Dietary Patterns and Cognitive Health in Older Adults: Findings from the Sydney Memory and Ageing Study

Abstract

Objectives

Systematic reviews report dietary patterns may be associated with cognitive health in older adults. However, inconsistent findings have been reported and relevant research lacks large scale studies. This study aims to examine the associations of dietary patterns and cognitive function among older adults in an Australian ageing cohort.

Design

A population-based, cross-sectional analysis of the baseline phase of the Sydney Memory and Ageing Study, a well-characterised Australian ageing study.

Setting

The Sydney Memory and Ageing Study was initiated in 2005 to examine the clinical characteristics and prevalence of mild cognitive impairment (MCI).

Participants

Non-demented community-dwelling individuals from English-speaking background (N = 819) aged 70–90 recruited from two areas of Sydney, following a random approach to 8914 individuals on the electoral roll in the Sydney Memory and Ageing study.

Measurements

The Cancer Council of Victoria Food Frequency Questionnaire was used to assess dietary intake. Scores for Mediterranean diet, Dietary Approaches to Stop Hypertension (DASH) diet and the Dietary Guidelines Index (DGI 2013) were generated. Two patterns — a Prudent healthy and a Western dietary pattern — were derived using principal components analysis (PCA). Neuropsychological tests were used to assess global cognition and six cognitive domains. Multivariate linear modelling assessed the relationship between dietary patterns and cognitive domain scores.

Results

Mediterranean diet and DASH diet were both positively linked to visuospatial cognition (P=0.002 and P=0.001 respectively). Higher intake of legumes and nuts was related to better performance in global cognition (β=0.117; 95% CI:0.052, 0.181; P<0.001) and language and visuospatial cognitive domains.. The Prudent healthy diet was associated with better global cognition (β=0.307; 95% CI: 0.053, 0.562; P=0.019) in women and a Western diet was related to poorer global function (β=-0.242; 95% CI: -0.451,-0.034; P=0.023) and executive function (β=-0.325; 95% CI: -0.552,-0.099; P=0.005) in men.

Conclusion

In this analysis, higher adherence to the Mediterranean diet, DASH diet, Prudent healthy diet and greater consumption of legumes and nuts were associated with better cognition among older adults.

Introduction

Dementia is a global concern. Over 50 million people worldwide are currently living with dementia, and this is estimated to triple by 2050 (1). Cardiovascular risk factors, psychosocial factors, lifestyle behaviours, education and social networking, have been consistently linked to cognitive health among older adults (2, 3). Of the modifiable factors, nutrition has been recognised as a possible strategy for the prevention of cognitive decline among older adults (4, 5, 6, 7, 8, 9, 10).

A considerable body of research has examined the relationships between single nutrients or foods and cognitive decline among older adults (11). However, nutrients or specific foods are not consumed in isolation (4, 6) but within dietary patterns, where the synergies and interactions between multiple nutrients and foods may play an important role to prevent or slow cognitive decline (4, 5, 6, 7). Our systematic review reported that dietary patterns with positive effects on cognition in older adults are mostly plant-based, rich in poly-/mono-unsaturated fatty acids and low in processed foods (11). However, the association has not been fully investigated due to limited number of large-scale studies. The review revealed mixed findings with respect to Western dietary environments, with 5 out of 14 cohort studies from Western countries finding that Mediterranean diets had little to no effect on cognitive health, while 1 out of 3 cohort studies observing Dietary Approaches to Stop Hypertension (DASH) diets were not associated with cognition. By contrast the majority of studies conducted in Mediterranean regions reported significant associations between dietary patterns and at least one cognitive domain, or the incidence of Mild Cognitive Impairment (MCI) or dementia. These differences suggest culture and eating habits may play a role, and further investigation of the relationship between diet and cognition is needed, particularly in western countries (11).

Our study examines the associations of dietary patterns and cognitive function among Sydney Memory and Ageing Study (MAS) participants. Specific dietary patterns assessed include the Mediterranean diet, DASH diet, healthy diet as recommended by the 2013 Australian Dietary Guidelines, and dietary patterns derived by principal components analysis (PCA).

