Association of MIND diet with cognitive decline among Black and White older adults

Abstract

INTRODUCTION

We examined the Mediterranean–Dietary Approaches to Stop Hypertension Intervention for Neurodegenerative Delay (MIND) diet's association with cognitive decline by race among older adults in the Chicago Health and Aging Project.

METHODS

Five thousand two hundred fifty‐nine participants (73.5 [± 6.0] years, 62% Black participants, 62% female) completed a food frequency questionnaire, and two or more cognitive assessments over 7.8 ± 4.6 years.

RESULTS

Overall, higher MIND diet was associated with slower cognitive decline (p for trend = 0.0025). The MIND score (range:0‐15) was different between Black and White older adults(6.97 vs. 7.12, p = 0.010). Compared to the lowest tertile, among White participants, the two highest tertiles (MIND score –7: β = 0.0121 [95% confidence interval [CI]: 0.0006, 0.0237]; MIND score –8.5: β = 0.0146 [95% CI: 0.0003, 0.0260]) and among Black participants, only the highest tertile (MIND score –8.5: β = 0.0088 [95% CI: 0.0003, 0.0172]) had association with cognitive decline. Vascular and lifestyle factors attenuated the association only for Black older adults.

DISCUSSION

The MIND diet was associated with slower cognitive decline in Black and White older adults, but this may vary with other lifestyle and vascular factors. Further research is warranted on race‐specific cultural diets considering other risk factors for cognitive decline.

Highlights

• The intake of Mediterranean–Dietary Approaches to Stop Hypertension Intervention for Neurodegenerative Delay (MIND) diet components varies by race.

• The MIND diet may slow cognitive decline in both Black and White older adults.

• This association may vary with other lifestyle and vascular risk factors.

1. BACKGROUND

The role of modifiable lifestyle risk factors, such as diet, in slowing cognitive decline and reducing Alzheimer's disease (AD) dementia risk among increasingly diverse aging populations is of great public health interest. People living with AD dementia are projected to increase in the coming years, with those from underrepresented racial and ethnic backgrounds having a higher disease burden. ¹ Black and Hispanic older adults have a higher incidence and prevalence of dementia compared to White older adults. ² This health disparity is thought to be due to various reasons, including a higher vascular risk burden, lower education and socioeconomic status (SES), limited access to health care, and other social determinants of health (SDOH), ³ , ⁴ , ⁵ which can also impact lifestyle factors like diet. Various environmental, metabolic, and nutritional risk factors for dementia are different between Black and White adults. ⁶ In addition, diet may vary by race and ethnicity and be affected by factors such as residential neighborhood or geographic location, food accessibility, and food environment. ⁷ , ⁸ , ⁹ Various dietary patterns, including Mediterranean, ¹⁰ , ¹¹ , ¹² , ¹³ Dietary Approaches to Stop Hypertension (DASH), ¹² Prudent, ¹⁴ Nordic‐prudent, ¹⁵ and MIND ¹⁶ , ¹⁷ , ¹⁸ (Mediterranean–DASH Intervention for Neurodegenerative Delay) diets have been investigated for their association with changes in cognition among older adults. ¹⁹ , ²⁰ However, most of these studies were conducted in predominantly White populations, and even among the multiethnic cohorts that have reported diet associations with cognitive decline, they have not reported stratified analyses by race and sex. ²¹ Thus, it is important to understand how existing established dietary patterns, the MIND diet in particular, which is known to be associated with not only slower cognitive decline ¹⁶ but also with reduced AD dementia risk, ²² fewer AD pathologic changes, ²³ and less disability, ²⁴ is associated with cognitive decline in diverse populations. Further, it is important to understand whether the association varies by race or within race‐specific sex groups to ultimately design interventions that are tailored to specific groups with intersecting identities (e.g., Black women, White men, etc.). In this study, we assessed the association of the MIND diet with cognitive decline among a diverse population‐based sample of older adults from a longitudinal cohort study of aging and AD, the Chicago Health and Aging Project (CHAP) and further assessed whether the association varied by race or race‐specific sex groups.

