several limitations. First, experimental biases may have arisen from human factors, as both staff and participants could infer the study’s objectives. Exercise interventions, unlike pharmaceutical ones, are especially challenging to double-blind. Future studies should ensure the research hypothesis remains unknown to all parties, ideally with third-party supervision. Second, moderating factors like gender and age were assessed at the study level, not the individual level, which may explain some heterogeneity. To more accurately evaluate these factors, future reviews ought to take individual participant meta-analyses into account.
5 Conclusion
This research delineates the therapeutic effect of exercise on CF in individuals with MDD, identifying dose–response connections between exercise kinds and cognitive outcomes. Aerobic exercise requires a minimum of 60 min per week to improve CF, a feasible amount for all age groups, while strength training needs about 210 min per week for similar benefits. These findings offer valuable exercise-related recommendations to help address the global challenge of depression.
Data availability statement
The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding authors.
Author contributions
JG: Writing – review & editing, Writing – original draft. QW: Writing – review & editing, Writing – original draft. SW: Writing – review & editing, Writing – original draft. XX: Writing – original draft, Writing – review & editing. YZ: Writing – original draft, Writing – review & editing.
Funding
The author(s) declared that financial support was not received for this work and/or its publication.
Conflict of interest
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that Generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Supplementary material
The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpubh.2025.1662778/full#supplementary-material
Abbreviations
MDD, Major depressive disorder; CF, Cognitive function; SD, Standard deviation; MET, Metabolic equivalent of task; MBNMA, Model-based network meta-analysis; MCID, Minimum clinically important difference; REML, Restricted maximum likelihood; CI, Confidence intervals; MoCA, Montreal cognitive assessment; MMSE, Mini-mental state examination; SMD, Standardized mean difference; SE, Standard error; LRTs, Likelihood ratio tests; CrI, Credible intervals; GAPPA, Global Action Plan on Physical Activity.
References
1. Belmaker, RH, and Agam, G. Major depressive disorder. N Engl J Med. (2008) 358:55–68. doi: 10.1056/NEJMra073096,
2. Cuijpers, P, Noma, H, Karyotaki, E, Vinkers, CH, Cipriani, A, and Furukawa, TA. A network meta-analysis of the effects of psychotherapies, pharmacotherapies and their combination in the treatment of adult depression. World Psychiatry. (2020) 19:92–107. doi: 10.1002/wps.20701,
3. Marx, W, Penninx, BWJH, Solmi, M, Furukawa, TA, Firth, J, Carvalho, AF, et al. Major depressive disorder. Nat Rev Dis Primers. (2023) 9:44. doi: 10.1038/s41572-023-00454-1,
4. Birnie, MT, Eapen, AV, Kershaw, YM, Lodge, D, Collingridge, GL, Conway-Campbell, BL, et al. Time of day influences stress hormone response to ketamine. J Neuroendocrinol. (2022) 34:e13194. doi: 10.1111/jne.13194,
5. Kawilapat, S, Maneeton, B, Maneeton, N, Prasitwattanaseree, S, Kongsuk, T, Arunpongpaisal, S, et al. Comparison of unweighted and item response theory-based weighted sum scoring for the nine-questions depression-rating scale in the northern Thai dialect. BMC Med Res Methodol. (2022) 22:268. doi: 10.1186/s12874-022-01744-0,
6. Spinhoven, P, Van Hemert, AM, and Penninx, BW. Repetitive negative thinking as a predictor of depression and anxiety: a longitudinal cohort study. J Affect Disord. (2018) 241:216–25. doi: 10.1016/j.jad.2018.08.037
7. Rock, PL, Roiser, JP, Riedel, WJ, and Blackwell, AD. Cognitive impairment in depression: a systematic review and meta-analysis. Psychol Med. (2014) 44:2029–40. doi: 10.1017/S0033291713002535,
