Ageing, Wealth, and Diet: A Global Cross-National Analysis of Cereal Consumption and Dementia Risk

Abstract

Introduction

Global variation in dementia incidence reflects demographic and socioeconomic forces, yet the role of staple dietary patterns remains less defined. While population ageing is a key determinant of dementia burden, differences in cereal consumption, particularly rice, wheat, and maize, have received limited attention. This ecological study examined whether national cereal consumption patterns are associated with dementia incidence across countries independent of confounding factors.

Methods

Country-level data from 204 nations were compiled, with complete case analyses conducted in 184 countries. Alzheimer’s disease and other dementias incidence in 2021 served as the outcome variable. Predictors included per capita consumptions of total cereals, rice, wheat, and maize, alongside genetic predisposition, economic affluence, urban living, ageing indexed by life expectancy at age sixty, and meat consumption. Pearson and partial correlations, principal component analysis, and stepwise multiple regression were applied.

Results

Wheat consumption was positively associated with dementia incidence, affluence, and longevity, whereas rice and maize consumption showed inverse associations. Partial correlations confirmed a persistent inverse association for rice consumption and a weaker inverse association for total cereal consumption after adjustment. Principal component analysis identified a socioeconomic component aligned with wheat consumption and ageing, while rice and maize loaded inversely. Stepwise regression demonstrated that ageing was the strongest predictor of dementia incidence, while rice and total cereal consumption retained small independent inverse associations.

Conclusions

Global dementia incidence is driven primarily by population ageing and socioeconomic development. Cereal type reflects distinct developmental contexts, with rice-based patterns associated with a modestly lower dementia burden.

Plain Language Summary

Dementia is becoming more common worldwide, mainly because people are living longer. However, dementia rates vary widely between populations, and diet may play a supporting role alongside ageing and economic development. This study examined whether national patterns of cereal consumption are linked to differences in dementia incidence across countries. Using data from more than 200 populations, we compared dementia rates with consumption of total cereals and specific staple grains (rice, wheat, and maize), while also accounting for key factors, such as population ageing, economic affluence, urbanization, genetic predisposition, and meat consumption. We found that ageing was by far the strongest predictor of dementia rates globally, explaining most of the differences between countries. Cereal consumption showed much weaker associations. Countries with higher rice consumption tended to have slightly lower dementia incidence, even after adjusting for ageing and socioeconomic factors. In contrast, wheat consumption was more common in wealthier, older populations and was associated with higher dementia rates, likely reflecting broader dietary and lifestyle patterns rather than an effect of wheat itself. Maize showed no consistent relationship with dementia. Overall, these findings suggest that cereal type mainly reflects broader social and dietary contexts rather than directly influencing dementia risk. Protecting cognitive health worldwide will depend primarily on addressing population ageing and health inequalities, while supporting culturally appropriate, balanced dietary patterns as part of healthy ageing strategies.

Key Message

• Ageing and affluence are the strongest global predictors of dementia prevalence, together explaining more than 60% of the observed variance.

• Rice and total cereal consumption show modest inverse ecological associations with dementia incidence after adjustment for ageing and socioeconomic indicators, without implying causality or individual-level protection.

• Wheat and maize consumption primarily act as socioeconomic markers, reflecting dietary modernization and affluence rather than independent cognitive effects.

• Global dietary transitions parallel dementia patterns, with wheat-based diets more common in higher prevalence, affluent settings and rice-based diets prevalent in lower prevalence contexts.

• Supporting traditional plant-based dietary patterns, alongside policies addressing population ageing and health inequities, may contribute to global strategies for dementia prevention and healthy ageing.

Introduction

Dementia is a major public health challenge in the 21st century. Globally, an estimated 57 million people live with dementia, with nearly 10 million new cases occurring each year, and projections suggest that this number will triple by 2050 [1]. The socioeconomic burden of dementia affects both high-income (HI) and low- and middle-income countries (LMICs), intensifying as population ageing accelerates [2]. In many regions, rising life expectancy, urbanization, and dietary transitions are accompanied by shifts in dementia prevalence; however, the specific contributions of staple dietary components remain underexplored [3].

Age is the strongest known risk factor for dementia, and demographic ageing is widely regarded as the central driver of its increasing prevalence [4]. However, ageing alone does not fully explain the marked cross-national variation in dementia burden [5]. Economic affluence, urbanization, and lifestyle transformations, such as reduced physical activity, dietary shifts, and rising cardiometabolic risk, further shape the context in which dementia develops [6, 7]. The transition from traditional diets to more processed, energy-dense food patterns has been implicated in the growing prevalence of non-communicable diseases, many of which also increase dementia risk [8, 9]. Examining how staple dietary patterns, particularly the consumption of rice, wheat, and maize, covary with dementia prevalence at the global level may yield new insights into the nutritional and socioeconomic determinants of cognitive health.

Most nutritional epidemiology research on cognition and dementia has concentrated on composite dietary indices, such as the Mediterranean, Mediterranean-DASH Intervention for Neurodegenerative Delay (MIND), or Nordic diets, rather than individual staple foods [10, 11]. A growing body of evidence suggests that adherence to plant-based dietary patterns is associated with improved cognitive outcomes and reduced dementia risk [12, 13]. In contrast, higher meat consumption has been linked to adverse cardiometabolic profiles and dementia risk in prior research [14, 15]. However, randomized controlled trial evidence remains inconsistent, and many studies rely on surrogate outcomes such as cognitive performance or neuroimaging markers rather than incident dementia [3]. Moreover, the role of staple cereals, which form the foundation of most national diets, has been largely overlooked in cross-national analyses integrating demographic and economic dimensions [16].

Cereals, particularly rice, wheat, and maize, serve as dominant global energy sources, but they also reflect broader dietary cultures and stages of economic development [17–19]. In many Asian and sub-Saharan contexts, rice remains the principal staple and is often consumed within traditional, minimally processed dietary patterns. In contrast, wheat consumption is highest in affluent nations, frequently in refined or processed forms embedded within Western dietary systems [20]. Maize occupies an intermediate position, functioning as a major staple in several lower and middle-income regions while also serving as livestock feed and industrial input in wealthier countries [21, 22]. These variations offer a unique lens to examine how staple consumption corresponds with dementia risk, both as a nutritional exposure and as a marker of socioeconomic development.

Importantly, analyses of cereal-dementia associations must account for key confounding variables such as genetic predisposition, ageing, and economic affluence [23]. Countries with higher life expectancy naturally have a larger proportion of older adults and, consequently, a higher dementia burden. Similarly, greater affluence and urbanization often coincide with risk-enhancing conditions, including hypertension, diabetes, sedentary behaviour, and dietary transitions characterized by increased animal-source food consumption [24]. In multivariate ecological analyses, the independent contribution of staple dietary patterns can be meaningfully assessed only after adjusting for these demographic, economic, and dietary-transition factors [25].

The present study undertakes a cross-national ecological analysis encompassing 204 countries to examine relationships among cereal consumption (total cereals, rice, wheat, and maize), ageing, economic affluence, genetic predisposition, and the incidence of Alzheimer’s disease and other dementias (ADODs hereafter). Meat consumption was included as an adjustment variable to account for broader dietary-transition effects rather than as a primary exposure of interest. A multi-step statistical approach comprising Pearson and Spearman correlations, partial correlations, principal component analysis (PCA), and stepwise multiple regression was employed to (1) identify bivariate associations, (2) isolate independent cereal-related effects after controlling for confounders, and (3) quantify the variance in ADOD incidence explained by demographic, socioeconomic, and dietary predictors.