RESEARCH IN CONTEXT

1. Systematic review: The Mediterranean–Dietary Approaches to Stop Hypertension Intervention for Neurodegenerative Delay (MIND) diet is associated with slower cognitive decline; however, most findings have been reported among predominantly White populations. This study aimed to determine the association of the MIND diet with cognitive decline in a population‐based study of older Black and White participants.

2. Interpretation: The MIND diet was associated with a slower rate of cognitive decline. Among White participants, the middle or highest group of the MIND diet score was associated with cognitive decline; however, among Black participants, only the highest group of the MIND diet score has this association compared to the lowest. Further, this association was attenuated among Black, but not White older adults, when other lifestyle and vascular factors were considered.

3. Future directions: Further research is warranted on race‐specific cultural diets considering other lifestyle factors, vascular health, or other correlated risk factors that impact cognitive decline differently by race.

2. METHODS

2.1. Subjects and methods

2.1.1. Study population

The study was conducted with participants in the CHAP, a population‐based cohort study of older adults residing in the Chicago area ²⁵ who were followed longitudinally from 1993 to 2012. The study consists of 10,802 individuals aged ≥ 65 years from four geographically defined communities of Black and White population. Detailed study design and objectives of CHAP have been reported previously. ²⁵ Informed consent was obtained from all participants. Data collection within the population occurred every 3 years, and participants received brief cognitive tests and risk factor assessments, which included diet assessments using a modified Harvard Food Frequency Questionnaire (FFQ). Of 10,802 participants, 6997 participants had at least one FFQ. We excluded those who completed the FFQ > 2.5 years after the baseline cognitive assessment (n = 171). We further excluded participants from races other than Black or White, that is other groups (Asians, Native Americans, Pacific Islanders, and others [n = 22]), and those without covariate data (n = 41) or follow‐up cognitive assessments (deceased or did not participate in follow‐up assessments or lost to follow‐up, n = 1387). We also excluded those with invalid dietary assessments including those with (1) < 700 kcal or > 4200 kcal for men and < 600 kcal or > 3700 kcal for women ²⁶ or (2) entire pages or more than half of the items missing from the FFQ (n = 68); or those with a baseline Mini‐Mental State Examination (MMSE) score < 10 (n = 49) as done previously. ²⁷ , ²⁸ The final analytic sample included 5259 participants with a mean follow‐up of 7.85 (± 4.57) years. Among analyzed participants, the baseline FFQ was completed by 0.82 (± 0.94)/median years after the initial cognitive test. Demographically, our analytical sample was like the overall cohort (mean age: 73.5 ± 6.7 vs. 73.4 ± 7.1 years, 62% vs. 63% Black, 62% vs. 61.5% female). The institutional review board of Rush University Medical Center approved the study.

2.1.2. Dietary assessment

The FFQ used to assess diet was validated in CHAP participants and has 144 items, including questions on various foods and supplement use. ²⁹ , ³⁰ The daily nutrient and total calorie intake were computed using the Harvard nutrient database. One serving size of all the food items was defined as the average amount consumed in the United States according to the US Department of Agriculture database as described previously. ²⁹ The frequency of food intake was multiplied by the nutrient composition of one serving of that food item and then summed over all the foods and supplements to estimate the total daily intake.

The MIND diet was computed as the sum of 15 dietary components, as previously described. ¹⁶ The score ranged from 0 to 15; a higher score indicates higher adherence. ¹⁶ The diet includes 10 brain‐healthy food groups (green leafy vegetables, other vegetables, nuts, berries, beans, whole grains, fish, poultry, olive oil, and wine) and five unhealthy food groups (red meats, butter and stick margarine, cheese, pastries and sweets, and fried/fast food) that were reverse coded, that is, “zero” if consumed more and “one” if consumed less. ¹⁶ Further, we also used each of these 15 dietary component servings per week to assess dietary intake by race. There are repeated measures of FFQ in the cohort. However, when we looked at the average change of the MIND diet score per year using a mixed effects model controlled for age, sex, education, and race, we found only a 0.24% change in the MIND diet score per year, that is diet is not changing substantially among these older adults. Thus, we have only considered the first MIND diet score for this analysis.

