Three research questions were formulated through preliminary literature review: (1) What core components are included in multidomain dementia-prevention interventions, and what are the distribution characteristics of their implementation frequencies? (2) What are the differences in outcome indicators between multidomain interventions and single-domain interventions? (3) What barriers are encountered during the implementation of multidomain interventions, and what coping strategies have been proposed in existing studies?

PubMed, Cochrane Library, Embase, Web of Science, CINAHL, and PsycInfo databases were systematically searched from inception to March 12, 2025. A combination of Medical Subject Headings (MeSH) and free-text terms was employed for the search. The PubMed search strategy is detailed in Table 1.

Two researchers independently screened titles/abstracts and full texts and extracted data. Retrieved citations were imported into EndNote 21 to remove duplicates, followed by eligibility screening based on inclusion/exclusion criteria. Full texts of potentially eligible studies were reviewed. Discrepancies were resolved by a third researcher.

Data extraction was performed using a standardized Excel form, with cross-verification by a second researcher. Extracted data included: authors, country, study objectives, participant characteristics, intervention duration, multidomain components and frequencies, outcome measures, assessment tools, and implementation barriers.

A PRISMA 2020 flow diagram (adapted for scoping reviews) was used to document the screening process (Figure 1).

Initial searches identified 2,968 articles, including PubMed (n = 316), Web of Science (n = 569), Embase (n = 1,123), CINAHL (n = 121), Cochrane Library (n = 822), and PsycInfo (n = 17). After removing 915 duplicates, 2,053 articles remained. A total of 1,923 articles were excluded after title/abstract screening (including 469 duplicates, 362 ineligible study types, 1,085 irrelevant research topics, and 7 non-English articles). Full texts of 130 potentially eligible articles were reviewed, and 112 were excluded (including 26 articles with unrecoverable full texts, 52 ineligible study types, 12 ineligible study populations, 3 with insufficient intervention duration, and 19 with irrelevant study content). Finally, 18 studies were ultimately included (11–28). The detailed screening flowchart is presented in Figure 1.

The 18 included studies originated from 11 countries: Finland (n = 2), France (n = 2), the Netherlands (n = 2), China (n = 2), Germany (n = 1), Singapore (n = 1), the United States (n = 3), South Korea (n = 3), the United Kingdom (n = 2), Spain (n = 1), and Latin America (n = 1). Key characteristics of the included studies are summarized in Table 2.

Multidomain interventions have been widely applied in dementia prevention research. The 18 included studies identified four core components: physical exercise, cognitive training, nutritional guidance, and cardiovascular risk factor management.

Physical exercise (100% of studies) primarily included aerobic exercise, strength training, and balance training, delivered via group or individualized programs. The more often reported was 3–5 sessions per week (20–60 min/session). The FINGER trial (11) demonstrated that high-intensity exercise (5 sessions/week) significantly improved cognitive function. Cognitive training (94.4% of studies) employed computer-based programs targeting memory, executive function, and other domains, typically delivered 2–3 times weekly (15–30 min/session). Nutritional guidance (88.9% of studies) focused on Mediterranean diet recommendations, Omega-3 supplementation, and personalized dietary advice. Monthly consultations (1–2 sessions/month) showed efficacy, though Omega-3 adherence was lower in Latin American populations (24). Cardiovascular risk management (77.8% of studies) involved regular blood pressure and metabolic parameter monitoring. The eMIND trial (16) highlighted the feasibility of remote monitoring technologies.

The overall intervention duration ranged from 6 months to 2 years. Ngandu et al. (11) reported optimal outcomes with ≥3 weekly sessions combining physical and cognitive training over 2 years.

For detailed information on the specific intervention objectives, component combinations, and implementation frequencies of each included study, refer to the Appendix. This table systematically extracts the core intervention-related data from Table 2, thereby providing a specific basis for direct comparison of multidomain intervention protocols across different regions and study designs.