8. Kriesche, D, Woll, CFJ, Tschentscher, N, Engel, RR, and Karch, S. Neurocognitive deficits in depression: a systematic review of cognitive impairment in the acute and remitted state. Eur Arch Psychiatry Clin Neurosci. (2023) 273:1105–28. doi: 10.1007/s00406-022-01479-5,
9. Pan, Z, Park, C, Brietzke, E, Zuckerman, H, Rong, C, Mansur, RB, et al. Cognitive impairment in major depressive disorder. CNS Spectr. (2019) 24:22–9. doi: 10.1017/S1092852918001207,
10. Biringer, E, Mykletun, A, Sundet, K, Kroken, R, Stordal, KI, and Lund, A. A longitudinal analysis of neurocognitive function in unipolar depression. J Clin Exp Neuropsychol. (2007) 29:879–91. doi: 10.1080/13803390601147686,
11. Hasselbalch, BJ, Knorr, U, and Kessing, LV. Cognitive impairment in the remitted state of unipolar depressive disorder: a systematic review. J Affect Disord. (2011) 134:20–31. doi: 10.1016/j.jad.2010.11.011,
12. Keefe, RSE, Cañadas, E, Farlow, D, and Etkin, A. Digital intervention for cognitive deficits in major depression: a randomized controlled trial to assess efficacy and safety in adults. AJP. (2022) 179:482–9. doi: 10.1176/appi.ajp.21020125,
13. Rosenblat, JD, Kakar, R, and McIntyre, RS. The cognitive effects of antidepressants in major depressive disorder: a systematic review and meta-analysis of randomized clinical trials. IJNPPY. (2016) 19:pyv082. doi: 10.1093/ijnp/pyv082,
14. North, TC, McCULLAGH, P, and Tran, ZV. Effect of exercise on depression. Exerc Sport Sci Rev. (1990) 18:379???416. doi: 10.1249/00003677-199001000-00016,
15. Ren, FF. Effects of aerobic exercise on cognitive function in adults with major depressive disorder: a systematic review and meta-analysis. Int J Clin Health Psychol. (2024) 24:447. doi: 10.1016/j.ijchp.2024.100447,
16. Garrett, JG. Characterizing the Influence of Perturbations in Global State on Working Memory (Doctoral dissertation). Santa Barbara: University of California (2024).
17. Lavretsky, H, Milillo, MM, Kilpatrick, L, Grzenda, A, Wu, P, Nguyen, SA, et al. A randomized controlled trial of Tai Chi Chih or health education for geriatric depression. Am J Geriatr Psychiatry. (2022) 30:392–403. doi: 10.1016/j.jagp.2021.07.008,
18. Liu, Y, Zhao, G, Guo, J, Qu, H, Kong, L, and Yue, W. The efficacy of exercise interventions on depressive symptoms and cognitive function in adults with depression: an umbrella review. J Affect Disord. (2025) 368:779–88. doi: 10.1016/j.jad.2024.09.074,
19. Dong, H, Liu, C, and Qin, M. The intervention effect of exercise on the attention of patients with depression: a systematic review. Front Psychol. (2025) 16:1536262. doi: 10.3389/fpsyg.2025.1536262,
20. Northey, JM, Cherbuin, N, Pumpa, KL, Smee, DJ, and Rattray, B. Exercise interventions for cognitive function in adults older than 50: a systematic review with meta-analysis. Br J Sports Med. (2018) 52:154–60. doi: 10.1136/bjsports-2016-096587,
21. Bull, FC, Al-Ansari, SS, Biddle, S, Borodulin, K, Buman, MP, Cardon, G, et al. World Health Organization 2020 guidelines on physical activity and sedentary behaviour. Br J Sports Med. (2020) 54:1451–62. doi: 10.1136/bjsports-2020-102955,
22. Erickson, KI, Hillman, C, Stillman, CM, Ballard, RM, Bloodgood, B, Conroy, DE, et al. Physical activity, cognition, and brain outcomes: a review of the 2018 physical activity guidelines. Med Sci Sports Exerc. (2019) 51:1242–51. doi: 10.1249/MSS.0000000000001936,
23. Xu, L, Gu, H, Cai, X, Zhang, Y, Hou, X, Yu, J, et al. The effects of exercise for cognitive function in older adults: a systematic review and meta-analysis of randomized controlled trials. Int J Environ Res Public Health. (2023) 20:1088. doi: 10.3390/ijerph20021088,
24. Page, MJ, McKenzie, JE, Bossuyt, PM, Boutron, I, Hoffmann, TC, Mulrow, CD, et al. Updating guidance for reporting systematic reviews: development of the PRISMA 2020 statement. J Clin Epidemiol. (2021) 134:103–12. doi: 10.1016/j.jclinepi.2021.02.003,
25. Shahid, A, Wilkinson, K, Marcu, S, and Shapiro, CM. Beck depression inventory In: A Shahid, K Wilkinson, S Marcu, and CM Shapiro, editors. STOP, THAT and one hundred other sleep scales [internet]. New York, NY: Springer New York (2011). 63–4.
