Professional Documents
Culture Documents
Nutrition
journal homepage: www.nutritionjrnl.com
Review
a r t i c l e i n f o a b s t r a c t
Article history: Objective: Epidemiologic investigations evaluating the association of fruit and vegetable con-
Received 13 August 2015 sumption with depression risk have yielded controversial results. Therefore, a meta-analysis was
Accepted 13 September 2015 carried out to qualitatively summarize the evidence regarding association of fruit and vegetable
intake with risk of depression in the general population.
Keywords: Methods: PubMed, Embase, and Web of Knowledge were searched for relevant articles published
Fruit
up to June 2015. To evaluate the association of fruit and vegetable intake with depression risk,
Vegetable
combined relative risks were calculated with the fixed or random effects model. Meta-regression
Depression
Meta-analysis was conducted to explore potential sources of heterogeneity. Publication bias was estimated by the
Egger’s test and the funnel plot.
Results: Ten studies involving 227 852 participants for fruit intake and eight studies involving 218
699 participants for vegetable intake were finally included in this study. The combined relative risk
(95% confidence interval) of depression for the highest versus lowest category of fruit and vege-
table intake was 0.86 (0.81, 0.91; P < 0.01) and 0.89 (0.83, 0.94; P < 0.01), respectively. In subgroup
analyses stratified by study design, the inverse association of fruit (0.83 [0.77, 0.91; P ¼ 0.006]) and
vegetable (0.88 [0.79, 0.96; P ¼ 0.007]) intake with risk of depression was also observed in the
cohort study.
Conclusions: This meta-analysis indicated that fruit and vegetable consumption might be inversely
associated with the risk of depression, respectively.
Crown Copyright Ó 2016 Published by Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.nut.2015.09.009
0899-9007/Crown Copyright Ó 2016 Published by Elsevier Inc. All rights reserved.
X. Liu et al. / Nutrition 32 (2016) 296–302 297
in three studies [12,13,18], no association was found in the others Statistical analyses
[9,10,14–16].
Pooled measure was calculated as the inverse-variance weighted mean of the
Therefore, a meta-analysis was systematically performed to logarithm of RR (95% CI) of depression for the highest versus lowest category of
evaluate the association of fruit and vegetable consumption with fruit and vegetable intake, respectively. The I2 [19] was used to assess hetero-
the risk of depression. geneity among studies. If I2 50% [20], the random effect model (REM) was
adopted as the pooling method; otherwise (I2 < 50%), the fixed effect model
Materials and methods (FEM) was employed. Meta-regression was performed to access the potentially
important covariates (including publication year, continent, study design, status
Literature search strategy for physical activity adjustment, and ascertainment method of depression) that
might exert substantial impacts on between-study heterogeneity [21]. To assess
A literature search was performed on the databases of PubMed, Embase, and whether the results could have been affected distinctly by a single study, an in-
Web of Knowledge for relevant articles published up to June 2015, using the fluence analysis was carried out with one study removed at a time [22]. The
following search terms: “depression,” “depressive disorder,” “depressive symp- leave-one-out sensitive analysis was conducted to evaluate the key studies that
toms,” “fruit,” “vegetable,” and “diet” without restrictions. Moreover, the refer- have remarkable impact on the between-study heterogeneity. Publication bias
ences of related reviews and original articles were also reviewed. The detailed was evaluated with the visual inspection of funnel plot and Egger regression
steps of the literature search are shown in Figure 1. asymmetry test [23]. Subgroup analysis was performed by study design (cross-
sectional or cohort study). All statistical analyses were conducted using Stata
V.12.0 (Stata Corp., College Station, TX, USA). A two-tailed P < 0.05 was consid-
Inclusion criteria ered statistically significant.
Quantitative synthesis
Author [Ref.] Study design Country Year Age range Sex Sample RR (95% CI) Method Depression method
or mean age
Allgower et al. [17] Cross-sectional 16 countries 2001 21.6 F 3438 0.81 (0.66, 1.01) for fruit Interview 13-item short BDI 5
study in Europe 21.6 M 2091 0.85 (0.62, 1.17) for fruit
Hintikka et al. [15] Cross-sectional Finland 2005 44.04 FþM 2011 0.92 (0.79, 1.07) for fruit FFQ 21-item BDI 15
BDI, Beck Depression Inventory; BMI, body mass index; CES-D, Center for Epidemiologic Studies Depression Rating Scale; CI, confidence interval; CIDI-SF, Composite International Diagnostic Interview-Short form; DSM,
Diagnostic and Statistical Manual of Mental Disorders; FFQ, Food Frequency Questionnaire; GDS, Geriatric Depression Scale; RR, relative risk.
