Abstract
Background:
A few studies have linked dietary patterns and sleep to cognitive decline.
Objective:
To examine the independent and joint associations of dietary patterns and sleep with cognitive decline.
Methods:
Our analysis included 2,307 participants aged 55– 89 years at baseline from the China Health and Nutrition Survey. Dietary intake was assessed using weighing methods in combination with 24 h dietary recalls for three consecutive days. Exploratory factor analysis was applied to identify major dietary factors. Cognition was assessed in 1997, 2000, 2004, 2006, and 2015.
Results:
Five dietary patterns were identified: dairy-fruits-fast foods, grains-vegetables-pork, plant-based food, beans-mushroom, and beverages-nuts patterns. Beans-mushroom pattern and sleep duration of 8 h/day were defined as healthy habits. There was a positive association between the beans-mushroom pattern and change in the global cognitive Z-score over seven years (β (95% CI) for quintile 5 versus quintile 1:0.17 (0.05, 0.30)). Compared to individuals with sleep duration of 8 h/day, those with sleep duration of≤5 h/day (β (95% CI): – 0.23 (– 0.45, – 0.00)) or > 10 h/day (– 0.52 (– 0.73, – 0.32)) had a greater decrease in global cognitive Z-score. Compared to individuals with no healthy patterns, those with a healthy dietary pattern only (β (95% CI): 0.18 (0.08, 0.28)), healthy sleep pattern only (0.13 (0.04, 0.23), and both healthy dietary and sleep patterns (0.19 (0.08, 0.31)) had a relative increase in global cognitive Z-score.
Conclusion:
Our findings highlight the importance of involving both diet and sleep as intervention priorities for the potential prevention of cognitive decline.
INTRODUCTION
Diet is among the modifiable factors that play an important role in the development of dementia [1–3]. Although findings for the association between intakes of individual foods or nutrients and cognitive decline are inconsistent [4], healthy dietary patterns including the Mediterranean, Dietary Approaches to Stop Hypertension (DASH), and Mediterranean-DASH Intervention for Neurodegenerative Delay (MIND) diets have been associated with less cognitive decline in many studies [5, 6]. However, in a longitudinal study of 8,225 participants with a follow-up of 24.8 years, dietary quality assessed in midlife was not significantly associated with the risk of dementia [7]. Furthermore, a cohort study of 1,650 adults aged≥55 years from the China Health and Nutrition Survey (CHNS) reported that an adapted Mediterranean diet was associated with a slower rate of cognitive decline among those aged≥65 years only [8]. This suggests that the beneficial effects of some specific healthy dietary patterns observed in some countries may not apply to all individuals. Most of the studies investigating the association between dietary patterns and cognitive decline assessed dietary intakes with food frequency questionnaires (FFQs), but FFQs are limited with respect to the variety of foods assessed [5]. Therefore, more longitudinal studies with comprehensive dietary assessment tools are needed to explore associations of dietary patterns with cognition.
Sleep is an emerging modifiable factor that may influence cognitive decline. Although sleep is not currently well recognized as an intervention priority for the prevention of cognitive decline [9], a recent meta-analysis of 7 longitudinal studies showed that both long and short sleep durations were associated with lower cognitive performance [10]. Short sleep duration is associated with higher energy and fat intake and lower quality diets, indicating that unhealthy sleep and dietary patterns maybe clustered [11]. Evidence also shows that healthy dietary patterns may promote sleep quality [12]. Unhealthy sleep and diet patterns are shared risk factors for numerous chronic conditions including obesity, hypertension, diabetes, and cardiovascular disease [11]. This suggests that sleep and diet may work simultaneously on cognitive function. Due to preventative potential for risk mitigation, it is important to investigate the combination of multiple health behaviors and the association with cognitive decline [13].
This study aimed to examine whether dietary patterns and sleep duration are independently associated with cognitive decline. We also analyzed whether the combination of these two factors was associated with cognitive decline. We hypothesized that healthy dietary patterns and normal sleep duration might help to reduce cognitive decline in older adults and the combined potential protective effects of two healthy habits were larger than those of one individual healthy habit alone.