2.1.3. Cognitive function assessment

There were four cognitive tests, including the East Boston Memory Test (Immediate and Delayed Recall), ³¹ MMSE, ³² and Symbol Digit Modalities Test ³³ administered during the in‐home interviews at baseline, and at 3, 6, and 9 year follow‐up. For the East Boston Memory Test, participants were read a story and asked to recall it immediately (immediate recall) and then after 3 minutes of distractor tasks (delayed recall). Based on the number of story elements recalled, both immediate and delayed recall conditions had scores ranging between 0 and 12. The MMSE was used to measure overall cognitive function, and the score ranged from 0 to 30. The Symbol Digit Modalities Test assessed perceptual speed; participants were asked to match the numbers and symbols as fast as possible within 90 seconds (possible scores ranged from 0 to 96 as per the correctly matched items). All four tests are highly correlated, and to reduce measurement error along with any floor and ceiling artifacts, we computed a global composite score for cognitive function. Using the baseline population mean and standard deviation, we converted all the raw scores of each test to standardized z scores and averaged them to obtain the overall global measure of cognitive function. ³⁴

2.1.4. Covariates

Information on date of birth, sex, years of education, and self‐reported race was collected at participants’ baseline interviews. Age was calculated based on the date of birth and date of the first cognitive assessment. Race was determined by questions and categories from the 1990 US Census. SES was a composite of education, income, and occupation. ³⁵ Education included the number of years of formal education. Income was collected by a show‐card method with 10 categories as a continuous measure. The occupation was based on a prestige score for the primary occupation during most of their life. Education, income, and occupation continuous values were converted into a z score to compute a composite for SES. All lifestyle and vascular factors were obtained at baseline. Physical activity was based on self‐reported minutes spent walking, exercising, yard work, calisthenics, biking, and water exercise. ³⁶ The mean frequency of time spent participating in cognitive activities such as reading newspapers, and books, playing board games or crossword puzzles, or going to religious services was constructed as a composite score as described previously, such that a higher score means activities are done more often. ³⁷ Hypertension was present if there was a self‐reported history or any use of antihypertensive medication or measured systolic/diastolic blood pressure ≥ 160/90 mm Hg. Information on diabetes (self‐reported or taking any diabetic medications), stroke (based on medical history and neurological examination), and myocardial infarction (self‐reported history or use of digoxin) were also collected. The study staff measured weight and height during the home visit. Body mass index (BMI) was calculated using height in centimeters and weight in kilograms.

2.2. Statistical analysis

We examined the baseline characteristics of study participants overall (described in Table 1 as means or percentages) and by tertile of MIND diet score based on overall analytical sample distribution (described in Supplementary tables in supporting information). We used Spearman correlations to determine whether the MIND diet score was correlated with other lifestyle factors (physical and cognitive activity), SES, and BMI. Fifteen MIND diet components were compared between Black and White participants using a non‐parametric test, and Bonferroni correction was applied (0.05/15 = 0.0033, p < 0.003 was significant). We used separate linear mixed‐effects models to examine the association of MIND diet score with change in cognitive function in the overall analytical sample, then used an interaction term to test the effect modification by race (used two‐way [diet x race] and three‐way [diet x race x time (in years)] interaction term in the same model). Given we have limited literature on diet and cognitive decline among diverse populations, and there are significant racial differences in baseline cognition, diet, and covariates including age and socioeconomic status, as well as other lifestyle and vascular risk factors in this sample, we further stratified models by race (White [N = 1997] and Black [N = 3262]) to explore a deeper understanding of diet and cognition within each racial group. ³⁸ The MIND diet score was modeled as a continuous variable and also with an indicator variable for the middle (MIND diet score = 7.0) and highest (MIND diet score = 8.5) tertiles, which were compared to the lowest tertile (MIND diet score = 5.5) as the referent group. The same tertile cutoffs were used for all stratified analyses to compare high, middle, and low diet score groups. The linear trend of the MIND diet score association was also examined, in which all the observations within the tertile were assigned the median value and modeled as a categorical variable. The core model included multivariable adjustments for established risk factors, including a term for age (years), sex, race, composite SES (including education), and total calories. Additionally, for secondary analyses, we also tested if adding BMI (may be considered clinical sequelae of cognitive decline or dementia) or an interaction term of age x race in the models modified the association as there was a significant BMI and age difference by race in our analytical sample and age is a well‐established risk factor for cognitive decline. We also tested our models by adding baseline lifestyle (physical and cognitive activities) and cardiovascular risk factors (hypertension, diabetes, stroke, and myocardial infarction; all measured at baseline). All the model diagnostics were evaluated both analytically and graphically and no violations of model assumptions were found. Additionally, we calculated the estimate of the equivalent age difference in years to the difference in decline rates with one unit increase in the MIND diet score, by computing the ratio of the beta coefficients (β [time x age]/β [time x MIND score]) in the model adjusted for demographic, lifestyle, and vascular risk factors.

