Association of Adherence to a MIND-Style Diet With the Risk of Cognitive Impairment and Decline in the REGARDS Cohort

Abstract

Background and Objectives

Diet may influence the development of cognitive impairment and affect cognitive decline, but whether this relationship varies between Black American and White American people is unclear. This study examined the association of Mediterranean-Dietary Approaches to Stop Hypertension Intervention for Neurodegenerative Delay (MIND) and incident cognitive impairment and cognitive trajectories in a biracial prospective cohort study.

Methods

Using data derived from the Food Frequency Questionnaire in the REasons for Geographic and Racial Differences in Stroke study, we compared MIND diet adherence with incident cognitive impairment and cognitive trajectory in Black participants and White participants. Logistic regression was used to model MIND diet score (continuous variable and using tertiles) and incident cognitive impairment after adjusting for age, sex, race, region, education, income, total energy, hypertension history, dyslipidemia, diabetes, estimated glomerular filtration rate, ischemic heart conditions, atrial fibrillation, and lifestyle factors including sedentary, obesity, and smoking. Mixed-effects models were used to examine the association between cognitive trajectory and MIND diet adherence.

Results

Dietary data to calculate the MIND diet score and cognitive data were available on 14,145 participants with a mean age of 64 years (SD 9.0 years) that was 56.7% female. Greater MIND diet adherence was associated with a decreased incidence of cognitive impairment (odds ratio [OR] 0.96, 95% CI 0.93–0.99, p = 0.02) after adjusting for all covariates. In the fully adjusted model, greater MIND diet adherence was associated with decreased risk of cognitive impairment in female participants (OR 0.92, 95% CI 0.89–0.96, p < 0.001) but not in male participants (OR 1.01, 95% CI 0.97–1.06, p = 0.64). In all models, greater MIND diet adherence was associated with decreased risk of cognitive decline. MIND diet adherence was a better predictor of cognitive decline in Black participants (β = 0.04, SE = 0.007, p < 0.001) than in White participants (β = 0.03, SE = 0.004, p < 0.001).

Discussion

Greater MIND diet adherence was associated with decreased risk of cognitive impairment in female participants but not male participants, with no difference between Black participants and White participants. However, MIND diet adherence was a better predictor of cognitive trajectory in Black participants than in White participants.

Introduction

In 2015, approximately 47 million people worldwide were diagnosed with dementia, making it the seventh leading cause of death worldwide.¹ Owing to the global aging population, the number of people living with dementia is expected to increase to 75 million by 2030.¹ By focusing on cognitive decline rather than cognitive impairment and dementia, researchers may identify early lifestyle and pharmacologic interventions which delay the development of cognitive impairment and subsequent dementia.

Cognitive decline may be a better endpoint than cognitive impairment because cognitive decline can affect emotional and social functioning and overall quality of life before it reaches the severity of mild cognitive impairment.² Cognitive decline has also been shown to be a good predictor of underlying Alzheimer disease³ and can be easily assessed with focused cognitive batteries.⁴ An effective intervention to delay or prevent pathologic cognitive decline may best be targeted at the earliest symptomatic disease stages such as the cognitive decline before mild cognitive impairment during which cognitive functioning is still relatively preserved.²

Lifestyle modifications are key steps that can reduce the expected rise in cognitive decline and dementia cases. Previously, the role of specific nutrients and foods in the prevention of cognitive decline and dementia was extensively studied.^5,6 More recently, research has focused on dietary patterns and their implications in cognitive decline. Previous research has demonstrated that higher accordance to either the Mediterranean diet or the Dietary Approaches to Stop Hypertension (DASH) diet was associated with better cognitive function in cross-sectional and longitudinal studies.^1,7-9 Similarly, higher accordance to the Mediterranean-DASH Intervention for Neurodegenerative Delay (MIND) diet, which combines features' components of the Mediterranean and DASH diets, is associated with better cognition in cross-sectional studies^1,10 and longitudinal studies^1,11 as well as resistance to neuropathology.^12,13 Although a recent clinical trial of MIND diet intervention was not associated with delayed cognitive slowing in cognitively unimpaired participants with a family history of dementia when compared with those who followed the control diet with mild caloric restriction.¹⁴

Although diet is associated with delayed onset of cognitive decline, these studies had several limitations, including limited representation of Black American people. The REasons for Geographic and Racial Differences in Stroke (REGARDS) study was created to examine why Southern American and Black American people have a higher incidence of stroke and has been a robust source of research on the interplay between diet, stroke, and cognition.^15-17 Studying the association between diet and cognitive decline in the REGARDS cohort will overcome these limitations and provide robust data to inform clinical decision-making regarding dietary recommendations to attenuate cognitive decline across Black and White American people.

Methods

Standard Protocol Approvals, Registrations, and Patient Consents

REGARDS was conducted according to guidelines in the Declaration of Helsinki, and all procedures involving human participants were approved by the relevant Institutional Review Board at participating universities. Verbal and written informed consent were obtained from all participants.