The 18 included studies evaluated 18 outcome measures, covering cognitive function, quality of life, dementia risk, and physical function. Specifically, 83.3% of the studies reported outcomes related to cognitive function. Tools used to assess cognitive function included the Mini-Mental State Examination (MMSE) (29), neuropsychological battery (30), and Montreal Cognitive Assessment (MoCA) (31), among others. Among the included studies, 11 articles (11, 12, 14–18, 20, 21, 24, 25) evaluated the cognitive function of participants at three time points, namely baseline, mid-intervention, and post-intervention, while the remaining 6 articles (13, 19, 22, 23, 26, 27) only assessed cognitive function at baseline and post-intervention endpoint. In addition, 5 studies (11, 12, 22, 24, 26) conducted regular follow-up after the end of the intervention to observe the long-term effects of multidomain interventions on cognitive function. Moreover, the results of all the above-mentioned articles indicated a significant improvement in cognitive function. Additionally, 55.6% of studies showed improvements in quality of life, as measured by the EuroQol Five-Dimensional Descriptive System (EQ-5D) (32), with an average increase of 7.3 points. Long-term interventions lasting at least 2 years were associated with a 40% reduction in risk of developing dementia (33).

Multidomain interventions were significantly more effective than single-domain interventions in slowing cognitive decline. For example, the multidomain FINGER trial (11) reported a cognitive composite score improvement of +0.21 standard deviations (SD), which was significantly higher than the +0.08 SD observed in the single-domain EXERT trial (34). However, direct head-to-head comparisons were lacking, and differences may also be attributed to intervention frequency and participant characteristics.

Multidomain interventions commonly face barriers such as low participant adherence (18, 20), resource limitations (16, 20), and cultural disparities (14, 20).

Low adherence manifests as failure to adhere to intervention protocols or complete follow-ups, attributable to factors such as forgetfulness (12), lack of motivation (13), side effects (14, 23), or intervention complexity (19). Strategies to improve adherence include setting personalized goals (21) and providing material incentives. Crivelli et al. (24) highlighted that adherence varies across cultural contexts, emphasizing the need to consider cultural influences on participant engagement. Resource constraints, including shortages of healthcare professionals (12–15, 18–20, 28), limited facilities (12, 13, 15, 16, 19–24), recruitment challenges (14, 23), and funding limitations (15, 17), restrict the scope and depth of interventions. These issues can be mitigated through the use of digital technologies for remote monitoring (16) and community-based collaborative models (13). Cultural differences also impact intervention efficacy. Crivelli et al. (24) observed cultural diversity across Latin American countries and adapted interventions by substituting local staples (e.g., corn/beans) to enhance acceptability.

Furthermore, as multidomain interventions are implemented over the long term, dropout rates tend to increase. Tainta et al. (23) addressed this by shortening session durations, while Wang et al. (13) promoted family involvement to strengthen support systems, thereby enhancing the long-term sustainability of multidomain interventions.

Two Chinese studies (13, 20) and 16 studies from non-Chinese countries/regions were included in the subgroup analysis, with key differences as follows:

Multidomain interventions encompass various components, including nutritional guidance, physical exercise, cognitive training, and cardiovascular risk factor monitoring and management. These interventions demonstrate significantly superior efficacy in improving cognitive function and preventing dementia compared to single-domain approaches. Studies have shown that single-domain interventions, such as the Mediterranean diet (35), aerobic and strength training (34), and cognitive training targeting memory, reasoning, and processing speed (36), can yield positive effects.

However, given the multifactorial and complex etiology of dementia, multidomain interventions achieve optimal outcomes through synergistic targeting of multiple pathways: the Mediterranean diet inhibits β-amyloid deposition (37), physical exercise enhances hippocampal neurogenesis and cerebral blood flow (38), cognitive training activates prefrontal-parietal network connectivity (39), and cardiovascular risk management reduces the risk of vascular dementia (4). These combined effects comprehensively address multiple risk factors, resulting in superior preventive efficacy.

The effectiveness and sustainability of multidomain interventions are influenced by implementation models. For example, Ngandu et al. (11) demonstrated that high-frequency interventions (3–5 sessions/week) significantly outperformed the low-frequency model employed by Yaffe et al. (18) in improving cognitive and physical function. Therefore, future research should investigate the relationship between intervention dosage and outcomes to define the optimal multidomain intervention model.