TABLE 1.

Baseline characteristics among 5259 Chicago Health and Aging participants.

	Overall	Black partiicpants	White participants	p value (Black vs. White participants)
N	5259	3262	1997
MIND diet score, Mean ± SD	7.01 ± 1.67	6.97 ± 1.59	7.09 ± 1.79	0.010 *
Age, Mean ± SD, years	73.5 ± 6.0	72.7 ± 5.4	74.7 ± 6.7	<0.0001
Sex, female (%)	62%	63%	62%	0.447 **
Education, Mean ± SD, years	12.5 ± 3.6	11.5 ± 3.7	14.1 ± 3.3	<0.0001
Adult SES score, mean ± SD ^	0.21 ± 0.83	−0.06 ± 0.74	0.67 ± 0.79	<0.0001 *
BMI, Mean ± SD	27.7 ± 5.75	28.50 ± 6.04	26.54 ± 5.02	<0.0001
Physical activity, mean ± SD, hours/week	3.3 ± 5.3	2.8 ± 4.9	4.2 ± 5.7	<0.0001 *
Cognitive activity score, mean ± SD a	0.21 ± 0.65	0.06 ± 0.64	0.46 ± 0.58	<0.0001
Calories, Mean ± SD, Kcal/day	1725 ± 602	1682 ± 628	1795 ± 550	<0.0001
Diabetes (%)	19%	23%	12%	<0.0001 **
Hypertension (%)	68%	73%	60%	<0.0001 **
Stroke (%)	8%	9%	7%	0.108 **
Myocardial infarction (%)	13%	14%	13%	0.312 **

Abbreviations: BMI, body mass index; MIND, Mediterranean–Dietary Approaches to Stop Hypertension Intervention for Neurodegenerative Delay; SD, standard deviation; SES, socioeconomic status.

^{^} Adult SES is composed score of three items: education, income, and occupation.

^a Higher score means cognitive activities are done more often.

^*Non‐parametric tests; ^**chi‐square test.

For each race, we tested for effect modification by sex in separate multivariable core models that included terms for tertile of MIND, sex, and other covariates of our core model, with time and a three‐way interaction term between diet score, sex, and time. All the stratified models were further controlled for vascular risk factors, physical activity (hours/week), late‐life cognitive activity, and an interaction term between time and each variable. We also conducted a sensitivity analysis for the core model after removing participants whose baseline cognitive scores were in the lowest 10% of the distribution. Standardized betas were calculated by dividing the beta by standard error to compare the effect estimates by race. Statistical analysis was performed using SAS version 9.4 (SAS Institute Inc.).

3. RESULTS

The baseline characteristics of 5259 study participants (mean age = 73.5 ± 6.7 years, 62% Black, 62% females) and by race are presented in Table 1. Black participants were a few years younger, had fewer years of education, lower SES, lower frequency of physical or cognitive activities, and had higher BMI and more vascular risk factors (diabetes and hypertension) compared to White participants (Table 1). Overall, the MIND diet score was weakly correlated with other lifestyle factors, BMI, education, or SES (composite score; 0.04 < rho < 0.20, p < 0.05 for all correlations with MIND diet score, Table S1 in supporting information). Baseline characteristics of Black and White participants by tertile of MIND diet score are presented in Table S2 in supporting information. Among both Black and White participants, those in the highest tertile of the MIND diet score were younger, were more likely to be female, engaged in more physical and cognitive activities, and had better SES than those in the lowest MIND diet tertile.

