Development of the cMIND Diet and Its Association with Cognitive Impairment in Older Chinese People

Abstract

Objectives

Cognitive impairment commonly occurs among older people worldwide. Although the Mediterranean-DASH Intervention for Neurodegenerative Delay (MIND) diet was associated with better cognitive function and lower risk of cognitive impairment, it could not be applied to older Chinese due to the traditional dietary characteristics in China. We aimed to develop the Chinese version of the MIND (cMIND) diet and verify its association with cognitive impairment among older Chinese individuals.

Design

A cross-sectional study.

Setting and Participants

We included a total of 11,245 participants from the Chinese Longitudinal Healthy Longevity Study (CLHLS) follow-up survey in 2018. The mean age of the participants at study baseline was 84.06 (±11.46) years.

Measurements

We established the cMIND diet based on current evidence in the diet-cognition field, combined with Chinese dietary characteristics. The verification of its association with cognitive impairment was conducted using the data from the CLHLS follow-up survey. Adherence to the cMIND diet was assessed by the cMIND diet score, which was calculated from a food frequency questionnaire. Cognitive impairment was identified by the Mini-Mental State Examination. Instrumental activities of daily living (IADL) disability was defined according to the self-reported performance of eight activities.

Results

The cMIND diet comprised 11 brain-healthy food groups and 1 unhealthy food group. The median cMIND diet score of all participants was 4.5 (from a total of 12 points) and the prevalence of cognitive impairment was 15.2%. Compared with the lowest tertile, the highest tertile score was associated with lower odds of cognitive impairment (odds ratio (OR)=0.60, 95% confidence interval (CI): 0.51–0.72) and IADL disability (OR=0.86, 95% CI: 0.75–0.98) in the full-adjusted model.

Conclusion

We developed the cMIND diet that was suitable for older Chinese individuals, and our results suggested that higher adherence to the cMIND diet was associated with reduced odds of cognitive impairment and IADL disability. In view of the limitations of cross-sectional design in the study, further research is clearly warranted.

Introduction

The cognitive function of older people declines gradually with increasing age and eventually develops into mild cognitive impairment (MCI) and even dementia. MCI is recognized as the intermediate stage between normal cognition and dementia. The prevalence of MCI across the world varies broadly, since it was estimated to be 42% in France, 7.7% in Italy and 9.8–28.3% in United States (1, 2). The global population of dementia was 46.8 million in 2015 and estimated to increase to 74.7 million by 2030 (3). China is the most populous country in the world, and the population of Chinese individuals aged 60 years and older was 264 million in 2020. In addition, the prevalence of dementia and MCI in Chinese people aged 60 years and older was estimated to be 6.0% and 15.5%, respectively (4). Moreover, cognitive impairment is associated with disability in activities of daily living, which is another common problem in older people. The prevention and management of MCI and dementia should be of great importance because either of them brings a heavy burden to individuals or families all over the world.

Many factors could increase the risk of cognitive impairment, including unmodifiable factors (e.g., aging and female sex) and modifiable factors (e.g., obesity, smoking, malnutrition, hypertension, diabetes and heart disease), and nutrition is an essential factor that could be ameliorated (4, 5). Nutrition, including certain nutrients (e.g., folate, vitamin D and omega-3 fatty acids) and food groups (e.g., seafood, vegetables, fruits, nut and tea), had a positive effect on promoting healthy cognitive aging and decreasing the risk of cognitive impairment (6, 7). However, considering the complicated interactions between various nutrients and food groups, the whole dietary pattern may provide stronger neuroprotection and could be employed more easily. Older people with a higher risk of cognitive impairment ought to adhere to healthy dietary patterns to assist in healthy cognitive aging, so there is a need to develop a practical dietary pattern to delay the progression of cognitive decline for them. Aiming at protecting the brain and reducing cognitive decline, Morris M. C. and his colleagues developed the Mediterranean-DASH Intervention for Neurodegenerative Delay (MIND) diet in 2015, which was modified from the Mediterranean diet and Dietary Approach to Stop Hypertension (DASH) diet and was found to be more neuroprotective than other dietary patterns (8, 9, 10). Observational studies suggested that higher adherence to the MIND diet was associated with better cognitive function, a reduced rate of cognitive decline and a lower risk of cognitive impairment in older people (11, 12, 13, 14).

Previous studies investigating the association between the MIND diet and cognition were carried out in North America (8, 10, 11), Europe (12) and Australia (9, 15), rather than Asia, so the evidence from Eastern countries was restricted. It is imperative to assess the neuroprotective effect of MIND diet across various countries, especially in China. Generally, plant-based foods (e.g., grains, vegetables, soybeans and fruits) were the major components in Chinese dietary patterns, and Chinese adults consumed less meat, dairy products and wine than Western people. However, the MIND diet features Western food such as olive oil, wine and cheese, thus causing limited application in China. In view of the large population of cognitively impaired individuals in China as well as the lack of essential research, it is extremely urgent to establish a modified version of the MIND diet to cater to Chinese dietary characteristics.

The main purpose of our study was to establish an appropriate Chinese version of the MIND (cMIND) diet and verify its association with cognitive impairment among older Chinese individuals. In addition, to discover potential health benefits, we explored the possible relation between the cMIND diet and instrumental activities of daily living (IADL)(16, 17). We aimed to develop a healthy dietary pattern that is beneficial to the cognitive function and IADL of older Chinese individuals.