Methods

Participants

The Sydney Memory and Ageing Study (MAS) commenced in 2005 to investigate the rate of cognitive decline, predictors and protectors for cognitive health, as well as incidence and prevalence of Mild Cognitive Impairment (MCI) and dementia among older adults. Participants were non-demented, community dwelling (n=1037) persons aged 70–90 years at baseline assessment from 2005 to 2007, recruited from two areas of Sydney, Australia. Participants underwent detailed neuropsychological and medical assessments and donated blood for clinical chemistry, proteomics and genomics (12).

Those with a previous diagnosis of dementia, psychotic symptoms, schizophrenia or bipolar disorder were excluded. Other exclusionary conditions were multiple sclerosis, motor neuron disease, developmental disability, and progressive malignancy. Our study population consisted of 819 MAS participants from English-speaking background (able to speak English at a basic conversational level by the age of 10 years) with complete neuropsychological testing and dietary assessment.

The study was approved by the Ethics Committees of the University of New South Wales and the South Eastern Sydney and Illawarra Area Health Service.

Cognitive Assessment

A comprehensive neuropsychological battery was administered according to standard protocols. Six major cognitive domains were assessed: attention/processing speed, language, executive function, verbal memory, global memory (consisted of verbal and visual memory) and visuospatial function. Tests for each domain were: Attention/processing speed — Digit Symbol-Coding (13) and Trail Making Test (TMT) A (14); global memory — Logical Memory Story A delayed recall (15), Rey Auditory Verbal Learning Test (RAVLT) (14) and Benton Visual Retention Test recognition (16); Language — Boston Naming Test (17) and Semantic Fluency (Animals) (14); visuo-spatial— Block Design test (18) and executive function — Controlled Oral Word Association Test (14) and Trail Making Test (TMT) B (14). Global cognition scores were calculated using composite z scores in all domains (12).

Dietary Assessment

The Dietary Questionnaire for Epidemiological Studies Version 2 (DQES v2) was used to assess dietary intake at baseline. This 80-item Food Frequency Questionnaire (FFQ), developed by the Cancer Council of Victoria, includes 74 food items and six alcoholic beverages and has been validated against weighed food records (19, 20, 21).

DQES v2 was self-administered. Nutrient intakes were calculated by the Cancer Epidemiology Centre of the Cancer Council in Victoria using an Australian food composition NUTTAB database (22).

Dietary Patterns

A priori patterns

Mediterranean diet scores were constructed following commonly used scoring systems, i.e. the 0–9 scoring system by Trichopolou et al (23, 24) and the 0–55 scoring system by Panagiotakos et al (23, 24, 25) (Supplementary Table 1). Both systems score food group components including fruit, vegetable, legumes and alcohol intake, using population sex-specific cut-offs (23, 24) or pre-defined cut-offs based on recommended food group amounts from the Mediterranean diet pyramid (25). Intakes from food groups were categorised as “beneficial” or “detrimental” according to characteristics of a Mediterranean dietary pattern. We used monounsaturated: saturated fat ratio (MUFA: SFA) in both systems to replace olive oil. Food intake (g) was adjusted for total energy intake using the residual method (26). For both scores a higher Mediterranean score represented higher adherence to the Mediterranean diet (see Supplementary Table 1).

Table 1.

Demographic and clinical characteristics of participants from Sydney Memory and Ageing study (N=819)

