Abstract
While the role of diet and nutrition in cognitive health and prevention of dementia in older adults has attracted much attention, the efficacy of different dietary patterns remains uncertain. Previous reviews have mainly focused on the Mediterranean diet, but either omitted other dietary patterns, lacked more recent studies, were based on cross-sectional studies, or combined older and younger populations. We followed PRISMA guidelines, and examined the efficacy of current research from randomized controlled trials and cohort studies on the effects of different dietary patterns. We reviewed the Mediterranean diet, Dietary Approach to Stop Hypertension (DASH) diet, the Mediterranean-DASH diet Intervention for Neurodegenerative Delay (MIND) diet, Anti-inflammatory diet, Healthy diet recommended by guidelines via dietary index, or Prudent healthy diets generated via statistical approaches, and their impact on cognitive health among older adults. Of 37 studies, the Mediterranean diet was the most investigated with evidence supporting protection against cognitive decline among older adults. Evidence from other dietary patterns such as the MIND, DASH, Anti-inflammatory, and Prudent healthy diets was more limited but showed promising results, especially for those at risk of cardiovascular disease. Overall, this review found positive effects of dietary patterns including the Mediterranean, DASH, MIND, and Anti-inflammatory diets on cognitive health outcomes in older adults. These dietary patterns are plant-based, rich in poly- and mono-unsaturated fatty acids with lower consumption of processed foods. Better understanding of the underlying mechanisms and effectiveness is needed to develop comprehensive and practical dietary recommendations against age-related cognitive decline among older adult.
Keywords
INTRODUCTION
Dementia is a global concern, placing a significant financial and social burden on patients, carers, and health care systems [1, 2]. Cardiovascular risk factors, psychosocial factors, lifestyle behaviors, education and social networking, have been consistently linked to cognitive health among older adults [3, 4]. Approximately 30% of the population attributable risk for Alzheimer’s disease (AD), the most common cause of dementia, has been calculated to be determined by modifiable environmental factors [5]. In addition, the incidence of neurocognitive decline in the older population appears to be declining, suggesting a cohort effect with lifestyle factors having an impact, and diet may also be a promising strategy to postpone, slow, or prevent cognitive decline [6–12].
Despite research into the relationships between single nutrients or food with cognitive decline among older adults, it has been suggested that single nutrients or food are not as important as an integral dietary pattern [6, 8]. Within dietary patterns, the synergies and interactions between multiple nutrients and foods may play an important role to prevent or slow cognitive decline [6–9]. An important question is, whether and how effective different types of dietary patterns are in protecting against neurocognitive decline in older adults?
One of these dietary patterns, the Mediterranean diet originated among countries bordering the Mediterranean Sea in southern Europe. It is characterized by high intake of vegetables, fruits, olive oil, legumes, fish, whole grains, nuts and seeds, moderate wine consumption and low consumption of processed foods, dairy products, red meat, and vegetable oils [13]. A variety of tools to score adherence to the Mediterranean diet exist. Commonly used scoring systems are the 0–9 scoring system by Trichopolou et al. [14] or the 0–55 scoring system by Panagiotakos et al. [15]. The systems are similar in that they both score food component characteristics and include fruit, vegetable, legume, and alcohol intake. However, Trichopolou et al. used population sex-specific cut-offs around the median, while Panagiotakos et al. used pre-defined cut-offs based on frequency of consumption of foods relative to recommended amounts from the Mediterranean diet pyramid [15]. Furthermore, the 0–9 scoring system uses monounsaturated: saturated fat ratio (MUFA: SFA) while the 0–55 scoring system uses olive oil consumption. The 0–9 scoring system scores meat and meat products and fish as two components while the 0–55 system scores fish, poultry, red meat and meat products separately. Instead of non-specific intake of “cereal” and “dairy intake” in the 0–9 scoring system, the 0–55 scoring system redefined the component characteristics as “non-refined cereal” and “full fat dairy”.
The DASH diet, the dietary pattern proven to be effective in lowering blood pressure in patients with hypertension or those at high risk [16, 17], is defined as a diet with high consumption of fruits, vegetables, low-fat dairy products, whole grains, poultry, fish and nuts, and low consumption of red meat, fats, sweets, and sugary beverages. DASH requires low consumption of saturated fat but high consumption of low-fat dairy when compared to the Mediterranean diet [18].
The MIND diet, on the other hand, specifies intake of 10 brain healthy food groups, including berries, nuts, beans, whole grains, seafood, poultry, green leafy vegetables, other vegetables, wine, and olive oil. There is little emphasis on overall fruit or dairy intake, compared to the Mediterranean diet. The MIND diet also scores a low intake of 5 unhealthy foods (red meats, butter and margarine, cheese, pastries and sweets, fried/fast food) [19].
Overall, the Mediterranean, MIND, and DASH diets have been associated with lower plasma levels of inflammatory markers [20–22], which suggests diets impact inflammation and therefore may also indirectly affect cognitive health in older adults [23]. Inflammation has been viewed as an important risk factor of neurodegenerative diseases including cognitive impairment and dementia [24], as higher levels of circulating inflammatory markers, especially interleukin-6 (IL-6), tumor necrosis factor-α [TNF-α], and C-Reactive protein (CRP), are associated with brain atrophy and greater cognitive decline [25–28]. In general, anti-inflammatory diets include foods such as vegetables, legumes, fruits, whole grains, and seafood. These foods are naturally rich in vitamins, bioactive nutrients including antioxidants and poly-/mono- unsaturated fatty acids, which have been reported to reduce systemic inflammation [29, 30]. By contrast, an inflammatory diet is characterized by high consumption of red and processed meats, sweets, desserts, fries, and refined grains which may increase inflammation [31, 32].
