Effects of multidomain lifestyle interventions on cognitive decline and Alzheimer's disease prevention: A literature review and future recommendations

Alzheimer's disease (AD) is the most common form of dementia, affecting over 50 million people worldwide, with projections rising to 152 million by 2050, creating substantial public health and economic burdens. Although there is no cure, numerous modifiable risk factors (MRFs) such as physical inactivity, low cognitive activity, smoking, and poor diet have been identified. Meta-analyses suggest that a substantial fraction of AD cases may be attributable to MRFs, underscoring prevention as a key strategy. While single-domain interventions (e.g., physical activity, cognitive training) show associations with improved cognitive outcomes, other single-risk-factor trials (e.g., hypertension, obesity) have yielded mixed or discouraging results. Consequently, targeting multiple risk factors simultaneously via multidomain lifestyle interventions has been proposed as a more effective preventive strategy. Previous reviews up to 2019 indicated promising potential of multidomain interventions to enhance cognitive reserve and reduce AD risk, but several additional trials have since been completed. The aim of this review is to provide an updated overview and evaluation of completed randomized controlled multidomain lifestyle intervention studies published up to 31 May 2021, assessing effects on cognitive decline and AD prevention in adults aged 45 years and older.

Prior literature has identified multiple MRFs for AD and suggested that up to one-third of AD cases may be attributable to factors such as diabetes, midlife hypertension/obesity, physical inactivity, depression, smoking, and low education. Nineteen modifiable factors have been proposed more recently for AD prevention. Reviews prior to 2019 (including RCTs, cohort, cross-sectional, and experimental studies) concluded that multidomain lifestyle interventions may build cognitive reserve and reduce AD risk, but evidence was limited by heterogeneity and the paucity of long-term outcomes. Observational and trial data on single domains (physical activity, cognitive training) generally support cognitive benefits, though results are not uniformly positive, strengthening the rationale for multidomain approaches. This review builds on and updates prior syntheses by including RCTs completed up to May 2021 and focusing on cognitive and AD-specific outcomes.

- Databases: PubMed and Scopus searched for English-language publications up to 31 May 2021 using keywords related to multidomain interventions, AD or cognitive decline, and prevention (details in Appendix 1). Bibliographies of prior reviews were also screened, yielding five additional studies. - Study selection: Inclusion criteria were (1) randomized design with a control group, (2) at least three combined intervention components, (3) duration ≥6 months, (4) clear cognitive outcomes (neuropsychological tests, AD incidence, or risk scores), and (5) participants aged ≥45 years. Exclusions: protocols, non-human studies, reviews/meta-analyses, irrelevant exposures/outcomes, uncommon populations or severe cognitive impairment/diagnosed AD. - Screening/results: Of 379 records (after duplicates), fifteen were potentially relevant on title/abstract; nine RCTs met full-text eligibility (including three pilot studies). Most were conducted in Europe, with others in USA/Puerto Rico, Taiwan, and China. Interventions commonly included physical activity, diet, cognitive training, vascular/metabolic risk factor management; some included social engagement, medications, or supplementation. - Outcomes: Global cognition (MMSE, MoCA, or composite Z scores), specific cognitive domains, AD incidence, and AD risk scores were extracted. Cognitive domains followed DSM-5 neurocognitive domains. Due to few AD incidence reports, cognitive decline was considered as an indicator of future AD risk. - Data extraction: Conducted by one researcher (SN), exported and summarized per study. - Quality assessment: Two reviewers (BW, SN) independently applied Cochrane RoB 2 to assess randomization process, deviations from intended interventions, missing outcome data, outcome measurement, and selective reporting. Six studies had low risk of bias; three had some concerns; full details in Table 1.