Variables	All (n=819) Mean±SD Or N(%)	Female(n=459) Mean±SD Or N(%)	Male(n=360) Mean±SDOr N(%)
Age (year)	78.6±4.8	78.7±4.9	78.4±4.6
Years of education	11.6±3.5	11.0±3.0	12.5±3.9
CVD risk scorea	4.1±3.1	4.2±3.4	4.0±2.6
Sum of physical activity	1.6±1.1	1.5±1.1	1.7±1.1
BMI, kg/m2	25.7±4.5	25.3±4.6	26.2±4.2
Vitamin D level (nmol/L)	62.7±25.1	58.0±24.0	68.0±25.0
Total cholesterol (mmol/L)	4.8±1.0	5.0±1.0	4.5±0.9
Triglyceride (mmol/L)	1.1±0.6	1.1±0.5	1.1±0.6
HDL-chol (mmol/L)b	1.5±0.5	1.6±0.4	1.3±0.4
LDL-chol (mmol/L)c	2.8±0.9	3.0±0.9	2.7±0.8
CRP (mg/L)d	3.1±5.7	3.0±5.0	3.0±6.0
Vitamin A (umol/L)	3.1±0.8	3.0±0.8	3.2±0.8
Vitamin E (umol/L)	35.4±12.9	37.5±12.4	32.9±13.0
Beta- carotene (umol/L)	0.7±0.6	0.9±0.7	0.6±0.5
Total energy intake (KJ/day)	6943.2±2265.0	6107.6±1858.9	8026.4±2287.7
History of hypertension, N (%)	499(60.9%)	295(64.3%)	204(57.1%)
History of diabetes, N (%)	88(10.7%)	32(7.0%)	56(15.6%)
History of hyperlipidaemia, N (%)	478(58.4%)	262(57.3%)	216(60.3%)
History of depression, N (%)	129(15.8%)	81(18.2%)	48(13.7%)
History of stroke, N (%)	33(4.0%)	11(2.4%)	22(6.2%)
History of TIA, N (%)	57(7%)	34(7.6%)	23(6.5%)
Current smoker, N (%)	252(30.8%)	139(38.6%)	113(24.6%)
APOE4 genotype, N (%)	183(22.3%)	93(20.3%)	90(25.0%)

Notes: SD= Standard deviation; a. The CVD Risk Factor data is computed based on the research of the Framingham Stroke Study (http://www.framinghamheartstudy.org/index.html) and based on the 10-year risk prediction of general cardiovascular disease (http://www.framinghamheartstudy.org/risk/gencardio.html); b. How-density-lipoprotein cholesterol c. Low-density-lipoprotein cholesterol d. C-reactive protein

To construct DASH diet scores, dietary and nutrient intake were grouped into nine categories (fruits, vegetables, legumes and nuts, red and processed meat, whole grain, low fat dairy, sodium intake, sugar intake and sum of monounsaturated fat and polyunsaturated fat intake) (27, 28). Food intake (g) was adjusted for total energy intake (26). Participants were classifed into quintiles according to consumption of each food category. Among the nine food categories, beneficial food categories were scored 1 to 5, with higher scores indicating greater adherance to the DASH diet and three detrimental factors received reverse scoring (sodium intake, sugar intake, red and processed meat intake) (27, 28, 29, 30).

The Dietary Guideline Index (DGI) 2013 is developed to reflect diet quality and adherence to the 2013 Australian Dietary guidelines (31). It includes the Five Food Groups (grains and cereals, vegetables and legumes, fruits, dairy products or alternatives, lean meats or alternatives) as well as components to limit (including discretionary foods high in sugar, salt or saturated fat), with detrimental factors reversely scored (Supplementary Table 2).

Table 2.

Cognitive function at baseline in the Sydney Memory and Ageing study (N=819)

Cognition Domains	All (n=819)	Female(n=459)	Male(n=360)	P value*
Attention/Processing Speed	-0.31±1.16	-0.29±1.14	-0.34±1.18	0.605
Language	-0.49±1.36	-0.59±1.37	-0.37±1.33	0.022
Executive	-0.40±1.25	-0.45±1.26	-0.33±1.24	0.180
Visuo-Spatial	-0.27±1.08	-0.44±1.04	-0.06±1.09	<0.0001
Global Memory	-0.42±1.17	-0.22±1.18	-0.66±1.11	<0.0001
Verbal Memory	-0.37±1.16	-0.14±1.15	-0.68±1.10	<0.0001
Global Cognition	-0.53±1.30	-0.57±1.29	-0.48±1.30	0.333

Notes: statistical significance by gender using independent t test. Raw scores were converted to Z-scores using the baseline mean and standard deviation (SD) values for a reference group (reference group was selected from 504 MAS participants with fluent English before 10 years old and classified as cognitive normal at baseline). If necessary, the signs of the z-scores were reversed so that higher scores reflect better performance. Domain scores were calculated by averaging z-scores of the component tests with the exception of the visuo-spatial domain represented by a single test. Global cognition scores were calculated by averaging the domain scores. All domain and global cognition scores were standardised. *P < 0.05 is significant.