X. Liu et al. / Nutrition 32 (2016) 296–302 299
Fig. 2. Meta-analysis of the association between fruit intake and depression risk. The size of gray box is positively proportional to the weight assigned to each study, which is
inversely proportional to the standard error of the relative risks (RRs), and horizontal lines represent the 95% confidence intervals (95% CI).
Vegetable intake and the risk of depression influence on the pooled effects (P > 0.05). For the leave-one-out
For vegetable intake, four cohort studies [9,10,12,16] and four sensitivity analysis, one study [14] was found to contribute to the
cross-sectional studies [13–15,18] involving 218 699 participants between-study heterogeneity. After excluding this study, the
were included. Among these studies, three were conducted in heterogeneity (I2 ¼ 27.5%, P ¼ 0.200) was reduced and the pooled
Asia [13,16,18], two in Europe [10,15], one in Oceania [9], one in RR was 0.83 (0.78, 0.88).
North America [14], and one in South America [12]. With regard
to the depression diagnostic criteria, six studies used the scales
or questionnaires [9,12,13,15,16,18], and two were ascertained by Influence analysis and publication bias
interview or self-reported physician diagnosis or use of regular
antidepressant medication [10,14]. The major confounding fac- Influence analysis showed that no individual study had an
tors adjusted in studies included age, sex, education, body mass excessive influence on the pooled association of fruit and vege-
index, and physical activity. table intake with the risk of depression. The visual inspection of
Among the eight studies, three reported that high vegetable the funnel plot (Fig. 5) and Egger’s test showed no evidence of
intake could decrease the risk of depression [12,13,18], whereas publication bias for the analysis of vegetable intake with
the other five indicated no significant association [9,10,14–16]. depression (P ¼ 0.053). For fruit intake, publication bias was
The pooled RR (95% CI) of depression for the highest versus found in both the Egger’s test (P ¼ 0.022) and the visual in-
lowest category of vegetable intake was 0.89 (0.83, 0.94; spection of the funnel plot. However, after excluding one study
Pfor significance < 0.01, FEM, I2 ¼ 14.1%, Pfor heterogeneity ¼ 0.319; [14], the publication bias was not evident with the Egger’s test
Fig. 4). The pooled RRs for cross-sectional and cohort studies (P ¼ 0.078).
were 0.83 (0.68, 1.02; Pfor significance ¼ 0.078, REM, I2 ¼ 60.4%,
Pfor heterogeneity ¼ 0.056) and 0.88 (0.79, 0.96; Pfor significance ¼
Discussion
0.007, FEM, I2 ¼ 0.0%, Pfor heterogeneity ¼ 0.923), respectively, for
subgroup analysis stratified by study design.
To our knowledge, this is the first meta-analysis to explore the
relationship between fruit and vegetable intake and depression
Sources of heterogeneity and sensitive analysis risk. A total of 227 852 participants for fruit intake and 218 699
participants for vegetable intake were included in this meta-
As shown in Figure 2, moderate heterogeneity (I2 ¼ 48.2%, analysis. Findings from the meta-analysis indicated that both
Pfor heterogeneity ¼ 0.043) was found in the analysis of fruit intake fruit intake and vegetable intake were significantly associated
and depression. Therefore, univariate meta-regression was per- with the decreased risk of depression. Subgroup analysis was
formed with the covariates of publication year (P ¼ 0.513), also conducted by study design. The significantly inverse asso-
continent (P ¼ 0.089), study design (P ¼ 0.610), status for ciation was observed in both cross-sectional and cohort studies
physical activity adjustment (P ¼ 0.134), and depression ascer- for fruit intake and depression risk. For vegetable intake, the
tainment (P ¼ 0.751) to investigate potential sources of the significantly inverse association was also found in the cohort
heterogeneity. None of these covariates had an excessive studies.
300 X. Liu et al. / Nutrition 32 (2016) 296–302
Fig. 3. Forest plot of the relative risks (RRs) with corresponding 95% confidence intervals (95% CI) of studies on fruit intake and depression for cross-sectional and cohort
studies. The size of gray box is positively proportional to the weight assigned to each study, which is inversely proportional to the standard error of the RRs, and horizontal
lines represent the 95% CIs.
The exact biological mechanisms underlying fruit and vege- varied among the studies could be the sources, such as publi-
table intake and depression risk are still not fully understood. cation year, continent, study design, status for physical activity
First, various minerals and vitamins, such as magnesium, zinc, adjustment, depression ascertainment, and other covariates.