METHODS
Participants
The CHNS is an ongoing open-cohort study initiated in 1989 and followed up in 1991, 1993, 1997, 2000, 2004, 2006, 2009, 2011, and 2015 [14, 15]. The study was conducted in nine provinces from northeast to southwest in China: Heilongjiang, Liaoning, Jiangsu, Shandong, Henan, Hubei, Hunan, Guizhou, and Guangxi. Two cities and four counties were randomly selected in each province based on their income levels. Four communities in each city or county and 20 households in each community were then randomly selected. Data from 2004 to 2015 were used in the analysis as data on sleep were not collected between 1989 and 2000. Of the 38,536 individuals who participated in any of the ten surveys, 14,827 attended at least one of the surveys between 2004 and 2015. The following were excluded from the present analysis: those aged less than 55 years (n = 11,442), those who did not have cognitive function assessed (n = 795), those who completed cognitive assessment at only one survey (n = 55), those who had stroke, heart disease, or cancer at baseline (n = 150), those who did not have data on sleep or diet (n = 78) or those with psychiatric conditions (n = 0, the steps above excluded all those with psychiatric conditions). A total of 2,307 participants were included in the final analysis (Supplementary Figure 1).
The survey was approved by the institutional review committees of the University of North Carolina at Chapel Hill and the National Institute of Nutrition and Food Safety, Chinese Center for Disease Control and Prevention (R01-HD30880, DK056350, and R01-HD38700). Written informed consent was obtained from all participants prior to any procedure.
Dietary assessment
Diet intake at the household level was assessed using weighing methods for three consecutive days at each wave and diet intake at the individual level was assessed with three-day 24 h dietary recalls accordingly. All foods, beverages, and condiments were measured and recorded using scales by trained interviewers at the beginning and end of the three-day survey period. Household food consumption was computed by subtracting the amount of weighed foods at the end of the three-day survey period from that at the beginning. Proportions of foods, beverages, and condiments consumed at the household level were allocated to each individual based on the data reported [16]. This was achieved by asking all household members on a daily basis to report all foods including snacks and shared dishes consumed at home and away from home. Intake of foods and nutrients at the individual level were then calculated. Nutrients and energy intake was calculated based on the China Food Composition [17].
Foods were divided into 24 groups: low fiber grains, high fiber grains, deep fried grains, starchy roots and tubers, beans, vegetables, edible fungi and algae, pickled vegetables, fresh fruits, dried or canned fruits, nuts, pork, red meat other than pork, offal, processed meat, poultry, fish and seafood, eggs, dairy, snacks, beverages, fast-foods, candy and sugar, and sauces/jams (Food items for each group are detailed in Supplementary Table 1). The average amount of each food group consumed was calculated for each individual (per day per person).
The assessment of energy intake has been validated by using the doubly labeled water method with a correlation coefficient of 0.56 for men and 0.60 for women [18]. Another independent validation analysis showed that the correlation coefficients between dietary intake estimated by weighing with 24 h recalls and urine excretions measured from 24 h urine samples for sodium and potassium were 0.58 and 0.59, respectively [19].
Sleep duration
Participants were asked about sleep duration: “How many hours each day do you usually sleep, including during both daytime and nighttime?” No consensus on the cutoff points for short or long sleep duration has been achieved, but≤5 or > 10 h/day of sleep were shown to be associated with cognitive decline or dementia in many studies [10, 20]. Therefore, sleep duration per day was divided into five groups in our study: ≤5 h, 6– 7 h, 8 h, 9– 10 h, and > 10 h.
Cognitive assessment
A subset of the items from the Telephone Interview for Cognitive Status (modified version) was used to assess cognitive function in 1997, 2000, 2004, 2006, and 2015 [21]. The tool has been applied in other population studies in China [22, 23]. Cognitive screening through face-to-face interviews included four tasks: included immediate (immediate memory) and delayed recall of a 10-word list (delayed memory), counting backward from 20 (attention), and serial 7 subtraction five times (working memory). The total score for immediate and delayed recall ranged from 0 to 20 with each correctly recalled word assigned a score of 1. For counting backward, a score of 2 was given to those who counted backwards correctly on the first try and 1 to those who counted backwards correctly on the second try. The total score for serial 7 subtraction ranged from 0 to 5 with a score of 1 assigned to each of the 5 serial 7 subtractions.