The low GI diet, classified as food choices with glycemic index (GI) lower than 55 such as legumes, low GI whole grains and fruits, has been shown to assist in managing blood glucose and insulin levels [33].
Other dietary patterns that utilize data reduction methods (e.g., using principle components analysis (PCA), factor analysis, cluster analysis, or reduced rank regression) [34] have also been studied. An example is the Prudent healthy diet, characterized by high intake of vegetables and fruits, nuts, whole grains, fish, poultry, and low-fat dairy. The Prudent healthy diet is a healthier pattern, and contrasts with the Western diet which is characterized by high intake of red meat and processed foods, refined grains, high fat dairy, and high saturated/trans-fat [35–37].
There is growing evidence that the Mediterranean diet may protect against cognitive decline [38–41]. The MIND diet and the DASH diet, have also shown promising results [42]. On the other hand, the effects of diets such as the low GI diet, commonly used to treat diabetes [43], and to reduce cardiovascular risk factors [44, 45], on cognitive decline in the older population have been little studied. Similarly, there is limited research on the effects of the Anti-inflammatory diet, the Prudent healthy diet as opposed to the Western diet, and diets recommended by the World Health Organization or national peak bodies such as the Dietary Guidelines for Americans and the Australian Dietary Guidelines on cognitive health of older people [46–49].
Most reviews on the association of dietary patterns with cognitive function have been from studies focusing on the Mediterranean diet [38–41]; very few have reviewed the full variety of dietary patterns. The most recent published systematic review [42] covered Mediterranean, DASH, and MIND diets but only assessed cross-sectional and longitudinal cohort studies, and omitted randomized control studies (RCTs). No recent review has synthesized findings from studies with higher level evidence [50–52] including cohort studies and RCTs, nor examined the full range of dietary patterns and association with cognitive function.
The aim of this systematic review is to provide a comprehensive update on the topic by collating and evaluating the evidence from all human studies of RCTs and prospective cohorts conducted on a variety of dietary patterns and the outcome of cognitive function and/or dementia.
METHODS
Literature search
Process and search terms
We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [53]. An electronic literature search was conducted for articles published between January 1997 and September 2017. Databases PubMed, Medline, Cochrane Library, Embase, and Scopus were searched. Key search terms were Diet/ Dietary pattern/ Mediterranean diet/ DASH diet/ MIND diet/ low GI diet/ low fat diet/ low calorie diet/ healthy diet/ prudent diet/ anti-inflammatory diet/ Western diet AND cognition/ cognitive function/ memory/ cognitive decline/ dementia/ MCI and Alzheimer’s disease. Systematic reviews and meta-analyses were included. References were managed using the Endnote X8 referencing software. Papers and systematic literature reviews were hand searched for additional relevant studies.
Study selection
Abstracts, keywords, and titles were screened by XC, if unable to obtain adequate information then full texts were assessed. We included peer reviewed studies from level I and II levels of evidence ranking, i.e., RCTs and longitudinal studies according to NHMRC level of Evidence [51], if they measured cognitive function or brain morphology and provided follow up regardless of the time frame, to test cognitive function or incident cases of mild cognitive impairment (MCI) or dementia in older populations (>50 years at baseline). We excluded studies that were cross-sectional, not published in English, did not have full-text available, or used non-human participants.
Data extraction and quality assessment
Study and participant characteristics, exposure assessment, length of follow-up, confounders, cognitive assessment methods, key statistical results, and overall quality rating were inserted into a customized extraction table. An extraction table was generated for recent systematic reviews, included as a supplementary tables.
Data extraction and quality assessment were first assessed by a single reviewer (XC), cross checked by a second reviewer (FOL), with discrepancies discussed with a third reviewer (HB). Cochrane Risk of Bias tool and the SIGN 50 checklist [54] were used.
The Cochrane Risk of Bias tool [55], rates studies as having low, unclear, or high risk of bias (ROB) based on several criteria: random sequence generation, allocation concealment, selective reporting, blinding, and incomplete outcome data reporting and other biases.
The SIGN 50 checklist assesses selection, performance, attrition and detection bias, confounding and overall methodological quality in longitudinal studies [54]. This assessment tool has 14 questions targeting internal validity and four questions covering overall evaluation of quality. The questions prompt a “Yes”, “No”, or “Can’t Say” answer [56]. Overall, articles were scored Low Quality (–) if they had 1–6 “yes” scores, Acceptable (+) for 7–9 “yes” scores and High Quality (++) for 10–14 “yes” scores.
RESULTS
Study selection
Study selection flow chart is shown in Fig. 1 and PRISMA Checklist is provided as Supplementary Material. From 1,765 articles obtained, 37 studies were eligible for inclusion. Study characteristics and outcomes can be found in Tables 1 and 2. Thirty-one were cohort studies emanating from the US (n = 17), France (n = 2), Australia (n = 3), UK (n = 2), Italy (n = 1), Sweden (n = 2), Asia (n = 2), Greece (n = 1), and Spain (n = 1). Six studies were RCTs; they came from Spain (n = 2), Finland (n = 1), Australia (n = 1), US (n = 1), and Hong Kong (n = 1). Most papers (5/6 for RCTs and 31/31 for cohorts) assessed community-dwelling older adults.
Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram.