selenium, and vitamin B12, are present in fruits and vegetables. Thus, meta-regression for fruit intake analysis was performed to
Some of them were found to exert influence on the mechanisms explore the potentially important causes for between-study
of depression. For instance, magnesium intake could reduce heterogeneity. Meta-regression analysis did not find any of the
plasma concentrations of C-reactive protein, which is a marker of above-mentioned covariates as the important contributors to the
low-grade inflammation, and depression was found to be related between-study heterogeneity. To further explore the potential
to chronic inflammation [14]. Vitamin B12 affects the biochem- sources of between-study heterogeneity, the leave-one-out
ical processes in the central nervous system. Vitamin B12 defi- sensitivity analysis was carried out. After excluding one study
ciency could lead to hyperhomocysteinemia by activation of N- that had a strong effect on the heterogeneity, the heterogeneity
methyl-D-aspartate receptors, oxidative stress, and lesions in decreased to 27.5% and no publication bias was found in the
vascular endothelium, which could result in neurotoxicity and analysis between fruit intake and depression risk. What’s more,
then lead to the depression incident [28]. Zinc deficiency induces the results still remained significant after reducing the hetero-
neurologic and somatic symptoms as well as psychopathological geneity in the analysis of fruit intake and depression, strongly
symptoms that are connected with depressive disorder [29,30]. identified the stability of results.
In addition, antioxidants such as vitamin C, vitamin E, and folic The study has several strengths. First, a large number of par-
acid, which play an important role in the endothelial cell ticipants were included in this study, allowing a much greater
signaling cascades, could dampen the detrimental effects of possibility of achieving reasonable conclusions. Second, an
oxidative stress on mental health [31,32]. As a good source of inverse association was found in cohort studies, indicating a po-
antioxidants, fruit and vegetable intake might be beneficial to tential causal relationship between fruit and vegetable intake and
protect against depression. depression. Third, RRs with the most adjusted covariates were
Between-study heterogeneity occurs frequently in meta- extracted, which could reduce the confounding. Fourth, after
analysis [20], and it is essential to explore the potential sources excluding one study, the heterogeneity was decreased, and the
of between-study heterogeneity. In this meta-analysis, moderate summary result of sensitive analysis did not change substantially.
heterogeneity was found in the analysis between fruit intake and However, this meta-analysis was limited in some aspects as
depression risk. An indeterminate number of characteristics that well. First, confounders adjusted in each study were inconsistent.
X. Liu et al. / Nutrition 32 (2016) 296–302 301
Fig. 4. Meta-analysis of the association between vegetable intake and depression risk. The size of gray box is positively proportional to the weight assigned to each study,
which is inversely proportional to the standard error of the relative risks (RRs), and horizontal lines represent the 95% confidence intervals (95% CI).
For example, body mass index and physical activity were be accessed because of the limited information in the included
adjusted in some studies, whereas not adjusted in other studies. studies.
Second, publication bias was observed in the analysis of fruit In summary, the present meta-analyses indicated that fruit
intake and depression. However, after removing one study that and vegetable intake was inversely associated with the risk of
had a strong effect on the heterogeneity, the bias was not evident depression.
anymore, and the overall summary RR did not substantially
change, suggesting that the result was stable. Third, the diag-
nostic criteria of depression were inconsistent. Different scales,
including Beck Depression Inventory, Geriatric Depression Scale, References
and Center for Epidemiologic Studies Depression Rating Scale,
[1] World Health Organization. The global burden of disease:2004 update.
were used in some studies, and depression was ascertained by Available at: http://www.who.int/healthinfo/global_burden_disease/GBD_
physician diagnosis or beginning use of regular antidepressant report_2004update_full.pdf. Accessed October 27, 2008.
[2] World Health Organization. Mental health action plan:2013-2020. Available
medication in the others. Fourth, diet assessment methods var-
at: http://apps.who.int/iris/bitstream/10665/89966/1/9789241506021_eng.
ied among the studies and dose-response relationship could not pdf?ua=1. Accessed June 3, 2013.
[3] Malhi GS, Parker GB, Greenwood J. Structural and functional models of
depression: from sub-types to substrates. Acta Psychiatr Scand
2005;111:94–105.
[4] Kempermann G, Kronenberg G. Depressed new neuronsdadult hippo-
campal neurogenesis and a cellular plasticity hypothesis of major depres-
sion. Biol Psychiatry 2003;54:499–503.
[5] Smith RS. The macrophage theory of depression. Med Hypotheses
1991;35:298–306.
[6] Zhai L, Zhang Y, Zhang D. Sedentary behaviour and the risk of depression: a
meta-analysis. Br J Sports Med 2015;49:705–9.
[7] Slavich GM, Irwin MR. From stress to inflammation and major depressive
disorder: a social signal transduction theory of depression. Psychol Bull
2014;140:774–815.
[8] Zhai L, Zhang H, Zhang D. Sleep duration and depression among adults: a
meta-analysis of prospective studies. Depress Anxiety 2015;32:664–70.
[9] Mihrshahi S, Dobson AJ, Mishra GD. Fruit and vegetable consumption and
prevalence and incidence of depressive symptoms in mid-age women:
results from the Australian longitudinal study on women’s health. Eur J Clin
Nutr 2015;69:585–91.