26. Shephard, RJ. 2011 compendium of physical activities: a second update of codes and MET values. Yearb Sports Med. (2012) 2012:126–7. doi: 10.1249/MSS.0b013e31821ece12
27. Higgins, JPT, Jackson, D, Barrett, JK, Lu, G, Ades, AE, and White, IR. Consistency and inconsistency in network meta-analysis: concepts and models for multi-arm studies. Res Synth Methods. (2012) 3:98–110. doi: 10.1002/jrsm.1044,
28. Mawdsley, D, Bennetts, M, Dias, S, Boucher, M, and Welton, N. Model-based network Meta-analysis: a framework for evidence synthesis of clinical trial data. CPT Pharmacom Syst Pharma. (2016) 5:393–401. doi: 10.1002/psp4.12091,
29. Hedges, LV, and Olkin, I. Meta-analysis in the physical and biological sciences In: Statistical methods for meta-analysis. New York, NY, USA: Elsevier (1985). 311–25.
30. Etzioni, RD, and Kadane, JB. Bayesian statistical methods in public health and medicine. Annu Rev Public Health. (1995) 16:23–41. doi: 10.1146/annurev.pu.16.050195.000323,
31. Pedder, H, Dias, S, Bennetts, M, Boucher, M, and Welton, NJ. Modelling time-course relationships with multiple treatments: model-based network meta-analysis for continuous summary outcomes. Res Synth Methods. (2019) 10:267–86. doi: 10.1002/jrsm.1351,
32. Evans, NJ. Assessing the practical differences between model selection methods in inferences about choice response time tasks. Psychon Bull Rev. (2019) 26:1070–98. doi: 10.3758/s13423-018-01563-9,
33. Hammar, Å, Ronold, EH, and Rekkedal, GÅ. Cognitive impairment and neurocognitive profiles in major depression—a clinical perspective. Front Psych. (2022) 13:764374. doi: 10.3389/fpsyt.2022.764374,
34. Bernstein, JA, and Mauger, DT. The minimally clinically important difference (MCID): what difference does it make? J Allergy Clin Immunol Pract. (2016) 4:689–90. doi: 10.1016/j.jaip.2016.04.022,
35. Freitas, S, Simões, MR, Alves, L, Vicente, M, and Santana, I. Montreal cognitive assessment (MoCA): validation study for vascular dementia. J Int Neuropsychol Soc. (2012) 18:1031–40. doi: 10.1017/S135561771200077X,
36. Molloy, DW, and Standish, TIM. A guide to the standardized Mini-mental state examination. Int Psychogeriatr. (1997) 9:87–94.
37. Hengartner, MP, and Plöderl, M. Estimates of the minimal important difference to evaluate the clinical significance of antidepressants in the acute treatment of moderate-to-severe depression. BMJ EBM. (2022) 27:69–73. doi: 10.1136/bmjebm-2020-111600,
38. Antoch, J. Environment for statistical computing. Comput Sci Rev. (2008) 2:113–22. doi: 10.1016/j.cosrev.2008.05.002
39. Pedder, H, Dias, S, Boucher, M, Bennetts, M, Mawdsley, D, and Welton, NJ. Methods to assess evidence consistency in dose-response model based network meta-analysis. Stat Med. (2022) 41:625–44. doi: 10.1002/sim.9270,
40. Assink, M, and Wibbelink, CJM. Fitting three-level meta-analytic models in R: a step-by-step tutorial. TQMP. (2016) 12:154–74. doi: 10.20982/tqmp.12.3.p154
41. Bürkner, PC. Brms: an R package for Bayesian multilevel models using Stan. J Stat Softw. (2017) 80:1–28. doi: 10.18637/jss.v080.i01
42. Wilkinson, L. ggplot2: elegant graphics for data analysis by Wickham, H. Biometrics. (2011) 67:678–9. doi: 10.1111/j.1541-0420.2011.01616.x
43. Maher, CG, Sherrington, C, Herbert, RD, Moseley, AM, and Elkins, M. Reliability of the PEDro scale for rating quality of randomized controlled trials. Phys Ther. (2003) 83:713–21. doi: 10.1093/ptj/83.8.713,
44. Jørgensen, L, Paludan-Müller, AS, Laursen, DRT, Savović, J, Boutron, I, Sterne, JAC, et al. Evaluation of the Cochrane tool for assessing risk of bias in randomized clinical trials: overview of published comments and analysis of user practice in Cochrane and non-Cochrane reviews. Syst Rev. (2016) 5:80. doi: 10.1186/s13643-016-0259-8,