We hypothesized that (1) ADOD incidence would be positively associated with life expectancy and economic affluence, reflecting demographic transition, and (2) traditional cereal-based dietary patterns, particularly rice consumption, would show inverse population-level associations with ADOD incidence, reflecting developmental and dietary context rather than causal effects. By integrating demographic, economic, and nutritional dimensions, this study aims to provide a holistic understanding of how ageing, affluence, and staple dietary composition collectively shape global ADOD risk. The findings may inform public health strategies that address not only population ageing and economic development, but also culturally grounded dietary practices as part of comprehensive dementia prevention efforts.

Materials and Methods

Study Design and Data Sources

This ecological cross-national study examined the relationship between cereal consumption and ADOD incidence across 204 countries. The dependent variable was ADOD incidence in 2021 (per 100,000 people), sourced from the Institute for Health Metrics and Evaluation (IHME) Global Health Data Exchange [26]. Predictor variables included national mean per-capita annual consumption (2019–2021) of total cereals, rice, wheat, and maize, derived from the Food and Agriculture Organization (FAO) food balance sheets [27].

To account for major demographic, socioeconomic, and dietary-transition influences, the following covariates were included:

• 1.Ageing: indexed by life expectancy at age 60 (e₆₀) in 2018, used as a proxy indicator of population ageing. Data were sourced from the World Health Organization (WHO) [28].
• 2.Economic affluence: represented by gross domestic product based on purchasing power parity (GDP PPP, international dollars per capita) in 2018. This variable reflects national economic capacity and living standards and was sourced from the World Bank database [29].
• 3.Urban living: defined as the proportion of the total population residing in urban areas in 2018, reflecting the degree of urbanization and associated lifestyle transitions. Data were extracted from the World Bank datasets [30].
• 4.Genetic predisposition: represented by the Biological State Index (Ibs), which reflects the retention of deleterious alleles owing to reduced natural selection pressure. Data were extracted from previously published sources (2018) [31].
• 5.Meat consumption: defined as national mean per-capita meat consumption (kg/person/year; 2019–2021 average), derived from FAO food balance sheets. Meat consumption was included as an adjustment variable to account for broader dietary transition and modernization rather than as a primary exposure.

Data were organized in Microsoft Excel^® 2016 for analysis, with each country serving as a distinct unit. Sample sizes varied by variable due to inconsistent data availability across international databases.

Data Classification and Subgroup Analyses

Countries were categorized using multiple international classification systems to enable subgroup analyses:

• Development level: According to the United Nations classification (developed vs. developing countries) [32].

• Income level: based on the World Bank taxonomy: low income (LI), lower middle income (LMI), upper middle income, and HI [1].

• The WHO regional divisions include the African Region (AFR), Region of the Americas (AMR), Eastern Mediterranean Region (EMR), European Region (EUR), South-East Asian Region (SEAR), and Western Pacific Region (WPR) [33].

• Geopolitical and cultural groupings include the Arab World [34], OECD [35], Asia Cooperation Dialogue (ACD) [36], Asia-Pacific Economic Cooperation (APEC) [37], Southern African Development Community (SADC) [38], Shanghai Cooperation Organization (SCO), English as an Official Language (EOL) country (based on government data), and Latin American subregions (LA, LAC) [39].

These subgroup analyses were conducted to capture contextual variations in diet, economic structure, and ageing patterns. Each subgroup was analysed separately using both Pearson and Spearman correlation coefficients, with N values varying according to data availability. This stratified design allowed the assessment of whether the direction or strength of the associations between cereal consumption and dementia prevalence differed across development and regional contexts.

Statistical Analyses

All analyses were performed using SPSS version 31. Multicollinearity among predictors was assessed using tolerance (≥0.10) and variance inflation factor (VIF ≤10) criteria [40]. Pearson’s correlation coefficient (r) was used to assess linear relationships, and Spearman’s rho (ρ) was used to examine monotonic associations. Correlation strength was classified as weak (|r| < 0.3), moderate (0.3 ≤ |r| < 0.5), or strong (|r| ≥ 0.5).

Partial correlations were computed for total cereal, rice, wheat, and maize consumption, controlling sequentially for genetic predisposition, economic affluence, urban living, ageing (e₆₀), and meat consumption. PCA with varimax rotation was used to identify latent structures among dietary and socioeconomic variables. Sampling adequacy was evaluated using the Kaiser-Meyer-Olkin measure and Bartlett’s test of sphericity. Components with eigenvalues >1 and factor loadings ≥0.5 were retained. Stepwise multiple linear regression was then conducted to identify independent predictors of ADOD incidence, with entry and removal probabilities set at p < 0.05 and p > 0.10, respectively. Stepwise regression was used for exploratory model building in a highly collinear ecological dataset; coefficient stability was examined in alternative specifications.

Statistical Significance and Interpretation

All tests were two-tailed, with significance set at p < 0.05. Correlations and regression coefficients were interpreted based on the strength, direction, and consistency across subgroup analyses. By comparing global, developmental, income-based, and regional datasets, this study aimed to determine whether cereal-ADOD associations remained robust across diverse demographic and socioeconomic contexts.

Results

All correlation, principal component, and regression analyses were conducted using data from 184 countries with complete information across dietary, demographic, and socioeconomic variables. Although data were initially compiled for 204 countries, analyses requiring full covariate availability were restricted to this complete-case sample to ensure consistency across models.

Comparative Analysis of Cereal-Type Consumption and Dementia Burden

Figure 1 depicts the relationships between consumption of four staple cereals (rice, wheat, maize, and total cereals) and the incidence of ADODs across the 184 countries. Visual inspection of the scatterplots revealed heterogeneous and predominantly nonlinear associations across cereal types.

Fig. 1.

Global associations between cereal-type consumption and dementia incidence across 184 countries. Notes: Alzheimer’s disease and other dementias (ADOD) incidence data are from 2021. Consumption of meat, cereals, maize, rice, and wheat is expressed as kg per capita per year and represents the mean values for 2019–2021.

Rice consumption showed a weak inverse association with dementia incidence, characterized by a shallow downward power trend (R² = 0.06). Wheat consumption demonstrated a more pronounced curvilinear pattern (R² = 0.21), with moderate wheat consumption corresponding to higher dementia rates, while both lower and higher levels of wheat consumption were associated with lower dementia incidence. Maize consumption exhibited a modest inverse quadratic relationship (R² = 0.08), indicating generally lower dementia rates at higher levels of maize consumption. Total cereal consumption displayed a mild inverted U-shaped association (R² = 0.06), with dementia incidence peaking at intermediate levels of consumption. Although coefficients of determination were relatively small, these patterns highlighted substantial heterogeneity across cereal types, with wheat showing the strongest positive association with dementia incidence among the cereals examined.

Global Correlations between Cereals, Dementia (ADOD), and Socioeconomic Indicators

Table 1 presents the Pearson and Spearman correlation coefficients for cereal consumption, ADOD incidence, and major demographic and socioeconomic predictors, including genetic predisposition, economic affluence, urban living, ageing, and meat consumption. Wheat consumption showed consistently strong positive associations with ADOD (r = 0.417, ρ = 0.551, p < 0.001), economic affluence (r = 0.411, ρ = 0.611, p < 0.001), and ageing (r = 0.570, ρ = 0.585, p < 0.001). Wheat was also positively correlated with meat consumption (r = 0.365, ρ = 0.457, p < 0.001). These patterns suggest that wheat aligns with modernization and affluence, reflecting dietary transitions typical of HI nations.

