Intake of fruit and vegetables and the incident risk of cognitive disorders: A systematic review and meta-analysis of cohort studies

Abstract

Objectives

No quantitative assessment has been performed to specifically link the consumption of fruit and vegetables with the incident risk of cognitive disorders.

Methods

We searched the PubMed and the Embase databases (both from the inception to June 13th, 2016) for records that report the intake of fruit and vegetables and the risk of developing cognitive disorders (Alzheimer's disease, dementia, and cognitive decline/ impairment). A generic inverse-variance method (random-effects model) was used to combine the relative risks (RRs) and 95% confidence intervals (CIs). To explore the potential sources of heterogeneity, we performed the subgroup and meta-regression analyses by pre-specified characteristics.

Results

We identified 6 cohorts involving a total of 21,175 participants. The pooled analysis showed that consumption of fruit and vegetables was inversely associated with the incident risk of cognitive disorders, and the pooled RR (95% CI) was 0.74 (0.62, 0.88), with evidence of significant heterogeneity (I² =68%). Furthermore, we found that the significant heterogeneity might be attributed to the ethnic difference.

Conclusion

Further large prospective studies should be performed to quantify the potential dose-response patterns of fruit and/or vegetables intake and to explore the role of fruit or vegetables consumption separately on cognitive disorders in different populations.

Abbreviations

AD

Alzheimer's Disease

FFQ

food-frequency questionnaire

DSM

Diagnostic and Statistical Manual of Mental Disorders

NINCDS-ADRDA

National Institute of Neurological and Communicative Disorders and Stroke-Alzheimer's Disease and Related Disorders Association

FAQ

Functional Activities Questionnaire

DECO

Observed Cognitive Deterioration

NOS

Newcastle-Ottawa Scale

RR

relative risk

HR

hazard ratio

OR

odds ratios

95% CI

95% confidence interval

Introduction

With the acceleration of the aging process, cognitive disorders (Alzheimer's disease, dementia, cognitive decline, and cognitive impairment) have become important public health issues around the world (1, 2, 3). Matthews et al. estimated that 8.3% of the individuals aged 65 years or older would be projected to have dementia in 2011 in 6 areas of England and Wales (4). The effective treatment of neurodegenerative disease is limited, and thus the modifiable environmental elements, such as dietary patterns, sleeping habits, and tobacco and alcohol consumption are considered as important strategies in the prevention of cognitive disorders (5, 6, 7, 8, 9).

In order to prevent the progressing of the chronic noncommunicable diseases, the World Health Organization (WHO) has promoted the individuals to daily consume 5 portions of fruit and vegetables (“Five a day”) since 1990 (10). In recent decades, accumulating evidence of meta-analyses has demonstrated that higher intake of fruit and vegetables may cause a protective role against a large number of chronic diseases, including hypertension, type II diabetes mellitus, cardiovascular disease (CVD), and some cancer (11, 12, 13, 14, 15, 16, 17). However, no quantitative assessment has been performed to specifically link the consumption of fruit and vegetables with the risk of cognitive disorders (18, 19). A meta-analysis by Cao et al. investigated the association of dietary patterns and dementia, but the pooled analysis showed non-significant result involving the intake of fruit and vegetables (20).

Previous epidemiological studies have explored the association between intake of fruit and vegetables and the risk of cognitive disorders (21, 22, 23, 24, 25, 26), but the conclusions remain inconsistent. Therefore, a systematic review and meta-analysis was conducted to summarize the accumulating evidence from prospective cohort studies on the relationship of fruit and vegetables consumption with the incident risk of cognitive disorders (Alzheimer's disease, dementia, cognitive decline, and cognitive impairment).

Materials and methods

Literature search

The present study was conducted according to the standardized guidelines (27, 28). We searched the PubMed and the Embase databases (both from the inception to June 13th, 2016) for records reporting the consumption of fruit and vegetables and the risk of incident cognitive disorders. No language restriction was set. Both free search terms and MeSH terms were included in the search strategy, such as “fruit*”, “vegetable*”, “fruit(MeSH Terms)”, “vegetables(MeSH Terms)”, “AD”, “hypertensi*”, “Alzheimer*”, “aphronesia”, “cognitive*” and “cognition”. Supplementary Table 1 presents the detailed search strategy. The article with the longest duration of follow-up or reported the most relevant result was included in the pooled analysis, if multiple publications from the same study were identified. In addition, we manually searched the reference lists of the relevant original articles and systematic reviews in order to identify more potential eligible articles.