45. Brush, CJ, Hajcak, G, Bocchine, AJ, Ude, AA, Muniz, KM, Foti, D, et al. A randomized trial of aerobic exercise for major depression: examining neural indicators of reward and cognitive control as predictors and treatment targets. Psychol Med. (2022) 52:893–903. doi: 10.1017/S0033291720002573,
46. Buschert, V, Prochazka, D, Bartl, H, Diemer, J, Malchow, B, Zwanzger, P, et al. Effects of physical activity on cognitive performance: a controlled clinical study in depressive patients. Eur Arch Psychiatry Clin Neurosci. (2019) 269:555–63. doi: 10.1007/s00406-018-0916-0,
47. Foley, LS, Prapavessis, H, Osuch, EA, De Pace, JA, Murphy, BA, and Podolinsky, NJ. An examination of potential mechanisms for exercise as a treatment for depression: a pilot study. Ment Health Phys Act. (2008) 1:69–73. doi: 10.1016/j.mhpa.2008.07.001
48. Hoffman, BM, Blumenthal, JA, Babyak, MA, Smith, PJ, Rogers, SD, Doraiswamy, PM, et al. Exercise fails to improve neurocognition in depressed middle-aged and older adults. Med Sci Sports Exerc. (2008) 40:18580416:1344–52. doi: 10.1249/MSS.0b013e31816b877c
49. Imboden, C, Gerber, M, Beck, J, Holsboer-Trachsler, E, Pühse, U, and Hatzinger, M. Aerobic exercise or stretching as add-on to inpatient treatment of depression: similar antidepressant effects on depressive symptoms and larger effects on working memory for aerobic exercise alone. J Affect Disord. (2020) 276:866–76. doi: 10.1016/j.jad.2020.07.052,
50. Khatri, P, Blumenthal, JA, Babyak, MA, Craighead, WE, Herman, S, Baldewicz, T, et al. Effects of exercise training on cognitive functioning among depressed older men and women. J Aging Phys Act. (2001) 9:43–57. doi: 10.1123/japa.9.1.43
51. Krogh, J, Saltin, B, Gluud, C, and Nordentoft, M. The DEMO trial: a randomized, parallel-group, observer-blinded clinical trial of strength versus aerobic versus relaxation training for patients with mild to moderate depression. J Clin Psychiatry. (2009) 70:790–800. doi: 10.4088/JCP.08m04241,
52. Krogh, J, Videbech, P, Thomsen, C, Gluud, C, and Nordentoft, M. DEMO-II trial. Aerobic exercise versus stretching exercise in patients with major depression—a randomised clinical trial. PLoS One. (2012) 7:e48316. doi: 10.1371/journal.pone.0048316,
53. Krogh, J, Rostrup, E, Thomsen, C, Elfving, B, Videbech, P, and Nordentoft, M. The effect of exercise on hippocampal volume and neurotrophines in patients with major depression–a randomized clinical trial. J Affect Disord. (2014) 165:24–30. doi: 10.1016/j.jad.2014.04.041,
54. Lavretsky, H, Alstein, LL, Olmstead, RE, Ercoli, LM, Riparetti-Brown, M, St Cyr, N, et al. Complementary use of Tai Chi Chih augments escitalopram treatment of geriatric depression: a randomized controlled trial. Am J Geriatr Psychiatry. (2011) 19:839–50. doi: 10.1097/JGP.0b013e31820ee9ef,
55. Neviani, F, Belvederi Murri, M, Mussi, C, Triolo, F, Toni, G, Simoncini, E, et al. Physical exercise for late life depression: effects on cognition and disability. Int Psychogeriatr. (2017) 29:1105–12. doi: 10.1017/S1041610217000576,
56. Oertel-Knöchel, V, Mehler, P, Thiel, C, Steinbrecher, K, Malchow, B, Tesky, V, et al. Effects of aerobic exercise on cognitive performance and individual psychopathology in depressive and schizophrenia patients. Eur Arch Psychiatry Clin Neurosci. (2014) 264:589–604. doi: 10.1007/s00406-014-0485-9,
57. Olson, RL, Brush, CJ, Ehmann, PJ, and Alderman, BL. A randomized trial of aerobic exercise on cognitive control in major depression. Clin Neurophysiol. (2017) 128:903–13. doi: 10.1016/j.clinph.2017.01.023,
58. Sharma, VK, Das, S, Mondal, S, Goswami, U, and Gandhi, A. Effect of SAHAJ YOGA on neuro-cognitive functions in patients suffering from major depression. Indian J Physiol Pharmacol. (2006) 50:375–83.