[10] Sanchez-Villegas A, Delgado-Rodriguez M, Alonso A, Schlatter J, Lahortiga F,
Serra Majem L, et al. Association of the Mediterranean dietary pattern with
the incidence of depression: the Seguimiento Universidad de Navarra/
University of Navarra follow-up cohort. Arch Gen Psychiatry
2009;66:1090–8.
[11] Liu C, Xie B, Chou CP, Koprowski C, Zhou D, Palmer P, et al. Perceived stress,
depression and food consumption frequency in the college students of
Fig. 5. Funnel plot for the analysis of vegetable intake and depression risk. China Seven Cities. Physiol Behav 2007;92:748–54.
302 X. Liu et al. / Nutrition 32 (2016) 296–302
[12] Gangwisch JE, Hale L, Garcia L, Malaspina D, Opler MG, Payne ME, et al. [22] Tobias A. Assessing the influence of a single study in the meta-analysis
High glycemic index diet as a risk factor for depression: analyses from the estimate. Stata Tech Bull 1999;47:15–7.
Women’s Health Initiative. Am J Clin Nutr 2015;102:454–63. [23] Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis
[13] Woo J, Lynn H, Lau WY, Leung J, Lau E, Wong SY, et al. Nutrient intake and detected by a simple, graphical test. BMJ 1997;315:629–34.
psychological health in an elderly Chinese population. Int J Geriatr Psy- [24] Woo J, Ho SC, Yu AL. Lifestyle factors and health outcomes in elderly Hong
chiatry 2006;21:1036–43. Kong chinese aged 70 years and over. Gerontology 2002;48:234–40.
[14] McMartin SE, Jacka FN, Colman I. The association between fruit and [25] Tsai AC, Chang TL, Chi SH. Frequent consumption of vegetables predicts
vegetable consumption and mental health disorders: evidence from five lower risk of depression in older Taiwanesedresults of a prospective
waves of a national survey of Canadians. Prev Med 2013;56:225–30. population-based study. Public Health Nutr 2012;15:1087–92.
[15] Hintikka J, Tolmunen T, Honkalampi K, Haatainen K, Koivumaa- [26] Rienks J, Dobson AJ, Mishra GD. Mediterranean dietary pattern and prev-
Honkanen H, Tanskanen A, et al. Daily tea drinking is associated with a low alence and incidence of depressive symptoms in mid-aged women: results
level of depressive symptoms in the Finnish general population. Eur J from a large community-based prospective study. Eur J Clin Nutr
Epidemiol 2005;20:359–63. 2013;67:75–82.
[16] Chi SH, Wang JY, Tsai AC. Combined association of leisure-time physical [27] Mihrshahi S, Dobson A, Mishra G. Fruit consumption is associated with
activity and fruit and vegetable consumption with depressive symptoms in lower risk of depression in mid-age women. Ann Nutr Metab 2013;63:830.
older Taiwanese: results of a national cohort study. Geriatr Gerontol Int; [28] Karakula H, Opolska A, Kowal A, Domanski M, Plotka A, Perzynski J Does
2015 [Epub ahead of print]. diet affect our mood? The significance of folic acid and homocysteine. Pol
[17] Allgower A, Wardle J, Steptoe A. Depressive symptoms, social support, and Merkuriusz Lek 2009;26:136–41.
personal health behaviors in young men and women. Health Psychol [29] Murakami K, Mizoue T, Sasaki S, Ohta M, Sato M, Matsushita Y, et al. Dietary
2001;20:223–7. intake of folate, other B vitamins, and omega-3 polyunsaturated fatty acids in
[18] Shahar S, Hassan J, Sundar VV, Kong AY, Ping Chin S, Ahmad SA, et al. relation to depressive symptoms in Japanese adults. Nutrition 2008;24:140–7.
Determinants of depression and insomnia among institutionalized elderly [30] Nowak G. Zinc, future mono/adjunctive therapy for depression: mecha-
people in Malaysia. Asian J Psychiatr 2011;4:188–95. nisms of antidepressant action. Pharmacol Rep 2015;67:659–62.
[19] Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. [31] Maes M, De Vos N, Pioli R, Demedts P, Wauters A, Neels H, et al. Lower
Stat Med 2002;21:1539–58. serum vitamin E concentrations in major depression. Another marker of
[20] Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in lowered antioxidant defenses in that illness. J Affect Disord 2000;58:241–6.
meta-analyses. BMJ 2003;327:557–60. [32] Khanzode SD, Dakhale GN, Khanzode SS, Saoji A, Palasodkar R. Oxidative
[21] Higgins JP, Thompson SG. Controlling the risk of spurious findings from damage and major depression: the potential antioxidant action of selective
meta-regression. Stat Med 2004;23:1663–82. serotonin re-uptake inhibitors. Redox Rep 2003;8:365–70.