Overall, the MIND diet score was marginally lower (p = 0.01, Table 1) in Black participants (mean ± standard deviation [SD], range [highest to lowest]: 6.97 ± 1.59, 2.0 to 13.5) than the White particiapnts (mean ± SD, range [highest to lowest]: 7.09 ± 1.79, 1.5 to 14). The MIND diet overall and tertiles distribution by race are presented in Figure S1 in supporting information. Black participants consumed fewer healthy dietary components such as vegetables, berries, chicken, olive oil, and wine compared to White participants (Table S3 in supporting information). In contrast, Black participants consumed more of other healthy components, including fish and whole grains (Table S3). The intake of green leafy vegetables, nuts, legumes, and beans was similar by race. We also found that Black participants consumed fewer unhealthy components of the MIND diet as recommended (i.e., foods that should be avoided or limited, red and processed meat, pastries and sweets, butter/margarine, and full‐fat cheese) but more fried foods compared to White participants.

In our analysis of the MIND diet score as a continuous variable and change in cognitive function over the mean follow‐up time (SD) of 7.85 (± 4.57) years, we found that a higher MIND diet score was associated with slower cognitive decline (β = 0.0024, 95% confidence interval [CI]: 0.001, 0.004, Table 2) after adjusting for demographic characteristics and total calories. This association was retained (β = 0.002, 95% CI: 0.0003, 0.003) after including BMI in the model. When the model was controlled for physical activity, cognitive activity, hypertension, stroke, myocardial infarction, and diabetes, the association remained (β = 0.0019, 95% CI: 0.003, 0.004, Table 2). With a one‐unit increase in the MIND diet score, the difference in the rates of cognitive decline was almost equivalent to being 2.1 years younger.

TABLE 2.

Association of MIND diet with cognitive decline among 5259 Chicago Health and Aging Participants.

		MIND Diet sore (tertiles)
	Continuous MIND diet score	Tertile 1	Tertile 2	Tertile 3	p for trend
		5.5	7.0	8.5
N	5259	1822	1745	1692
beta (95% confidence interval)
Model (A)	0.0024 (0.001, 0.004)	Reference	0.0049 ( −0.002,0.011)	0.0104 (0.004, 0.017)	0.0025
Model (B)	0.0019 (0.0003,0.004)	Reference	0.0044 ( −0.002,0.012)	0.0083 (0.002,0.015)	0.0170

Abbreviation: MIND, Mediterranean–Dietary Approaches to Stop Hypertension Intervention for Neurodegenerative Delay.

Model (A) adjusted for age, sex, socioeconomic status (includes education), race, calories, lag, and lag x each covariate.

Model (B) adjusted for age, sex, socioeconomic status (includes education), race, calories, physical activity, cognitive activity, hypertension, stroke, myocardial infarction diabetes, lag, and lag x each covariate.

In a separate model including tertiles, we found that participants in the highest tertile score (n = 1692) of the MIND diet had slower cognitive decline compared to those in the lowest tertile (n = 1822; Table 2, core and further adjusted models). In our study population, Black particiapnts were younger than White participants, so we adjusted our core model with an age x race interaction term, and there was no change in the effect estimate for the association of MIND diet score and cognitive decline (β = 0.0024, 95% CI: 0.0007, 0.004). We repeated our core models for sensitivity analysis after removing the participants with baseline cognitive scores in the lowest 10% of the distribution. The findings of the MIND diet with cognitive decline remained similar (β = 0.0018, 95% CI: 0.0001, 0.0035).

Next, we investigated whether race was an effect modifier for the association of diet with cognitive decline. The global interaction between diet and race approached significance in this model (p for interaction = 0.067), but the three‐way interaction (MIND diet x race x time) was not significant (p for interaction = 0.900) for the overall sample. The three‐way interaction for the tertile comparison for MIND diet association with cognitive decline was also insignificant (the middle MIND tertile x race x time had p for interaction = 0.129).