Methods

Establishment of cMIND diet

We developed the cMIND diet based on the primary MIND diet to better meet the dietary characteristics of older Chinese individuals. The cMIND diet was made up of 12 food items altogether and designed on the basis of current evidence (18, 19, 20) in the diet-cognition field.

The MIND diet recommended 10 brain-healthy food groups (green leafy vegetables, other vegetables, nuts, berries, soybeans, whole grains, not fried fish, not fried poultry, olive oil, and wine) while avoiding five unhealthy groups (red meat and products, butter/margarine, cheese, pastries and sweets, and fast fried foods). Refined cereals, such as polished rice and white wheat, were the dominant types of staple food for most Chinese people currently. The proportion of coarse cereals (such as the whole grains, tubers/roots and legumes) in all staple food was rather low and the consumption of them declined gradually in China over the last several decades (21). In the cMIND diet, we recommended the whole grains which are the important sources of dietary fiber and vitamin B6. But the source and the total amount of staple food were taken into consideration simultaneously. According to the Chinese Dietary Guidelines, we recommended 250–400 g intake of staple food every day while discouraging either excessive or insufficient intake of staple food. Consumption of the mushroom/algae was also included in the cMIND diet. The findings of observational and experimental studies indicated that higher mushroom/algae consumption contributed to better performance in cognitive tests and less risk of MCI or dementia (20, 22, 23). Olive oil was rarely used among the Chinese, so in terms of the cooking oil component, the maximum score was attributed to vegetable oil instead of animal oil. Vegetable oil contains more unsaturated fatty acids (monounsaturated and polyunsaturated fatty acids), which are commonly linked to greater cognition, and less saturated fatty acids, which cause worse memory and learning ability (19). Garlic extract is a potential treatment for Alzheimer's disease and has been widely investigated in animal experiments. We recommended the intake of garlic because the garlic extract reduced the amyloid-β protein concentrations and had a protective effect on oxidative damage and neuroinflammation (24, 25). A glass of wine every day was recommended in the MIND diet, but it was substituted with tea in the cMIND diet. Tea is the second most consumed beverage globally and prospective cohort studies found that a higher intake of tea was associated with a lower risk of cognitive decline (18, 26). We emphasized the effect of green tea since the neuroprotective effect of tea was only observed in green tea consumers in a few studies (26, 27). However, the intake of butter/margarine and cheese was not considered in the cMIND diet since they were seldom consumed in older Chinese individuals. Red meat was also removed because it was not especially pointed out in the Chinese Dietary Guidelines. The other components in the cMIND diet were similar to those in the MIND diet, the consumption of fresh vegetables, fruit, fish, soybeans and nuts was encouraged while the sugar/sweets were discouraged.

For all the food items, nine of them (fresh vegetables, mushroom/algae, fresh fruit, fish, soybeans, nuts, garlic, tea and sugar/sweets) were scored based on consumption frequency as 0, 0.5 or 1, while the other three items (type of staple food, amount of staple food and cooking oil) were marked either 0 or 1 (Table 1). The total score was calculated by adding up all individual component scores so the maximum score was 12. A higher score indicated greater adherence to the cMIND diet.

Table 1.

Comparison of the dietary components and scores between the original MIND diet and the cMIND diet

MIND diet				cMIND diet
Components	Score			Components	Score
	0	0.5	1		0	0.5	1
Whole grains	<1 serving/day	1–2/day	≥3 servings/day	Type of staple food	Rice/wheat		Whole grains
				Amount of staple food	<250g or >400g		250g-400g
Green leafy vegetables	≤2 servings/week	>2 to <6/week	6 servings/week	Fresh vegetables	≤2 servings/week	3–5/week	≥6 servings/week
Other vegetables	<5 serving/week	5 to <7 week	≥1 serving/day	Mushroom or algae	≤1 meal/week	1–3/week	≥4 meals/week
Berries	<1 serving/week	1/week	≥2 servings/week	Fresh fruit	≤2 servings/week	3–5/week	≥6 servings/week
Olive oil	Not primary oil		Primaiy oil used	Cooking oil	Animal oil		vegetable oil
Fish (not fried)	Rarely	1–3/month	≥1 meals/week	Fish	<1/month	1–3/month	≥1 meal/week
Beans	<1 meal/week	1–3/week	>3 meals/week	Soybeans	<1 meal/week	1–3/week	≥4 meals/week
Nuts	<1/month	1/month to <5/week	>5 servings/week	Nuts	<1 serving/week	1–4/week	≥5 servings/week
				Garlic	<1 meal/week	1–3/week	≥4 meals/week
				Tea	Not almost every day	Other types of tea (almost every day)	Green tea (almost every day)
Poultry (not fried)	<1 meal/week	1/week	≥2 meals/week
Wine	>1 glass/day or never	1/month-6/week	1 glass/day
Pastries and sweets	≥7 servings/week	5–6/week	<5 servings/week	Sugar or sweets	≥2 servings/week	1/month–1/week	<1 serving/month
Butter, margarine	>2 servings/day	1–2/day	<1 serving/day
Cheese	≥7 servings/week	1–6/week	<1 serving/week
Red meat and products	≥7 meals/week	4–6/week	<4 meals/week
Fast fried foods	≥4 times/week	1–3/week	<1 time/week

Abbreviation: MIND, Mediterranean-DASH Intervention for Neurodegenerative Delay; cMIND, Chinese version of the MIND.