Supplementary table 1.

: Search strategy

Source: PubMed (Searched on: June 13th, 2016) #1 fruit*[Title/Abstract] #2 vegetable*[Title/Abstract] #3 fruit[MeSH Terms] #4 vegetables[MeSH Terms] #5 AD[title/abstract] #6 Alzheimer*[title/abstract] #7 aphronesia[title/abstract] #8 cognitive*[title/abstract] #9 cognition[title/abstract] #10 prospective[Title/Abstract] OR cohort[Title/Abstract] OR “follow up” [Title/Abstract] OR following[Title/Abstract] OR longitudinal[Title/Abstract] OR incidence[Title/Abstract] OR «Prospective Studies»[Mesh] OR «Cohort Studies»[Mesh] OR «Longitudinal Studies»[Mesh] #11 #1 OR #2 OR #3 OR #4 #12 #5 OR #6 OR #7 OR #8 OR #9 #13 #10 AND #11 AND #12 Source: Embase (Searched on: June 13th, 2016) #1 fruit*:ti,ab #2 vegetable*:ti,ab #3 dementia:ti,ab #4 AD:ti,ab #5 Alzheimer*:ti,ab #6 aphronesia:ti,ab #7 cognitive*:ti,ab #8 (‘prospective’ OR ‘cohort’ OR ‘follow up’ OR ‘longitudinal’ OR ‘incidence’):ti,ab #9 #1 OR #2 #10 #3 OR #4 OR #5 OR #6 OR #7 #11 #8 AND #9 AND #10

Selection criteria and data extraction

The initial search was conducted independently by two authors. We removed the duplicate records, and then the titles and abstracts of each article were screened. Each article was assessed independently as exclusion or requiring further assessment. Any disagreements were resolved by consensus with the third author.

Articles met criteria as follows were included in the present study. (1) studies reported relative risks (RRs) or hazard ratios (HRs) or odds ratios (ORs) and their corresponding 95% confidence intervals (CIs) of the incident cases of cognitive disorders involving the consumption of fruit and vegetables. (2) a study design based on prospective cohort. Articles were excluded if: (1) data described a specific type of fruit or vegetables, such as blackberries, garlic, or beans. (2) data described surrogate nutrients of fruit or vegetables, such as protein or fibre.

Data extraction was performed by two authors independently. Disagreements were resolved by discussion with the third author. The following data were extracted from each article: the first author, published year, number of study participants, race, gender and baseline age of participants, duration of follow-up, method of exposure and outcome measurements, types of exposure and outcome, category of exposure, number of cases, the largest adjusted RRs or HRs or ORs with the corresponding 95% CIs of the incident cognitive disorders.

Quality assessment

Two authors performed the quality assessments independently. The Newcastle-Ottawa Scale (NOS) was used to assess the quality of the included cohorts according to the following 3 domains: selection (0-4 points), comparability (0-2 points), and ascertainment of outcome (0-3 points) (29). A score of 9 points reflected the highest study quality.

Statistical analysis

We calculated the pooled RRs (95% CIs) for the relationship of the highest vs. lowest category of fruit and vegetables consumption with the incident risk of cognitive disorders. A generic inverse-variance method (random-effects model) was used to combine the outcome data. Between-study heterogeneity was examined by the I2 statistic. An I2 statistic higher than 50% indicated that significant heterogeneity was existed. To explore the potential sources of heterogeneity, we performed the subgroup and meta-regression analyses by pre-specified characteristics (race, gender, duration of followup, number of cases, exposure assessment method, and type of exposure). Begg's and Egger's tests were used to test the potential presence of publication bias (30, 31). If a publication bias was detected, we performed a sensitivity analysis to estimate the possible source and effect of the bias. It is noted that when the total number of included articles is small, the test for asymmetry process limits the interpretability of the finding (27). By omitting one study in every turn, sensitivity analysis was also conducted to assess the influence of a single study on the overall combined outcome.

We used the Stata, version 12.0 (StataCorp LP, College Station, Texas, USA) and the Review Manager software, version 5.2 (The Nordic Cochrane Centre, Copenhagen, Denmark) to perform the statistical analyses of our study. All reported two-sided P values of less than 0.05 were considered statistically significant.