Table 1.

Pearson and Spearman correlations among cereal consumption, dementia, and socioeconomic predictors

Variable pair	Pearson r	p (2-tailed) value	Spearman ρ	p (2-tailed) value	Direction
Cereal ↔ rice	0.575**	<0.001	0.338**	<0.001	Positive
Cereal ↔ wheat	0.205**	0.005	0.199**	0.007	Positive
Cereal ↔ maize	0.074	0.326	0.132	0.079	None
Cereal ↔ ADOD	−0.128	0.084	−0.097	0.193	Negative
Cereal ↔ genetic predisposition	0.013	0.863	−0.152*	0.042	Weak negative
Cereal ↔ economic affluence	−0.112	0.138	−0.099	0.189	None
Cereal ↔ urban living	−0.081	0.279	−0.095	0.201	None
Cereal ↔ ageing (e60)	0.041	0.580	0.037	0.615	None
Cereal ↔ meat intake	−0.273**	<0.001	−0.281**	<0.001	Negative
Rice ↔ wheat	−0.474**	<0.001	−0.497**	<0.001	Negative
Rice ↔ maize	−0.123	0.103	−0.013	0.867	None
Rice ↔ ADOD	−0.331**	<0.001	−0.464**	<0.001	Negative
Rice ↔ genetic predisposition	−0.172*	0.021	−0.387**	<0.001	Negative
Rice ↔ economic affluence	−0.289**	<0.001	−0.385**	<0.001	Negative
Rice ↔ urban living	−0.255**	<0.001	−0.186*	0.012	Negative
Rice ↔ ageing (e60)	−0.282**	<0.001	−0.430**	<0.001	Negative
Rice ↔ meat intake	−0.353**	<0.001	−0.330**	<0.001	Negative
Wheat ↔ maize	−0.353**	<0.001	−0.322**	<0.001	Negative
Wheat ↔ ADOD	0.417**	<0.001	0.551**	<0.001	Positive
Wheat ↔ genetic predisposition	0.532**	<0.001	0.556**	<0.001	Positive
Wheat ↔ economic affluence	0.411**	<0.001	0.611**	<0.001	Positive
Wheat ↔ urban living	0.410**	<0.001	0.449**	<0.001	Positive
Wheat ↔ ageing (e60)	0.570**	<0.001	0.585**	<0.001	Positive
Wheat ↔ meat intake	0.365 **	<0.001	0.457**	<0.001	Positive
Maize ↔ ADOD	−0.252**	<0.001	−0.174*	0.020	Negative
Maize ↔ genetic predisposition	−0.313**	<0.001	−0.253**	<0.001	Negative
Maize ↔ economic affluence	−0.274**	<0.001	−0.215**	0.005	Negative
Maize ↔ urban living	−0.172*	0.023	−0.110	0.147	Negative
Maize ↔ ageing (e60)	−0.283**	<0.001	−0.195**	0.009	Negative
Maize ↔ meat intake	−0.288**	<0.001	−0.273**	<0.001	Negative
Dementia ↔ genetic predisposition	0.604**	<0.001	0.846**	<0.001	Positive
Dementia ↔ economic affluence	0.597**	<0.001	0.761**	<0.001	Positive
Dementia ↔ urban living	0.498**	<0.001	0.527**	<0.001	Positive
Dementia ↔ ageing (e60)	0.741**	<0.001	0.766**	<0.001	Strong positive
Dementia ↔ meat intake	0.581**	<0.001	0.670**	<0.001	Positive
Genetic predisposition ↔ economic affluence	0.571**	<0.001	0.896**	<0.001	Positive
Genetic predisposition ↔ urban living	0.533**	<0.001	0.631**	<0.001	Positive
Genetic predisposition ↔ ageing (e60)	0.670**	<0.001	0.809**	<0.001	Positive
Genetic predisposition ↔ meat intake	0.641**	<0.001	0.752**	<0.001	Positive
Economic affluence ↔ urban living	0.647**	<0.001	0.718**	<0.001	Positive
Economic affluence ↔ ageing (e60)	0.636**	<0.001	0.778**	<0.001	Positive
Economic affluence ↔ meat intake	0.617**	<0.001	0.753**	<0.001	Positive
Urban living ↔ ageing (e60)	0.511**	<0.001	0.495**	<0.001	Positive
Urban living ↔ meat intake	0.563**	<0.001	0.579**	<0.001	Positive
Ageing (e60) ↔ meat intake	0.567**	<0.001	0.589**	<0.001	Positive

ADOD incidence (2021, IHME); dietary consumption (kg per capita per year, 2019–2021 average, FAOSTAT); population ageing (life expectancy at age 60, WHO); economic affluence (GDP per capita, PPP, 2018, World Bank); urbanization (2018, World Bank); and genetic predisposition (Biological State Index, 2018, published literature). Pearson and Spearman correlations were two-tailed, used pairwise deletion, and are interpreted as direct (positive) or inverse (negative) associations (*p < 0.05; **p < 0.01).

In contrast, rice and maize consumption were inversely associated with ADOD and most socioeconomic indicators. Rice was moderately negatively associated with dementia (r = −0.331, ρ = −0.464, p < 0.001), affluence (r = −0.289, ρ = −0.385, p < 0.001), life expectancy (r = −0.282, ρ = −0.430, p < 0.001), and meat consumption (r = −0.353, ρ = −0.330, p < 0.001). Maize showed weaker but significant negative associations with dementia (r = −0.252, ρ = −0.174, p < 0.05), socioeconomic indicators, and meat consumption (p < 0.001). Total cereal consumption exhibited only weak correlations with ADOD (r = −0.128, p = 0.084), suggesting that aggregate measures obscured distinct patterns among individual grains.

As anticipated, ADOD incidence correlated most strongly with ageing and affluence (r = 0.741 and 0.597, respectively; both p < 0.001) and was also positively associated with meat consumption (r = 0.581, ρ = 0.670, p < 0.001), reaffirming their dominant roles in global dementia distribution. Collectively, the data indicate that while wheat consumption clusters with high-affluence, high-meat, and high-dementia contexts, rice and maize align with lower affluence, lower meat, and lower dementia regions.

Partial Correlations Controlling for Confounders

Table 2 reports the partial correlations between cereal consumption and ADOD after sequentially controlling for genetic predisposition, economic affluence, urban living, ageing (life expectancy at age 60), and meat consumption. After adjusting for genetic predisposition (model 1), rice retained a significant negative correlation with dementia (r = −0.288, p < 0.001), as did total cereal consumption to a lesser extent (r = −0.186, p = 0.015). Adding economic affluence (model 2) attenuated most associations; however, the rice-dementia relationship remained significant (r = −0.240, p = 0.002). Including urban living (model 3) produced minimal further change (r = −0.236, p = 0.002).

Table 2.