A posterori patterns

All 74 FFQ food items were classified into 40 food groups (Supplemental Table 3) based on the similarity of their nutrient profiles as described in the Australian Food Composition Database (22) to reduce the number of input variables before analysis. Intake of each food group was calculated by adding the intakes of member food items. We used PCA for dimension reduction so that linear combinations of food group clusters with fewer underlying components were identified. Varimax rotation was applied to improve the separation of components and interpretability of the pattern derived, resulting in higher factor loadings for a smaller number of food groups and lower factor loadings for the rest. We considered components with an eigenvalue of >2 for female and male participants separately (32), based on scree plots (Supplementary figure 1 and 2). Food components with a factor loading ≥0.2 (33) were considered as important contributors to each pattern.

Table 3.

Significant associations of food group and cognition in fully adjusted model: Sydney Memory and Ageing study (N=819)

Cognition Domains	Food Groups	β	95% CI		P value*
Attention/ Processing Speed	Legumes and Nuts	.035	-.048	.119	.407
Language	Legumes and Nuts	.113	.038	.189	.003
Visuo-Spatial	Legumes and Nuts	.105	.047	.163	<.001
Executive	Legumes and Nuts	.041	-.051	.133	.364
Global Memory	Legumes and Nuts	.077	.006	.160	.068
Verbal Memory	Legumes and Nuts	.055	-.013	.123	.111
Global cognition	Legumes and Nuts	.117	.052	.181	<.001

Notes: Values are β(95% CI), n = 819. β — Coefficients show a 1 serve increase in food intake is associated with higher cognitive score (positive β) or lower cognitive score (negative β). CI= Confidence Interval. Results were fully adjusted with age, gender, education, as well as physical activity, BMI, metabolic syndrome, hypertension, diabetes, hypercholesterolemia, history of stroke/ transient ischaemic attack (TIA), physical activity, smoking, depression and APOE ξ4 genotype. *P value<.05 is significant for global cognition and p <.01 for subdomains.

Other Measurements

At baseline and subsequent interviews, participants provided information on medical history including cardiovascular diseases and related risk factors (including hypertension, hypercholesterolemia, diabetes, atrial fibrillation, smoking, obesity, stroke or transient ischemic attack etc) and alcohol consumption (12). Assessment of levels of physical activities was also undertaken using self-report questionnaires, where types of physical activities included walking, gardening, yoga, gym work, bowls, golf, tennis, swimming, dancing, bicycling, dancing, aerobics and other sports.

Statistical Analysis

Means, standard deviations and percentages are provided for the entire cohort as well as stratification by sex. Two sample t tests (for continuous variables) and chi-square tests (for categorical variables) were used to compare means, and proportions of demographic, clinical characteristics, dietary intake, cognitive functions, and other variables between female and male participants.

A multivariable linear regression model was used to analyze the associations between global cognition and dietary patterns scores. The relationship between diet and six separate cognitive domains were explored as secondary outcomes. The basic model adjusted for age, gender and education (basic model). The fully adjusted model additionally adjusted for confounding variables, namely physical activity, Body Mass Index (BMI), metabolic syndrome, hypertension, diabetes, hypercholesterolemia, history of stroke/transient ischaemic attack (TIA), smoking, depression, ethnicity and apolipoprotein E (APOE) ξ4 genotype. To determine the effect of covariates, changes in the β-coefficient were examined. Significance level of 0.05 was set for global cognition as primary outcomes, and level of 0.01 was set for multiple secondary outcomes including individual cognitive domains to adjust for multiple testings. Statistical analyses were performed with IBM SPSS statistics 23.0 software.