Quality assessment of RCTs included in the review (Cochrane Risk of Bias Tool)
Characteristics of RCTs included in the review
AD, Alzheimer’s Disease; APOE, Apolipoprotein E; BMI, body mass index; CI, confidence interval; CDT, the Clock Drawing Test; CVD, cardiovascular disease; EVOO, extra virgin olive oil; FFQ, Food frequency questionnaire; GI, glycemic index; HTN, hypertension; HDL, high-density lipoprotein; LDL, low-density lipoprotein; MCI, mild cognitive impairment; Med Diet, Mediterranean diet; MMSE, Mini-Mental Status Examination; PREDIMED, the PREvención con DIeta MEDiterránea study; RAVLT, Rey Auditory Verbal Learning Test; WAIS-IV, the Wechsler Adult Intelligence Scale.
A priori and a posteriori studies
Most cohort studies included in this review used a priori (n = 26) approaches by assessing dietary adherence scores to a specific dietary pattern based on consumption of key food components. Patterns included the Mediterranean diet, which was the most studied dietary pattern with three RCTs (two conducted in the Mediterranean area) and 19 cohort studies (4 conducted in Mediterranean countries) using 0–9 [14] or 0–55 scoring system [15]. Other a priori studies were the DASH diet (n = 3) using 8–40 [37] or 0–10 DASH scoring system [57], and the MIND diet (n = 3 studies) using a 0–15 scoring system [58]. Other less studied patterns were the Anti-inflammatory diet (n = 2), the Low GI diet (n = 1), and the Healthy diet recommended by dietary guidelines (n = 7) (see Fig. 2).
By contrast, only seven studies (n = 2 overlapped with a priori group as reported both a priori and a posteriori patterns) used a posteriori approaches to investigate dietary profiles of a target population (see Fig. 2), including studies on the Prudent healthy diet compared to the Western diet (n = 3 using PCA [59] or factor analysis [35, 36]), and wheat based diet (n = 1, using PCA [60]). Other studies used reduced rank regression [61], cluster analysis [62], and factor analysis [63] to derive population-specific dietary patterns (see Table 3 for details).
Dietary patterns and study outcomes.
Characteristics of cohort studies in this review
3MS, Modified Mini mental State Examination; AD, Alzheimer’s disease; AHEI, Alternate Healthy Eating Index; APOE, Apolipoprotein E; BMI, body mass index; CHO, carbohydrate; CVD, cardiovascular disease; CI, confidence interval; CRP, C-reactive protein; DASH, Dietary Approach to Stop Hypertension; DSM, the Diagnostic and Statistical Manual of Mental Disorders; DSM-III-R, the Diagnostic and Statistical Manual of Mental Disorders, Revised Third Edition (DSM-III-R); FFQ, Food frequency questionnaire; GI, glycemic index; HEI, Healthy Eating Index; HDI, Healthy Diet Indicator; HTN, hypertension; HR, hazard ratio; MMSE, Mini-Mental Status Examination; Med Diet, Mediterranean diet; MCI, mild cognitive impairment; MIND, the Mediterranean-DASH diet Intervention for Neurodegenerative Delay; MDS, Mediterranean Diet Score; MSDPS, Mediterranean-Style Dietary Pattern Score; NINCDS-ADRDA, National Institute of Neurological and Communicative Disorders and Stroke-Alzheimer’s Disease and Related Disorders Association; OR, odds ratio; PCA, principal component analysis; PD, probable dementia; RFS, Recommended Food Score; SD, standard deviation; SFFQ, semi-quantitative food frequency questionnaire; SMMSE, Standardized Mini-Mental State Examination; TIA, transient ischemic attack; TICS, Telephone Interview for Cognitive Status.
Overall eight articles published results of more than one dietary pattern from the same cohort, and these are discussed separately under the appropriate dietary patterns.
Exposure and outcome assessment
Nutritional assessment tools to determine dietary pattern exposure varied. The most commonly used were the semi-quantitative Food Frequency Questionnaire (FFQ), or a food record or 24-h recall, either self-administered or researcher-administered, for both RCTs and cohort studies.
Most studies focused on cognitive change and used the Mini-Mental State Examination (MMSE) score or a neuropsychological test battery. Tests generally covered a broad range of cognitive skills including language, memory (short-term and working), visual perception, executive function, cognitive flexibility, global cognitive function, and cognitive processing. In detail, memory tests included recognition, immediate recall, delayed recall, face-name recall, paired associates, and semantic memory. Executive function tests included working memory, verbal fluency, reasoning, attention, and processing speed. Composite measures of episodic memory (e.g., immediate and delayed recall) were also included. Global cognition was measured using composite measures of cognitive function. Studies also investigated brain morphology using magnetic resonance imaging, or diagnosis of incident cases of MCI or dementia as assessed by neurologists or neurophysiologists.
Quality assessment
Except for one study with unclear risk of bias, all other RCTs received a high risk of bias due to the long-term nature of nutrition interventions and impossibility of complete double blinding. Among longitudinal studies thirteen received “High Quality” with remaining nineteen articles assessed as “Acceptable” (see Table 1 for results from the Cochrane Risk of Bias and table 3 and Supplementary data for the SIGN 50 quality assessment).
Dietary patterns and cognitive health: A priori patterns
Mediterranean diet
RCTs in Mediterranean countries: Two RCTs which investigated the Mediterranean diet in Spain and reported positive outcomes, targeted those at high cardiovascular risk but without cardiovascular disease at baseline [64, 65], from the multicenter randomized prevention trial PREDIMED (PREvención con DIeta MEDiterránea) study. Both studies had high risk of bias due to the difficulty of blinding.
