Effects of aerobic exercise on cognitive function and quality of life in patients with Alzheimer’s disease: a systematic review and meta-analysis

Alzheimer’s disease (AD) accounts for the vast majority of dementia in older adults and imposes a growing global health and socioeconomic burden. With no curative pharmacologic therapies and concerns about adverse effects of some agents, non-pharmacological interventions such as exercise are of increasing interest. Prior studies suggest that physical activity may benefit cognition and health-related quality of life, but findings in AD populations are mixed, and many reviews combine heterogeneous exercise types. This study asks whether aerobic exercise specifically improves cognitive function, quality of life, and depressive symptoms in people with AD, and whether effects vary by intervention duration, session length, frequency, and country, focusing on commonly used cognitive assessment tools (MMSE and ADAS-cog).

Evidence on exercise in dementia and AD is mixed. Several trials and reviews report that physical activity slows cognitive decline or improves cognition in older adults and individuals with cognitive impairment (e.g., Lautenschlager, Kramer, Kemoun, Heyn). Some studies of non-aerobic exercise report benefits on certain cognitive domains, while other trials and reviews (e.g., Cochrane review by Forbes; Littbrand et al.) are inconclusive or find no effect on cognition. Prior meta-analyses often combined various exercise modalities; some evidence suggests aerobic exercise may be the key driver of benefits in AD (e.g., López-Ortiz; Jia; Zhang). Mechanistically, exercise may promote neuroplasticity, hippocampal neurogenesis, cerebral blood flow and metabolism, and reduce AD-related pathology (amyloid-β, tau), neuroinflammation, and cardiometabolic risk factors. Despite these findings, variability in study design, intervention parameters, and outcomes contributes to inconsistent results, underscoring the need for modality-specific synthesis focusing on aerobic exercise and standardized assessment tools.

Protocol and reporting: Registered in PROSPERO (CRD42024526067) and reported per PRISMA 2020. Data sources and search: Systematic searches in PubMed, Cochrane Library, Embase, Web of Science, Scopus, CINAHL, and CNKI from inception to 12 March 2024, using MeSH and free-text terms. Reference lists of relevant trials and meta-analyses were screened. Study selection: Inclusion criteria were RCTs in English or Chinese enrolling patients with AD; interventions consisted of aerobic exercise; comparators included routine care or non-aerobic activities (e.g., stretching); outcomes included cognitive function (MMSE, ADAS-cog), quality of life (e.g., SF-36, QoL-AD, EQ-5D, DAD), or depressive symptoms (e.g., BDI, HAMD-17, Cornell Scale). Excluded were non-RCTs, commentaries, case reports, conference abstracts, unpublished studies, studies lacking quantitative outcomes, inaccessible or low-quality reports, and those combining aerobic with anaerobic or cognitive training. Data extraction: Two reviewers independently screened titles/abstracts and full texts, extracted data (authors, year, country, sample size, demographics, intervention characteristics: type, frequency, duration, intensity; outcomes), and resolved disagreements with a third reviewer. Risk of bias: Assessed using Cochrane RoB2 across five domains (randomization, deviations from intended interventions, missing outcome data, outcome measurement, selective reporting). Certainty of evidence: GRADE rated each outcome as high, moderate, low, or very low based on RoB, inconsistency, indirectness, imprecision, and publication bias. Statistical analysis: Continuous outcomes were synthesized using standardized mean differences (SMDs) with 95% CIs. Heterogeneity was assessed with I² and Chi² tests; random-effects models were applied when I²≥50% and fixed-effects when I²<50%. Meta-regression (Stata MP v18.0/v14.0) explored heterogeneity sources. Sensitivity analyses used leave-one-out procedures. Publication bias was evaluated via funnel plots and Egger’s test; Duval and Tweedie trim-and-fill assessed robustness to potential bias. Planned subgroup analyses examined country (China vs other), total duration (≤16 vs >16 weeks), session duration (<30, 30–50, ≥50 min), and frequency (≤3 vs >3 sessions/week). Patient and public involvement: None.