The global cognitive score was computed by summing the scores of all three tasks ranging from 0 to 27. Cognition was analyzed as Z-scores, with a higher score represented better cognitive function. The change in global cognitive Z-score was computed by subtracting the score at baseline from that at follow-up. Cognitive decline was defined as a change in global cognitive Z-score during follow-up that fall below the mean minus one standard deviation (SD) of the change [24].
Physical examinations
Height was measured to the nearest 0.1 centimeter using a freestanding stadiometer and weight was measured to the nearest 0.1 kg using a balance-beam scale. Body mass index (BMI) was computed as weight in kilograms divided by the square of height in meters. Systolic and diastolic blood pressure were measured in the seated position using a standard mercury sphygmomanometer by trained nurses with at least 5 min rest before the measurement. Three measurements were taken to the nearest two mmHg and the average of the last two was used in the analysis.
Confounders
Data on age, gender, region (urban/rural areas), education (illiterate/primary school, middle school, and high middle school or higher), smoking (never smoked, former smoking and current smoking), and alcohol consumption (yes or no) were collected using a questionnaire. Physical activity was assessed based on hours per week spent in different occupational, household, transportation, and leisure-time activities, from which metabolic equivalent of task (MET) was calculated [25]. History of hypertension, diabetes, stroke, myocardial infarction, and cancer were also self-reported.
Statistical analysis
Data were expressed as frequency (percentage) and means±SDs for baseline characteristics. ANOVA for continuous variables and the Chi-square test for categorical variables was performed to compare the difference of baseline characteristics across the healthy dietary and sleep patterns.
Exploratory factor analysis was applied to the 24 food groups previously defined to identify the major factors accounting for main variance in dietary intake. Eigen value, scree test, and the interpretability of the factors were considered in determining the number of factors to retain. Food groups with factor loadings≥0.20 or ≤– 0.20 were considered as significant contributors to the factors [26]. A higher factor score represented a high intake of the foods constituting that food factor or lower intake of those with negative loadings. Compared with cluster or rank regression analysis, factor analysis can produce dietary scores as continuous variables that can be easily analyzed for their associations with sleep duration.
ANOVA was used to test the difference of dietary pattern scores as well as food and nutrient intakes between sleep duration groups, and Bonferroni Post-hoc test was used to conduct pairwise comparisons.
General linear regression models (GLM) were used to test the difference in changes in global cognitive Z-score between individuals in different quintiles of the main dietary pattern factors. Overall F-test was used to assess the fit of the GLM. We tested the following models: 1) age and gender; 2) model 1 plus education, living area, years of follow-up, smoking, alcohol intake, physical activity, global cognitive Z-score, diabetes, BMI, systolic and diastolic blood pressure at baseline; 3) model 2 plus and intake of energy, sodium, potassium, fat, protein, carbohydrate, and fiber (these nutrients are related to dementia risks) at baseline. GLM was also used to test the association of changes in global cognitive Z-score with sleep duration. We repeated the analysis for the association between dietary patterns and sleep duration and change in Z-scores of different cognitive domains.
Healthy dietary and sleep patterns that were associated with a lower rate of cognitive decline were identified. Individuals were divided into four groups: no healthy patterns, healthy dietary patterns only, healthy sleep patterns only, and both healthy dietary and sleep patterns. GLM was then used to test whether the change in global cognitive Z-score differed between different combinations of healthy dietary and sleep patterns.
Logistic regression models were used to examine the association of sleep duration, dietary patterns, and healthy and sleep patterns with cognitive decline. The same covariates in the three models above were adjusted for in the analysis. The Wald test was used to assess the fit of the logistic regression models.
Repeated-measures generalized estimating equations were used to test the association of dietary patterns/sleep duration with cognitive Z-scores over time. Three models with the same adjusted covariates mentioned above were tested.
GLM was also used to test whether the association of dietary patterns/sleep duration with cognitive decline was mediated by sleep duration, dietary patterns, systolic blood pressure, smoking, or physical activity.
Sensitivity analysis for the association of dietary patterns/sleep duration with mild cognitive impairment was conducted among individuals with raw global cognitive score≥7 at baseline using logistic regression models. Mild cognitive impairment was defined as global cognitive score < 7 based on previous studies among Chinese populations [27, 28].
Missing values for categorical variables were assigned as a single category and those for continuous variables were given the corresponding means.