Partial correlations between cereal intake and ADOD incidence, with sequential covariate adjustment

Model (control variables added sequentially)	Variable	Partial r	p value	N	Interpretation
Model 1: + genetic predisposition	Total cereal	−0.186	0.015	172	Weak negative correlation; significant after controlling for genetic influence
	Rice	−0.288	<0.001	172	Moderate negative correlation; highly significant
	Wheat	0.122	0.113	172	Non-significant positive association
	Maize	−0.076	0.323	172	No significant correlation
Model 2: + economic affluence (GDP PPP)	Total cereal	−0.139	0.074	168	Weak negative correlation; marginal significance
	Rice	−0.240	0.002	168	Moderate, significant inverse correlation persists
	Wheat	0.100	0.198	168	Positive but not significant
	Maize	−0.035	0.658	168	No relationship observed
Model 3: + urban living (% urban)	Total cereal	−0.136	0.082	168	Weak, non-significant inverse relationship
	Rice	−0.236	0.002	168	Significant negative association maintained
	Wheat	0.095	0.226	168	Non-significant positive trend
	Maize	−0.036	0.650	168	No relationship
Model 4: + ageing (life expectancy at age 60, e60)	Total cereal	−0.212	0.007	163	Significant negative correlation re-emerges after full adjustment
	Rice	−0.263	<0.001	163	Weak but significant negative correlation remains
	Wheat	0.113	0.156	163	Non-significant weak inverse trend
	Maize	−0.075	0.349	163	No relationship detected
Model 5: + meat intake (kg/cap/year, outliers excluded)	Total cereal	−0.152	0.059	161	Weak negative correlation; marginal significance
	Rice	−0.247	0.002	161	Moderate, significant inverse correlation persists
	Wheat	0.158	0.048	161	Weak positive association; borderline significance
	Maize	−0.078	0.332	161	No relationship detected

ADOD incidence (2021, IHME); dietary consumption (kg per capita per year, 2019–2021 average, FAOSTAT); population ageing (life expectancy at age 60, WHO); economic affluence (GDP per capita, PPP, 2018, World Bank); urbanization (2018, World Bank); and genetic predisposition (Biological State Index, 2018, published literature). Two-tailed partial correlations with listwise deletion were conducted, with covariates entered sequentially: genetic predisposition, economic affluence, urban living, ageing (e₆₀), and meat consumption. Pearson and Spearman correlations used pairwise deletion. Sample sizes varied by model due to covariate availability (N = 161–172; exact Ns reported in Table 2).

p < 0.05; p < 0.01.

In the fully adjusted model controlling for ageing (model 4), both total cereal consumption (r = −0.212, p = 0.007) and rice consumption (r = −0.263, p < 0.001) remained significantly and inversely associated with ADOD, indicating modest but persistent associations independent of major demographic and socioeconomic confounders. After additional adjustment for meat consumption (model 5), the inverse association for rice remained robust (r = −0.247, p = 0.002), while the association for total cereal consumption was attenuated to marginal significance (r = −0.152, p = 0.059). Wheat and maize were not consistently associated with ADOD across models, although wheat showed a weak positive association after adjustment for meat consumption.

Principal Component Analysis

Table 3 summarizes the results of the PCA examining underlying structures among cereal consumption and major socioeconomic variables, including genetic predisposition, economic affluence, urban living, ageing (e₆₀), and meat consumption. Full component loadings for each model are presented in online supplementary Table S1 (for all online suppl. material, see https://doi.org/10.1159/000551241). Depending on cereal type, PCA identified either one or two dominant components, explaining approximately 59–75% of the total variance.

Table 3.

Principal component analysis (PCA) of cereal intake and socioeconomic variables

Model	Variables included	Components	Variance, %	Key loadings (≥0.70)	Lowest loading	Interpretation
Model 1	Total cereal, genetic predisposition, economic affluence, urban living, ageing (e60), total meat	2	75.24	Affluence (0.84), ageing (0.86), genetic (0.85), urban (0.77), meat (0.80), cereal (0.98)	–	Two components explain ∼75% variance; cereal forms a distinct axis, while remaining variables cluster along a shared socioeconomic–ageing dimension
Model 2	Rice, genetic predisposition, economic affluence, urban living, ageing (e60), Total meat	1	59.73	Affluence (0.84), ageing (0.83), genetic (0.82), urban (0.78), meat (0.84)	Rice (−0.47)	Single component (∼60% variance); rice loads inversely relative to socioeconomic development and ageing
Model 3	Wheat, genetic predisposition, economic affluence, urban living, ageing (e60), total meat	1	63.73	Ageing (0.85), genetic (0.84), affluence (0.82), urban (0.77), meat (0.80), wheat (0.69)	–	One dominant component (∼64% variance); wheat aligns positively with affluence, ageing, and dietary transition
Model 4	Maize, genetic predisposition, economic affluence, urban living, ageing (e60), total meat	1	59.02	Affluence (0.83), ageing (0.83), genetic (0.84), urban (0.76), meat (0.82)	Maize (−0.44)	One component (∼59% variance); maize loads negatively, indicating weaker integration with developed socioeconomic profiles

Principal component analysis (PCA) with Varimax rotation (where applicable) was used, with components retained based on eigenvalues >1. Loadings ≥0.70 indicate strong contributions. Components reflect underlying socioeconomic and demographic structures rather than direct biological or nutritional mechanisms. ADOD incidence (2021, IHME); dietary consumption (kg per capita per year, 2019–2021 average, FAOSTAT); population ageing (life expectancy at age 60, WHO); economic affluence (GDP per capita, PPP, 2018, World Bank); urbanization (2018, World Bank); and genetic predisposition (Biological State Index, 2018, published literature).

p < 0.05; p < 0.01.

In model 1 (total cereal), two components together explained 75.2% of the variance. Total cereal consumption loaded strongly and independently, forming a distinct axis, while genetic predisposition, economic affluence, urban living, ageing, and meat consumption clustered along a shared socioeconomic-ageing dimension. This suggests that aggregate cereal consumption captures variation partially independent of broader development-related factors.

In models 2–4, a single dominant component explained approximately 59–64% of the variance. In these models, affluence, ageing, genetic predisposition, urban living, and meat consumption consistently loaded positively, reflecting a common developmental and dietary-transition structure. Rice and maize loaded inversely on this component, indicating weaker alignment with developed socioeconomic profiles, whereas wheat loaded positively alongside affluence, ageing, and meat consumption, consistent with its association with more affluent and urbanized contexts.

Across all models, meat consumption clustered with affluence, ageing, and urban living, reinforcing its role as a marker of broader socioeconomic and dietary transition rather than an isolated nutritional dimension. Collectively, the PCA results support the correlation findings by demonstrating that cereal type, rather than total cereal consumption, reflects distinct global dietary and developmental patterns relevant to dementia epidemiology.

Stepwise Multiple Regression Analysis

Table 4 presents the results of the stepwise multiple regression analyses conducted to identify the strongest predictors of ADOD incidence across countries. The dependent variable was ADOD incidence (2021), while predictor variables included cereal consumption (total, rice, wheat, and maize), meat consumption, economic affluence, urban living, and ageing (life expectancy at age 60).

Table 4.