Participants

The REGARDS study is an ongoing national, prospective cohort study examining stroke mortality and cognitive function in Black and White adults aged 45 years and older.¹⁸ A computer-assisted telephone interview (CATI) and an in-home examination were completed at baseline (2003–2007).^18,19 The study was designed to be roughly half from the Southeastern United States, an area known as the stroke belt (includes Louisiana, Arkansas, Mississippi, Alabama, Tennessee, Georgia, North Carolina, and South Carolina), and specifically within the stroke “buckle” region along the coastal plains of North Carolina, South Carolina, and Georgia.²⁰ The study was also designed to have half the population who identified as Black and half female to allow for the study of racial and sex differences in stroke and cognition. During the in-home visit, a self-administered questionnaire was provided to collect data on dietary intake, height/weight, electrocardiogram, blood pressure measurements, and blood work. A second CATI and in-home assessment like the first one was conducted approximately 10 years later (2013–2016). This study was conducted according to guidelines in the Declaration of Helsinki, and all procedures involving human participants were approved by the Institutional Review Board at all participating universities. Verbal and written informed consent were obtained from all participants.

MIND Diet Score Determination

The Block 98 Food Frequency Questionnaire (FFQ) was included as part of the baseline self-administered questionnaires that were left with participants.²⁰ The Block 98 FFQ was developed by Block Dietary Data Systems (Berkeley, CA) and validated in populations similar to REGARDS.^21,22 The questionnaire includes more than 150 multiple-choice questions based on 107 food items and can be completed in about 30–40 minutes. Participants were asked to recall usual dietary intake from the past year and mail the completed form along with the other questionnaires to the REGARDS coordinating center. The MIND diet score was calculated based on a scoring system of the components of the diet. One point was given for the following: whole grains ≥3 per day, green leafy vegetables ≥6 per week, other vegetables ≥1 per day, berries ≥2 per week, fish ≥1 per week, poultry ≥2 per week, beans >3 per week, nuts ≥5 per week, red meats/their products <4 per week, fast/fried food <1 per week, olive oil primary oil, and butter/margarine <1 tablespoon per day.⁹ See eTable 1 for MIND diet scoring.

Other Dietary Score Determination

In addition to the MIND diet score, other dietary scores were calculated using the above information. These dietary scores and their calculation methods in REGARDS include the DASH,²³ Mediterranean,²⁴ Dietary Inflammatory Index (DII),²⁵ dietary inflammation score (DIS),²⁵ Healthy Eating Index,²⁵ Life's Essential 8,²⁶ and Southern-style diet.²⁰

Cognitive Outcomes and Their Assessment

The Six-Item Screener (SIS), Word List Learning (WLL), Delayed Recall (WLD), Animal Fluency (AF), and Letter “F” Fluency (LF) were measured at baseline and CATI follow-up.^19,27-29 Based on the Mini-Mental State Examination, the SIS is a test of global cognitive function.^19,27,29 The SIS is comprised of a 3-item word recall and 3-item temporal orientation, it has a score range of 0–6.^19,27,29 The SIS has a 74.2% sensitivity and 80.2% specificity for clinically confirmed cognitive impairment in community samples with a score less than 5.^19,27,29-31 The number of instances of cognitive assessment varied by participant, ranging from one instance of cognitive assessment at baseline to 9 instances.

Using a robust norms approach, incident cognitive impairment was determined. First, a normative sample was selected from the REGARDS cohort who completed the second in-home assessment.²⁹ This normative sample was selected using a pseudorobust norms approach incorporating outcomes and the Ascertain Dementia 8-item and assessments of the instrumental activities of daily living derived from the Minimum Data Set to infer the absence of clinically significant cognitive decline at the second in-home visit.^29,32-34 Using linear regression of the first assessments of participants in the normative sample, estimated scores and SDs were calculated for each assessment by age/sex groups.²⁹ Age groups were 45–54, 55–64, 65–74, 75–84, and 85+.²⁹ A total of 6,264 participants were included in the normative sample. For this study, a participant had incident cognitive impairment if their 3 of 4 assessment (WLL, WLD, LF, and AF) scores were more than 1.5 SDs below the expected score.²⁹ If a participant did not complete 3 of the 4 assessments on the same call, incident cognitive impairment was based on the SIS.^15,29 If a participant's incident cognitive impairment was based on the SIS, the SIS had to be < 4 once or = 4 twice. If the participant was free from cognitive impairment (≤4 on baselined SIS), the cognitive impairment was considered incident. Next a composite index for cognitive function was then created as previously described.^29,35 Briefly, the composite score averaged Z-scores from WLL, WLD, AF, LF, and SIS which were calculated relative to the normative sample.²⁹