Data analyses were conducted using SAS 9.4 for Windows (SAS Institute Inc.) and two-sided p values < 0.05 were considered statistically significant.
RESULTS
Participant characteristics
A total of 2,307 participants (50.8% women) with complete data on variables of interest were included in the analysis. They were aged 55– 89 (63.3±7.0) years at baseline and 57– 94 (70.2±6.9) years at follow-up. Men, those with low education or living in urban areas, were more likely to have both healthy dietary and sleep patterns (the diet and sleep patterns that were associated with least cognitive decline). Healthy dietary and sleep patterns were associated with lower systolic blood pressure, a higher intake of energy, fiber, sodium, potassium, fat, and protein and a lower intake of carbohydrate at baseline. Individuals with healthy dietary and sleep patterns had a higher global cognitive Z-score at baseline (Table 1).
Baseline characteristics according to healthy dietary and sleep patterns*
Baseline characteristics according to healthy dietary and sleep patterns*
*Individuals in quintiles 4 and 5 of the beans and mushroom pattern score were considered to have a healthy dietary pattern. We defined healthy sleep pattern as sleep duration of 8 h/day. †ANOVA was used to test the difference of continuous variables across meal patterns and Chi-square for categorical variables. ‡All such data were means±standard deviations. §All such data were frequency (percentage).
Four factors with Eigen values > 1.10, accounting for 24.5% (7.7%, 6.6%, 5.1%, 5.1%, and 4.7% for factor 1, factor 2, factor 3, factor 4, and factor 5, respectively) of the total variation, were extracted on the basis of the scree plot and evaluation of the factor loading matrix after orthogonal rotation. Factor 1 with high positive loadings on dairy, fruits, fast foods, eggs and beverages was labeled as ‘dairy-fruits-fast foods dietary pattern’. Factor 2 with high positive loadings on low fiber grains, vegetables, and pork, was labeled as ‘grains-vegetables-pork dietary pattern’. Factor 3 with high positive loadings on deep fried grains, higher fiber grains, fruits, and beans, and high negative loadings on low fiber grains, candies and sugar, was labeled as ‘plant dietary pattern’. Factor 4 with high positive loadings on beans, edible fungi and algae (mainly mushroom), and snacks, was labeled as ‘beans and mushroom dietary pattern’. Factor 5 with high positive loadings on beverages, nuts, and poultry, was labeled as ‘beverages-nuts dietary pattern’ (Supplementary Table 2).
The change in global cognition and cognitive decline
During a median follow-up of 7 (2– 11) years, the global cognitive score decreased from 14.1±5.4 at baseline to 12.6±5.7 at follow-up (mean change: – 1.5±6.6). The prevalence of cognitive decline was 15.8%.
Sleep duration and dietary intakes at baseline
Sleep duration greater than 10 h/day was associated with a lower score on the beans and mushroom dietary pattern. Individuals with sleep duration greater than 10 h/day had a lower intake of fiber, potassium, beans, and fruits whereas those with sleep duration of≤5 h/day had a higher offal intake compared to those with sleep duration of 8 h (Table 2).
Sleep duration and dietary intakes at baseline
Sleep duration and dietary intakes at baseline
*ANOVA was used to test the difference of dietary intakes between groups of sleep duration. Bonferroni Post-hoc test was used to examine the difference between each two groups with aindicating significance compared with sleep duration of≤5 h, bindicating significance compared with sleep duration of 6– 7 h, cindicating significance compared with sleep duration of 8 h, dindicating significance compared with sleep duration of 9– 10 h. †Factors 1 to 5 were identified using exploratory factor analysis. Factor 1 had high positive loadings on dairy, fruits, fast foods, eggs and beverages; Factor 2 had high positive loadings on low fiber grains, vegetables, and pork; Factor 3 had high positive loadings on deep fried grains, higher fiber grains, fruits, and beans, and high negative loadings on low fiber grains, candies and sugar; Factor 4 had high positive loadings on beans, edible fungi and algae (mainly mushroom), and snacks; Factor 5 had high positive loadings on beverages, nuts, and poultry.