Stepwise multiple regression predicting ADODs (2021) from cereal intake, meat intake, and demographic-socioeconomic predictors

Model	Predictors included (stepwise entry order)	Standardized coefficients (β)	t	p value	R 2	ΔR2	Adjusted R2	Std. error of estimate	Interpretation
Model A: total cereal	Step 1: ageing (e60)	0.792	16.98	<0.001	0.628	–	0.626	57.83	Ageing alone explains ∼63% of variance in ADOD incidence
	Step 2: + total cereal	−0.162	−3.59	<0.001	0.654	0.026	0.650	55.91	Higher total cereal intake is independently associated with lower ADOD incidence
	Step 3: + economic affluence	0.144	2.40	0.018	0.665	0.011	0.659	55.15	Economic affluence adds modest explanatory power beyond ageing and cereal intake
Model B: rice	Step 1: ageing (e60)	0.792	16.98	<0.001	0.628	–	0.626	57.83	Ageing is the dominant predictor of ADOD incidence
	Step 2: + meat intake	0.175	3.13	0.002	0.648	0.020	0.644	56.40	Meat intake contributes additional explanatory variance after ageing
	Step 3: + rice	−0.113	−2.30	0.022	0.659	0.011	0.653	55.70	Higher rice intake remains weakly and inversely associated with ADOD after accounting for ageing and meat intake
Model C: wheat	Step 1: ageing (e60)	0.792	16.98	<0.001	0.628	–	0.626	57.83	Ageing strongly predicts ADOD incidence
	Step 2: + meat intake	0.175	3.13	0.002	0.648	0.020	0.644	56.40	Meat intake adds explanatory power, reflecting dietary-transition effects
	(Wheat excluded)	–	–	–	–	–	–	–	Wheat intake does not independently predict ADOD once ageing and meat intake are considered
Model D: maize	Step 1: ageing (e60)	0.783	16.13	<0.001	0.613	–	0.611	58.13	Ageing remains the primary determinant of ADOD incidence
	Step 2: + meat intake	0.175	3.04	0.003	0.634	0.021	0.630	56.73	Meat intake contributes modest additional explanatory variance
	(Maize excluded)	–	–	–	–	–	–	–	Maize intake shows no independent association with ADOD in adjusted models

ADOD incidence (2021, IHME); dietary consumption (kg per capita per year, 2019–2021 average, FAOSTAT); population ageing (life expectancy at age 60, WHO); economic affluence (GDP per capita, PPP, 2018, World Bank); urbanization (2018, World Bank); and genetic predisposition (Biological State Index, 2018, published literature). Stepwise multiple linear regression was used (entry p ≤ 0.05; removal p ≥ 0.10). Ageing (e₆₀) was the strongest predictor across all models.

Across all models, ageing emerged as the strongest predictor of ADOD incidence (β ≈ 0.78–0.79, p < 0.001), explaining approximately 61–63% of the total variance. This highlights the dominant influence of population ageing on the global burden of dementia. Economic affluence entered only the total cereal model and contributed modest additional explanatory power (ΔR² = 0.011), consistent with higher dementia incidence in more affluent and diagnostically developed settings.

Dietary variables contributed comparatively small but measurable effects. Total cereal consumption showed an independent inverse association with ADOD after adjustment for ageing and affluence (β = −0.162, p < 0.001). Rice consumption also retained a modest but significant inverse association after accounting for ageing and meat consumption (β = −0.113, p = 0.022). In contrast, wheat and maize consumption did not enter the final models, indicating no independent association with ADOD once ageing and dietary-transition effects were considered.

Meat consumption entered the rice, wheat, and maize models following ageing, contributing a small but consistent increase in explained variance (ΔR² ≈ 0.02). This suggests that meat consumption captures an additional dietary-transition dimension beyond ageing, rather than acting as a primary independent predictor of ADOD.

Overall, the regression results reinforce that ageing is the principal global determinant of ADOD incidence, while dietary factors exert secondary effects. Among cereals, total cereal and rice consumption showed small but independent inverse associations, whereas wheat and maize did not retain significance after adjustment. These findings support the interpretation that cereal-type consumption largely reflects broader socioeconomic and dietary-transition contexts rather than direct causal effects.

Subgroup Correlation Analyses

Table 5 presents subgroup correlation analyses examining associations between cereal consumption (total cereals, rice, wheat, and maize) and ADOD incidence across development levels, income classifications, and WHO regions. These analyses assessed whether the direction and strength of cereal-ADOD relationships observed globally were consistent across demographic and socioeconomic contexts.

Table 5.

Correlations between cereal consumption and Alzheimer’s disease and other dementias (ADOD) by development level, income classification, and WHO region