Results

Study identification and selection

Figure 1 shows the detailed flow diagram of studies involved in our systematic review and meta-analysis. A total of 797 articles were identified from the databases of the Pubmed (448 articles) and the Embase (349 articles). After removing duplicated articles from the 797 articles, 646 articles were included for further screening. And then, 619 articles were excluded by reading the titles and the abstracts. The remaining 27 full-text articles were assessed for eligibility. No additional records were identified from other resources. Finally, 6 cohort studies comprised 9 comparatives were selected for the present systematic review and meta-analysis (21, 22, 23, 24, 25, 26).

Figure 1.

Flow diagram of articles included in the present study

Study characteristics

Table 1 presents the main characteristics of the included studies. These studies were published between 2006 and 2010. Five of these studies were performed among Caucasians (French (21, 24, 26), American (25), and Swedish (23)), and 1 study was conducted among the Japanese Americans (22). Follow-up duration ranged between 2.2 years to 31.5 years. Five articles included participants with both genders (21, 22, 23, 24, 25), and 1 article included only females (26). The sample size ranged from 1,233 to 8,085 for a total of 21,175 participants. The fruit and/or vegetables intake was assessed by foodfrequency questionnaire (FFQ) in three articles (21, 22, 25), except for 3 studies (assessed by diet and nutritional questionnaires) (23, 24, 26).

Table 1.

Characteristics of included studies

First author, Published year	Country	Follow-up (yrs)	Male(%)	Baseline age (yrs) (min-max)	Participants, No.	Fruit and vegetables intake			Incident cognitive disorders			Adjustment*
						Method of assessment	Category	Type	Method of ascertainment	Type	Case, No.
Barberger-Gateau, 2007	France	3.5	36.4	65-	8,085	FFQ	Never, Frequent	FV	DSM-IV, NINCSD-ADRDA	Dementia, AD	281, 183	1-9
Dai, 2006	America (Japanese American)	6.3	45.6	65-	1,836	FFQ	<1 time /week, 1-2 times /week, >3 times/week	FV	NINCSD-ADRDA	AD	63	2, 3, 7, 18, 10-20
Hughes, 2010	Sweden	31.5	37.9	42-	3,779	Diet questionnaire	No/small, Great/medium	FV	DSM-IV, NINCSD-ADRDA	Dementia, AD	302, 201	1-3, 6, 8, 10, 16, 21, 22
Ritchie, 2010	France	7.3	39.8	65-	1,433	Nutritional questionnaire	<2 times /week, >2 times /week	FV	Score (>1 cognitive test) in the lowest baseline fifth.	Dementia or cognitive impairment	405	1, 2
Roberts, 2010	America	2.2	59.5	70-	1,233	FFQ	≤109.6 g/day, 109.7-191.0 g/d, >191.0 g/d	V	Cognitive concern by others; impairment in 1-4 domains from the cognitive testing battery; essentially normal functional activities from the CDR and FAQ	Cognitive impairment	163	1-3, 13, 23-25
Vercambre, 2009	France	13.0	0.0	63-	4,809	Diet questionnaire	Tertile	F, V	DECO score <33	Cognitive decline	598	1,3,8,10, 13, 16, 18, 26, 27

AD, Alzheimer’s Disease; FFQ, food-frequency questionnaire; DSM, Diagnostic and Statistical Manual of Mental Disorders; NINCDS-ADRDA, National Institute of Neurological and Communicative Disorders and Stroke-Alzheimer’s Disease and Related Disorders Association; FAQ, Functional Activities Questionnaire; DECO, Observed Cognitive Deterioration; * 1= age, 2= sex, 3= education, 4= city, 5= income, 6= marital status, 7= ApoE genotype. 8= body mass index, 9= diabetes, 10= physical activity, 11= baseline CASI score, 12= olfaction diagnostic group, 13= total energy intake, 14= intake of saturated, 15= polyunsaturated fatty acids, 16= smoking status, 17= alcohol drinking, 18= vitamin supplementation, 19= tea drinking, 20= dietary intake of vitamin, 21= angina pectoris, 22= total food compared to others, 23= stroke, 24= coronary heart disease, 25= depressive symptoms, 26= use of postmenopausal hormones, 27= history of chronic diseases

The incident risk of cognitive disorders was diagnosed from DSM-IV (Diagnostic and Statistical Manual of Mental Disorders, fourth edition) (21, 23), or score in the lowest baseline fifth for more than cognitive test (24) for dementia; diagnosed from NINCDS-ADRDA (National Institute of Neurological and Communicative Disorders and Stroke- Alzheimer's Disease and Related Disorders Association) for Alzheimer's disease (21, 22, 23); diagnosed from DECO (Observed Cognitive Deterioration) score <33 (26), or the criteria in brackets (cognitive concern by others; impairment in 1-4 domains from the cognitive testing battery; essentially normal functional activities from the CDR and FAQ) (25) for cognitive impairment/decline.