59. Zhang, L, Wang, D, Xie, C, Liu, S, Chi, L, Ma, X, et al. The effects of Tai Chi on the executive functions and physical fitness in middle-aged adults with depression: a randomized controlled trial. Evid Based Complement Alternat Med. (2022) 2022:1–16. doi: 10.1155/2022/1589106,
60. Tian, S, Liang, Z, Qui, F, Yu, Y, Wang, C, Zhang, M, et al. Optimal exercise modality and dose to improve depressive symptoms in adults with major depressive disorder: a systematic review and Bayesian model-based network meta-analysis of RCTs. J Psychiatr Res. (2024) 176:384–92. doi: 10.1016/j.jpsychires.2024.06.031,
61. Salehinejad, MA, Ghanavai, E, Rostami, R, and Nejati, V. Cognitive control dysfunction in emotion dysregulation and psychopathology of major depression (MD): evidence from transcranial brain stimulation of the dorsolateral prefrontal cortex (DLPFC). J Affect Disord. (2017) 210:241–8. doi: 10.1016/j.jad.2016.12.036,
62. Gallardo-Gómez, D, Del Pozo-Cruz, J, Noetel, M, Álvarez-Barbosa, F, Alfonso-Rosa, RM, and Del Pozo Cruz, B. Optimal dose and type of exercise to improve cognitive function in older adults: a systematic review and Bayesian model-based network meta-analysis of RCTs. Ageing Res Rev. (2022) 76:101591. doi: 10.1016/j.arr.2022.101591,
63. Lim, K, Pysklywec, A, Plante, M, and Demers, L. The effectiveness of Tai Chi for short-term cognitive function improvement in the early stages of dementia in the elderly: a systematic literature review. CIA. (2019) 14:827–39. doi: 10.2147/CIA.S202055,
64. Balbim, GM, Falck, RS, Barha, CK, Starkey, SY, Bullock, A, Davis, JC, et al. Effects of exercise training on the cognitive function of older adults with different types of dementia: a systematic review and meta-analysis. Br J Sports Med. (2022) 56:933–40. doi: 10.1136/bjsports-2021-104955,
65. Barha, CK, Galea, LA, Nagamatsu, LS, Erickson, KI, and Liu-Ambrose, T. Personalising exercise recommendations for brain health: considerations and future directions. Br J Sports Med. (2017) 51:636–9. doi: 10.1136/bjsports-2016-096710,
66. Gallardo-Gómez, D, Del Pozo-Cruz, J, Pedder, H, Alfonso-Rosa, RM, Álvarez-Barbosa, F, Noetel, M, et al. Optimal dose and type of physical activity to improve functional capacity and minimise adverse events in acutely hospitalised older adults: a systematic review with dose-response network meta-analysis of randomised controlled trials. Br J Sports Med. (2023) 57:37536984:1272–8. doi: 10.1136/bjsports-2022-106409
67. Ren, FF, Alderman, BL, Wang, WG, Chen, FT, Zhou, WS, Zong, WJ, et al. Effects of exercise training on executive functioning in adults with depression: a systematic review and Meta-analysis of randomized controlled trials. Sports Med. (2023) 53:1765–88. doi: 10.1007/s40279-023-01869-2,
68. Upadhayay, N. Comparison of cognitive functions between male and female medical students: a pilot study. J Clin Diagn Res. (2014) 8:BC12–5. doi: 10.7860/JCDR/2014/7490.4449
69. Jones, MD, Clifford, BK, Stamatakis, E, and Gibbs, MT. Exercise snacks and other forms of intermittent physical activity for improving health in adults and older adults: a scoping review of epidemiological. Exp Qual Stud Sports Med. (2024) 54:813–35. doi: 10.1007/s40279-023-01983-1,
Keywords: exercise, cognition, major depressive disorder, dose–response, Bayesian model
Citation: Ge J, Wang Q, Wang S, Xiong X and Zhai Y (2025) Optimal dose and type of exercise to improve cognitive function in adults with major depressive disorder: a systematic review and Bayesian model-based network meta-analysis. Front. Public Health. 13:1662778. doi: 10.3389/fpubh.2025.1662778
Edited by:
Wulf Rössler, Charité University Medicine Berlin, GermanyReviewed by:
José Augusto Marinho Alves, Polytechnic Institute of Santarém, PortugalAndrea Perrottelli, University of Campania “Luigi Vanvitelli”, Italy
Copyright © 2025 Ge, Wang, Wang, Xiong and Zhai. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Xuan Xiong, eGlvbmd4dWFuQG5qdS5lZHUuY24=; Yifei Zhai, enlmQG5qdS5lZHUuY24=
†These authors have contributed equally to this work