Dataset	N	Cereal → dementia	Rice → dementia	Wheat → dementia	Maiz → dementia
Worldwide	184	r = −0.128, p = 0.084	r = −0.331, p < 0.001	r = 0.417, p < 0.001	r = −0.252, p < 0.001
		ρ = −0.097, p = 0.193	ρ = −0.464, p < 0.001	ρ = 0.551, p < 0.001	ρ = −0.174, p = 0.020
United Nations common practice
Developed countries	45	r = −0.272, p = 0.071	r = 0.465, p = 0.001	r = −0.381, p = 0.010	r = −0.003, p = 0.983
		ρ = −0.269, p = 0.074	ρ = 0.273, p = 0.069	ρ = −0.358, p = 0.016	ρ = 0.043, p = 0.793
Developing countries	138	r = 0.118, p = 0.168	r = −0.065, p = 0.446	r = 0.385, p < 0.001	r = −0.132, p = 0.126
		ρ = 0.108, p = 0.208	ρ = −0.136, p = 0.111	ρ = 0.502, p < 0.001	ρ = −0.087, p = 0.312
World bank income classifications
Low income (LI)	27	r = 0.030, p = 0.881	r = −0.153, p = 0.445	r = 0.522, p = 0.005	r = 0.055, p = 0.787
		ρ = −0.162, p = 0.420	ρ = −0.363, p = 0.063	ρ = 0.396, p = 0.041	ρ = 0.168, p = 0.401
Lower middle income (LMI)	49	r = 0.168, p = 0.247	r = 0.021, p = 0.884	r = 0.258, p = 0.074	r = −0.144, p = 0.323
		ρ = 0.368, p = 0.009	ρ = −0.049, p = 0.740	ρ = 0.262, p = 0.068	ρ = −0.135, p = 0.355
Upper-middle income (UMI)	52	r = 0.059, p = 0.679	r = −0.268, p = 0.054	r = 0.350, p = 0.011	r = −0.170, p = 0.237
		ρ = −0.003, p = 0.981	ρ = −0.455, p < 0.001	ρ = 0.322, p = 0.020	ρ = 0.107, p = 0.461
High income (HI)	55	r = −0.199, p = 0.145	r = −0.483, p < 0.001	r = 0.085, p = 0.538	r = 0.006, p = 0.966
		ρ = −0.188, p = 0.169	ρ = −0.489, p < 0.001	ρ = 0.014, p = 0.918	ρ = 0.036, p = 0.803
Low- and middle-income countries (LMICs)	128	r = 0.099, p = 0.267	r = −0.172, p = 0.053	r = 0.482, p < 0.001	r = −0.180, p = 0.044
		ρ = 0.118, p = 0.184	ρ = −0.271, p = 0.002	ρ = 0.576, p < 0.001	ρ = −0.142, p = 0.114
WHO regions
African region countries	45	r = 0.146, p = 0.340	r = −0.126, p = 0.409	r = 0.708, p < 0.001	r = −0.078, p = 0.610
		ρ = −0.087, p = 0.571	ρ = −0.246, p = 0.103	ρ = 0.546, p < 0.001	ρ = 0.168, p = 0.270
American region countries	35	r = 0.070, p = 0.688	r = −0.148, p = 0.396	r = 0.447, p = 0.007	r = −0.187, p = 0.289
		ρ = −0.070, p = 0.689	ρ = −0.142, p = 0.415	ρ = 0.405, p = 0.016	ρ = −0.069, p = 0.699
Eastern Mediterranean region countries	21	r = 0.079, p = 0.735	r = −0.442, p = 0.045	r = 0.386, p = 0.084	r = 0.070, p = 0.762
		ρ = 0.134, p = 0.563	ρ = −0.377, p = 0.092	ρ = 0.422, p = 0.057	ρ = 0.045, p = 0.845
European region countries	49	r = −0.479, p < 0.001	r = 0.035, p = 0.812	r = −0.478, p < 0.001	r = −0.051, p = 0.744
		ρ = −0.484, p < 0.001	ρ = 0.136, p = 0.352	ρ = −0.493, p < 0.001	ρ = 0.007, p = 0.963
South-East Asian Region countries	10	r = −0.016, p = 0.966	r = 0.302, p = 0.397	r = −0.430, p = 0.215	r = −0.159, p = 0.661
		ρ = 0.164, p = 0.651	ρ = 0.539, p = 0.108	ρ = −0.636, p = 0.048	ρ = 0.091, p = 0.803
Western Pacific Region countries	23	r = 0.012, p = 0.958	r = −0.100, p = 0.649	r = 0.169, p = 0.441	r = 0.000, p = 1,000
		ρ = 0.318, p = 0.139	ρ = 0.124, p = 0.574	ρ = 0.309, p = 0.151	ρ = 0.446, p = 0.033
Countries grouped with various factors
Asia Cooperation Dialogue (ACD)	27	r = −0.186, p = 0.354	r = −0.046, p = 0.820	r = −0.141, p = 0.483	r = 0.007, p = 0.972
		ρ = 0.041, p = 0.839	ρ = 0.111, p = 0.583	ρ = −0.124, p = 0.538	ρ = 0.104, p = 0.606
Asia-Pacific Economic Cooperation (APEC)	18	r = −0.366, p = 0.135	r = −0.269, p = 0.281	r = 0.095, p = 0.707	r = −0.160, p = 0.525
		ρ = −0.393, p = 0.106	ρ = −0.251, p = 0.316	ρ = 0.296, p = 0.233	ρ = 0.075, p = 0.766
Arab World	21	r = 0.041, p = 0.860	r = −0.552**, p = 0.009	r = 0.391, p = 0.080	r = 0.118, p = 0.611
		ρ = 0.082, p = 0.724	ρ = −0.545*, p = 0.011	ρ = 0.486*, p = 0.026	ρ = 0.135, p = 0.559
English as official language (EOL)	51	r = 0.000, p = 0.998	r = −0.338*, p = 0.015	r = 0.686, p < 0.001	r = −0.294*, p = 0.036
		ρ = −0.062, p = 0.664	ρ = −0.269, p = 0.056	ρ = 0.768, p < 0.001	ρ = −0.222, p = 0.118
Latin America (LA)	23	r = 0.293, p = 0.174	r = −0.017, p = 0.938	r = 0.535, p = 0.009	r = −0.201, p = 0.369
		ρ = 0.120, p = 0.587	ρ = −0.094, p = 0.670	ρ = 0.280, p = 0.196	ρ = −0.208, p = 0.352
Latin America and the Caribbean (LAC)	33	r = 0.180, p = 0.316	r = −0.031, p = 0.863	r = 0.433, p = 0.012	r = −0.162, p = 0.377
		ρ = 0.005, p = 0.979	ρ = −0.091, p = 0.615	ρ = 0.331, p = 0.060	ρ = −0.067, p = 0.714
Organization for Economic Cooperation and Development (OECD)	37	r = −0.326, p = 0.049	r = 0.098, p = 0.565	r = −0.089, p = 0.602	r = −0.439, p = 0.013
		ρ = −0.298, p = 0.073	ρ = −0.028, p = 0.871	ρ = −0.138, p = 0.417	ρ = −0.202, p = 0.276
Southern African Development Community (SADC)	16	r = 0.527, p = 0.036	r = 0.160, p = 0.555	r = 0.898, p < 0.001	r = −0.442, p = 0.087
		ρ = 0.371, p = 0.158	ρ = 0.165, p = 0.542	ρ = 0.674, p = 0.004	ρ = −0.294, p = 0.269
Shanghai Cooperation Organization (SCO)	25	r = −0.103, p = 0.625	r = −0.106, p = 0.613	r = 0.059, p = 0.781	r = −0.258, p = 0.224
		ρ = 0.000, p = 1,000	ρ = −0.307, p = 0.136	ρ = 0.062, p = 0.767	ρ = −0.254, p = 0.231

Two-tailed significance tests were used.

Bold values indicate *p < 0.05; double-bold values (marked **) indicate p < 0.01. Sample sizes varied due to pairwise deletion. ADOD denotes Alzheimer’s disease and other dementias (incidence, 2021). Food consumption variables are expressed as kg per capita per year (2019–2021 mean).

At the global level (N = 184), total cereal consumption showed a weak and non-significant inverse association with ADOD incidence, indicating limited explanatory value when cereals were considered in aggregate. In contrast, cereal type demonstrated clearer and more consistent patterns. Wheat consumption was positively associated with ADOD incidence, whereas rice and maize consumption showed significant inverse associations. These findings suggest that cereal composition, rather than total cereal volume, is more relevant to global dementia patterns.

Stratification by development status revealed marked heterogeneity. In developed countries, total cereal consumption showed a moderate inverse trend with ADOD, whereas developing countries exhibited a weak positive association. Wheat consumption remained positively associated with ADOD in developing countries but showed an inverse association in developed countries, reflecting differences in dietary processing, population ageing, and epidemiological transition. Rice consumption demonstrated divergent patterns, with positive correlations in developed countries and weak or inverse associations in developing settings.

Income-based analyses further highlighted context dependence. Wheat-ADOD correlations were strongest in upper-middle-income and LI countries, while rice consumption was inversely associated with ADOD in upper-middle- and HI countries. In LMICs, wheat consumption showed a strong positive association with ADOD incidence, whereas rice and maize tended to show inverse or weak associations, consistent with younger population structures and lower diagnostic coverage.

Regional analyses demonstrated distinct dietary-dementia profiles. In Europe, total cereal and wheat consumption were both strongly and inversely associated with ADOD, whereas African, American, and several Asian regions showed weak or non-significant associations for total cereals. Across multiple regions, wheat consumption consistently aligned with higher ADOD incidence, while rice-dominant regions, particularly in the Eastern Mediterranean and Western Pacific, tended towards lower ADOD incidence, although not uniformly significant.

Analyses across geopolitical and cultural groupings reinforced these patterns. Wheat consumption showed strong positive correlations with ADOD in English-speaking, LMIC, African, and Southern African Development Community countries, whereas rice consumption was inversely associated with ADOD in Arab World, HI, and several Asian groupings. Maize consumption demonstrated generally weak or inconsistent associations across subgroups.

Overall, subgroup analyses confirm that cereal-dementia relationships are highly context dependent. Wheat consumption is most consistently associated with higher ADOD incidence in economically developing and transitioning regions, whereas rice and maize consumption tend to align with lower ADOD incidence, particularly in less affluent and agrarian settings. These findings reinforce the interpretation that cereal type functions primarily as a marker of socioeconomic development, dietary modernization, and population ageing rather than as a uniform dietary determinant of dementia risk. Subgroup findings should be interpreted cautiously where sample sizes were small and are presented to illustrate contextual heterogeneity rather than for formal inference.