Verification of cMIND diet

Participants and dietary assessment

We used data from the Chinese Longitudinal Healthy Longevity Study (CLHLS) follow-up survey in 2018 to verify the association between the cMIND diet and health outcomes. The CLHLS is an ongoing, longitudinal study focusing on the determinants of healthy longevity among older Chinese individuals. It is an international collaborative project between Peking University, the Chinese Academy of Social Science and Duke University. The CLHLS collected demographic and psychological characteristics, cognition, dietary nutrition, social contact, activities of daily living and other health-related information from the participants or their families. The baseline survey was undertaken in 1998 covering 22 provinces across China, with follow-up surveys every 2 or 3 years (28). The Biomedical Ethics Committee of Peking University approved the CLHLS (IRB00001052-13074). All the participants or their proxy respondents signed written informed consent.

A total of 11,245 participants with complete information were analyzed in this study after excluding those with stroke or dementia (4,629 participants were excluded). Participants with stroke were not included since the occurrence of stroke could affect cognitive function significantly and complicatedly. All information was collected by trained interviews using a well-designed questionnaire through face-to-face interviews in 2018. Interviewees were encouraged to answer the questions by themselves as far as possible. If they could not answer some questions, they may request a close family member or another knowledgeable proxy to answer instead. Researchers have conducted extensive assessments, which verified the high quality of the data in the CLHLS (28).

Dietary information was extracted from the questions in a simplified qualitative food frequency questionnaire (FFQ). These questions were designed based on international standards and Chinese dietary habits, and carefully tested by pilot studies/interviews, which supported the validity and reliability of measurements (28). Previous studies also recognized the use of the simplified FFQ in CLHLS (29, 30, 31). The self-administered FFQ investigated the usual dietary intake of a list of food groups in the past one year, which included meat, fish, eggs, bean products, milk products, sugar, tea and so on. The options representing consumption frequency for each food group were set as “almost every day”, “not every day, but at least once per week”, “not every week, but at least once per month”, “not every month, but occasionally” and “rarely or never”. The options regarding fresh fruits and vegetables were a little different, which were set as “almost every day”, “quite often”, “occasionally” and “rarely or never”. The type and amount of staple food were investigated by additional questions, so we could know the amount of staple food intake. We computed the cMIND diet score based on the FFQ data with slight adaptations. For example, we assigned 1 point for the “almost every day” or “not every day, but at least once per week” answer to the question about fish consumption, 0.5 point for “not every week, but at least once per month” and 0 point for the other answers.

Cognitive impairment and IADL disability

Cognitive function was evaluated by the Chinese version of the Mini-Mental State Examination (MMSE), which consists of 24 questions and covers five dimensions: orientation, registration, attention and calculation, recall, and language (28). The Chinese version of the MMSE was translated from the international MMSE questionnaire with adaptations to Chinese culture, whose validity and reliability have been verified in several studies (32, 33). Except for the question 'Please name as many kinds of food as possible in 1 minute”, each question was scored as “1” point for the correct answer and “0” point otherwise. With regard to the “naming kinds of food” question, the participants would get 1 point every time they named one kind of food successfully, with the upper limit of 7 points. The total score of the MMSE was 30, and we defined the participants as having “cognitive impairment” with a score below 18 for illiterate participants, 21 for those with 1–6 years of education, and 25 for those with education over 6 years (34).

IADL was assessed according to eight items: visiting neighbors, going shopping, making meals, doing laundry, walking one kilometer, lifting a five-kilogram object, crouching and standing up three times, and taking public transportation (28). The scale was adapted from the Lawton IADL Scale and was demonstrated to be a reliable and valid instrument to assess IADL disability among older Chinese (35). We treated an interviewee as having an “IADL disability” if he/she needed any help in any item. Besides, each item was scored ranging from 1 (completely dependently) to 3 (completely independently), and we generated the total IADL score by adding up scores of these eight items. The higher total score indicated the better performance in IADL.

Covariates

We adjusted the sociodemographic, health conditions and lifestyle covariates (5). The sociodemographic covariates were sex, age, residence and education. Age was categorized into three groups: 50–65 years, 66–80 years and >80 years. Residence was categorized into “urban” and “rural”. Education was categorized into two groups: illiterate (no formal schooling) and literate. The health conditions and lifestyle covariates included body mass index (BMI), diabetes, hearing impairment, hypertension, depression, smoking, drinking, exercise and social engagement. BMI was calculated by measured height and weight and then categorized into four groups: underweight (< 18.5), normal (18.5–23.9), overweight (24.0–27.9) and obese (≥28) (36). We identified diabetes and hearing impairment by self-report, as several research did previously (33, 37). Hypertension was classified into three types based on self-reported hypertension, measured systolic blood pressure (SBP) and diastolic blood pressure (DBP): non-hypertension (SBP<140 mmHg, DBP<90 mmHg, without self-reported hypertension), aware hypertension (with self-reported hypertension, regardless of the measured blood pressure), and unaware hypertension (SBP≥140 mmHg or DBP≥90 mmHg, without self-reported hypertension). Depression was screened by a short form of the Center for Epidemiologic Studies Depression Scale (38). Smoking history was categorized into “no” (never smoke) or “yes” (previously or currently smoke), as was drinking or exercise status. We evaluated social engagement by the following activities: housework, outdoor activities, garden work, reading newspapers/books, raising domestic animals, playing cards and/or mah-jong, watching TV and/or listening to the radio, and social activities (organized) (28). There were five frequency alternatives for each activity, and a score ranging from 1 to 5 was given: 1 for the lowest frequency (never) and 5 for the highest (almost every day). A total score of all items was computed and then divided into tertiles, reflecting three levels of social engagement.