The earlier PREDIMED study [64] reported a positive link between Mediterranean diet supplemented with either extra virgin olive oil (EVOO) 1 L/week or raw mixed nuts (walnuts, hazelnuts, and almonds) 30 g/day and cognition which was assessed by MMSE and Clock Drawing Test (CDT) after 6.5 years, as well as a lower incidence of MCI and dementia compared to a control diet. In the more recent clinical trial conducted in a subcohort from PREDIMED, two intervention groups followed the Mediterranean diet supplemented with either EVOO (1 L/week) or mixed nuts (30 g/d) [65]. They were compared to a control group of those receiving dietary advice on a reduced fat diet. After a median follow up of 4.1 years, participants allocated to a Mediterranean diet plus EVOO scored significantly better than controls on the Rey Auditory Verbal Learning Test (RAVLT) and Colour Trail Test part2. The Mediterranean diet plus nuts group had significantly improved memory composite, while Mediterranean diet plus EVOO group had significantly better performance of the frontal and global cognition composites than control group.
Cohort studies in Mediterranean countries: Four cohort studies evaluated the Mediterranean diet and were conducted in the Mediterranean area (France n = 2, Spain n = 1, and Greece n = 1). All had acceptable study quality, and all supported a positive link between Mediterranean diet and better cognitive health in older adults.
Among those, three studies used MMSE or Telephone Interview of Cognitive Status (TICS) to measure cognition. The Three-City study reported that a higher Mediterranean diet score was associated with better MMSE performance but not with other cognitive tests on verbal/visual function or with incidence of dementia [66]. Similarly, The European Prospective Investigation into Cancer and Nutrition (EPIC) [67] reported an inverse association between adherence to the Mediterranean Diet and decline in MMSE scores. The Spanish SUN (Seguimiento Universidad de Navarra) cohort study [68] reported greater cognitive decline (as assessed by the TICS) in participants with low or moderate adherence to the Mediterranean Diet compared to those with higher adherence.
In the SU.VI.MAX (Supplementation with Vitamins and Mineral Antioxidants) study, lower backward digit span and lower phonemic fluency scores were associated with poorer adherence to Mediterranean diet over 13 years [70], no effect was observed on composite cognition score.
RCTs in non-Mediterranean Western countries: The only RCT completed in a non-Mediterranean area was the Australian Medley study [71], which reported no evidence that Mediterranean diet benefited cognitive function compared to a habitual dietary intake after 6 months.
Cohort studies in non-Mediterranean Western countries: Fourteen cohort studies were undertaken in non-Mediterranean western countries, the majority from the US (n = 11). Five studies reported little to no association of adherence to the Mediterranean diet with cognitive decline [72–76]. However, the majority, nine studies [35, 77–84] provided evidence of statistically significant positive associations between the Mediterranean diet and the protection of cognitive health.
Among those studies finding no effect, most received “High Quality” (n = 3) while the rest were “acceptable” (n = 2). Two studies that measured cognitive function using a cognitive test battery, were the Women’s Antioxidant Cardiovascular Study (WACS) [72] which followed 2,504 participants with vascular disease or coronary risk factors for five years, and the US Nurses’ Health Study which followed 16,058 stroke free female nurses at baseline for six years [74]. Three of these five studies looked at incidence of MCI or dementia as the primary outcome, including the PATH Through Life study in Australia which followed 1528 community dwelling participants for 4 years [73], the Uppsala longitudinal study that followed older men for 12 years [75], and the US Women’s Health Initiative Memory (WHI) study which followed cognitively intact (at baseline) post-menopausal white women aged more than 65 years for 9 years [76]. Most results of the five studies were adjusted for apolipoprotein E (APOE) ɛ4 genotype except for Nurses’ Health Study [74] and WACS [72].
By contrast, nine cohort studies, of which five received “acceptable” study quality and four were ranked as “high quality”, revealed statistically significant associations that the Mediterranean diet protected against cognitive decline in older adults.
Four of the nine studies assessed cognitive decline using either cognitive tests averaged for composite measure of global cognition [83], or modified MMSE (3MS) [79], a comprehensive neuropsychological battery [35], or a Subjective cognitive function questionnaire [77]. The Chicago Health and Aging Project (CHAP) reported higher Mediterranean diet scores were associated with slower rates of cognitive decline after an average 7.6 years follow up on 3790 participants aged 75.4 years on average at baseline [83], while The Health, Aging and Body Composition study in the US [79] reported varied findings among racial groups and positive effects from the Mediterranean diet was only observed among black but not white participants. By contrast, the Australian Imaging Biomarkers and Lifestyle study of Ageing reported no significant effects in subjects without MCI or AD at baseline, except for APOE ɛ4 allele carriers, in whom higher Mediterranean diet scores were associated with less decline in executive functioning after 36 months [35]. The (male) Health Professionals’ Follow-up study [77] reported that long-term adherence to the Mediterranean diet was strongly linked to lower subjective ratings of change in cognitive function. A limitation of this study is that cognitive function results had relied on subjective self-reporting of six “yes” or “no” questions, rather than performance-based methods [77].
Among the nine studies, five assessed incidence of MCI or AD as the clinical outcome [78, 84]. The US study Washington/Hamilton Heights-Inwood Columbia Aging Project (WHICAP) 1992 and WHICAP 1999 [78, 82] found that higher adherence to the Mediterranean diet was associated with reduced risk for developing MCI or AD. Similar conclusions were drawn from The Reasons for Geographic and Racial Differences in Stroke Study (REGARDS) which followed up 17,478 participants, mean age 64.4 years and cognitively intact with no history of stroke at baseline, for 4 years [84], and reported high adherence to Mediterranean diet was associated with a lower likelihood of incident cognitive impairment in nondiabetic but not diabetic individuals. Moreover, in the Rush Memory and Aging Project (MAP), the highest tertile of Mediterranean diet scores were found to have lower AD rates [80].