Discussion and Public Health Implications

Taken together, these results indicate that cereal type aligns more strongly with demographic and socioeconomic structure than with independent dietary effects on dementia incidence.

Overview of Key Findings

This ecological analysis across 204 countries examined population-level associations between cereal consumption and ADOD incidence after accounting for ageing, economic affluence, urbanization, genetic predisposition, and broader dietary transition. The findings reaffirm that ADOD burden is primarily shaped by demographic ageing and socioeconomic development, while cereal type reflects broader dietary and economic contexts rather than direct nutritional effects. Wheat consumption was positively associated with ADOD incidence and economic affluence, consistent with modernization and Westernized dietary structures, whereas rice consumption showed a consistent inverse ecological association, aligning with traditional, less-industrialized dietary contexts. Maize demonstrated weak and inconsistent relationships with ADOD, and total cereal consumption alone was not a meaningful predictor once cereal composition and contextual variables were considered.

Novelty and Conceptual Contribution of the Cereal-Focused Analysis

The primary contribution of this study is not to propose cereals as direct causal dietary risk factors for dementia, but to demonstrate that cereal type – specifically rice, wheat, and maize – functions as a structural marker of dietary modernization and developmental context at the global level. This perspective provides epidemiological insight beyond analyses based on total cereal consumption or animal-based foods.

In contrast to total cereal consumption, which showed weak and inconsistent associations with ADOD incidence, individual cereal types exhibited distinct and systematic patterns aligned with socioeconomic development. Wheat consumption consistently clustered with economic affluence, urbanization, population ageing, and indicators of dietary transition, whereas rice and maize aligned with less affluent, more agrarian contexts. These contrasting patterns were evident across correlation analyses, principal component structures, and multivariable models, indicating that cereal composition reflects broader food system organization rather than isolated nutritional exposure.

Conceptually, this approach differs from prior global diet–dementia studies that emphasize animal protein or meat supply as risk-enhancing exposures within affluent, ageing populations. The cereal-focused framework captures an earlier structural dimension of the nutrition transition, differentiating traditional staple-based food systems from modernized dietary patterns. The persistence of an inverse ecological association between rice consumption and ADOD incidence after adjustment for ageing, affluence, urbanization, genetic predisposition, and meat consumption supports the interpretation that this pattern reflects developmental and dietary context rather than a direct dietary effect. Overall, cereal type serves as a proxy for long-term dietary organization and development trajectory, justifying the cereal-focused analysis as a standalone contribution to global dementia epidemiology.

Ageing and Affluence as Dominant Predictors

The dominance of ageing and economic affluence as predictors of ADOD incidence is consistent with extensive epidemiological evidence [41]. Life expectancy and economic affluence emerged as the strongest determinants of global ADOD burden, together explaining over 60% of the observed variance in regression models. Ageing (e₆₀) alone accounted for the largest share of explained variance, underscoring demographic ageing as the principal global driver of ADOD [42]. These findings indicate that differences in population age structure and socioeconomic development largely determine cross-national variation in ADOD incidence, whereas dietary and cultural factors exert comparatively modest effects.

This pattern aligns with World Health Organization projections attributing the growing global ADOD burden primarily to demographic ageing and economic development [1]. Although ADOD incidence rates are highest in affluent nations, the majority of people living with dementia currently reside in LMICs, a proportion expected to rise as longevity increases in populous countries such as China and India [43]. Higher reported incidence in wealthier countries also reflects longer lifespans and more comprehensive diagnostic systems [44].

Similar trends have been reported by Nichols and Lee [45] and Wu et al. [46], who found that national income and life expectancy jointly predicted dementia rates independent of genetic predisposition. Collectively, these findings reinforce the demographic transition model, in which ADOD incidence increases with age, modernization, and improved healthcare capacity.

Wheat as a Marker of Modernization

Wheat consumption showed strong positive associations with ADOD, economic affluence, and ageing, mirroring patterns observed in Westernized and industrialized food systems [18]. Rather than functioning as a direct dietary risk factor, wheat appears to serve as a nutritional proxy for modernization [47]. Refined wheat-based foods are typically embedded within dietary patterns characterized by higher energy density, lower fibre content, and increased cardiometabolic risk [20, 48, 49].

PCA clustered wheat with affluence and ageing, reinforcing its interpretation as an indicator of urbanized dietary transition rather than an independent etiological driver of ADOD. This finding aligns with evidence showing that refined grain consumption in affluent nations often coincides with sedentary lifestyles, high caloric consumption, and increased metabolic disorders [50], which are indirectly linked to cognitive decline [51].

It is important to note that wheat is not the sole staple carbohydrate in affluent societies. In many developed countries, potatoes and non-wheat grains (e.g., rye and barley) remain central components of the diet. These foods similarly reflect industrialized food systems and contribute to broader modernization-linked dietary patterns associated with higher dementia prevalence. Accordingly, wheat consumption in this study should be interpreted as a general indicator of dietary modernization rather than a singular determinant of cognitive health.

Meat and overall protein consumption represent important contextual factors in interpreting cereal-ADOD associations. In affluent societies, dietary transition is characterized by reduced reliance on staple grains and increased consumption of fats and animal-source foods. Although higher meat consumption has been associated with dementia risk in prior studies [15, 52], it also reflects socioeconomic affluence and longevity. In the present analyses, meat consumption was included as an adjustment variable to account for these broader dietary-transition effects rather than as a primary exposure.

Rice as a Marker of Traditional Dietary Contexts

Rice displayed a consistent inverse ecological association with ADOD incidence across correlation and adjusted models. This pattern should not be interpreted as evidence of a protective or causal effect of rice consumption, but rather as an indicator of broader dietary, cultural, and developmental contexts that differ systematically from modernized food systems.

These patterns are consistent with cohort evidence from Asian populations, where rice-centred diets often coexist with higher consumptions of legumes and vegetables and lower consumption of fats and animal products, and are associated with more favourable cardiometabolic profiles [53, 54]. At the ecological level, the observed inverse association likely reflects broader lifestyle, cultural, and food system characteristics rather than rice consumption per se.

From a public health perspective, cereal quality and processing may be more relevant than cereal quantity. Traditional staple-based diets emphasizing minimally processed grains tend to align with dietary patterns supportive of vascular and metabolic health, whereas refined cereal consumption often accompanies energy-dense, low-fibre diets. These distinctions reinforce the interpretation of rice as a contextual marker within traditional dietary systems rather than a direct determinant of cognitive outcomes.

Maize as a Socioeconomic and Regional Indicator

Maize demonstrated weak and inconsistent associations with ADOD incidence. While global analyses suggested a small inverse relationship, this association attenuated after adjustment for ageing and affluence. Regional patterns were similarly inconsistent, indicating that maize consumption primarily reflects subsistence-oriented food systems and regional economic conditions rather than cognitive outcomes.

This interpretation aligns with Fanzo et al. [55], who demonstrated that maize-based diets are predominantly concentrated in lower resource, subsistence-oriented settings characterized by shorter life expectancy and constrained health-system capacity. Such contexts are associated with weaker food and health systems [56], and dementia incidence in these settings is widely recognized to be underdiagnosed, yielding an apparently low – but likely underestimated – ADOD burden [57]. Accordingly, maize consumption in global ecological analyses functions primarily as a socioeconomic marker of subsistence food systems rather than as a reliable or independent predictor of cognitive outcomes.