Statistical analysis

Sociodemographic, health conditions and lifestyle characteristics of different diet score (categorized in tertiles) groups were compared by the Mantel-Haenszel chi-square test or Kruskal-Wallis H test. Logistics regression was used to explore the association between the cMIND diet score (modelled in tertiles and as a continuous variable) and cognitive impairment. The basic model (model 1) adjusted for sex, age, residence, education and BMI. Model 2 adjusted for the above-mentioned covariates and lifestyle factors: smoking, drinking, exercise and social engagement. Model 3 adjusted for all covariates in model 2 and health conditions: diabetes, hearing impairment, hypertension and depression. In addition, we performed the Mann-Whitney U test to compare the difference in each diet component score between the abnormal and normal cognition groups. The statistical methods we used to examine the association between the cMIND diet and IADL disability were the same as those in the analysis for cognitive impairment. Furthermore, we used general linear regression models to investigate the association between the cMIND diet score (modelled in tertiles) and the MMSE score, as well as IADL score. The adjusted variables in the linear regression models basically corresponded with those in logistics regression models.

We conducted several sensitivity analyses. In the first sensitivity analysis, we excluded participants with severe hypertension (SBP≥180 mmHg or DBP≥110 mmHg). In the second sensitivity analysis, the participants with the lowest 5.5% MMSE score were removed.

All analyses were conducted using IBM SPSS Statistics v25. Two-tailed p values less than 0.05 were considered statistically significant.

Results

Participants characteristics

Table 2 shows the baseline characteristics of the 11,245 participants. Of these participants with an average age of 84.06 (±11.46) years, 54.7% were female. The median diet score of all participants was 4.5; 9% of them scored over 6.5, while 15% of them scored below 2.5. All the baseline characteristics were associated with the tertiles of the cMIND diet score. In the third tertile of the cMIND diet score, most participants were male, more educated, lived in urban areas and kept regular exercise. On the contrary, the majority of participants in the first tertile were female, less educated, lived in rural areas and lacked regular exercise. The proportion of diabetes patients increased from the first to the last tertile, while the proportion of depressive and hearing-impaired individuals declined. The prevalence of cognitive impairment and IADL disability were 15.2% and 61.6%, respectively.

Table 2.

Baseline characteristics of analyzed participants according to tertiles of diet score (N=11,245)

Characteristics	All(N=11,245)	Diet score (range 0–12)			P value&
		Tertile 1 (N=4,110)	Tertile 2 (N=3,966)	Tertile 3 (N=3,169)
	N (%)	N (%)	N (%)	N (%)
Diet score median (range)	4.5(0–12)	3.0(0–3.5)	4.5(4–5)	6(5.5–12)
Gender					< 0.001
Male	5090(45.3%)	1585(38.6%)	1837(46.3%)	1668(52.6%)
Female	6155(54.7%)	2525(61.4%)	2129(53.7%)	1501(47.4%)
Age					< 0.001
50–65years	315(2.8%)	74(1.8%)	109(2.7%)	132(4.2%)
66–80years	4320(38.4%)	1213(29.5%)	1551(39.1%)	1556(49.1%)
> 80years	6610(58.8%)	2823(68.7%)	2306(58.1%)	1481(46.7%)
Residence					< 0.001
Urban	6114(54.4%)	1951(47.5%)	2077(52.4%)	2086(65.8%)
Rural	5131(45.6%)	2159(52.5%)	1889(47.6%)	1083(34.2%)
Education					< 0.001
Illiterate	4471(39.8%)	2086(50.8%)	1582(39.9%)	803(25.3%)
Literate	6774(60.2%)	2024(49.2%)	2384(60.1%)	2366(74.7%)
Body mass index (kg/m2)					< 0.001
< 18.5	1753(15.6%)	825(20.1%)	618(15.6%)	310(9.8%)
18.5–24	5984(53.2%)	2321(56.5%)	2089(52.7%)	1574(49.7%)
24–28	2600(23.1%)	717(17.4%)	932(23.5%)	951(30.0%)
≥28	908(8.1%)	247(6.0%)	327(8.2%)	334(10.5%)
Hypertension					< 0.001
Non — hypertension	3846(34.2%)	1456(35.4%)	1379(34.8%)	1011(31.9%)
Aware hypertension	4339(38.6%)	1402(34.1%)	1484(37.4%)	1453(45.9%)
Unaware hypertension	3060(27.2%)	1252(30.5%)	1103(27.8%)	705(22.2%)
Diabetes	968(8.6%)	206(5.0%)	307(7.7%)	455(14.4%)	< 0.001
Depression	2155(19.2%)	1041(25.3%)	749(18.9%)	365(11.5%)	< 0.001
Hearing impairment	3920(34.9%)	1653(40.2%)	1360(34.3%)	907(28.6%)	< 0.001
Smoking					< 0.001
No	7817(69.5%)	3004(73.1%)	2742(69.1%)	2071(65.4%)
Yes	3428(30.5%)	1106(26.9%)	1224(30.9%)	1098(34.6%)
Drinking					< 0.001
No	8237(73.3%)	3133(76.2%)	2874(72.5%)	2230(70.4%)
Yes	3008(26.7%)	977(23.8%)	1092(27.5%)	939(29.6%)
Regular exercise					< 0.001
No	6858(61.0%)	3016(73.4%)	2454(61.9%)	1388(43.8%)
Yes	4387(39.0%)	1094(26.6%)	1512(38.1%)	1781(56.2%)
Social engagement					< 0.001
Lower	3754(33.4%)	1822(44.3%)	1281(32.3%)	651(20.5%)
Middle	3767(33.5%)	1382(33.6%)	1415(35.7%)	970(30.6%)
Higher	3724(33.1%)	906(22.0%)	1270(32.0%)	1548(48.8%)
IADL disability	6924(61.6%)	2957(71.9%)	2395(60.4%)	1572(49.6%)	< 0.001
Cognitive impairment	1711(15.2%)	877(21.3%)	569(14.3%)	265(8.4%)	< 0.001