There was a positive association between the beans and mushroom dietary pattern and the change in the global cognitive Z-score after adjustment for age and sex (β (95% confidence interval [CI]) for quintile 5 versus quintile 1:0.27 (0.16, 0.39), p value for trend < 0.0001). This association was attenuated but remained significant after additional adjustment for socioeconomic status and lifestyle factors (β (95% CI) for quintile 5 versus quintile 1:0.20 (0.09, 0.32)). In the multivariable analysis (additionally adjusted for important nutrients intake), β (95% CI) for the change in the global cognitive Z-score was 0.17 (0.05, 0.30) for quintile 5 (with quintile 1 as reference) of the beans and mushroom dietary pattern score. The positive association between dairy-fruits-fast foods dietary pattern and the change in the global cognitive Z-score was significant before (p value for trend: 0.0036) but not after adjustment for intake of energy, fiber, fat, protein, potassium, and sodium (p value for trend: 0.0609). Individuals in the quintile 4 of the dairy-fruit-fast foods dietary pattern score had a higher increase in the global cognitive Z-score (β (95% CI): 0.15 (0.04, 0.27)) compared with those in the quintile 1 after adjustment for confounders. The corresponding β (95% CI) for the quintile 5 was 0.03 (– 0.10, 0.16). Individuals in the quintile 5 of the beverages-nuts dietary score had a relatively higher increase in global cognitive Z-score (β (95% CI): 0.12 (0.00, 0.24)) compared with those in the quintile 1.
The beans and mushroom dietary pattern was associated with a lower risk of cognitive decline before (OR (95% CI): 0.56 (0.37– 0.84)) but not after adjustment for confounders (0.75 (0.49– 1.16)). Similarly, the dairy-fruits-fast foods dietary pattern was associated with a lower risk of cognitive decline before but not after adjustment for confounders (Table 3).
The change in global cognitive Z-score and cognitive decline associated with dietary patterns
The change in global cognitive Z-score and cognitive decline associated with dietary patterns
*General linear regression models were used to obtain coefficients for the change in global cognitive Z-score for quintiles 2– 5 versus the quintile 1 of dietary pattern scores. The change in global cognitive Z-score was computed as the score at baseline subtracting from that at follow-up. †Model 1 was adjusted for age and sex. ‡Model 2 was adjusted for Model 1 plus education, living area, years of follow-up, smoking, alcohol intake, physical activity, global cognitive score, diabetes, BMI, systolic and diastolic blood pressure at baseline. §Model 3 was adjusted for model 2 plus intake of energy, sodium, potassium, carbohydrates, fat, protein, and fiber. ¶Logistic regression models were used to estimate ORs (95% CIs) for cognitive decline associated with dietary patterns. Cognitive decline was defined as the change in the global cognitive Z-score below the mean minus one standard deviation.
For individual domains of cognition, the beans-mushroom dietary pattern was associated with a relatively higher increase in working memory Z-score but not immediate memory, delayed memory, or attention Z-scores (Supplementary Table 3).
Compared with individuals with sleep duration of 8 h/day, those with sleep duration of≤5 h/day had a larger decrease in global cognitive Z-score (β (95% CI): – 0.23 (– 0.45, – 0.0004)). The corresponding β (95% CI) for sleep duration of > 10 h/day was – 0.52 (– 0.73, – 0.32). Sleep durations of≤5 h/day and > 10 h/day were associated with a higher risk of cognitive decline before, but not after, adjustment for confounders (Table 4). Compared to individuals with sleep duration of 8 h/day, those with sleep duration of≤5 h/day had a higher decrease in immediate memory Z-score, but not Z-scores of other domains. Sleep duration of > 10 h/day was independently associated with a higher decrease in Z-scores of immediate memory, delayed memory, and working memory but not attention (Supplementary Table 4).
The change in global cognitive Z-score and cognitive decline associated with sleep duration
The change in global cognitive Z-score and cognitive decline associated with sleep duration
*General linear regression models were used to obtain coefficients for the change in global cognitive Z-score for sleep duration ≤5, 6– 7, 9– 10, or > 10 h versus the 8 h. The change in global cognitive Z-score was computed as the score at baseline subtracting from that at follow-up. †Model 1 was adjusted for age and sex. ‡Model 2 was adjusted for Model 1 plus education, living area, years of follow-up, smoking, alcohol intake, physical activity, global cognitive Z-score, diabetes, BMI, systolic and diastolic blood pressure at baseline. §Model 3 was adjusted for model 2 plus intake of energy, sodium, potassium, carbohydrates, fat, protein, and fiber. ¶Logistic regression models were used to estimate ORs (95% CIs) for cognitive decline associated with sleep duration. Cognitive decline was defined as the change in the global cognitive Z-score below the mean minus one standard deviation.