Regional and Developmental Differences

Regional subgroup analyses further illustrate the contextual nature of cereal–ADOD associations [58]. In Europe, total cereal consumption showed inverse associations with ADOD, potentially reflecting dietary diversity and limited reliance on refined staples [59]. In contrast, African and American regions exhibited weaker and more inconsistent patterns, likely influenced by dietary heterogeneity and data limitations [60].

Rice-dominant populations in Asia and the Western Pacific tended to report lower ADOD incidence, consistent with traditional dietary structures, although underdiagnosis may contribute to these differences [61]. Collectively, these findings illustrate a developmental gradient contrasting wheat-dominant affluent societies with rice- or maize-dominant agrarian contexts, reflecting the global nutrition transition [62, 63].

Integration with Broader Literature

The findings align with literature emphasizing the interplay between demographic, economic, and nutritional factors in ADOD risk. Prior studies highlight the central role of ageing, cardiovascular health, and lifestyle factors in shaping cognitive outcomes [64–68]. This ecological analysis extends that work by positioning cereal composition as a macro-level indicator of dietary and developmental context, complementing protein- and meat-focused perspectives without implying direct dietary causation.

Public Health Implications: Promoting Healthier and Culturally Grounded Diets

These findings underscore that while population ageing remains the dominant driver of ADOD, dietary patterns and food-system context play an important supporting role at the population level. Preserving traditional, minimally processed dietary structures may align with population contexts associated with lower reported dementia burden, complementing broader strategies that primarily address demographic ageing and health inequities.

Traditional dietary patterns, often characterized by whole grains, legumes, and predominantly plant-derived foods, are commonly associated with more favourable vascular and metabolic profiles that support cognitive health across the life course. In contrast, dietary transition towards refined grain-based, energy-dense food systems in affluent settings parallels higher ADOD prevalence, mirroring trajectories observed for cardiovascular and metabolic diseases.

Public health initiatives should therefore prioritize dietary strategies that are nutritionally sound and culturally appropriate [56]. In low- and middle-income countries undergoing rapid economic transition, nutritional education and policy support that encourage the maintenance of traditional grain diversity may help sustain cognitive resilience as populations age. In HI countries, food system reforms that reduce reliance on refined grains and support whole-grain alternatives may contribute to healthier ageing environments.

Cross-sector collaboration among health, agriculture, and education sectors is essential to translate these findings into effective public health action. Promoting sustainable and culturally relevant food practices can support healthier ageing trajectories, particularly when integrated with policies addressing healthcare access and socioeconomic inequities.

Ultimately, encouraging balanced, plant-forward dietary patterns that retain traditional grains aligns with broader goals of nutritional sustainability and population-level dementia prevention [67]. Such culturally sensitive approaches acknowledge that healthy ageing is shaped not only by biological longevity but also by the preservation of food systems and cultural practices that support long-term cognitive health [5, 46].

Limitations and Future Research

This study had several limitations that warrant consideration. Firstly, as an ecological analysis, the findings reflect population-level associations and may not translate to individual risk, raising the possibility of an ecological fallacy. Accordingly, the observed associations do not support causal inference or individual-level dietary recommendations. Secondly, the use of secondary data from international databases introduces potential inconsistencies in measurement accuracy, diagnostic criteria, and reporting completeness, particularly in LMICs where ADOD may be underdiagnosed, especially in regions where dementia surveillance systems are still developing. Thirdly, cereal consumption data (averaged for 2019–2021) may not reflect long-term or lifetime dietary exposures that are more relevant to ADOD development. Fourthly, although the analysis controlled for major confounders, including ageing, affluence, urbanization, and genetic predisposition, residual confounding from factors such as education, physical activity, comorbidities, and healthcare access cannot be excluded.

In addition, stepwise regression is sensitive to collinearity and model specification, particularly in ecological datasets with correlated demographic and socioeconomic predictors. To address this, alternative regression models excluding correlated covariates were examined, and the direction and statistical significance of the key predictors remained stable, supporting the robustness of the reported associations.

Finally, cereal processing levels (e.g., whole vs. refined) were not distinguished, which may have influenced nutritional quality and cognitive outcomes. Future longitudinal and multilevel studies incorporating individual-level data are needed to clarify causal mechanisms and confirm these ecological associations.

Conclusion

This global ecological study demonstrates that ADOD incidence is primarily shaped by demographic ageing and socioeconomic development, with cereal consumption showing modest, context-dependent associations at the population level. Across 204 countries, life expectancy and economic affluence explained over 60% of the observed variation in ADOD burden. Wheat consumption was positively associated with ADOD, reflecting dietary modernization and longer lifespan rather than direct nutritional harm. In contrast, higher rice consumption – often characteristic of traditional and less-industrialized dietary systems – was inversely associated with ADOD incidence after adjustment for demographic, socioeconomic, genetic, and dietary-transition factors, including meat consumption. Maize exhibited weak and inconsistent associations, likely reflecting regional and cultural heterogeneity rather than systematic dietary effects.

Overall, while population ageing remains the dominant determinant of ADOD at the global level, cereal type appears to function as a marker of broader dietary and developmental context. Traditional cereal-based dietary patterns appear to co-occur with population contexts reporting lower dementia incidence and may therefore inform culturally grounded public health strategies without implying direct dietary protection or causation.

Acknowledgments

The authors appreciate Ms. Turi Christensen from the Institute for Health Metrics and Evaluation of the University of Washington for her assistance in locating and defining the data on cardiovascular disease incidence rates. During the preparation of this work, the authors used ChatGPT (GPT-5, OpenAI) via a web interface (https://chat.openai.com) for language refinement. AI was not used for content generation, data analysis, or interpretation. After using this tool, the authors reviewed and edited the content as required and took full responsibility for the integrity, originality, and accuracy of the manuscript.

Statement of Ethics

This study did not involve individual human participants or animals and was; therefore, exempt from formal ethical reviews under institutional and international human and animal research guidelines. The Ethics Committee at the University of Adelaide reviewed the study protocol and confirmed that formal ethical approval was not required for secondary analysis of publicly available data (Approval No. 36289; details will be provided upon manuscript acceptance).

Conflict of Interest Statement

The author declares no conflict of interest.

Funding Sources

There is no specific funding to support this study.

Author Contributions

Conceptualization and investigation: W.Y., L.G., S.G.K.R., B.S., M.H., S.F., and H.-C.R.C. Data curation: W.Y., B.S., M.H., and S.F. Formal analysis and writing – original draft: W.Y., M.H., and S.F. Methodology: W.Y., M.H., and H.-C.R.C. Project administration: W.Y. and S.F. Resources: W.Y., L.G., S.G.K.R., S.F., and H.-C.R.C. Software: W.Y. and M.H. Validation: W.Y., M.H., S.F., and H.-C.R.C. Visualization: W.Y., L.G., S.G.K.R., M.H., and H.-C.R.C. Writing – original draft: W.Y., M.H., and S.F. Writing – review and editing: W.Y., L.G., S.G.K.R., B.S., M.H., S.F., and H.-C.R.C.

Funding Statement

There is no specific funding to support this study.

Data Availability Statement

All data utilized in this study are secondary and publicly accessible and freely available from the publications of individual authors, IHME, and United Nations agencies. The use of these data complies with the terms and conditions outlined by these international organizations, and no formal permissions are required. Detailed descriptions of the data sources are provided in the “Materials and Methods.” Further enquiries can be directed to the corresponding author.