Abbreviations: IADL, instrumental activities of daily living; & Mantel-Haenszel chi-square test (nominal categorical variable) and Kruskal-Wallis H test (ordinal categorical variable) were used to compare distributions across the tertiles of diet score.

cMIND diet and cognitive impairment

In all three models, a higher diet score was significantly related to lower odds of cognitive impairment (Table 3). Compared to the lowest tertile, the highest tertile score was associated with reduced odds of cognitive impairment in both the basic model (odds ratio (OR)=0.52, 95% confidence interval (CI): 0.44–0.61) and the full-adjusted model (OR=0.60, 95% CI: 0.51–0.72). Similarly, in models where the cMIND diet score was regarded as a continuous variable, a higher cMIND diet score was also associated with lower odds of cognitive impairment (basic model, OR=0.82, 95% CI: 0.79–0.85; full-adjusted model, OR=0.86, 95% CI: 0.82–0.89) (Supplementary Table 2). We further distinguished every single component score between the abnormal and normal cognition groups. As showed in Table 4, all items differed significantly except “Type of staple food” and “Cooking oil”.

Table 3.

Odds ratio (OR) and 95% confidence interval (CI) of estimated effects for tertiles of the diet score of cognitive impairment&

Variables	Model 1a	Model 2b	Model 3c
	OR (95% CI)	OR (95% CI)	OR (95% CI)
Diet score
Tertile 1	1(reference)	1(reference)	1(reference)
Tertile 2	0.75(0.66,0.85)*	0.82(0.72,0.93)'	0.81(0.71,0.92)'
Tertile 3	0.52(0.44,0.61) *	0.62(0.53,0.73) *	0.60(0.51,0.72) *
Gender(Female)	1.51(1.33,1.71) *	1.41(1.22,1.63) *	1.54(1.32,1.79) *
Age
50–65years	1(reference)	1(reference)	1(reference)
66–80years	1.31(0.57,3.00)	1.08(0.47,2.49)	0.97(0.42,2.23)
> 80years	11.88(5.27,26.77) *	5.70(2.51,12.93) *	3.74(1.63,8.52)'
Residence(Rural)	0.92(0.82,1.03)	0.89(0.79,1.00)#	0.91(0.80,1.02)
Education(Literate)	0.84(0.74,0.95)'	1.08(0.95,1.22)	1.19(1.04,1.36) #
Body mass index (kg/m2)
< 18.5	1.32(1.15,1.51) *	1.24(1.08,1.43)'	1.16(1.00,1.34) #
18.5–24	1(reference)	1(reference)	1(reference)
24–28	0.74(0.63,0.86) *	0.78(0.66,0.92)'	0.82(0.70,0.98) #
≥28	0.85(0.67,1.08)	0.90(0.70,1.16)	1.00(0.77,1.29)
Smoking		0.97(0.83,1.14)	0.97(0.82,1.14)
Drinking		0.99(0.85,1.16)	0.99(0.84,1.16)
Regular exercise		1.06(0.93,1.20)	1.06(0.92,1.21)
Social engagement
Lower		1(reference)	1(reference)
Middle		0.27(0.23,0.31) *	0.30(0.26,0.35) *
Higher		0.15(0.13,0.19) *	0.18(0.15,0.23) *
Hypertension
Non — hypertension			1(reference)
Aware hypertension			0.69(0.60,0.80) *
Unaware hypertension			0.89(0.77,1.03)
Diabetes			1.07(0.83,1.37)
Depression			1.00(0.86,1.15)
Hearing impairment			3.23(2.85,3.66) *
Sensitivity analysis 1d
Diet score
Tertile 1	1(reference)	1(reference)	1(reference)
Tertile 2	0.75(0.66,0.86) *	0.76(0.66,0.86) *	0.80(0.70,0.92)'
Tertile 3	0.51(0.43,0.60) *	0.52(0.44,0.61) *	0.60(0.50,0.72) *
Sensitivity analysis 2e
Diet score
Tertile 1	1(reference)	1(reference)	1(reference)
Tertile 2	0.79(0.68,0.91) *	0.80(0.69,0.93)'	0.84(0.72,0.98) #
Tertile 3	0.52(0.43,0.62) *	0.53(0.44,0.64) *	0.60(0.49,0.73) *

&Multivariable logistics regression was used to test the association between the diet score (modelled in tertiles) and cognitive impairment; *P<0.001; P<0.01; #P<0.05; a. Model 1: adjusted for gender, age, region, education and body mass index; b. Model 2: adjusted for gender, age, region, education, body mass index, smoking, drinking, exercise and social engagement; c. Model 3: adjusted for gender, age, region, education, body mass index, smoking, drinking, exercise, social engagement hypertension, diabetes, depression and hearing impairment; d. Sensitivity analysis 1: Remove those with severe hypertension (systolic blood pressure ≥180mmHg or diastolic blood pressure ≥110mmHg) from all the participants (N=10,390); e. Sensitivity analysis 2: Remove those with the lowest 5.5% on the MMSE score from all the participants (N=10,619).