- Study pool: 21 RCTs with 1286 participants (53% female), ages 60–90 years, from seven countries (notably China, USA, Italy, Denmark, France, Brazil, Saudi Arabia). Interventions typically lasted 8–64 weeks, with 2–5 sessions/week and 30–90 min/session. - Cognitive function (MMSE): 17 study effects from 14 articles showed a significant improvement with aerobic exercise vs control (SMD=0.95, 95% CI 0.58 to 1.32; Z=5.06; p<0.00001; I²=83%). - Cognitive function (ADAS-cog): 9 study effects from 7 articles showed significant improvement (lower scores better) (SMD=-0.67, 95% CI -1.15 to -0.20; Z=2.77; p=0.006; I²=84%). - Quality of life: 9 study effects from 8 articles showed significant improvement (SMD=0.36, 95% CI 0.08 to 0.64; Z=2.51; p=0.01; I²=65%). - Depressive symptoms: 3 studies showed no significant effect (SMD=-0.25, 95% CI -0.63 to 0.13; Z=1.27; p=0.21; I²=56%). - Subgroup insights: • MMSE: Greater benefits with total duration >16 weeks and session duration <50 min; ≤30 min per session (SMD≈1.44) outperformed 30–50 min (SMD≈0.91) and >50 min (SMD≈0.76; not significant). • ADAS-cog: Benefits prominent with duration >16 weeks and session duration ≥30 min; effects significant in studies from China but not in other countries. • Quality of life: Improvements more likely with duration >16 weeks, frequency >3 sessions/week, and session duration 30–50 min; effects significant in China-based studies but not others. - Heterogeneity: High for MMSE (I²=83%) and ADAS-cog (I²=84%); moderate for quality of life (I²=65%). - Sensitivity analyses: Results were robust; exclusion of individual studies did not materially change pooled estimates. - Publication bias: Egger’s test suggested potential bias for MMSE (t=3.72, p=0.002) and ADAS-cog (t=-2.78, p=0.027); trim-and-fill adjustments did not negate significance. No evidence of bias for quality of life (Egger’s p=0.534). - Evidence quality (GRADE): Very low for MMSE and ADAS-cog; moderate for quality of life.

The review specifically targeted aerobic exercise to address mixed findings from prior work that pooled heterogeneous exercise modalities. The meta-analysis shows that aerobic exercise significantly improves global cognitive function (by MMSE and ADAS-cog) and quality of life in people with AD, but does not significantly alleviate depressive symptoms. The findings support the hypothesis that sustained aerobic training confers cognitive and functional benefits in AD. Subgroup patterns suggest that intervention dosing matters: total program duration exceeding 16 weeks, session lengths around 30–50 minutes (and often ≥30 minutes for ADAS-cog), and higher weekly frequency (>3 sessions/week for quality of life) are associated with greater improvements. These dosing patterns align with plausible neurobiological mechanisms—enhanced neuroplasticity, hippocampal neurogenesis, cerebral perfusion and metabolism, and modulation of AD-related pathophysiology (amyloid-β, tau, inflammation, oxidative stress). The absence of a clear effect on depressive symptoms may reflect insufficient study power (only three RCTs), heterogeneous depression measures, and interventions not optimized for mood outcomes. High heterogeneity across many analyses indicates variability in study design, intervention delivery, participant characteristics, and measurement tools. Despite this, sensitivity analyses and publication bias adjustments support the overall robustness of cognitive and quality of life findings. The results emphasize the potential clinical utility of prescribing structured aerobic exercise in AD care while highlighting the need for standardized protocols and better reporting (e.g., intensity) to refine dose–response recommendations.

Aerobic exercise is associated with meaningful improvements in cognitive function and health-related quality of life among individuals with Alzheimer’s disease. Programs extending beyond 16 weeks, with session durations of approximately 30–50 minutes and higher weekly frequency, appear most beneficial for quality of life, while >16 weeks with <50 minutes per session improves MMSE and >16 weeks with ≥30 minutes per session improves ADAS-cog. No significant effect on depressive symptoms was detected. Given the high heterogeneity and very low to moderate certainty of evidence for many outcomes, future research should prioritize well-powered, rigorously designed RCTs with standardized intervention parameters (including clear intensity reporting), consistent outcome measures, and longer follow-up to validate and refine optimal dosing guidelines.

- Blinding of participants and intervention providers was infeasible, introducing unavoidable performance bias across trials. - High heterogeneity across studies due to differences in designs, samples, intervention modalities/delivery, durations, frequencies, and session lengths. - Inadequate reporting of exercise intensity in several studies precluded intensity-based subgroup analyses. - Small numbers of studies within some subgroups may bias subgroup estimates. - Limited number of RCTs assessing depressive symptoms reduced power to detect effects on mood. - Some studies had unclear randomization reporting and missing data issues (one study with higher attrition), contributing to risk of bias. - Overall GRADE ratings were very low for cognitive outcomes, indicating imprecision, inconsistency, and potential publication bias.