Results

Baseline Characteristics

This cohort of 819 participants (56% female) had an average age of 78.7 years, mean education level of 11.6 years (11.0 years for women and 12.6 years for men), and mean BMI of 25.7 kg/m² (25.3 kg/m² for women and 26.2 kg/m² for men). Of the participants 22.3% were carriers of the APOE ξ4 allele (genotypes ξ2/4, ξ3/4 or ξ4/4), and 30.8% were current smokers (Table 1). Cognitive scores are described in Table 2. Male participants performed better on the visuo-spatial function tests and females on memory tests.

Dietary pattern scores and intake from food groups for the whole group and stratified for men and women are presented in Supplementary Tables 4 and 5. Mean Mediterranean diet scores were in the mid-range on both scoring systems (4.4 for men and 4.3 for women using 0–9 scoring system (23, 24), and 27.9 for men and 27.5 for women on a 0–55 scoring system (25). Mean DASH diet scores, 26.6 for male and 27.3 for female participants (27, 28), were also about half of the maximum DASH score (range from 9 to 45).

Table 4.

Major dietary patterns derived by PCA*: Prudent Healthy Diet (female) and Western Diet (male): Sydney Memory and Ageing study (N=819)

Prudent Healthy Diet (female)		Western Diet (male)
Food items	Factor Loading	Food items	Factor Loading
Yellow vegetables	0.342	Yellow vegetables	-0.401
Green vegetables	0.429	Cruciferous vegetables	-0.310
Cruciferous vegetables	0.391	Nuts	0.234
Other vegetables	0.395	Legumes	-0.302
Tomato	0.222	Low fat dairy	-0.482
Legumes	0.204	Full fat dairy	0.356
Nuts	0.270	Margarine	-0.456
Garlic	0.244	Butter	0.270
Full fat dairy	-0.217	Chocolate	0.226
Poultry	-0.420	Vegemite	0.214
Meat pies	-0.228	Cakes and biscuits	0.759
Fried fish	-0.603	Flavoured milk	0.238
Chips	-0.366	Fruit juice	0.271
Sugar	-0.211	Tinned fruit	0.357

Notes: *PCA Principal Components Analysis

Table 5.

A-posterori dietary patterns and cognition by sex: Sydney Memory and Ageing study (N=819)

Cognition Domains	DP by PCA	β	95% CI		P value*
Attention Processing Speed	Prudent healthy Diet (female)	.451	.08	.823	.018
	Western Diet(male)	-.221	-.498	.055	.114
Language	Prudent healthy Diet (female)	.25	-.042	.543	.093
	Western Diet(male)	-.176	-.439	.087	.186
Executive	Prudent healthy Diet(female)	.082	-.203	.366	.57
	Western Diet(male)	-.325	-.552	-.099	.005
Visuospatial	Prudent healthy Diet(female)	.265	.043	.488	.02
	Western Diet(male)	-.146	-.268	-.024	.02
Global Memory	Prudent healthy Diet(female)	.107	-.193	.407	.479
	Western Diet(male)	.008	-.268	.285	.954
Verbal Memory	Prudent healthy Diet(female)	.101	-.191	.393	.492
	Western Diet(male)	.111	-.179	.040	.451
Global cognition at baseline	Prudent healthy Diet(female)	.307	.053	.562	.019
	Western Diet(male)	-.242	-.451	-.034	.023

Notes: Results were fully adjusted for age, sex, education, as well as physical activity, BMI, metabolic syndrome, hypertension, diabetes, hypercholesterolemia, history of stroke/transient ischaemic attack (TIA), physical activity, smoking, depression and APOE ξ4 genotype.

*P < 0.05 for global cognition or P<0.01 for individual cognitive domains, is significant.

A -priori diets and cognition

A higher Mediterranean diet score constructed by the 0–55 scoring system, was associated with better visuospatial cognition (β=0.045; 95% CI:0.017, 0.072; P=0.002) (supplementary table 7). Further analysis of food groups (25) showed that after full adjustment, consumption of the legume and nuts group was positively related to better performance in global cognition (β=0.117; 95% CI:0.052, 0.181; P<0.001) and cognitive domains of language and visuospatial (Table 3). A positive association between alcohol intake and baseline cognition observed in the basic model did not reach significance after full adjustment of covariates (Supplementary Table 8).