- Studies: Nine RCTs (three pilot) of multidomain lifestyle interventions; components included physical activity (9/9), diet (8/9), cognitive training (6/9), metabolic/cardiovascular risk factor management (8/9), social activity (2/9), medications (2/9), and supplementation (1/9; omega-3 PUFAs). - Global cognition (8 studies assessed): - Significant improvements in 4/8 studies. - FINGER (n=1260; at-risk older adults): 2-year intervention (diet, physical activity, cognitive training, social activities, vascular/metabolic risk management) vs general health advice improved global cognition (mean difference 0.022 NTB total Z score per year; p=0.03) and reduced risk of cognitive decline (OR 1.31, 95% CI 1.01–1.71) in the intervention vs control group. - MAPT (n=1525; older adults with memory complaints): Multidomain with/without omega-3 PUFAs vs omega-3 alone or placebo for 3 years. Individual multidomain arms showed nonsignificant trends, but pooled multidomain groups had significantly less cognitive decline vs omega-3 and placebo (p=0.015). - Taiwan Multidomain Intervention Efficacy Study (n=1082 prefrail/frail ≥65 y with SMC/IADL loss/slow gait): 12-month diet, physical activity, cognitive training improved MoCA among participants ≥75 years (mean difference 1.96, 95% CI 0.25–3.68; p=0.027); overall trend (adj 1.03, 95% CI −0.19 to 2.24; p=0.094). - Xu et al. pilot (n=19 MCI): Risk factor modification (diet, metabolic/vascular management) improved HK-MoCA vs control (Group × time p<0.001); CPR arm trends not uniformly significant. - Null effects in several studies: preDIVA (6-year vascular care; MMSE/VAT), Age Well (12 months; MoCA), eMIND (6 months web-based; composite Z and individual tests), HATICE (18 months web-based; global MMSE and composite Z) showed no significant cognitive benefit. - Specific cognitive domains (3 studies): - Significant improvements in 2/3 studies. - FINGER: Processing speed (estimated mean difference 0.030, 95% CI 0.003–0.057; p=0.029) and executive function (0.027, 95% CI 0.001–0.052; p=0.039) improved in intervention vs control. - Taiwan study: Concentration improved at 6 months (interaction 0.23, 95% CI 0.04–0.42; p=0.019) and 12 months (0.46, 95% CI 0.22–0.70; p<0.001). - AD outcomes: - Incidence: No significant reduction in AD/dementia incidence. - preDIVA: AD (DSM-IV) HR 0.92 (95% CI 0.71–1.19), NS. - SPRINT MIND: Probable dementia HR 0.83 (95% CI 0.67–1.04; p=0.10), NS; however, reduced MCI HR 0.81 (95% CI 0.69–0.95; p=0.007) and composite of probable dementia/MCI HR 0.85 (95% CI 0.74–0.97; p=0.01). - Risk scores: HATICE reduced 20-year dementia risk by CAIDE score (mean difference −0.15, 95% CI −0.28 to −0.03; p=0.02) vs control. - Overall synthesis: Approximately half the studies showed significant global cognitive benefits; two of three showed domain-specific improvements. AD incidence reductions were not demonstrated, though dementia risk scores improved in at least one large RCT.

The review’s findings suggest multidomain lifestyle interventions can improve global cognition and certain cognitive domains among older adults, indicating potential to reduce future risk of cognitive decline and, indirectly, AD. However, clear evidence for reducing AD incidence is lacking, likely due to methodological constraints: heterogeneous populations (age, frailty, cognitive status), varied intervention components and intensities, differing outcome measures, short follow-up periods, and limited statistical power in several trials. Composite cognitive scores appear more sensitive than single tests, supporting their use in future studies. The absence of significant AD incidence effects in trials such as preDIVA and SPRINT MIND may relate to high-quality usual care (preDIVA), early termination and loss to follow-up (SPRINT MIND), and insufficient time horizons to capture dementia endpoints. Web-based interventions showed limited cognitive effects, potentially due to shorter durations, lower intensity, or adherence challenges. The positive CAIDE risk score reduction in HATICE suggests that targeting cardiovascular/metabolic risk factors may contribute to lowering long-term dementia risk, consistent with vascular contributions to cognitive impairment and dementia, though causality for AD prevention remains uncertain. Notably, AD outcome studies did not include cognitive training components, and sleep—another promising modifiable factor—was not included in interventions, highlighting avenues for enhanced multidomain designs. For policy and practice, the data support promoting healthy, multidomain lifestyles, while acknowledging that statistically significant cognitive benefits may not always translate into large clinical effects and that more robust evidence is needed to inform guidelines.

Multidomain lifestyle interventions may reduce the risk of cognitive decline and improve aspects of cognitive function, but current evidence does not demonstrate prevention of AD incidence. Positive changes in AD risk scores indicate potential for long-term risk reduction. Future research should prioritize long-term randomized trials (>10 years), standardized and sensitive cognitive outcomes (including composite scores), well-defined and comprehensive intervention components (considering cognitive training and sleep), adequate power, and inclusion of diverse populations, including those in low- and middle-income countries.

- The review is not a systematic review; searches did not strictly follow Cochrane guidelines, so relevant studies may have been missed. - In four included studies, cognition or AD outcomes were not primary endpoints, potentially reducing statistical power; sample size calculations were sometimes absent or based on other outcomes. - Substantial heterogeneity existed in sample sizes, adherence, outcome measures, and intervention components; some components (especially metabolic/cardiovascular risk reduction) were not well-defined, complicating comparisons and introducing potential overlap with diet and physical activity components. - Participants’ baseline cognitive status varied across studies, limiting comparability and generalizability. - Most studies were conducted in Europe or other high-income settings and only English-language articles were included, limiting applicability to low- and middle-income countries.