Given that there were differences in the MIND diet score by race, including some of its prominent dietary components, and there are limited studies investigating diet and cognition in diverse populations of this size, we also examined models stratified by race. In the stratified linear mixed models adjusted for age, sex, SES, and total calories, we found a MIND diet association with slower cognitive decline in both Black and White participants. However, among Black participants, the association was only significant among those with high adherence to the MIND diet and not among those with moderate adherence (Figure 1, Table 3). Meanwhile among White participants, the association was significant for moderate and high adherence groups (Figure 1, Table 3). Comparing the standardized betas (β/standard error [SE]), White participants with a MIND diet score of 7 or 8.5 (middle and highest tertile median scores) compared to a MIND diet score of 5.5 (lowest tertile median score), had standardized betas (β/SE) of 2.05 and 2.51, respectively. In contrast for Black participants, only those in the highest tertile (i.e., MIND diet score of 8.5) had a significant relationship with change in cognition with a standardized beta (β/SE) of 2.05. However, once further adjusted for other lifestyle factors and cardiovascular risk factors, the association only remained for White but not for Black older adults (Table 3).

FIGURE 1.

MIND diet association with cognitive decline among Black and White participants. Model adjusted for age, sex, socioeconomic status (includes education), calories, lag, and lag x covariates. MIND, Mediterranean–Dietary Approaches to Stop Hypertension Intervention for Neurodegenerative Delay; t1, Tertile 1; t2, Tertile 2; t3, Tertile 3

TABLE 3.

Estimated effects (β coefficient [95% CI]) of MIND diet on the rate of change in global cognitive score among CHAP participants stratified by race over the average follow‐up of 7.85 (± 4.57) years.

	Black participants					White participants
	Continuous Mind diet score	MIND diet score (tertile)			p value for trend	Continuous MIND diet score	MIND diet score (tertile)			p value for trend
		T1	T2	T3			T1	T2	T3
		5.5	7.0	8.5			5.5	7.0	8.5
N	3262	1138	1139	985		1997	684	606	707
Model (A)	0.0026 (0.0004, 0.005)	Reference	0.0007 ( −0.0007, 0.0087)	0.0088 (0.0003, 0.0172)	0.044	0.0024 ( −0.0003, 0.005)	Reference	0.0121 (0.0006, 0.0237)	0.0146 (0.0003, 0.0260)	0.012
Model (B)	0.0019 ( −0.0003, 0.004)	Reference	0.0008 ( −0.0007, 0.0088)	0.0064 ( −0.0002, 0.0149)	0.144	0.0019 ( −0.0007, 0.005)	Reference	0.0104 ( −0.0008, 0.0217)	0.0118 ( 0.0007, 0.0229)	0.040

Abbreviations: MIND, Mediterranean–Dietary Approaches to Stop Hypertension Intervention for Neurodegenerative Delay; SES, socioeconomic status.

Model (A) adjusted for age, sex, socioeconomic status (includes education), calories, lag, and lag x covariates.

Model (B) adjusted for age, sex, socioeconomic status (includes education), calories, physical activity, cognitive activity, hypertension, stroke, myocardial infarction diabetes, lag, and lag x covariates.

Within each race, we further assessed the interaction with sex in each group. The global interaction with sex in Black participants (p for interaction [diet x sex x time] = 0.927), and White participants (p for interaction [diet x sex x time] = 0.212) was not significant. However, the diet x sex interaction in the models was significant for White (p for interaction = 0.016) but not Black participants (p for interaction = 0.317). For the diet tertile comparison, MIND diet x sex x time interaction was significant only among White participants for those in the middle tertile of the MIND diet score (p for interaction, MIND diet tertile 2 x sex x time = 0.045; MIND diet tertile 3 x sex x time = 0.113); there were no sex interactions for Black participants (p for interaction, MIND diet tertile 2 x sex x time = 0.896; MIND diet tertile 3 x sex x time = 0.952). We further stratified the models among White participants by sex and found that there was no significant association between the MIND diet and cognitive decline among White men. However, among White women, being in the highest or middle tertile of the MIND diet was associated with slower cognitive decline (p for trend = 0.012, Table S4 in supporting information). This association was retained when further adjusted for lifestyle and vascular risk factors (data not shown).