Cohort study in Asian countries: Only one longitudinal cohort study conducted in Asia [60] with acceptable study quality found a benefit for a Mediterranean-like diet, modified to suit local eating habits (Table 5). The diet shared similar characteristics with the Mediterranean diet such as high consumption of fruits, vegetables, and grains, as well as low consumption of meat and dairy foods. Qin et al. reported that among the 1,650 community dwelling persons more than 65 years old followed for 5.3 years, those in the highest tertile had a slower rate of cognitive decline than people in the lowest tertile of the adapted Mediterranean diet.
DASH diet
The three studies that examined the effect of adherence to the DASH diet on cognitive health in later life reported mixed results [18, 80]. The majority of the studies received a “high quality” rating (n = 2).
Two studies reported positive effects from the DASH diet [18, 80]. The Rush Memory and Aging Project followed-up 923 elderly men and women mean aged 81.4 years old, and reported modest but positive links between the highest tertile of DASH diet adherence and lower rates of AD [80]. The Nurse’s Health Study computed a long-term DASH score from five previous dietary assessments and reported greater adherence was associated with better composite scores of global cognition and verbal memory irrespective of APOE ɛ4 [16]. However, during the next six years, no association between DASH scores and change in cognition were found [18]. By contrast, the Women’s Health Initiative Memory Study (WHIMS) [76] reported that DASH scores were not associated with incidence of MCI or dementia in older women generally or in those with hypertension. Various cut-off scores for highest quintiles were selected including 28 [76] and 31 [18], respectively, for the 8–40 DASH scoring system [37] and 5 [80] was used for 0–10 DASH scoring system [57].
MIND diet
Three cohort studies, all with high study quality, found protective effects of the MIND diet on cognition.
The Rush Memory and Aging Project reported that moderate adherence to the MIND diet was associated with lower rates of AD [80], slower decline in a global cognitive score and five cognitive domains (episodic memory, semantic memory, perceptual organization, perceptual speed, working memory) [58]. The difference between highest and lowest quintiles in MIND scores was calculated to be equivalent to being 7.5 years younger in cognitive health [58, 80]. In addition, the US Nurse’s Health Study [19], where 16,058 older women aged 70 and over were followed up for 6 years with multiple assessments of dietary intake and cognition, reported long-term adherence to the MIND diet was moderately associated with better verbal memory in later life, but not with global cognition, verbal memory or TICS. All studies adjusted for multiple covariates including APOE ɛ4 and used the same 0–15 MIND scoring system [58], but had different MIND score cut-offs for highest quintiles [58, 80].
Anti-inflammatory/inflammatory diet
Two studies that investigated dietary patterns associated with inflammation, characterized by high consumption of foods such as red and processed meats, sweets, deserts, chips, and refined grains [31, 32], found significant impacts on cognitive function; however, divergence in results were reported for different age groups [23, 85]. Among the two studies, one was of high quality while the other received an “acceptable” quality assessment rating.
The Whitehall II cohort study [23] followed 5,083 participants whose median age was 56 years for 10 years. They reported higher intake of an inflammatory dietary pattern was associated with accelerated cognitive decline. The greatest decline in global cognition and reasoning was found in the highest tertile of inflammatory diet when compared to the lowest tertile. In an age-stratified analysis, higher inflammatory scores were linked to significantly faster cognitive decline in reasoning in the age <56 group, while no significant association was found among those aged 56 and older.
The WHIMS researched the inflammatory pattern [85] by following up 7,085 women aged at 65–79 years for 9.7 years with annual cognitive function assessments. They found that higher dietary pro-inflammatory scores were associated with greater cognitive decline and earlier onset of MCI, after adjustment for multiple covariates. This study suggested the existence of a possible threshold effect.
Healthy diet recommended by dietary guidelines
Two RCTs [86, 87] (both with high risk of bias) and five cohort studies (high quality n = 2 and acceptable quality n = 3) investigated diets complying with the dietary guidelines of national peak bodies or the World Health Organization’s Healthy Diet Indicator tool. Mixed results were reported and three studies [86, 89] found associations with cognitive benefits.
The Finnish Geriatric Intervention Study to Prevent Cognitive Impairment and Disability (FINGER), a randomized 2-year multi-domain lifestyle intervention trial [86, 90], assessed cognitive decline in 1,260 participants aged 60–77 years at baseline. This study reported cognitive benefits in groups which received dietary intervention together with exercise, cognitive training, and monitoring of vascular risk. In FINGER, dietary counselling was based on the Finish Nutrition Recommendations and a composite dietary intervention adherence score was generated based on the consumption of fruit, vegetables, fish, whole grain cereals, low fat milk, and meat products, as well as the limiting of sucrose intake and using vegetable margarine and rapeseed oil instead of butter [90] (see Table 2 for details).
A Hong Kong RCT [87] included 419 participants with mean age of 83 years living in old age hostels, and provided nutrition intervention including counselling and menu change, with a mean follow up time of 25 months. They reported that although nutrition interventions were effective in maintaining fruit and fish intake compared to control group, this did not result in a significant reduction in cognitive decline.