The DASH diet score, was associated with better performance on the visuospatial test (β=0.053; 95% CI: 0.023, 0.083; P=0.001). No significant associations were found between level of adherence to the Dietary Guideline Index 2013 with global cognition or performance in any other cognitive domains (Supplementary Tables 6 and 7).

A posteriori diets and cognition

Analysis of PCA dietary patterns were stratified by participants' sex, given significant differences in food group consumption and cognitive performances between men and women (Table 2 and Table 4). One major dietary pattern was generated and examined for each sex, the Prudent healthy diet for women and the Western diet for men (see Table 4 and Supplementary figure 1 and 2).

For women, the Prudent healthy diet pattern was associated with better performance in global cognition (β=0.307; 95% CI: 0.053,0.562; P=0.019) (Table 5). In men, a Western dietary pattern appeared to relate to poorer performance in global function (β=-0.242; 95% CI: -0.451, -0.034; P=0.023), as well as in executive function tests (β=-0.325; 95% CI: -0.552, -0.099; P=0.005) (Table 5).

Discussion

This study investigated the cross-sectional associations of dietary patterns and cognition in the Sydney Memory and Ageing Study. Main findings were that both the Mediterranean and DASH diets were positively linked to visuospatial cognition, and that higher consumption of the legumes and nuts food group was associated with better overall performance in global cognition and in multiple cognition domains. The Prudent healthy diet, derived by PCA, was positively associated with global cognition among women while the Western diet was associated with poorer global cognition and executive function in men.

Our analysis of food groups found significant positive associations between legume and nut consumption and better global cognition and higher scores in multiple cognitive domains, consistent with other research reporting positive links between high legume (34) and nut intake with cognition of older adults (35, 36). The underlying mechanism may be due to their low glycaemic index (GI) properties which benefit cognition by stabilizing brain glucose levels (37). Research also suggests the brain-gut-microbiome connection and interplay between food and gut microbiota influences neurocognitive health (38). Nuts and legumes are rich in plant based protein, fibre, anti-inflammatory agents such as polyphenols, poly-/mono-unsaturated fatty acids and may improve intestinal microbiota composition, positively affecting cognition (38).

Despite reports of cognitive benefits from the Mediterranean diet, putatively explained by the diet's anti-inflammatory, anti-oxidant properties (11), no associations were found in global cognition or cognition in individual domains apart from visuospatial cognition. It may be that visuospatial cognition is more susceptible to diet influence (39). Our dietary assessment tool was limited in measuring Mediterranean diet adherence, as it did not assess intake of olive oil which is a key component of the Mediterranean diet. In addition, the mostly Australian-born study population in Sydney, differed greatly from Mediterranean populations in various aspects including lifestyle, attitude towards foods and eating habits which could have impacted the overall outcome (11). Furthermore, higher adherence to the Mediterranean diet may be critical (40) and adherence in this study was low when compared to Mediterranean countries, with a mean score at only half of highest possible Mediterranean diet score (Supplementary Table 4).

Similarly, the link between higher adherence to DASH diet and visuospatial cognition (28, 41) was observed. However this result should be interpreted with caution. We performed multiple analyses and did not adjust for anti-hypertensive medications which may be linked to dementia-related pathophysiological pathways (42). Possible effects may be explained through the low sodium, low sugar and high vegetable focus of the DASH diet (30) with richness in mono-/poly-unsaturated fats, beneficial for vascular health and insulin sensitivity (43, 44) for stable brain glucose levels (37).

A Prudent healthy diet generated using dimension reduction methods has been associated with improved cognition among older adults in our systematic review (11). Our results from MAS, again showed association of a Prudent healthy dietary pattern with better performance in global cognition in older women. This cognition-friendly diet was high in yellow, green leafy, cruciferous and other vegetables, as well as nuts, legumes and garlic. By contrast, a western dietary pattern found to be associated with worse global cognition among men, comprised foods high in saturated fat and sugar, including full fat dairy, butter, flavoured milk and cakes.