Being associated with the lowest risk of cognitive decline, sleep duration of 8 h/day was defined as healthy sleep pattern. The beans and mushroom pattern score was used as a marker of healthy diet as it was more predictive of the change in global cognitive Z-score and cognitive decline than other patterns. Individuals in quintiles 4 and 5 of the beans and mushroom pattern score were considered to have a healthy dietary pattern.
Compared with individuals with no healthy patterns, those with a healthy dietary pattern only (β (95% CI): 0.18 (0.08, 0.28)), healthy sleep pattern only (0.13 (0.04, 0.23), and both healthy dietary and sleep patterns (0.19 (0.08, 0.31)) had a greater increase in global cognitive Z-score in the multivariable analysis.
In the multivariable analysis, individuals with both healthy dietary and sleep patterns had a 37% (95% CI: 4% – 58%) lower risk of cognitive decline and those with a healthy dietary pattern only had a 39% (95% CI: 11% – 58%) lower risk of cognitive decline compared to those with no healthy patterns (Table 5).
The change in global cognitive Z-score and cognitive decline associated with healthy dietary and sleep duration patterns
The change in global cognitive Z-score and cognitive decline associated with healthy dietary and sleep duration patterns
*General linear regression models were used to obtain coefficients for the change in global cognitive Z-score for healthy dietary pattern only, healthy sleep pattern only, and both healthy dietary and sleep patterns versus none healthy patterns. The change in global cognitive Z-score was computed as the score at baseline subtracting from that at follow-up. †Model 1 was adjusted for age and sex. ‡Model 2 was adjusted for Model 1 plus education, living area, years of follow-up, smoking, alcohol intake, physical activity, global cognitive Z-score, diabetes, BMI, systolic and diastolic blood pressure at baseline. §Model 3 was adjusted for model 2 plus intake of energy, sodium, potassium, carbohydrates, fat, protein, and fiber. ¶Logistic regression models were used to estimate ORs (95% CIs) for cognitive decline associated with healthy dietary and sleep patterns. Cognitive decline was defined as the change in the global cognitive Z-score below the mean minus one standard deviation.
Compared with individuals with no healthy pattern, those with both healthy dietary and sleep patterns had a relatively higher increase in working memory Z-score (Supplementary Table 5).
The beans-mushroom dietary pattern was positively associated with change in the Z-scores of global cognition, immediate memory, and working memory (Supplementary Table 6). Individuals with sleep duration of > 10 h/day had a higher decrease in Z-scores of the global cognition and all domains (Supplementary Table 7). Both healthy dietary and sleep patterns were associated with a relatively higher increase in Z-scores of the global cognition and immediate memory, delayed memory, and working memory (Supplementary Table 8).
Mediation analysis
The percentage of the association between the beans and mushroom pattern and the change in global cognitive Z-score (two time points) explained by sleep duration was 6.4% (95% CI: 0.6% – 12.7%) (p-value for mediation = 0.0478, Supplementary Figure 2). Physical activity was a marginally significant mediator for the association between sleep duration and cognitive decline (3.2% (– 0.2%, 6.7%), p = 0.0649). No other significant mediators were found for the association between dietary patterns/sleep duration and cognitive decline.
Sensitivity analysis
The beans-mushroom dietary pattern, but not other dietary pattens, was associated with a lower risk of mild cognitive impairment (odds ratio (95% CI): 0.49 (0.30– 0.81), Supplementary Table 9). Compared to individuals with sleep duration of 8 h/day, those with sleep durations of > 10 h/day had a higher risk of mild cognitive impairment (odds ratio (95% CI): 2.25 (1.18– 4.27), Supplementary Table 10). Individuals with both healthy diet and sleep patterns had a lower risk of mild cognitive impairment compared to those with no healthy pattern (odds ratio (95% CI): 0.53 (0.34– 0.84), Supplementary Table 11).