4. DISCUSSION

In this large prospective population‐based cohort study, we found that the MIND diet was associated with slower cognitive decline in older Black and White adults when controlled for various socioeconomic, demographic, lifestyle, and vascular risk factors. Adding one MIND diet recommended component resulted in the rate reduction of cognitive decline for an individual equivalent to being 2 years younger in age. Overall Black older adults consumed less of certain MIND diet components including other vegetables (excluding green leafy vegetables), berries, chicken, olive oil, and wine, and more fish and whole grains than White older adults. Given the MIND diet score marginally differed by race and the limited literature on race‐specific associations of diet and cognitive decline, we examined race‐stratified analyses and found that Black older adults who followed more than eight dietary component recommendations of the MIND diet versus five dietary components had slower cognitive decline whereas among White older adults, following just seven or more dietary components versus five dietary components as recommended for the MIND diet was associated with slower cognitive decline. Considering other lifestyle and vascular risk factors, the MIND diet association with cognitive decline was attenuated significantly among Black but was retained in White older adults. However, among White older adults, the MIND diet association with cognitive decline was present only among women and not men. These novel findings in a diverse population‐based sample that is predominantly Black are important to inform future studies on diet and cognition and address a significant deficiency of diversity in AD/AD and related dementias (ADRD) research. We speculate that various established lifestyle and vascular risk factors among underrepresented populations may play a critical role in understanding the relationship of diet with cognition among diverse populations.

Our results support most existing literature from other longitudinal cohort studies ¹⁵ , ¹⁶ , ¹⁸ , ³⁹ , ⁴⁰ but not all ⁴¹ , ⁴² and a meta‐analysis demonstrating an association of the MIND diet with cognitive decline. ¹⁸ These findings also update previous findings from our group in which, in a subsample of CHAP participants with genomic data, we did not find any significant association between MIND and cognitive decline. However, effect estimates were in the right direction. This was possibly due to a smaller sample size than the present study (n = 2449 vs. n = 5259). ⁴³ The discrepancies in the literature with other cohorts can be due to fewer mean years of follow‐up, smaller sample sizes, brief assessment of cognitive battery tests, use of dietary assessment tools that were not designed or validated for the specific study population of interest, or limited foods included in the MIND diet scoring. Our findings also support existing literature on the MIND diet association with other AD/ADRD traits such as reduced AD dementia risk, ²² disability risk, ²⁴ fewer AD pathologic changes, ²³ higher brain volume, ⁴¹ and cognitive resilience. ⁴⁰ The findings from this study indicate that for a population‐based diverse sample, achieving a MIND diet score of ≈ 8.5 may benefit brain health compared to those with a lower MIND diet score of 5.5. In the recently published MIND trial, we found that the MIND diet intervention group and the mild caloric restriction group showed no significant group difference in cognition after 3 years of intervention. ⁴⁴ Although the trial consisted of predominantly White participants, cognitive performance in the mild calorie restriction group with the average MIND diet score of 8.3 supports the finding that a MIND diet score of > 8 benefits cognition.