Table 4.

Comparison of different diet components score between cognitive impairment group and normal cognition group

Components	Normal cognition (mean±standard deviation)	Cognitive impairment (mean±standard deviation)	P value&
Type of staple food	0.04±0.19	0.03±0.18	0.312
Amount of staple food	0.49±0.50	0.33±0.47	< 0.001
Fresh vegetables	0.68±0.46	0.55±0.50	< 0.001
Mushroom or algae	0.12±0.24	0.08±0.21	< 0.001
Fresh fruit	0.23±0.42	0.17±0.38	< 0.001
Cooking oil	0.89±0.32	0.89±0.32	0.524
Fish	0.60±0.44	0.49±0.45	< 0.001
Soybeans	0.33±0.34	0.30±0.34	< 0.001
Nuts	0.14±0.29	0.06±0.21	< 0.001
Garlic	0.35±0.40	0.27±0.38	< 0.001
Tea	0.12±0.30	0.06±0.21	< 0.001
Sugar or sweets	0.56±0.41	0.52±0.42	< 0.001

&Mann-Whitney U test was used to compare the difference in each diet component score between the abnormal and normal cognition groups.

In linear regression models, we observed the association between higher adherence to the cMIND diet and better cognitive function. Compared to the first tertile of the cMIND diet score, the second (full-adjusted model, β=0.60, 95% CI: 0.37–0.82) and the third tertile (full-adjusted model, β=1.01, 95% CI: 0.76–1.26) were both associated with higher MMSE score (Supplementary Table 3).

Sensitivity analyses

Two sensitivity analyses were conducted and the results were presented in Table 3. The results of sensitivity analyses were generally in line with those for all participants, indicating that the association between the cMIND diet and cognitive impairment did not change after removing participants with severe hypertension and with the lowest 5.5% MMSE score.

cMIND diet and IADL disability

In the full-adjusted model, moderate and high adherence to the cMIND diet were related to 14.9% (95% CI: 3.8%–24.6%) and 14.2% (95% CI: 2%–24.9%) decreased odds of IADL disability, respectively (Table 5). Meanwhile, per point increase in the cMIND diet score was associated with the reduced odds of IADL disability (basic model, OR=0.88, 95% CI: 0.85–0.90; full-adjusted model, OR=0.96, 95% CI: 0.93–0.99) (Supplementary Table 2). Additionally, we considered cognition status as a covariate in the IADL analysis model and verified the association between cognitive impairment and IADL disability.

Table 5.

Odds ratio (OR) and 95% confidence interval (CI) of estimated effects for tertiles of the diet score of instrumental activities of daily living disability&

Variables	Model 1a	Model 2b	Model 3c
	OR (95% CI)	OR (95% CI)	OR (95% CI)
Diet score
Tertile 1	1(reference)	1(reference)	1(reference)
Tertile 2	0.74(0.66,0.82) *	0.82(0.73,0.92)'	0.85(0.75,0.96) #
Tertile 3	0.62(0.55,0.70) *	0.82(0.72,0.93)'	0.86(0.75,0.98) #
Gender(Female)	1.98(1.79,2.18) *	1.96(1.72,2.21) *	1.93(1.70,2.20) *
Age
50–65years	1(reference)	1(reference)	1(reference)
66–80years	3.18(2.26,4.48) *	2.89(2.04,4.10) *	2.72(1.91,3.88) *
> 80years	29.09(20.60,41.08) *	18.48(13.00,26.28) *	14.23(9.94,20.35) *
Residence(Rural)	0.89(0.81,0.98)#	0.83(0.75,0.91) *	0.86(0.77,0.95)'
Education(Literate)	0.52(0.46,0.58) *	0.64(0.58,0.72) *	0.65(0.58,0.73) *
Body mass index(kg/m2)
< 18.5	1.24(1.08,1.44)'	1.17(1.01,1.37) #	1.15(0.98,1.35)
18.5–24	1(reference)	1(reference)	1(reference)
24–28	0.97(0.86,1.09)	1.06(0.94,1.19)	1.07(0.94,1.21)
≥28	1.46(1.22,1.74) *	1.51(1.26,1.80) *	1.53(1.27,1.83) *
Smoking		0.97(0.85,1.10)	0.94(0.82,1.07)
Drinking		0.91(0.80,1.03)	0.92(0.81,1.05)
Regular exercise		0.96(0.86,1.06)	0.95(0.85,1.05)
Social engagement
Lower		1(reference)	1(reference)
Middle		0.28(0.25,0.33) *	0.36(0.30,0.40) *
Higher		0.13(0.11,0.15) *	0.16(0.14,0.19) *
Hypertension
Non — hypertension			1(reference)
Aware hypertension			1.12(0.99,1.26)
Unaware hypertension			0.99(0.86,1.12)
Diabetes			1.45(1.22,1.72) *
Depression			1.69(1.48,1.94) *
Hearing impairment			2.29(2.03,2.58) *
Cognitive impairment			4.58(3.52,5.96) *

&Multivariable logistics regression was used to test the association between the diet score (modelled in tertiles) and instrumental activities of daily living disability; *P<0.001; P<0.01; #P<0.05; a. Model 1: adjusted for gender, age, region, education and body mass index; b. Model 2: adjusted for gender, age, region, education, body mass index, smoking, drinking, exercise, social engagement; c. Model 3: adjusted for gender, age, region, education, body mass index, smoking, drinking, exercise, social engagement, hypertension, diabetes, depression, hearing impairment and cognitive impairment.