Data Analysis

Analyses were conducted using SAS, version 9.4 (SAS Institute, Cary, NC). Statistical significance was set at p < 0.05. Baseline characteristics were collected on all study participants. Age, race, sex, region (stroke belt vs nonstroke belt), level of education (at least a college degree or less than a college degree), income (refused and less than $75,000 annually, or $75,000 annually or above), and current smoking status were self-reported. Total energy (calories) was determined using components from the FFQ collected at baseline. Body mass index (BMI) and waist circumference were collected during the in-person examination at baseline. Sedentary behavior used as a binary variable and defined as watching more than 4 hours of television and participating in no daily physical activity, all other combinations were classified as “not sedentary behavior.” History of ischemic heart conditions includes self-reported myocardial infarction, coronary bypass surgery, angioplasty or stenting, or evidence of a myocardial infarction from an ECG obtained at baseline. The history of atrial fibrillation was self-reported or determined by evidence from the ECG. Participants reporting use of pills or insulin or with fasting glucose levels greater than 125 mg/dL or nonfasting levels at or greater than 200 mg/dL were considered diabetic at baseline. Hypertension history was defined as systolic blood pressure ≥140 mm Hg, diastolic blood pressure ≥90 mm Hg, or self-reported current medication used to control blood pressure. A participant was considered to have dyslipidemia if they reported current use of lipid medications, cholesterol ≥240 mg/dL, low-density lipoprotein ≥160 mg/dL, or 0 < high-density lipoprotein ≤ 40. An estimated glomerular filtration rate (eGFR) was calculated using the Chronic Kidney Disease Epidemiology Collaboration equation,³⁶ which takes race into account.

Summary statistics for baseline demographic characteristics were compared according to MIND diet score tertile using the Mantel-Haenszel χ² test of proportions for categorical variables and F-statistic for continuous variables. Logistic regression was used to model MIND diet score (as a continuous variable and using tertiles) and incident cognitive impairment. Four models were examined. The crude model was unadjusted with the MIND diet score. Model 1 adjusted for age group, sex, race, region, education, income, and total energy. Model 2 also adjusted for hypertension history, dyslipidemia, diabetes, eGFR, ischemic heart conditions, and atrial fibrillation. The fully adjusted model 3 included sedentary, obesity, and smoking. To test the hypothesis that the effect of MIND diet accordance differed by race, sex, or age group, an interaction term was tested in the fully adjusted model before stratification. When stratified, group-specific tertiles were used. This process was repeated for the other dietary scores. In addition, Pearson correlation was completed to compare each dietary score with the MIND diet score.

Mixed-effects models examined the association between cognition and MIND diet accordance and whether cognition worsened over time. Participants were included in analyses if they had at least ≥2 cognitive scores. Follow-up time was calculated from the baseline cognitive assessment, and random effects, a random intercept for each participant and a random slope time, were included in all models. The crude model was unadjusted with the MIND diet score. Adjustment for time, time-squared (to account for nonlinear trends in cognitive scores over time), interaction of time and diet, interaction of time and age, and indicators for first assessments were included in model 1. Model 2 added adjustment for the presence of age, sex, race, region, education, income, and total energy. Additional adjustment for the presence of hypertension history, dyslipidemia, diabetes, eGFR, ischemic heart conditions, and atrial fibrillation were included in model 3. Finally, the fully adjusted model 4 included sedentary, obesity, and smoking. To test the hypothesis, the association of MIND diet accordance differed by race, a 3-way interaction term, between race, MIND diet accordance, and time, as well as the 2-way interactions, were tested in the full model. Age and sex were also analyzed for interactions.

Data Availability

This study uses data from the REGARDS cohort. To abide by its obligations with NIH/National Institute of Neurological Disorders and Stroke and the Institutional Review Board of the University of Alabama at Birmingham, REGARDS facilitates data sharing through formal data use agreements. Any investigator is welcome to access the REGARDS data through this process. Requests for data access may be sent to regardsadmin@uab.edu.

Results

Dietary data available to calculate the MIND diet score and cognitive data were available on 14,145 participants as seen in Figure 1. White participants were more likely to be older, male, earn ≥ $75,000 per year, and be a college graduate (Table 1). White participants were also more likely to have ischemic heart conditions, despite being less likely to have a history of hypertension and diabetes mellitus. In addition, White participants were more likely to have a lower BMI and exercise ≥4 times per week.

Figure 1. Flowchart of Enrollment.

Table 1.