Four studies researched the effect of adherence to the dietary guidelines for Americans, using either the Recommended Food Score (RFS) [88], modified Alternative Healthy Eating Index (mAHEI) [89], HEI (Healthy Eating Index) -2005 [83], or HEI-2010 [76, 89]. The Cache County Study on Memory and Aging in US [88] evaluated effects on cognition of adherence to the RFS, which was developed to assess diet quality and food variety related to the dietary guidelines for Americans [91]. After 11 years follow up, those in the highest RFS quartile declined by 3.4 points compared with the 5.2 point decline in those in the lowest RFS quartile [88]. With regards to the mAHEI, lower risk of cognitive decline was found among those in the healthiest dietary quintile, compared with lowest quintile, regardless of baseline cognitive level [89].
No association with cognition change was observed in the CHAP study using HEI-2005 scores [83]. Similarly, the WHIMS reported no significant relationship between adherence to HEI-2010 or Alternate HEI-2010, with incidence of MCI or probable dementia in older women [76].
One of the five studies, the Uppsala longitudinal study, examined the HDI by following 1038 older men over 12 years and found no association between HDI and any of the cognition outcomes [75].
Low GI diet
Only one RCT (unclear risk of bias) examined the effect of GI on cognition by comparing healthy older adults or those with amnestic MCI allocated to high saturated fat, high GI or low saturated fat, low GI diets [92]. Both cognitive healthy and amnestic MCI participants had better scores on delayed visual recall with the low saturated fat, low GI combined diet, although not in immediate memory, executive and motor speed domains. However, a major limitation of this study is the small sample size (n = 49) and short intervention time (4 weeks) and the combination of effect from saturated fat and GI.
Dietary patterns and cognitive health: A posterori patterns
Prudent healthy diet versus Western diet
Findings from three cohort studies (high quality n = 2 and acceptable quality n = 1) of the Prudent healthy diet (see Table 3) reported mixed findings [35, 59].
Two studies used factor analysis (principle components) to extract dietary patterns [35, 36]. The Australian Imaging, Biomarkers and Lifestyle (AIBL) Study of Ageing [35] reported no significant relationships between adherence to a computer generated Prudent healthy diet and cognition, measured by a comprehensive battery of neuropsychological tests either for global or single domain scores (see Table 3 for details) [93]. By contrast, The Swedish National study on Aging and Care-Kungsholmen (SNAC-K) [36] found that the highest adherence to their Prudent healthy diet was related to less MMSE decline, and highest adherence to a Western dietary pattern was associated with greater MMSE decline. The decline associated with the Western diet was attenuated when accompanied by higher adherence to Prudent healthy diet. However, as these diet patterns were computer generated and based on the subjects consumption patterns, the definition of the Prudent healthy diets differed between studies with AIBL emphasizing nuts, tomatoes, potatoes and garlic as single food groups whereas SNAC-K study defined a Prudent healthy diet by the inclusion of cereals legumes, rice/pasta, water and cooking/dressing oil [35, 36].
In addition, the first and only human longitudinal study that investigated the impact of diet on changes in magnetic resonance imaging (MRI) over time is the Personality and Total Health Through Life Study, which focused on a subsample of 255 participants aged 60–64 years at baseline with two MRI scans 4 years apart. This study reported that a Western diet, characterized by lower intakes of naturally nutrient-dense foods and higher intakes of unhealthy foods was associated with smaller left hippocampal volume [59].
Wheat-based diverse diet
The China Health and Nutrition Survey, with acceptable study quality, assessed a wheat-based diverse diet, derived by factor analysis. The top tertile was associated with slower annual decline in global cognitive function [60] among adults 65 years and older. The wheat based diverse diet shared some features of the Mediterranean diet, and was characterized by high intakes of wheat buns, deep-fried wheat, nuts, fruits, moderate- to high-fat red meat, poultry and game, egg, fish, dairy, sugar, vinegar, soy sauce, plant oil, and with low intake of animal-source cooking fat.
Others
Three cohort studies [61–63] investigated the impact of other dietary patterns from a posteriori statistical approaches, and reported mixed results. Overall dietary patterns that shared features of a typical Western diet were linked to greater cognitive decline consistently across all studies.
Among the prospective cohort studies, Gu et al. reported in 2010 that the dietary pattern which was found to be protective against the development of AD had higher consumption of salad dressing, nuts, fish, tomatoes, poultry, cruciferous vegetables, fruits, dark and green leafy vegetables, with lower intake of high-fat dairy, red meat, organ meat, and butter [61]. Dietary patterns were derived by reduced rank regression, using predetermined food groups as predictor variables and seven potentially AD-related nutrients as response variables (see Table 3 for details).
In the UK Newcastle 85+ cohort study, dietary patterns (DPs) derived by cluster analysis that differed in intake of red meat, potato, gravy, and butter, were studied for their effect on cognition [62]. When compared with DP2 group (low meat intake, high intake of fruits, vegetables, fish, nuts, whole grains and dairy), men in DP1 group (high red meat, gravy and potato dishes, low in butter) scored worse in initial attention and MMSE, and DP3 (high butter intake and low in unsaturated fats, moderate intake of red meat) was associated with a 3.2-fold increased risk of cognitive decline despite APOE ɛ4 status. Both DP1 and DP3 had overall worse concentration and focused attention.
A prospective cohort study in Taiwan researched three dietary patterns characterized as vegetable, meat based, or traditional. High score “traditional” pattern was found to protect against decline in logical memory-recall I [63]. Mixed results were reported on the other two patterns: the moderate or high score “vegetable” pattern was significantly associated with less decline of logical memory, although high score “vegetable” pattern was linked to greater decline in executive function. A high-score “meat” pattern was related to decline of verbal fluency, but protection against attention decline [63].