Quality assessment

The quality score of the included studies ranged from 7 to 9 stars (Table 2). The main quality issues were listed as below. Three articles measured the consumption of fruit and/ or vegetables by diet/nutritional questionnaire rather than FFQ (23, 24, 26). The rate of loss to follow-up was more than 20% in 1 included study (23). The duration of follow-up was less than 5 years in 2 included studies (21, 25). One article only adjusted for the covariates of age and gender (24).

Table 2.

The Newcastle-Ottawa Scale used to grade the quality of each study (maximum=9 stars)

First author, published year	Selection	Comparability	Outcome	Total
Barberger-Gateau, 2007	****	**	**	********
Dai, 2006	****	**	***	*********
Hughes, 2010	***	**	**	*******
Ritchie, 2010	***	*	***	*******
Roberts, 2010	****	**	**	********
Vercambre, 2009	***	**	***	********

Selection column includes four items: 1) representativeness of the exposed cohort, 2) selection of the non-exposed cohort, 3) ascertainment of exposure to implants, and 4) demonstration that outcome of interest was not present at start of study. Comparability column includes two items: study controls for important and any additional covariates. Outcome column includes three items: 1) assessment of outcome, 2) follow-up long enough for outcomes to occur, and 3) adequacy of follow up of cohort.

Association between fruit and vegetables intake and cognitive disorders

Six cohorts comprised 9 comparatives reported the relationship of fruit and vegetables intake with the incident risk of cognitive disorders (Figure 2). Consumption of fruit and vegetables was inversely associated with the incident risk of dementia or cognitive impairment and the incident risk of AD, and the pooled RRs (95% CIs) were 0.79 (0.65, 0.95) and 0.57 (0.37, 0.87), respectively. The overall pooled outcome of cognitive disorders was 0.74 (0.62, 0.88), with evidence of significant heterogeneity (I2 =68%, P =0.002).

Figure 2.

Meta-analysis of the association between fruit and vegetables consumption and the incident risk of cognitive disorders

Subgroup analysis and meta-regression

As presented in Figure 3, analysis stratified by race (P-value for meta-regression =0.049) partly explained the heterogeneity between fruit and vegetables intake and incident cognitive disorders. No statistically significant sources of heterogeneity were found for the association in the subgroup analyses stratified by gender, duration of follow-up, number of cases, exposure assessment method and type of exposure (P >0.05 for each).

Figure 3.

Subgroup analysis of the relationship of fruit and vegetables consumption with the incident risk of cognitive disorders

Publication bias and sensitivity analysis

As shown in Supplementary Figure 1, visual inspection of the funnel plot suggests an evidence of publication bias among the articles for fruit and vegetables consumption (Egger's test, P =0.005; Begg's test, P =0.016). A sensitivity analysis excluding the study by Dai et al. (22), which caused asymmetry of the funnel plot, yielded the P values for the publication bias of 0.063 by Begg's test and 0.133 by Egger's test. In the sensitivity analysis, exclusion of each article in turn did not change the combined results, and the pooled RRs (95% CIs) ranged from 0.71 (0.59, 0.87) to 0.75 (0.62, 0.91).

Supplementary figure 1.

Funnel plot to explore publication bias.

Discussion

In the present systematic review and meta-analysis, we identified 6 prospective cohort studies involving a total of 21,175 participants. The pooled analysis showed that the intake of fruit and vegetables was inversely associated with the occurrence of cognitive disorders after adjustment for the largest number of possible confounders. Compared with the lowest level of consumption, the incident risk of cognitive disorders was decreased by 26% with the highest level of fruit and vegetables consumption. Furthermore, we found that the significant heterogeneity might be attributed to the ethnic difference.