Baseline REGARDS Characteristics by Race

	Overall (N = 14,145)	Black participants (N = 4,244)	White participants (N = 9,901)
Age	64.0 (9.0)	62.8 (8.6)	64.5 (9.1)
Sex, female	8,014 (56.7)	2,876 (67.8)	5,138 (51.9)
Income
Refused	1,516 (10.7)	397 (9.4)	1,119 (11.3)
<$20k	1,812 (12.8)	895 (21.1)	917 (9.3)
$20k–$34k	3,219 (22.8)	1,113 (26.2)	2,106 (21.3)
$35k–$74k	4,748 (33.6)	1,330 (31.3)	3,418 (34.5)
≥$75k	2,850 (20.2)	509 (12.0)	2,341 (23.6)
Education
<High school	1,066 (7.5)	540 (12.7)	526 (5.3)
High school graduate	3,458 (24.5)	1,116 (26.3)	2,342 (23.7)
Some college	3,860 (27.3)	1,246 (29.4)	2,614 (26.4)
College graduate	5,761 (40.7)	1,342 (31.6)	4,419 (44.6)
Region
Belt	4,855 (34.3)	1,372 (32.3)	3,483 (35.2)
Buckle	3,127 (22.1)	834 (19.7)	2,293 (23.2)
Nonbelt	6,163 (43.6)	2,038 (48.0)	4,125 (41.7)
Southeast	7,982 (56.4)	2,206 (52.0)	5,776 (58.3)
Urban group
Rural	1,583 (12.4)	157 (4.0)	1,426 (16.1)
Mixed	1,591 (12.5)	231 (5.9)	1,360 (15.4)
Urban	9,601 (75.2)	3,545 (90.1)	6,056 (68.5)
Total energy	1,711.7 (693.5)	1,674.3 (765.9)	1,727.7 (659.5)
Waist circumference	95.0 (14.9)	97.5 (14.7)	93.9 (14.9)
Comorbid conditions
Ischemic heart conditions	2,042 (14.4)	473 (11.2)	1,569 (15.9)
Atrial fibrillation	1,026 (7.3)	269 (6.3)	757 (7.7)
Hypertension history	7,625 (53.9)	2,909 (68.5)	4,716 (47.6)
Dyslipidemia	8,160 (57.7)	2,225 (52.4)	5,935 (59.9)
Diabetes	2,217 (15.7)	1,020 (24.0)	1,197 (12.1)
BMI
Normal	3,569 (25.2)	661 (15.6)	2,908 (29.4)
Overweight	5,424 (38.4)	1,462 (34.5)	3,962 (40.0)
Obese	5,152 (36.4)	2,121 (50.0)	3,031 (30.6)
eGFR	86.0 (18.2)	91.0 (21.7)	83.9 (16.1)
Lifestyle factors
Exercise
None	4,308 (30.5)	1,448 (34.1)	2,860 (28.9)
1–3 times per week	5,493 (38.8)	1,698 (40.0)	3,795 (38.3)
≥4 times per week	4,344 (30.7)	1,098 (25.9)	3,246 (32.8)
Sedentary	1,584 (11.2)	713 (16.8)	871 (8.8)
Current smoking	1,744 (12.3)	643 (15.2)	1,101 (11.1)
Incident cognitive impairment
3 of 4 assessments	534 (4.7)	125 (3.8)	409 (5.2)
Six-Item Screener	1,171 (8.3)	425 (10.1)	746 (7.6)

Abbreviations: BMI = body mass index; eGFR = estimated glomerular filtration rate; REGARDS = REasons for Geographic and Racial Differences in Stroke.

The lowest tertile of the MIND diet (N = 4,456) had a mean MIND diet score of 5.16 (SD = 0.81) and median (interquartile range [IQR]) of 5.5 (4.5–6.0), the middle tertile (N = 5,602) had a mean MIND diet score of 7.24 (SD = 0.55) and median (IQR) of 7.0 (7.0–7.5), and the highest tertile of MIND diet accordance (N = 4,087) had a mean score of 9.45 (SD = 0.90) and a median (IQR) of 9.5 (8.5–10.0). Compared with the lowest tertile of the MIND diet, participants in the highest tertile were more likely to be older, female, White participants, earn ≥ $75,000 per year, and be a college graduate (Table 2). Participants with the highest tertile of MIND diet accordance were less likely to have ischemic heart conditions, history of hypertension, dyslipidemia, diabetes mellitus, and a BMI ≥30 kg/m² compared with the lowest tertile. Regarding lifestyle factors, the participants in the highest tertile of MIND diet accordance were less likely to be sedentary and a current smoker.

Table 2.

Examining REGARDS Characteristics by Mind Diet Score Tertile

	Mind diet score			p Value
	T1 (N = 4,456)	T2 (N = 5,602)	T3 (N = 4,087)
Age <65 y	2,504 (56.2)	3,020 (53.9)	2,157 (52.8)	0.005
Male	2,221 (49.8)	2,437 (43.5)	1,473 (36.0)	<0.001
Black participants	1,571 (35.3)	1,676 (29.9)	997 (24.4)	<0.001
Income over $75,000	652 (14.6)	1,147 (20.5)	1,051 (25.7)	<0.001
College graduate	1,277 (28.7)	2,315 (41.3)	2,169 (53.1)	<0.001
Residence in the southeast	2,697 (60.5)	3,155 (56.3)	2,130 (52.1)	<0.001
Total energy	1,728.2 (715.4)	1,691.5 (712.0)	1,721.2 (641.1)	0.02
Waist circumference	97.9 (14.8)	95.3 (15.0)	91.5 (14.1)	<0.001
Comorbid conditions
Ischemic heart conditions	696 (15.6)	774 (13.8)	572 (14.0)	0.02
Atrial fibrillation	329 (7.4)	413 (7.4)	284 (7.0)	0.67
Hypertension history	2,567 (57.6)	3,026 (54.0)	2,032 (49.7)	<0.001
Dyslipidemia	2,688 (60.3)	3,234 (57.7)	2,238 (54.8)	<0.001
Diabetes	803 (18.0)	904 (16.1)	510 (12.5)	<0.001
Obese (BMI ≥30 kg/m2)	1,878 (42.2)	2,065 (36.9)	1,209 (29.6)	<0.001
eGFR	85.7 (19.8)	85.8 (18.2)	86.7 (16.5)	0.01
Lifestyle factors
Sedentary	755 (16.9)	580 (10.4)	249 (6.1)	<0.001
Current smoking	821 (18.4)	655 (11.7)	268 (6.6)	<0.001

Abbreviations: BMI = body mass index; eGFR = estimated glomerular filtration rate; REGARDS = REasons for Geographic and Racial Differences in Stroke.