DISCUSSION
Of studies included in this review, the Mediterranean diet was the most investigated with evidence supporting protection against cognitive decline among older adults. Research on other diets such as the MIND, DASH, and Anti-inflammatory diets was more limited but showed promising beneficial results, especially for the MIND diet which received a high quality rating for all three studies [94].
Within studies on the Mediterranean diet, there were differences in outcomes between Mediterranean and non-Mediterranean countries. This suggests that the effects of diet on cognition are complex and likely to vary across geographic, cultural, or sociodemographic contexts [95–98]. For example, cultural values and lifestyle in the Mediterranean area, such as social connection and sense of community, meals being traditionally cooked at home using slower cooking methods such as boiling and stewing, enjoyed slowly and mindfully, meal times shared with family and friends rather than rushing through the meals or eating in front of a screen [8], may partly explain the discrepancies observed in studies. The supportive evidence of a modified Mediterranean diet in Asian countries, suggests the possibility of adapting the principles of a Mediterranean diet to suit local foods, culture, and eating habits [60].
Discrepancies of results in western countries may result from cultural and demographic diversity of the populations being studied [95]. For example, the Australian RCT Medley study, while based on a multicultural population, included only 70 of 137 (51%) participants born locally, and the comparison habitual diet may not have been a typical Western diet. Secondly, more highly educated participants (such as in [72, 99]), are more likely to be living a healthier lifestyle and perform better on cognitive tasks. Many studies controlled for education and lifestyle factors but residual confounding and the impact of clustering of health behaviors may remain. Cognitive decline may be harder to detect in these groups requiring sensitive cognition tests and longer follow up [71]. Studies that excluded people with potential underlying health issues, targeting generally healthy participants [74, 79] may be less likely to detect a protective effect from the Mediterranean diet on cognition than those that included participants at risk of cardiovascular disease, as modification of cardiovascular risk factors may alter rates of cognitive decline [76]. In summary, differences in protective outcomes with the Mediterranean diet may result from differences in populations with respect to their culture or education, general health, level of physical activity, and specifically their risk of cardiovascular disease, and baseline cognition.
One potential mechanism by which the Mediterranean diet may protect against cognitive decline is through improving vascular health and preventing cardiovascular diseases due to its richness in poly-/mono- unsaturated fats [100]. A diet rich in poly-/mono-unsaturated fat improves insulin sensitivity, has an anti-diabetic effect [101–103] and lowers cardiovascular disease risk. Anti-diabetic effects of diet may also benefit cognition by maintaining relatively stable brain glucose levels; even small changes in glucose levels can alter metabolic homeostasis, which has been consistently linked to insulin resistance and cognitive impairment in older individuals [33, 104]. In support of the impact of fat quality, the positive association between the Mediterranean diet and cognition disappeared when monounsaturated fat: saturated fat ratio (MUFA:SFA) was excluded from the Mediterranean diet scoring system [105], suggesting that there is an important role of high MUFA:SFA ratio for protection of cognition [68].
The positive link between the Mediterranean diet and cognition protection may be also due to the higher consumption of antioxidants, as brain oxidative stress is associated with neurodegeneration [38]. Antioxidants in Mediterranean diets include phenolic compounds, and anti-inflammatory agents such as omega 3 fatty acids. Importantly urinary polyphenol excretion, a biomarker of adherence to the Mediterranean diet [106], has been associated with better memory, indicating the likelihood that phenolic compounds may benefit cognition in older adults [107]. In addition, EVOO consumption, compared to regular olive oil and other vegetable oils [108, 109], has resulted in significantly lower plasma inflammatory markers and increased anti-oxidant capacity [110, 111]. This is consistent with the PREDIMED study [64, 65] findings that a Mediterranean diet plus EVOO, or nuts, resulted in better cognition performance, indicating that the quality of oil as determined by the level of anti-oxidant and anti-inflammatory agents is important. It remains unclear as to what is an adequate therapeutic amount of EVOO and how long it needs to be taken to protect against cognitive decline, despite a recent trial reporting that 12 months of EVOO (26 g) to replace all vegetable oil (olive oil, high-oleic safflower oil, high-oleic sunflower oil, canola oil and hydrogenated vegetable oils) in Mediterranean diet may benefit-cognition [112].
Likewise, long-term inflammation might damage the blood-brain barrier leading to cognitive impairment, and increase the risk of neurodegenerative diseases [24], as higher levels of circulating inflammatory markers are associated with greater cognitive decline [113, 114]. Anti-inflammatory diets were associated with slower cognitive decline in older adults, however, the effect varied with age [23, 85]. Ozawa et al. reported a significant protective effect only in those under 56 years [23] while Hayden et al. reported protection against cognitive decline for those above 65 years [85]. Contributing factors to the difference may include different research methods such as selection of inflammatory diet scoring system. While one is based on specific food-based loading factors associated with one inflammatory marker [23], the other used the DII (Dietary Inflammatory Index) system, which includes a set of eight pro-inflammatory nutrients, 19 anti-inflammatory nutrients, and “10 whole foods and spices, caffeine, flavones, flavonols, flananones, anthocyanidins and isoflavones” [32, 85]. Selection of cut-points for tertiles may also make a difference as a possible threshold effect was suggested [85].