The complexity of multidomain interventions necessitates efficacy evaluations encompassing physiological, psychological, and social functional dimensions. Among the 18 included studies, a total of 18 outcome measures were examined; however, 83.3% relied on a single metric for assessment. For instance, the SUPERBRAIN trial (19) and GOIZ ZAINDU trial (23) evaluated cognitive function solely using the MMSE, while the U.S. POINTER trial (17) and APPLE TREE trial (21) adopted a multidimensional approach integrating biomarkers, mental health, and quality of life.

Although standalone cognitive tests (e.g., MMSE or MoCA) can assess cognitive improvements, they fail to capture the holistic effects of multidomain interventions on physiological parameters, mental health, and social functioning (4). Neuroimaging data and blood biomarkers (e.g., plasma Aβ42/Aβ40 ratio) can elucidate mechanisms underlying neurodegenerative inhibition (40). The U.S. POINTER trial (17) demonstrated that multidomain interventions significantly reduced Alzheimer’s disease pathological burden through Aβ42/Aβ40 analysis. Mental health and social impacts can be evaluated using tools such as the Geriatric Depression Scale (GDS) (41), EQ-5D, and social participation questionnaires. The APPLE TREE trial (21) revealed a significant positive correlation between quality of life improvements and family support intensity in the intervention group via multidimensional assessments.

Therefore, future studies should adopt multidimensional evaluation systems, integrating core metrics [e.g., MMSE/MoCA for cognition, ADL/IADL for daily functioning (42, 43), Aβ/inflammatory biomarkers for physiology (44, 45)] with extended assessments (e.g., GDS/HADS for mental health, social function questionnaires, and qualitative interviews) to comprehensively evaluate the efficacy of multidomain interventions.

Multidomain interventions encounter multiple barriers during implementation, including low participant adherence, resource constraints, and cultural disparities. Adaptive strategies must be developed to ensure intervention feasibility and sustainability.

Low adherence is a pervasive challenge (44). For instance, in the SMARRT trial (18), 28% of participants withdrew due to insufficient motivation or health status changes. Adherence issues may arise from intervention complexity, lack of immediate feedback, or health fluctuations (46). Strategies to enhance adherence include simplifying protocols, setting personalized goals, and offering material incentives (6). For example, the GOIZ ZAINDU trial (23) increased completion rates to 89% by reducing session duration from 60 to 30 min. The APPLE TREE trial (21) achieved higher adherence in the intervention group (72% vs. 55% in controls) through personalized goal-setting, feedback, and £20 vouchers.

Resource constraints—such as shortages of healthcare professionals, inadequate facilities, and limited funding (47)—hinder intervention scalability. These challenges can be addressed via digital alternatives and leveraging existing community resources (48). The eMIND trial (16) reduced staffing needs by 50% and costs by 30% using digital health technologies for remote cognitive training and monitoring. Similarly, the MIND-China trial (13) minimized facility costs by collaborating with rural and community health centers.

Cultural differences also affect intervention acceptability (49). The LATAM-FINGERS trial (24) addressed dietary disparities in Latin America by substituting Nordic fish with local staples (e.g., corn/beans), increasing participant acceptance from 45 to 82%. Thus, multidomain interventions must be culturally adapted to local dietary habits, educational levels, and sociocultural contexts to enhance feasibility.

This review identified three key evidence gaps:

Insufficient evidence on Chinese rural populations: Only 1 study (13) focused on rural China, and no studies explored interventions for low-income or ethnic minority groups, limiting the generalizability of findings to diverse Chinese populations.

Unclear dose–response relationship: The association between intervention frequency/duration and outcomes remains unclear, and no studies have determined the “minimum effective dose” for Chinese populations, making it difficult to balance intervention effectiveness and feasibility.

Limited long-term follow-up data: Most Chinese studies have a duration of ≤6 months, and long-term outcomes (e.g., dementia incidence) are lacking, making it impossible to evaluate the sustainability of intervention effects.