The lowest tertile of MIND diet score had 532 occurrences of incident cognitive impairment, the middle tertile had 617 occurrences of incident cognitive impairment, and the highest tertile had 402 occurrences of incident cognitive impairment. After adjusting for all covariates, greater MIND diet adherence was associated with a decreased risk for incident cognitive impairment (full adjusted odds ratio [OR] 0.96, 95% CI 0.93–0.99, p = 0.02) as in Table 3. Interaction testing was used to understand if greater adherence to the MIND diet had a differential association with incident cognitive impairment by race, sex, and different age groups. For the continuous MIND diet score, a significant interaction was found between sex and MIND diet adherence (p = 0.02), but not for race (p = 0.87) or age (p = 0.45). There were no significant interactions found using the MIND diet tertiles. As presented in Table 4 using logistic regression, greater MIND diet adherence was associated with decreased risk of incident cognitive impairment in female participants (OR 0.92, 95% CI 0.89–0.96, p < 0.001) but not in male participants (OR 1.00, 95% CI 0.96–1.04, p = 0.91) in the unadjusted model. This relationship persisted in female participants in all models (fully adjusted OR 0.92, 95% CI 0.88–0.96, p < 0.001).

Table 3.

Odds Ratio for MIND Diet Score Adherence and Incident Cognitive Impairment

	MIND diet score tertiles					MIND diet score continuous
	T1	T2	p Value	T3	p Value		p Value
Ncog	532	617	—	402	—	1,551	—
Crude	Ref.	0.91 (0.81–1.03)	0.75	0.81 (0.70–0.92)	0.005	0.95 (0.93–0.98)	0.001
Model 1	Ref.	0.92 (0.81–1.05)	0.89	0.84 (0.72–0.97)	0.03	0.96 (0.93–0.99)	0.008
Model 2	Ref.	0.93 (0.82–1.06)	0.92	0.85 (0.74–0.99)	0.06	0.96 (0.93–0.99)	0.02
Model 3	Ref.	0.93 (0.82–1.06)	0.91	0.85 (0.74–0.99)	0.06	0.96 (0.93–0.99)	0.02

Abbreviations: eGFR = estimated glomerular filtration rate; MIND = Mediterranean-DASH Intervention for Neurodegenerative Delay.

Odds ratio and 95% CIs

Model 1 adjusts for age group, sex, race, region, education, income, and total energy. Model 2 adjusts for age group, sex, race, region, education, income, total energy, hypertension history, dyslipidemia, diabetes, eGFR, ischemic heart conditions, and atrial fibrillation. Model 3 adjusts for age group, sex, race, region, education, income, total energy, hypertension history, dyslipidemia, diabetes, eGFR, ischemic heart conditions, atrial fibrillation, sedentary, obesity, and smoking.

Table 4.

Odds Ratio for MIND Diet Score Adherence and Incident Cognitive Impairment Stratified by Sex

Male participants	MIND diet score continuous		Female participants	MIND diet score continuous
		p Value			p Value
Ncog	746	—	Ncog	805	—
Crude	1.00 (0.96–1.04)	0.91	Crude	0.92 (0.89–0.96)	<0.001
Model 1	1.00 (0.96–1.05)	0.93	Model 1	0.92 (0.88–0.96)	<0.001
Model 2	1.01 (0.96–1.05)	0.83	Model 2	0.93 (0.89–0.97)	<0.001
Model 3	1.01 (0.97–1.06)	0.64	Model 3	0.92 (0.88–0.96)	<0.001

Abbreviations: eGFR = estimated glomerular filtration rate; MIND = Mediterranean-DASH Intervention for Neurodegenerative Delay.

Odds ratio and 95% CIs

Model 1 adjusts for age group, race, region, education, income, and total energy. Model 2 adjusts for age group, race, region, education, income, total energy, hypertension history, dyslipidemia, diabetes, eGFR, ischemic heart conditions, and atrial fibrillation. Model 3 adjusts for age group, race, region, education, income, total energy, hypertension history, dyslipidemia, diabetes, eGFR, ischemic heart conditions, atrial fibrillation, sedentary, obesity, and smoking.

Figure 2 demonstrates the relationship between MIND diet adherence and incident cognitive impairment compared with other diets using tertiles of diet scores. There were strong Pearson correlations between other dietary patterns and the MIND diet: DASH = 0.6472, Mediterranean = 0.5569, DII = −0.5483, DIS = −0.6071, Health Eating Index = 0.4867, Southern-style = −0.2948, and Life's Essential 8 = 0.6242 all with a p < 0.001. Greater adherence to the Mediterranean diet was associated with a reduced risk of incident cognitive impairment after adjusting for all covariates (fully adjusted OR 0.97, 95% CI 0.93–0.99, p = 0.04) (eTable 2). Conversely, greater adherence to a Southern-style diet was associated with an increased risk of incident cognitive impairment (fully adjusted OR 1.11, 95% CI 1.04–1.19, p = 0.003). The DASH, DII, DIS, Healthy Eating Index, and Life's Essential 8 scores were not associated with incident cognitive impairment, as presented in eTable 2.