Similarly, the benefits of MIND and DASH diets could also be related to their richness in mono-/poly-unsaturated fats, anti-oxidant, anti-inflammatory, and anti-diabetic effects. The low GI diet has also been commonly used for its anti-diabetic effects [33]. Additionally, the DASH diet is effective in managing hypertension [17], which may confer greater benefit in older people at higher cardiovascular risk [16]. Interestingly, when comparing low-fat dairy intake, the DASH diet requires “moderate to high consumption, 2-3 serves daily”, while the Mediterranean diet requires “low consumption”. The role of low-fat dairy with a Mediterranean diet remains uncertain, and differences in dairy intake might contribute to inconsistent results from the Mediterranean diet in western countries when compared to DASH diet, thus further research on the role of low-fat dairy on cognitive health is required [115]. When compared to the Mediterranean diet, the MIND diet emphasizes berry and nut intake, as well as scoring unhealthy food groups, and whilst it has shown consistent protective effects [58, 80], this requires further investigation.
Divergent results from research on the Prudent healthy diet may be due to the nature of a posteriori studies, as dietary patterns were generated by data reduction methods, based on different populations and different subjects consumption patterns. Other factors may include dietary assessment tools and differences in cut-off and factor loadings selected [35, 36]. In brain imaging studies, longer follow-up time may be needed particularly to separate dietary and aging effects on brain atrophy [59].
Encouragingly, the potential detrimental effects on cognition in older adults from unhealthy foods in a Western diet such as fast food meals, sugary drinks, and fatty snacks may be corrected greatly by adding healthier options such as fruits, vegetables, nuts, and whole grains [36]. This suggests the importance of more healthy dietary patterns with foods naturally rich in nutrients, even when Western diets have been mostly followed previously.
Overall, the dietary patterns that were found to be protective against cognitive decline, are plant-based, rich in poly- and mono-unsaturated fatty acids with reduced consumption of processed foods. As we focused on dietary patterns, a comprehensive review of the effects of single food groups on cognition among older adults is outside the scope of this article. However, several studies of dietary patterns further analyzed their results to determine which components may be key to cognitive protection, and mixed results have been reported. For example, vegetable consumption, compared to other Mediterranean food components [105], was shown to have a significant inverse association with cognitive decline [67], which is in line with earlier research on vegetables and cognition [116, 117]. On the other hand, moderate alcohol consumption and a high ratio of MUFA:SFA were also found to have a weak association with cognition. Of DASH diet components, vegetables, nuts, and legumes appear to be the key components [18]. Examination of MIND food group components and their impact on outcomes demonstrated that excluding high saturated fat components attenuated the association between the MIND score and verbal memory, perhaps because less saturated fat results in a higher MUFA:SFA [19]. Relationships of other foods with anti-inflammatory effects and cognitive health were also studied [116–121]. A link between fermented food and pickles and protection against logical memory decline is currently under investigation [63]. The effect may be seen as an anti-inflammatory effect [122], or explained by anti-oxidants and possibly probiotics like lactic acid bacteria in pickled foods, highlighting the need for further research on the relationship between probiotics, gastrointestinal health, and cognition [63].
We identified limitations common to many studies. Firstly, long-term medications, e.g., hypolipidemic, anti-hypertensive, anti-diabetic, and anti-cholinergic drugs, were only adjusted for in a few studies [64, 65]. Secondly, most studies only assessed dietary intake at baseline, which introduces performance bias, as dietary intake may change over the years of a trial, due to a change of eating habits secondary to medications, influences from society, friends, and family members and high risk for or diagnoses of medical conditions such as diabetes, hypertension, and cardiovascular disease [72]. In some studies, cognitive function was not assessed at baseline precluding assessment of the impact of the intervention on cognitive change over the years, although inferences may be drawn from looking at differences in cognitive function between groups at follow up [64]. Thirdly, for those with shorter follow up times the findings may reflect reverse causality, as cognitive impairment and AD are preceded by relatively long periods of subclinical cognitive decline which could influence eating patterns [70]. Fourthly, only 20 of the 38 studies adjusted for the APOE ɛ4 allele which is a major genetic risk factor for more rapid cognitive decline and earlier onset of dementia. Fifthly, it is not possible to adjust for all known risk factors, and future studies should consider including physical activity levels, obesity, diabetes, smoking, and alcohol as covariates. Sixthly, there are no standard cut-offs for diet adherence scores, for example, Mediterranean diet adherence scores. As tertiles were used, cut-off points varied [19, 78–80]. Future research is needed to determine the cut-off points for the level of adherence required to generate an effect [60, 79]. Seventhly, a common limitation among all RCT studies is, the single blinded nature, as due to the nature of nutrition trials using foods, it is not feasible to double blind RCTs for long term studies into nutrition interventions. Finally, there is limited evidence from longitudinal or intervention studies on associations between dietary patterns and neuro imaging outcomes [59]. Continuing research with longer follow up of older people may provide insight and better understanding on the changes in brain morphology, activity and function, in addition to cognitive function tests outcomes.
Future trials and observational studies should investigate the effectiveness of food components as well as the whole diet, take APOE ɛ4 genotype into consideration as well as multiple covariates, such as gender, age group (from >50 years to >85), medical conditions, gastrointestinal health, and report use of medications or nutrition supplements due to chronic conditions. Validated dietary assessment tools should be employed multiple times throughout long follow ups with analysis on dietary change over the years, and a comprehensive cognitive battery and neuro imaging included for outcome assessment.
Conclusion
This review adds to previous reviews in that it includes for the first time, higher level of evidence from six RCTs and 31 cohorts which were classified into different dietary patterns and investigated the potential effect of different diets on cognitive function among older adults. Overall, the findings support positive relationships between dietary patterns which are plant-based, rich in poly/monounsaturated fatty acids with reduced consumption of processed foods and cognitive health in older adults. More research is required for better understanding of the underlying mechanisms and effectiveness, in order to develop comprehensive and practical nutrition interventions and dietary recommendations to protect against cognitive decline and dementia with aging.