We conducted the general linear regression models to identify the association of the cMIND diet with IADL, where the IADL score was considered as a continuous variable. As shown in Supplementary Table 3, in contrast with the lower adherence, both the moderate (β=0.50, 95% CI: 0.30–0.71) and higher (β=0.73, 95% CI: 0.50–0.96) adherence to the cMIND diet were related to better IADL in basic model, even though the association became insignificant in adjusted models.

Discussion

We developed the cMIND diet, which agreed with Chinese dietary characteristics, and verified its applicability in older Chinese people. This was the first study to establish a cognitive-healthy dietary pattern, especially for older Chinese individuals, and we successfully proved its positive association with lower odds of cognitive impairment.

Before interpreting our results, it is extremely necessary to expound the major limitations existing in the study. We used MMSE to identity the individuals with cognitive impairment, which could be applied to determine cognitive function but not MCI adequately. Our main purpose was to distinguish participants with cognitive impairment from those with better cognition, rather than making a diagnosis of MCI, since an accurate diagnosis of MCI is complicated and requires a series of tests. More research focusing on the association between the cMIND diet and MCI needs to be carried out in the future. Besides, considering the limitations of the cross-sectional analysis, the reverse causality may partly account for our findings. Older people with cognitive impairment could suffer from other diseases which may impact their dietary behaviors. For example, olfactory dysfunction is commonly present in MCI patients and associated with the progression from MCI to Alzheimer's disease, which may act as a valuable predictor to screen individuals with higher risk of cognitive impairment (39, 40). Previous research also reported the high prevalence of depression and diabetes in patients with cognitive impairment (41, 42). People were likely to change their eating habits as the result of cognitive impairment and these relevant diseases. Therefore, we could not establish a causal relationship between the cMIND diet and cognitive function or IADL. Additionally, residual confounding was still possible even though we tried to control the potential confounders in the analyses.

In this research, the median diet score of all participants was 4.5. The relatively low level of diet score was reasonable considering the poor status of diet-related knowledge and behaviors in Chinese adults (43). In addition, people aged over 80 years accounted for 58.8% of all participants. They were more susceptible to diseases and more likely to have difficulty chewing, swallowing and digesting, which resulted in a worse dietary structure and lower diet score. The dietary quality of older Chinese individuals was unsatisfactory, thus calling for urgent adjustment to their dietary patterns.

The cMIND diet was styled after the original MIND diet but with modifications. The consumption of butter/margarine, cheese, fast fried foods, red meat and wine was removed and we specifically included the assessment of mushroom/algae, garlic and tea. Most items (amount of staple food, fresh vegetables, mushroom/algae, fresh fruit, fish, soybeans, nuts, garlic, tea and sugar/sweets) in the cMIND diet showed a significant association with cognitive impairment separately, further indicating the validity of the revised diet pattern. These food groups are beneficial to cognition owing to different components through various mechanisms. Given that previous studies demonstrated that either excessive or insufficient intake of carbohydrates was detrimental to cognitive function, it is more appropriate for older people to consume a moderate amount of staple food (19, 44). Fresh vegetables and fruit are good sources of folate, vitamin C, carotenes and flavonoids, which are related to slower cognitive decline and a lower incidence of dementia (7). Mushrooms have a positive influence on inhibiting the production of amyloid-β and phosphorylated tau protein, as well as stimulating neurite outgrowth and nerve growth factor synthesis (45). Fish are rich in omega-3 fatty acids, such as eicosapentaenoic acid and docosahexaenoic acid. Omega-3 fatty acids play an important role in neuronal membranes, and the literature suggested that a higher intake of omega-3 fatty acids was related to less brain atrophy and cognitive decline (6). Soybeans are good sources of plant protein containing all essential amino acids. The constituents of soybeans, including soy protein and nonprotein soy components (e.g., isoflavones), have many physiological functions, such as hypolipidemic, antihypertensive, anti-inflammatory, antioxidant and improving glycemic control (46, 47). Nuts contain many health-related compounds (e.g., vitamin B6, vitamin E, minerals and flavonoids) and possess the optimum proportion of fatty acids (high level of unsaturated fatty acids and low level of saturated fatty acids), which helps to reduce inflammation and oxidative stress (48). Tea polyphenols, especially EGCG, are neuroprotective through several mechanisms including antioxidant, iron chelation, and signal transduction modulation (49). EGCG also inhibits tau protein and amyloid-β protein, which are closely connected with Alzheimer's disease (50). Theanine, an amino acid of tea, assists in attentional processing and strengthens cognition of the human brain (49). As the only discouraging food in the cMIND diet, the excessive intake of sugar/sweets disturbs glucose and insulin metabolism and boosts neuroinflammation as well as oxidative stress, thus causing the structural changes in the normal brain (19). However, there was no significant difference between the abnormal and normal cognition groups, either for “Cooking oil” or “Type of staple food”. This is possibly because we only made a simple distinction between vegetable oil and animal oil, without differentiating various types of them. As for “Type of staple food”, it could be explained by the fact that the number of participants who consumed the whole grains was too few to show a positive association. Despite the negative results, we still recognized their benefit to cognitive function. Further studies are required to discover the exact association or explore a more suitable classification of cooking oil and staple food. Notably, we regarded the cMIND diet as a dietary pattern where food components interacted with one another complexly and variously, so more attention should be given to the effect of the whole dietary pattern rather than a certain food component.