Figure 2. Diet Type and Diet Adherence and Odds Ratio of Incident Cognitive Impairment by Tertiles.

DASH = Dietary Approaches to Stop Hypertension; DII = Dietary Inflammatory Index; DIS = dietary inflammation score; LE8 = Life's Essential 8; MIND = Mediterranean-DASH Intervention for Neurodegenerative Delay.

Interaction testing in the other diets uncovered 2 statistically significant relationships: an interaction between Southern diet tertiles and age group (p = 0.006), and an interaction between DII diet score tertiles and race (p = 0.009). In participants younger than 65 years, greater accordance to a Southern diet was associated with increased risk of incident cognitive impairment after adjusting for all covariates (fully adjusted OR 1.84, 95% CI 1.39–2.44, p < 0.001). However, in participants 65 years and older, the Southern diet was not associated with incident cognitive impairment (fully adjusted OR 0.96, 95% CI 0.80–1.15, p = 0.72), as presented in eTable 3. For the DII diet score, greater DII scores were associated with an increased risk of incident cognitive impairment in Black participants (fully adjusted OR 1.33, 95% CI 0.98–1.82, p = 0.005) but not White participants (fully adjusted OR 1.00, 95% CI 0.82–1.21, p = 0.68), as presented in eTable 4.

Next, to determine whether the greater accordance to the MIND diet was associated with cognitive trajectory, linear mixed models were used. Two hundred forty-five participants did not have a follow-up time and therefore did not contribute to cognitive trajectory analysis (N = 13,900 participants had follow-up times). Overall (N = 13,900), the mean follow-up time was 10.92 (SD = 4.70) years; no difference in follow-up time was observed in Black participants (N = 4,139, the mean follow-up time was 10.72 years, SD = 4.81 years) compared with White participants (N = 9,761, mean follow-up time of 11.00 years, SD = 4.65 years) with p = 0.36. In all models, greater MIND diet accordance was associated with decreased risk of cognitive decline (Table 5 and eFigure 1). Interaction analysis in the fully adjusted model 4 demonstrated a significant interaction between race, MIND diet accordance, and time (p = 0.005). MIND diet adherence was a better predictor of cognitive decline in Black participants (β = 0.04, SE = 0.007, p < 0.001) than in White participants (β = 0.03, SE = 0.004, p < 0.001) (eFigure 2). In addition, there was a significant interaction between sex and MIND diet accordance (p = 0.02), with MIND diet adherence a better predictor of cognitive decline in female participants (β = 0.04, SE = 0.005, p < 0.001) than in male participants (β = 0.02, SE = 0.005, p = 0.004) (eFigure 3).

Table 5.

Beta Estimates for MIND Diet Score and Cognitive Trajectory

	MIND diet score tertiles					MIND diet score continuous
	T1	T2	p Value	T3	p Value		p Value
N	4,344	5,527	4,029	—	13,900	—	—
Crude	Ref.	0.169 (0.014)	<0.001	0.302 (0.014)	<0.001	0.071 (0.003)	<0.001
Model 1	Ref.	0.176 (0.016)	<0.001	0.298 (0.017)	<0.001	0.070 (0.004)	<0.001
Model 2	Ref.	0.101 (0.015)	<0.001	0.153 (0.017)	<0.001	0.039 (0.003)	<0.001
Model 3	Ref.	0.098 (0.015)	<0.001	0.147 (0.017)	<0.001	0.038 (0.003)	<0.001
Model 4	Ref.	0.089 (0.015)	<0.001	0.129 (0.017)	<0.001	0.034 (0.003)	<0.001

Abbreviations: eGFR = estimated glomerular filtration rate; MIND = Mediterranean-DASH Intervention for Neurodegenerative Delay.

Beta estimates and SEs

Crude is the unadjusted model. Model 1 adjusts for time, time-squared, interaction of time and diet, interaction of time and age, and indicators for first assessments. Model 2 adjusts for time, time-squared, interaction of time and diet, interaction of time and age, indicators for first assessments, age, sex, race, region, education, income, and total energy. Model 3 adjusts for time, time-squared, interaction of time and diet, interaction of time and age, indicators for first assessments, age, sex, race, region, education, income, total energy, hypertension history, dyslipidemia, diabetes, eGFR, ischemic heart conditions, and atrial fibrillation. Model 4 adjusts for time, time-squared, interaction of time and diet, interaction of time and age, indicators for first assessments, age, sex, race, region, education, income, total energy, hypertension history, dyslipidemia, diabetes, eGFR, ischemic heart conditions, atrial fibrillation, sedentary, obesity, and smoking.