This study is among the first few, to our knowledge to investigate adherence to the Australian Dietary Guidelines and cognitive performance among older Australians. Similar to other international dietary guideline index studies we found no associations (Supplementary Table 6 and 7) (45, 46). The Dietary Guidelines are based on evidence for the prevention of chronic disease such as diabetes and obesity (31), and not specifically designed for the prevention of cognitive decline. Within the guidelines the mono-/poly- unsaturated fats including olive oil, because of their high energy density, have only a small allowance (approximately 20 g spread or 14g oil) (33), much lower than the level of daily consumption reported to be beneficial for cognitive health among older adults. For example, in the PREDIMED study 1 Litre/week extra virgin olive oils were provided (40). In addition, detrimental factors specified in the Mediterranean and DASH diets, such as red and processed meat, are counted positively as an essential protein source in the DGI and foods linked to better brain health such as fish, legumes and nuts are also counted within the protein group. This suggests that more specific dietary guidelines for cognition may be needed for education and policy around better cognitive health for adults.

Mild to moderate alcohol intake was positively related to better cognition at baseline in the basic model, but not after adding cardiovascular disease (CVD) risk factors as covariates (Supplementary Table 8). Whether this is due to confounding effects or CVD risk factors as a mediator between alcohol intake and cognition, is uncertain, especially lacking data of lifetime alcohol consumption in this analysis. Some previous reviews and meta-analysis have reported light to moderate alcohol intake may be associated with reduced risk (47), however, this requires further research (1, 48).

Our study has multiple strengths, including comprehensive neurocognitive tests in multiple cognition domains, a validated dietary assessment tool, dietary patterns constructed by a qualified dietitian, statistical adjustments for important confounding factors such as cardiovascular risk factors and APOE ξ4 genotype, analysis on both a-priori and a-posterori dietary patterns, as well as investigation into key Mediterranean food groups, rather than dietary patterns alone.

However, potential limitations suggest caution is needed in interpreting the findings. This is a cross-sectional analysis so causality cannot be determined. Our results may be impacted by limitations of the dietary assessment tool and scoring methods used. Firstly, DQES v2 did not assess olive oil intake, an essential component of Mediterranean diet, and our proxy of monounsaturated to saturated fat ratio may not fully compensate, as studies reported lower plasma inflammatory markers and increased anti-oxidant capacity when compared to other vegetable oils (49). Secondly, the Mediterranean diet 0–55 system scored alcohol intake linearly as a detrimental factor, while studies suggested a non-linear relationship and gender differences (50), and Mediterranean diet encourages mild to moderate alcohol intake (23). Thirdly, the study assessed current dietary patterns; lifetime patterns may be more revealing.

Conclusion

We found that greater adherence to the Mediterranean diet and DASH diet were both associated with better visuospatial function, and legumes and nuts were positively linked to better performance in multiple cognition domains and global cognition. Among dietary patterns derived by PCA, a Prudent healthy diet was positively associated with global cognition among older women. A Western diet was linked to poorer global function and executive among men. Future longitudinal analysis is needed to further investigate the relationship of diet and cognition change over time.

Conflict of interest

Henry Brodaty is an Advisory Board member for Nutricia. None declared by other authors.

Author contributions

XC designed the research protocol, conducted data analysis, drafted the protocol and report, including creating reference list and tables. ZXL was responsible for technical support on statistical analysis, interpreting results and providing comments on manuscript. PS and NK were responsible for project design, data collection and study coordination, critical revision of manuscript. FOL provided support on dietary pattern score construction, data analysis, interpretation of results, critical revision of manuscript and approval of report. HB was responsible for designing the research protocol, interpreting results, critical revision and final approval of report. All authors reviewed the final draft.

Funding

Sydney Memory and Ageing Study received funding from the National Health and Medical Research Council (NHMRC) Australia.

Declaration

This study complies with the current laws of Australia where research was performed.

Ethical standards

The study was approved by the Ethics Committees of the University of New South Wales and the South Eastern Sydney and Illawarra Area Health Service.

Electronic Supplementary Material

Supplementary material is available for this article at https://doi.org/10.1007/s12603-020-1536-8 and is accessible for authorized users.

Supplementary material, approximately 134 KB.