A prospective cohort study investigated the relations of five dietary patterns (alternate Mediterranean diet, DASH diet, the alternative Healthy Eating Index-2010, overall plant-based diet index, and healthful plant-based diet index) to cognitive impairment in Chinese adults (51). Although researchers concluded that higher adherence to these dietary patterns in midlife was associated with a lower risk of cognitive impairment in late life, they did not correspond to Chinese dietary habits well. Nevertheless, the cMIND diet was developed based on the comprehensive consideration of Chinese dietary patterns and we first confirmed the association between the cMIND diet and cognitive impairment among older Chinese individuals in the cross-sectional analysis. In the full-adjusted model, moderate and high adherence to the cMIND diet were related to 19.0% and 39.7% reduced odds of cognitive impairment, respectively. This finding was consistent with that of other studies, which suggested that higher adherence to the MIND diet was related to slower rates of cognitive decline (8, 10). The health benefits of the MIND diet were not limited to cognitive protection, and the MIND diet was associated with depression, psychological distress (52) and Parkinson's disease (53). Since prior evidence indicated that people who followed the Dietary Guidelines for Americans (16) or Australian Dietary Guidelines (17) were less likely to experience IADL disability, we also explored the relation between the cMIND diet and IADL, finding that a higher cMIND diet score was associated with lower odds of IADL disability. Therefore, the advantages of the cMIND diet are multifaceted and thus deserve to be discovered in future studies, and it is meaningful for older Chinese individuals to adhere to the cMIND diet earlier to obtain the maximum health benefits.

Our study has a number of strengths. First, we developed a cognitive-healthy dietary pattern especially for Chinese individuals, based on comprehensive reviews of previous studies. The cMIND diet originated from the MIND diet but was designed for better accordance with Chinese dietary habits. Second, we verified its association with cognitive impairment using the data obtained from the CLHLS, which was well organized and collected information by a high-quality questionnaire. The samples covering most provinces in China were widely representative of older Chinese individuals, which supported the generalization of our findings (28). We tried to statistically adjust the potential confounding factors to confirm the authentic association between the cMIND diet and health outcomes, which remained significant in sensitivity analyses, and our findings have compensated for the shortcomings of previous studies. Last but not least, medications have a limited effect on managing the neuropsychiatric symptoms of dementia patients and even lead to severe adverse reactions, which highlights the importance of preventing the progression of cognitive impairment at an earlier age (5).

Several limitations should be acknowledged in our study. First, as mentioned above, we drew conclusions based on the cross-sectional design so they could not be interpreted as a cause-and-effect relation. Second, the application of the FFQ relies on the generic memory of respondents and requires their ability to perform cognitively complex memory and averaging tasks. Thus, measurement error, which referred to the difference between reported dietary intake and true usual dietary intake, was another limitation for our study conducted among older people. Additionally, the CLHLS questionnaire only investigated the frequency of food consumption, so we could not evaluate the specific quantity of different food groups. We put forward a revised dietary pattern that was rather sketchy without the detailed definition of food quantity. In addition, fast fried foods were prevalent in the Chinese diet but they were not included in the current research due to the lack of available information. Notably, people aged over 80 years comprised a large proportion of all participants, so it is essential to conduct more research among younger people to completely discover the effect of the cMIND diet during early aging.

Conclusion

We develop the cMIND diet to protect cognition for older Chinese individuals and the results suggest that both the moderate and high adherence to the cMIND diet were associated with lower odds of cognitive impairment and IADL disability even after adjusting for covariates. In addition, higher cMIND diet score was also associated with better cognitive function and IADL. Overall, these findings provide further support for the benefit of the cMIND diet to cognitive function and IADL. Given the major limitations of our study, more longitudinal and interventional research is needed for better understanding of the association between the cMIND diet and cognitive impairment as well as IADL disability.

Conflict of interest

The authors declare no conflict of interest.

Author Contributions

Conceptualization: Xiaojie Huang, Sumiya Aihemaitijiang and Zhaofeng Zhang; methodology: Xiaojie Huang and Chen Ye; data analysis and data interpretation: Xiaojie Huang, Mairepaiti Halimulati and Ruoyu Wang; writing of the manuscript: Xiaojie Huang; supervision: Zhaofeng Zhang; project administration: Zhaofeng Zhang. All authors meet the criteria for authorship stated in the Uniform Requirements for Manuscripts Submitted to Biomedical Journals. All authors have read and agreed to the published version of the manuscript.

Sponsor's Role

The sponsors did not play a role in the design, methods, subject recruitment, data collection, analysis, or preparation of this study.

Ethical Standards

We obtained data from the CLHLS. The Biomedical Ethics Committee of Peking University approved the CLHLS (IRB00001052-13074).

Funding Statement

This research received no external funding.

Electronic Supplementary Material

Supplementary material is available for this article at https://doi.org/10.1007/s12603-022-1829-1 and is accessible for authorized users.

Supplementary material, approximately 29.8 KB.