Discussion

In this prospective cohort study of middle-aged and older US adults who were cognitively intact at baseline, greater accordance to the MIND diet was associated with reduced risk of incident cognitive impairment. This relationship was modified by sex differences; greater MIND diet accordance was associated with decreased risk of incident cognitive impairment in female participants but not in male participants. These associations persisted after adjusting for multiple demographic variables, clinical variables, and health behaviors. Similarly, greater MIND diet accordance was associated with slower rates of cognitive decline after adjusting for multiple demographic variables, clinical variables, and health behaviors. Although greater MIND diet accordance was associated with slower cognitive decline in both male and female participants, the association was greater in female participants. Similarly, although greater MIND diet accordance was associated with slower rate of cognitive decline in Black participants and White participants, the association was greater in Black participants.

The finding of greater MIND diet accordance associated with reduced risk of incident cognitive impairment aligns with most previous studies.^37,38 As with previous studies, our cohort did not have cognitive impairment at baseline^39,40 but had a smaller magnitude of risk reduction compared with comparably sized study by Hosking et al. (OR 0.47 vs 0.80 in our study). This may be related to demographic differences between the 2 studies. While income was included as a covariate in this study and the highest tertile of MIND diet accordance was associated with greater chance of higher income, unlike a previous study we did not find a significant interaction between income, MIND diet accordance, and incident cognitive impairment.⁴¹ Previous studies either were not powered to find whether biologic sex interacts with MIND diet accordance and incident cognitive impairment,³⁹ or did not specifically look for an interaction.^39-41 This contrasts with our study which found that greater MIND diet accordance was associated with reduced incident cognitive impairment in female participants but not in male participants. This may be due to differential effects of diet on cognitive reserve in women and men.

There are conflicting data about whether greater MIND diet accordance is associated with reduced cognitive decline. Some studies suggest greater MIND diet accordance is associated with reduced risk of cognitive decline,^10,42 while other studies suggest that greater MIND diet accordance is not associated with changes in risk of cognitive decline,^11,43,44 our study found that greater accordance to the MIND diet was associated with decreased risk of cognitive decline. Cognitive decline and MIND diet accordance were not associated with women older than 70 years in the Nurses' Health Study,⁴⁴ although in our study, the association between MIND diet accordance and cognitive trajectory was greater in female than male participants. Our study was significantly larger than previous studies which did not detect an association between MIND diet accordance and cognitive trajectory, which suggests that previous studies may have been too small to detect the difference. In addition, this study was adequately powered to examine MIND diet accordance and cognitive trajectories in a biracial cohort revealing that greater MIND diet accordance was associated with a slower rate of cognitive decline in Black participants compared with White participants.

Recently, a randomized clinical trial comparing the cognitive effects of the MIND diet with mild caloric restriction as compared with a control diet with mild caloric restriction in older adults without cognitive impairment but with a family history of dementia.¹⁴ A total of 604 participants were enrolled, and after 3 years of follow-up, improvements in global cognition scores were observed in both groups, with increases of 0.205 standardized units in the MIND-diet group and 0.170 standardized units in the control-diet group.¹⁴ Because the trial focused on dietary intervention, participants were required to have a suboptimal diet before enrollment and be overweight. The 3-year follow-up, while admirable for a randomized clinical trial provides less insights about the long-term benefits of diet compared with the 10+ years of follow-up in our study. In addition, the participants were generally well educated, and highly motivated as evidenced by the improvement in cognitive test scores across both groups over 3 years, which differs from the trend seen in the general population. Finally, only 66 were Black participants. These findings limit the generalizability of this randomized control study compared with our cohort study, although both offer important information.

In addition to our findings related to the MIND diet, we also observed that greater accordance to a Southern-style diet was associated with an increased risk of incident cognitive impairment in participants younger than 65 years but not older than 65 years after adjusting for all covariates. A Southern-style diet was derived from factor analysis within the REGARDS cohort and is characterized by high consumption of added fats, fried food, eggs and egg dishes, organ meats, processed meats, and sugar-sweetened beverages.²⁰ A previous study using REGARDS data found that a Southern-style diet was associated with lower scores on neuropsychologic testing, but did not stratify by race.¹⁷ The Atherosclerosis Risk in Communities (ARIC) study that found no association in 13,588 participants between a diet characterized by more meats and fried foods, similar to a Southern-style diet, in mid-life and late-life neuropsychologic assessments.⁴⁵ Differences between these findings and this study may be due to differences in dietary classification rather than age because the mean age in the ARIC study was 54 years old compared with 64 years old in our study.