Previously, we reported an association between the Mediterranean diet and cognitive decline in the CHAP cohort with no significant interaction between diet and race. ¹¹ In comparison, another cohort reported an association of the Mediterranean diet with cognition only among Black and not among White older adults. ¹³ Both the MIND and Mediterranean diets are plant‐based including more consumption of vegetables, nuts, legumes, whole grains, and olive oil, and in our sample, they were correlated (rho = 0.61, p < 0.0001). The Mediterranean diet emerged as an evidence‐based option to prevent cardiovascular disease. However, the MIND diet focuses on certain specific brain‐healthy components including berries and green leafy vegetables. Additionally, given cultural differences between the US and Mediterranean populations, the recommended MIND diet serving sizes for a few components (e.g., olive oil, fish, sugary and fried foods, processed meats) vary based on research on food and brain health outcomes as indicated in the original publication. ¹⁶ , ²² Recently, we also reported the association of a healthy plant‐based diet with cognitive decline among Black but not White older adults, ⁴⁵ even when models controlled for all the other lifestyle and vascular risk factors. These findings are similar to those reported for the Mediterranean diet in the Health, Aging, and Body Composition study among Black and White participants. ¹³ Healthy plant‐based food groups included whole grains, fruits, vegetables, vegetable oils, tea, coffee, nuts, and legumes. We postulate that it is a combination of certain specific food groups captured in the plant‐based diet that is associated with cognitive decline in Black but not White individuals. For example, we found whole grain consumption associated with slower cognitive decline among Black but not White participants. ⁴⁶ Thus, more studies that examine food component–based associations are needed to better understand how specific foods or combinations of foods may influence cognition in different populations. Additionally, diet choices may be driven by specific preexisting comorbidities that are more prevalent in older Black individuals and cause them to eat certain foods or limit others. Future studies are needed to investigate combinations of more culturally appropriate and more commonly consumed healthy foods to better understand these discrepant results. A recent report found that incorporating salient cultural characteristics and strategies is important for designing a dietary intervention that has the most adaptable MIND dietary model among older Black adults. ⁴⁷ Additionally, future studies are needed to investigate SDOH that shape food accessibility, nutrient environment, and dietary intake. To fully understand the role of diet in cognitive health among diverse older adults, we must consider different populations that consider existing food habits, food accessibility, food acculturation over time, and other preexisting risk factors that may put them at higher risk of cognitive decline, for example, obesity and vascular risk factors.

The mechanistic link between diet and cognition is not fully understood. However, given the MIND diet is primarily a plant‐based diet rich in essential nutrients and bioactive compounds, the role of these nutrients in maintaining neuronal health, synaptosome, blood–brain barrier, antioxidant, and anti‐inflammatory pathways are speculated to drive the MIND diet's relationship with brain health outcomes. ⁴⁸ , ⁴⁹ , ⁵⁰ Furthermore, the MIND diet recommends limiting processed meats, refined grains, high‐fat foods, and high‐sugar foods. Animal studies have shown that long‐term exposure to a high‐fat diet increases amyloid beta deposition ⁵¹ and neurofibrillary tangle formation while decreasing synaptic plasticity, effects that are accompanied by inflammatory and stress responses in whole brain lysate ⁵² and overall enhanced astrogliosis and microglia activation. ⁵³ Thus, further examining the mechanistic link between diet and cognition in different subgroups using novel biomarkers may help us better understand the role of nutrition in brain health and may help design more targeted nutrition interventions for dementia prevention.

This prospective study analysis has various strengths, including its large biracial sample of community older adults with long follow‐up time, a composite measure of standardized cognitive assessment at multiple time points, dietary information collected using an FFQ that was validated in this sample, and computation of a MIND diet score using all food categories captured in the detailed FFQ. Additionally, analyses considered adjustment of various confounding factors (demographic, lifestyle, and vascular factors) associated with cognition. The weak correlations of the MIND diet score with higher education and SES (composite score), raise the possibility that including additional unmeasured SDOH may have weakened its association with cognition. The study has a few limitations including the fact that given an observational study design, temporal causality cannot be established. Additionally, the study population is old and mostly White and Black participants; thus, the findings cannot be generalized to younger people or those from other racial and ethnic backgrounds.

The MIND diet is considered an important modifiable risk factor for cognitive decline and dementia in older adults. Our results extend these findings in a diverse population‐based sample that is mainly Black. The MIND diet is associated with slower cognitive decline in both Black and White older adults; however, this association may vary when considering other lifestyle and vascular risk factors. Whether our findings are due to lifestyle, vascular health, or other correlated risk factors that impact cognitive decline differently by race warrants further investigation.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflict of interest. Author disclosures are available in the supporting information.

CONSENT STATEMENT

All participants of this study provided informed consent.

Supporting information

Supporting Information

Supporting Information

Agarwal P, Barnes LL, Dhana K, et al. Association of MIND diet with cognitive decline among Black and White older adults. Alzheimer's Dement. 2024;20:8461–8469. 10.1002/alz.14277