Highlight
Most evidence is preliminary: five studies used mediation analysis to test whether cardiovascular disease (CVD) explains links between lifestyle and cognition. Two studies reported partial mediation for physical activity using composite CVD risk scores; evidence for diet and alcohol was inconclusive; no suitable studies assessed smoking. Overall, CVD-as-mediator remains tentative and under-researched.
Background: why this question matters
Dementia and cognitive impairment impose growing personal and societal burdens as populations age. Modifiable lifestyle factors—physical inactivity, unhealthy diet, smoking, and harmful alcohol use—are consistently implicated in dementia risk. Whether and to what extent cardiovascular disease mediates these associations is a key mechanistic and translational question. If cardiovascular health accounts for a substantial portion of lifestyle effects on cognition, interventions addressing vascular risk could be central to dementia prevention. Conversely, a weak or inconsistent mediating role would imply that lifestyle affects cognition through multiple, perhaps nonvascular, pathways (neuroinflammation, metabolic, direct neurotrophic effects).
Mediation analysis provides a statistical framework to quantify the contribution of an intermediate variable (mediator) — here, CVD or cardiovascular risk — to the total effect of an exposure (lifestyle factor) on an outcome (cognitive decline, mild cognitive impairment, dementia). Properly executed, mediation studies can inform causal models, prioritize interventions, and refine guidelines. However, mediation analyses demand rigorous design: temporality (exposure precedes mediator precedes outcome), adequate confounder control (including mediator–outcome confounders), measurement precision, and consideration of competing risks and time-varying processes.
Study design of the scoping review
The paper by Hensel et al. (Alzheimers Dement. 2025) conducted a scoping review to identify studies that applied mediation analysis to test whether cardiovascular disease mediates associations between lifestyle factors (smoking, alcohol, diet, physical activity) and cognitive outcomes in adults aged 45 years and older. The authors screened 1,309 records and included five studies meeting the prespecified criteria. Included studies were heterogeneous across exposures, mediators (clinical CVD events versus composite cardiovascular risk scores versus subclinical markers), cognitive outcomes (cognitive decline measures, mild cognitive impairment, dementia), analytic methods, and covariate adjustment strategies.
Key findings
Overview
– Evidence base: Very small—five studies met inclusion criteria out of 1,309 screened records. Heterogeneity in exposures (mostly physical activity), mediators, outcomes, and analytic approaches limited synthesis.
– Lifestyle exposures studied: Physical activity (4 studies), diet (2 studies), alcohol (1 study). No included studies assessed smoking as an exposure with CVD mediation analysis.
– Mediator operationalization: Two studies used composite CVD risk scores (e.g., Framingham-type), others used a mix of clinical CVD endpoints or vascular imaging markers.
– Outcomes: Cognitive test scores, incidence of cognitive impairment/dementia, and domain-specific changes.
Physical activity
Four of the five included studies examined physical activity as the exposure. Two reported evidence of partial mediation by composite CVD risk scores, suggesting that some of the protective association between higher physical activity and better cognitive outcomes operates via better cardiovascular health. Reported mediation proportions were generally partial (i.e., not accounting for the entire effect), indicating that nonvascular pathways likely also contribute to the relationship (e.g., neurotrophic effects, reduced inflammation, improved insulin sensitivity).
However, the two positive studies varied in key design elements: measurement of physical activity (self-report vs objective), timing and follow-up duration, adjustment for baseline cognitive function and relevant confounders, and definitions of the mediator (risk score vs diagnosed events). The other two studies of physical activity found no mediation or reported null/inconclusive mediation effects, reflecting inconsistency.
Diet and alcohol
Results for diet (two studies) and alcohol consumption (one study) were inconclusive. Diet studies used different indices of dietary quality (e.g., Mediterranean diet score vs other dietary patterns) and different cardiovascular mediators; neither provided compelling, replicable evidence that CVD mediates diet–cognition links. The single alcohol-focused study did not find clear evidence that measured CVD variables explained alcohol–cognition associations.
Smoking
Notably, no included study applied mediation analysis to test whether cardiovascular disease mediates the link between smoking and cognitive outcomes—an important gap given the well-established cardiovascular impacts of smoking and its association with dementia risk.
Quality, heterogeneity, and consistency
Across the five studies, methodological heterogeneity was substantial. Key limitations included cross-sectional or short follow-up designs, small samples limiting power for mediation decomposition, inconsistent or crude measurement of mediators (composite risk scores that mix risk factors and events), variable confounder adjustment, and limited attention to mediation assumptions (e.g., no unmeasured confounding of mediator–outcome paths). As a result, effect size estimates and mediation proportions were not directly comparable across studies, and pooled summary estimates were not feasible.
Clinical and statistical significance
Where partial mediation was reported (physical activity → CVD risk → cognition), the mediated proportion was limited—implying that cardiovascular pathways may account for part, but not all, of the relationship between activity and cognition. Confidence intervals and formal sensitivity analyses were often absent, so precision and robustness of mediation estimates remain uncertain.