We also found that greater DII scores were associated with increased risk of incident cognitive impairment in Black participants but not White participants. The DII is a summation of previously reported association of selected dietary factors with various inflammation biomarkers, and greater scores indicate a diet more associated with inflammatory markers.^25,46 Previously, greater scores on the DII have been associated with increased risk of cognitive impairment.^47,48 No previous cohort had an adequate number of Black American participants to evaluate whether DII scores are associated with risk for incident cognitive impairment, making our findings novel. Our findings of no association between DII and cognitive impairment is contrary to previous studies.^47,48 Differences between our study and previous literature may be due to differences in cognitive outcome measures rather than sample size because the 9,901 White participants in our study represent the largest sample with DII and cognitive outcomes data.^47,48

The Mediterranean diet is a diet rich in vegetables, fruits, legumes, cereals, monounsaturated fats, and fish which minimizes intake of meat and dairy productions. The DASH diet is mainly based on fruits, vegetables, low-fat or fat free dairy, whole grains, fish, poultry, legumes, and nuts. It recommends reducing sodium intake, sweets (in drinks and foods), and red meat.^1,7-9 Although both diets emphasize the consumption of whole grains, fruits, and vegetables, the Mediterranean diet differs from the DASH diet in fish, lean meat, and sweets consumption.^1,7-9 These 2 dietary approaches were combined to form the MIND diet.^1,7-9 The DII was created based on selected dietary factors associated with inflammation, which were primarily nutrient-based.²⁵ The Southern-style dietary pattern is characterized by fried foods, organ meats, processed meats, eggs and egg dishes, added fats, high-fat dairy foods, sugar-sweetened beverages, and bread.²⁰ It is important to acknowledge the social, cultural, economic factors that influence dietary patterns. For example, fresh fruits and vegetables are often more expensive than highly processed, shelf-stable foods. Whereas fast foods and fried foods tend to be inexpensive and readily available. Thus, individuals living in poverty may be eating Southern-style foods more often. This is relevant because in our data, the White participants earned more than Black participants.

The biologic plausibility of the MIND diet (and other healthy eating patterns) and neuroprotection, and by extension decreased cognitive impairment/cognitive decline, is supported by neuroimaging studies which demonstrated an association between healthy dietary patterns (such as the MIND diet) and grey matter volume, total brain volume, cortical thickness, white matter volume, and white matter integrity over time.⁴⁹

However, this study has certain limitations. First, because dietary intake information was based on FFQ data, which could lead to measurement error and misclassification of dietary intake.⁵⁰ In addition, the FFQ was not returned by all individuals, which could result in selection bias. Overall, those who did not return the FFQ were more likely to be Black participants, have lower income, and not have graduated from high school.⁵⁰ In addition, although the REGARDS cohort is more diverse than many cohorts, it is limited to Black and White participants of older age; thus, results may not be generalizable to more diverse and younger populations.⁵⁰ The analysis was stratified by binary sex, and sex was not the focus of the analysis. Furthermore, most REGARDS participants have 3 or more preexisting vascular risk factors, and their dietary behavior might differ from the general population. Another limitation is the variable number of cognitive assessments which were completed by each participant. Reverse causality, in which cognitive impairment and cognitive decline affect dietary habits, is also potential confounder; however, diet was quantified before assessments of cognition, and several confounding variables were controlled for.

In conclusion, the MIND diet has a differential effect on incident cognitive impairment and cognitive decline in different sexes and racial groups. This suggests that accordance to the MIND diet may affect cognitive reserve differently across races and warrants further research. Similarly, Southern diet and proinflammatory diets may have a differential effect on the development of cognitive impairment based on age and/or race. This finding needs verification in additional cohorts.

Glossary

AF

Animal Fluency

ARIC

Atherosclerosis Risk in Communities

BMI

body mass index

CATI

computer-assisted telephone interview

DASH

Dietary Approaches to Stop Hypertension

DII

Dietary Inflammatory Index

DIS

dietary inflammation score

FFQ

Food Frequency Questionnaire

IQR

interquartile range

LF

Letter “F” Fluency

MIND

Mediterranean-DASH Intervention for Neurodegenerative Delay

OR

odds ratio

REGARDS

REasons for Geographic and Racial Differences in Stroke

SIS

Six-Item Screener

WLD

World Delayed Recall

WLL

Word List Learning

Appendix. Authors

Name	Location	Contribution
Russell P. Sawyer, MD	Department of Neurology and Rehabilitation Medicine, University of Cincinnati College of Medicine, OH	Drafting/revision of the manuscript for content, including medical writing for content; study concept or design; analysis or interpretation of data
Jessica Blair, MS	Biostatistics Department, School of Public Health, University of Alabama at Birmingham	Major role in the acquisition of data; analysis or interpretation of data
Rhonna Shatz, DO	Department of Neurology and Rehabilitation Medicine, University of Cincinnati College of Medicine, OH	Study concept or design
Jennifer J. Manly, PhD	Taub Institute for Research on Alzheimer's Disease and the Aging Brain, G.H. Sergievsky Center, and Department of Neurology, Vagelos College of Physicians and Surgeons, Columbia University, New York Presbyterian Hospital, New York	Drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; analysis or interpretation of data
Suzanne E. Judd, PhD, MPH	Biostatistics Department, School of Public Health, University of Alabama at Birmingham	Drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design; analysis or interpretation of data

Study Funding

This research project is supported by cooperative agreement U01 NS041588, co-funded by the National Institute of Neurological Disorders and Stroke (NINDS) and the National Institute on Aging (NIA), NIH, Department of Health and Human Service. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NINDS or the NIA. Representatives of the NINDS were involved in the review of the manuscript but were not directly involved in the collection, management, analysis or interpretation of the data.

Disclosure

The authors report no disclosures. Go to Neurology.org/N for full disclosures.