DISCUSSION
In this large longitudinal study in older Chinese adults, we found a higher score for the beans and mushroom dietary pattern was associated with a lower rate of cognitive decline. Both short and long sleep duration were associated with a greater decrease in the global cognitive Z-score. Compared to individuals with no healthy dietary or sleep patterns, those with one or more healthy patterns had a relative increase in global cognitive Z-score.
The potential beneficial effects of plant-rich diets including Mediterranean, DASH, and MIND on cognitive decline and dementia have been observed in many studies [5]. A prospective study of 923 participants aged 58– 98 years from the US with a follow-up of 4.5 years found that adherence to MIND (HR (95% CI) for tertile 3 versus tertile 1 of diet score: 0.47 (0.26– 0.76)), DASH (0.61 (0.38– 0.97)), and Mediterranean (0.46 (0.26– 0.79)) diets were all associated with lower incidence of Alzheimer’s disease [29]. In another longitudinal analysis of 5,083 participants from the Whitehall II cohort study, inflammatory dietary pattern was associated with accelerated cognitive decline [30]. However, these studies were conducted in Western countries, and were limited by measurement errors due to self-reported dietary intakes. Diets in China and Western countries differ in general with main sources from plant-based foods in China and animal-based foods in Western countries. Data from the CHNS (diet assessed using weighing methods) of 1,650 adults surveyed in 1997– 2004 showed that an adapted Mediterranean diet may result in favorable effects on cognitive decline in individuals aged≥65 years but not in those aged 55– 64 years [8]. This suggests that diets with higher intakes of plant-based foods and limited intakes of animal-based foods may be beneficial for the prevention of cognitive decline, which is consistent with our study showing that the beans and mushroom, dairy-fruits-fast foods, and beverage and nuts dietary patterns were associated with a lower rate of cognitive decline (both two and multiple points). Our further analysis showed these dietary patterns were also associated with a lower risk of mild cognitive impairment. Likely, the potential benefits of beans, fruits, and nuts for cognitive health have been demonstrated by previous studies, which reported that Mediterranean diets high in plant foods (such as vegetables, fruits, nuts, and beans) and low in red meat were associated with a lower risk of cognitive impairment [5, 6]. However, high intake of fast foods and beverages may confer an increased risk of cognitive impairment [5]. It is unclear why dairy-fruits-fast foods and beverage-nuts dietary patterns with mixed healthy and unhealthy foods were associated with a lower risk of mild cognitive impairment. These dietary patterns may stand for a transition from traditional Chinese diets to Western diets. Individuals with these patterns (1997– 2006) were more likely to be socially active and have high levels of education and income, which were associated with better cognitive health. Both beans and mushrooms provide protein, minerals, and antioxidants, of which the higher intake may be associated with better cognitive health [4, 31]. Meanwhile, mushrooms are also rich in vitamin D, which has been shown to be potentially beneficial for the prevention of dementia [9]. Our study excluded individuals with well-known dementia risks including stroke, heart disease, or diabetes from the analysis, suggesting our findings maybe more reliable compared with most previous studies involving participants with existing chronic diseases. This highlights that high beans and mushrooms intake may help to protect against cognitive decline in older Chinese adults.
Sleep plays an important role on circadian rhythms, which are associated with metabolism [32]. The association between sleep duration and cognition may be explained by the fact that short or long sleep duration was associated with brain impairments including gray matter loss [33], hippocampal degeneration [32], and brain atrophy [34]. We found both short (≤5 h/day) or long (> 10 h/day) sleep durations were associated with a greater decrease in the global cognitive Z-score (two points). This is consistent with a longitudinal study of 1,010 Chinese adults aged 65– 80 years showing that those with sleep duration≤5 h/day had a higher risk of cognitive decline over one year (OR (95% CI): 1.95 (1.16– 3.27)), compared to those with sleep duration of 7 h/day [20]. A longitudinal study of 2,893 Korean adults aged≥60 years reported that long (≥7.95 h/day) but not short (< 5.05 h/day) sleep duration was related to the risk of cognitive decline over 4 years (OR (95% CI): 1.67 (1.18– 2.35)) [35]. A recent meta-analysis of 17 prospective studies found a U-shaped relationship that both short (< 4 h/day) or long (> 12.5 h/day) sleep durations were associated with a higher risk of cognitive disorders/dementia [36].