Expert commentary and methodological considerations
Biological plausibility
Cardiovascular disease is a plausible mediator. Vascular pathology (atherosclerosis, microvascular disease, ischemic injury) contributes to cognitive impairment and interacts with neurodegenerative processes. Interventions that reduce vascular risk (blood pressure control, lipid lowering, smoking cessation) are associated with lower rates of cognitive decline in some studies. Thus, demonstrating mediation would support vascular-targeted prevention strategies as mechanisms by which lifestyle modification reduces dementia risk.
Key methodological gaps identified
1. Temporality and study design: Mediation requires clear temporal ordering. Longitudinal cohorts with repeated measures of exposures, mediators, and outcomes are essential but underused.
2. Measurement of mediators: Composite risk scores (Framingham, SCORE) mix risk factors rather than directly measuring vascular damage; clinical events and subclinical markers (carotid atherosclerosis, white matter hyperintensities) might better capture the biologically relevant mediator.
3. Confounding and assumptions: Many studies did not explicitly address mediator–outcome confounding or use modern causal mediation methods that can quantify direct and indirect effects under explicit assumptions (e.g., counterfactual-based approaches, inverse probability weighting). Sensitivity analyses for unmeasured confounding were infrequent.
4. Competing risks and selection bias: Older cohorts face competing mortality and selective attrition, which can bias mediation estimates if not accounted for.
5. Power: Mediation decomposition reduces effective sample sizes—the included studies were often underpowered to detect modest mediated effects.
6. Exposure heterogeneity: Physical activity is multidimensional (intensity, duration, domain), and diet patterns vary across populations. Harmonized, validated exposure measures would enhance comparability.
Recommendations for future research
– Prioritize well-powered longitudinal studies with repeated measures and temporal separation of exposure, mediator, and outcome.
– Use more direct vascular mediators where possible (imaging markers, measures of endothelial function) alongside traditional risk scores and clinical events.
– Apply contemporary causal mediation methods (counterfactual frameworks), report natural direct and indirect effects, and conduct sensitivity analyses for unmeasured confounding.
– Account for competing risks (death) and informative loss to follow-up using appropriate techniques.
– Standardize exposure and outcome measurement across cohorts to enable pooled or consortia analyses.
– Consider triangulation: combine observational mediation analysis with randomized trial data where the exposure (e.g., exercise intervention) affects the mediator (vascular risk) and assess downstream cognitive outcomes.
Clinical implications
Current evidence does not yet allow firm claims that cardiovascular disease fully explains the protective effects of healthy lifestyle on cognition. Partial mediation for physical activity suggests vascular pathways are relevant but not sufficient to account for all benefits. Clinicians should continue to promote lifestyle modifications for their broad cardiometabolic and overall health benefits, recognizing potential cognitive advantages as one component of a multifactorial effect. Importantly, vascular risk modification remains an actionable target within dementia prevention strategies, but it should be complemented by interventions addressing other mechanisms (e.g., metabolic control, direct brain-targeted lifestyle effects).
Conclusion
The scoping review by Hensel et al. identifies a striking gap: despite strong rationale, very few studies have applied rigorous mediation analysis to quantify the role of cardiovascular disease in lifestyle–cognition links. Existing evidence is sparse and heterogeneous, with tentative support for partial vascular mediation of the physical activity–cognition relationship. Robust, longitudinal, and methodologically rigorous research is urgently needed to clarify vascular contributions and optimize preventive approaches that target modifiable lifestyle factors.
Funding and clinicaltrials.gov
See the original publication for detailed funding and conflict-of-interest disclosures: Hensel ALJ et al., Alzheimers Dement. 2025;21(10):e70843. No single clinical trial registration applies because the scoping review synthesizes observational mediation studies.
References
Hensel ALJ, Chan T, Ahmed R, et al. Cardiovascular disease as a mediator in the relationship between lifestyle risk factors and cognitive outcomes: a scoping review. Alzheimers Dement. 2025 Oct;21(10):e70843. doi: 10.1002/alz.70843. PMID: 41137621; PMCID: PMC12552895.
Livingston G, Huntley J, Sommerlad A, et al. Dementia prevention, intervention, and care: 2020 report of the Lancet Commission. Lancet. 2020;396(10248):413–446.
Gorelick PB, Scuteri A, Black SE, et al. Vascular contributions to cognitive impairment and dementia: a statement for healthcare professionals from the American Heart Association/American Stroke Association. Stroke. 2011;42(9):2672–2713.
VanderWeele TJ. Mediation analysis: a practitioner’s guide. Annu Rev Public Health. 2016;37:17–32.
World Health Organization. Risk reduction of cognitive decline and dementia: WHO guidelines. Geneva: WHO; 2019.