Previous qualitative reviews in this topic have explored the relationship of fruit and vegetables consumption with cognitive decline, dementia, and cognitive function (18, 19). Some of the previous findings are in agreement with our study. Lamport et al. found that consumption of fruit, vegetables, and juices played a beneficial role in cognition (19). Loef et al. reported that the increased consumption of vegetables was linked with a lower risk of dementia and cognitive decline (18). However, different from the work by Loef et al., the present study performed a quantitative research rather than a qualitative review in this topic. A meta-analysis by Cao et al. included only two studies and failed to discover a significant association in this issue (20). We extend the work of Cao et al. by confirming the protective role of vegetables and fruit against the occurrence of cognitive disorders. Fruit and vegetables consumption was demonstrated to be inversely associated with the risk of stroke in a meta-analysis, which could interpret the helpful aspect against the vascular component of dementia (32). It is worth noting that the majority of the included articles did not separately analyze the intake of fruit and vegetables. From the stratified analysis of the present study, we found that only the intake of total fruit and vegetables presented a significant result. Considering that some studies did not find the protective effects of vegetables intake (25, 26) or fruit intake alone (26, 33, 34), further studies are also needed to investigate the role of fruit or vegetables intake separately on the development of cognitive disorders.

Beneficial role of a diet rich in fruit and vegetables may be attributed to its containing of antioxidant nutrients, such as vitamin C, vitamin E, flavonoids, and carotenoids (35, 36). In addition, higher components of phytoestrogens and polyphenols in fruit and vegetables might also play an effective role in lowering the risk of cognitive disorders (37, 38). Studies in animal models have testified that chronic administration of apple juice protected against aging and cognitive impairment in mice (39, 40). Although no evidence from randomized controlled trials (RCTs) has investigated the preventive effect of fruit and vegetables consumption on the incident cognitive disorders in the elderly people, a trial among students has found the improvement of the capability of behaviour and observational learning through promoting intake of fruit and vegetables (41).

The present systematic review and meta-analysis obtained an important finding of a significantly inverse association between fruit and vegetables consumption and the incident cognitive disorders. However, our study has several limitations. The main limitation of the present study should be the presence of considerable heterogenicity and publication bias across the articles. Therefore, we did the stratified analysis and metaregression to investigate the possible explanation for the significant heterogeneity, and the results revealed that the heterogeneity might be associated with ethnicity. Exclusion of the study by Dai et al. (performed among Japanese Americans) did not change the overall pooled results, which illustrated the robustness of our finding. Moreover, the variety in the outcome measurement, the application of different dietary questionnaires, the diverse categories of fruit and vegetables intake, and the various adjusted confounders may also cause heterogeneity of the pooled analysis. The sensitivity analysis revealed that the study by Dai et al. might be the cause of publication bias. Exclusion of this study did not change the overall pooled results and caused symmetry of the funnel plot (P values of >0.05 in Begg's and Egger's tests). Another explanation of the publication bias is that smaller studies presenting non-significant result might be underreported in the published articles. Additionally, the small number of only 6 included articles limits the interpretability of our results. Second, all of the included articles were performed in the developed countries, and we did not find articles performed in Asia, Africa, etc. The findings of the present study may not be representative of other parts of the world. Third, changes of dietary patterns of fruit and vegetables consumption might be given during the long-term follow-up. These dynamic factors might have affected our findings. Lastly, the doses of the fruit and vegetables intake were reported in different measurement unit. For example, “times per week”, “gram per day”, “never vs. frequent”, “no/small vs. great/medium” and “tertiles” were used among various articles. Furthermore, when we attempted to extract the data of all categories of consumption, only three studies reported more than 3 categories of fruit and vegetables consumption (also in different measurement unit). Therefore, we could not quantify the possible dose-response patterns of the association. In the future, high quality studies with standardized measurement unit should be administrated to explore the potential dose-response effect of fruit and vegetables intake on the incident risk of cognitive disorders.

Conclusions

In summary, the present systematic review and meta-analysis of 6 studies shows an inverse association between fruit and vegetables consumption and the incident risk of cognitive disorders. Although we obtained a significant association, the present study is limited by its possible publication bias and considerable heterogeneity. Further large prospective studies should be performed to quantify the potential dose-response patterns of fruit and/or vegetables intake and to explore the role of fruit or vegetables consumption separately on cognitive disorders in different populations.

Conflict of interest: The authors declare no conflict of interest.

Author contributors: Lei Wu and Yan Tan were responsible for the conception and design. Lei Wu, Dali Sun, and Yan Tan took part in the acquisition, analysis and interpretation of data. All authors have read and approved the final manuscript.

Ethical standard: The authors declare that the study procedures comply with the ethical standards for involving human participants.