Sleep deprivation may alter individuals’ dietary behaviors by decreasing leptin and increasing ghrelin [37, 38], enhancing hedonic stimulus [39], and increasing time for intake, which have been associated with higher energy and total fat intake and lower fruit intake and quality diets [11]. Long sleep duration has been associated with decreasing time for physical activity resulting in less energy expenditure, which might also influence individuals’ dietary behaviors. Therefore, unhealthy sleep and dietary patterns may occur together [11]. We found sleep duration greater than 10 but not≤5 h/day was associated with a lower score on the beans and mushroom dietary pattern. It is also possible that the relationship between diet and sleep duration is in the other direction with diet affecting sleep [12, 40]. This could be explained by the fact that the intake of some foods exert influences on the availability of tryptophan, as well as the synthesis of serotonin and melatonin, which affect sleep duration and quality [12, 41]. We also observed that individuals with≥1 healthy patterns had a lower rate of cognitive decline (both two and multiple points) compared to those with none. The healthiest cognitive function was observed among those with both healthy dietary and sleep patterns, but this combined effect was just minimally larger than that of healthy dietary pattern only. This suggests the combined effect is largely dependent on dietary pattern. However, further analysis demonstrated that sleep mediated the association between beans and mushroom dietary pattern and cognitive decline. This indicates improvement in dietary habits may maximize cognitive health via the modification of sleep duration highlighting the importance of involving both behaviors as intervention priorities for the potential reduction of cognitive decline. Notably, cognitive mains including attention and working memory, and immediate and delayed memory are not likely to all decline at the same age and rate [42, 43], and these cognitive domains differ in the prediction performance of dementia [44, 45]. We found dietary patterns are more likely to affect working memory while long sleep duration is more likely to affect immediate and delayed memory (multiple time points). This is consistent with some previous studies showing that a healthy diet was associated with slower decline in executive functioning, but not episodic memory [46, 47]. While sleep is more likely to affect hippocampal-dependent processes of episodic memory [48]. Further study needs to verify our findings in other ethnic groups.
Strengths of the present study included the large sample size, long follow-up, and the measurement of dietary intakes using weighing methods in combination with 24 h dietary recalls for three consecutive days. In particular, the weighing methods might have reduced the reported bias due to the memory among those with low cognition at baseline, which is an advantage compared to FFQs used in previous studies. To our knowledge, this is the first study to investigate the association between the clustering of dietary and sleep patterns and cognitive decline in a longitudinal analysis. The present study has several limitations. Firstly, because of the observational nature, causal relationships could not be established based on our findings. Furthermore, the question used to assess sleep duration cannot differentiate between daytime napping and nighttime sleep and was not validated in previous surveys, which might have biased the association. Several aspects of sleep quality as well as sleep duration, such as late bed times, and sleep latency may all be related to cognitive decline, but only sleep duration was assessed in our study. More longitudinal studies with a comprehensive assessment of sleep quality are needed to examine the joint associations of sleep quality and dietary patterns with cognitive decline. Thirdly, it may be biased because of self-reported sleep duration in our study, although this has been shown to be a reliable predictor of cognitive decline and dementia [10].
In conclusion, healthy patterns of diet and sleep in combination were associated with a lower rate of cognitive decline over time. These results highlight the importance of involving both behaviors as intervention priorities for the potential reduction of cognitive decline.
Footnotes
ACKNOWLEDGMENTS
This research uses data from China Health and Nutrition Survey (CHNS). We thank the National Institute for Nutrition and Health, China Center for Disease Control and Prevention, Carolina Population Center (P2C HD050924, T32 HD007168), the University of North Carolina at Chapel Hill, the NIH (R01-HD30880, DK056350, R24 HD050924, and R01-HD38700) and the NIH Fogarty International Center (D43 TW009077, D43 TW007709) for financial support for the CHNS data collection and analysis files from 1989 to 2015 and future surveys, and the China-Japan Friendship Hospital, Ministry of Health for support for CHNS 2009, Chinese National Human Genome Center at Shanghai since 2009, and Beijing Municipal Center for Disease Prevention and Control since 2011. The publication of this article was supported by Fundamental Research Funds of the State Key Laboratory of Ophthalmology. The funding source had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
