Ageing Research Reviews: Annabel P. Matison, Karen A. Mather, Victoria M. Flood, Simone Reppermund

Ageing Research Reviews 70 (2021) 101403
Contents lists available at ScienceDirect
Ageing Research Reviews

journal homepage: www.elsevier.com/locate/arr
Review
Associations between nutrition and the incidence of depression in

middle-aged and older adults: A systematic review and meta-analysis of
prospective observational population-based studies
Annabel P. Matison a, *, Karen A. Mather a, Victoria M. Flood b, c, Simone Reppermund a, d
a
Centre for Healthy Brain Ageing, UNSW, Sydney, NSW, Australia
b
The University of Sydney, Faculty of Medicine and Health, Sydney, NSW, Australia
c
Westmead Hospital, Western Sydney Local Health District, Westmead, NSW, Australia
d
Department of Developmental Disability Neuropsychiatry, School of Psychiatry, UNSW, Sydney, NSW, Australia
A R T I C L E I N F O A B S T R A C T
Keywords: Aim: To systematically examine the longitudinal observational evidence between diet and the incidence of
Diet depression in adults aged 45 years and older.
Nutrition Method: Three electronic databases were searched for cohort studies published up to December 2020 that
Depression
investigated the association between baseline dietary intake and incidence of depression in community-dwelling
Older adults
Systematic review
adults aged 45+years. Combined odds ratios (OR) and 95% confidence intervals (95%CI) were calculated.
Meta-analysis Random-effects models were used.
Results: In total 33 articles were included, with 21 combined in meta-analyses. Both the Dietary Inflammatory
Index and the Western diet were associated with an increased odds of incident depression (Dietary Inflammatory
Index: OR 1.33; 95%CI 1.04, 1.70; P = 0.02; Western: OR 1.15 95%CI 1.04, 1.26; P = 0.005). Higher fruit and
vegetable intakes were associated with a reduced risk of incident depression (vegetables: OR 0.91; 95%CI 0.87,
0.96; P < 0.001; fruit: OR 0.85; 95%CI 0.81, 0.90; P < 0.001). No association was observed between the
Mediterranean diet, “healthy” diet, fish intake and incident depression.
Conclusions: Results suggest an association between higher consumption of pro-inflammatory diets and Western
diets and increased incidence of depression, while higher intake of fruit and vegetables was associated with
decreased incidence of depression. These results are limited by the observational nature of the evidence (results
may reflect residual confounding) and the limited number of studies. More high-quality intervention and cohort
studies are needed to confirm these associations and to extend this work to other food groups and dietary
patterns.
1. Introduction Modifiable lifestyle factors, such as diet, physical activity and social
participation, provide promising avenues to reduce the incidence of
Depression affects 264 million people globally and over the past 30 depression (Worrall et al., 2020). Over the past decade, a growing body
years has become the third leading burden of disease (years-lost-to- of evidence has accumulated suggesting a link between diet and
disease ~43 million; GBD, 2017 Disease and Injury Incidence and depression. A 2017 meta-analysis of 21 observational studies found that
Prevalence Collaborators, 2018). Compared with younger adults, healthy dietary patterns have the potential to decrease the risk of
depression in older adults is associated with greater impacts on physical depression by 36%, while western/unhealthy diets increase risk by 18%
performance and cognition and may impact on a person’s ability to live (Li et al., 2017). Similarly more recent meta-analyses have reported
independently. In older adults, depression is also associated with reduced risk of depressive outcomes with higher adherence to the
increased morbidity, suicide and increased non-suicide mortality Mediterranean diet (Relative Risk [RR] 0.67; 95%CI 0.55, 0.82) and a
(Blazer, 2003). lower Dietary Inflammatory Index (RR 0.76; 95%CI 0.63, 0.92) (Lassale
* Corresponding author at: Centre for Healthy Brain Ageing, UNSW, School of Psychiatry, Level 1, AGSM (G27) Gate 11, Botany Street., UNSW, Sydney, NSW,
2052, Australia.
E-mail address: [email protected] (A.P. Matison).
https://doi.org/10.1016/j.arr.2021.101403
Received 11 March 2021; Received in revised form 16 June 2021; Accepted 2 July 2021
Available online 8 July 2021
1568-1637/© 2021 Elsevier B.V. All rights reserved.
Downloaded for Fakultas Kedokteran Universitas Muslim Indonesia ([email protected]) at University of Muslim Indonesia from ClinicalKey.com
by Elsevier on September 22, 2021. For personal use only. No other uses without permission. Copyright ©2021. Elsevier Inc. All rights reserved.
A.P. Matison et al. Ageing Research Reviews 70 (2021) 101403
et al., 2019). A 2020 meta-analysis of 18 studies examining the rela The aim of this study is to summarise the longitudinal observational
tionship between diet and depression in adults 65+ years reported that a evidence between diet, including dietary patterns, foods/food groups
healthy dietary pattern was associated with a reduced risk of depression and nutrients, and the incidence of new cases of depression in
(OR 0.85; 95%CI 0.78, 0.92; Wu et al., 2020). There was, however, high community-dwelling adults aged 45 years and older.
heterogeneity between studies, the incidence of depression was not
examined and only a “healthy” dietary pattern was considered. 2. Methods
Only a limited number of intervention studies have examined the
effect of diet on depression, with these studies generally based on par This systematic review followed the Preferred reporting Items for
ticipants with specific health conditions. A recent meta-analysis of such Systematic and Meta-analysis (PRISMA) statement (Moher et al., 2009)
studies reported no statistically significant benefit of a healthy diet on and was registered with PROSPERO (Registration number
incidence (based on 2 interventions) or treatment (based on 4 in CRD42020202771). No significant deviations from the registered
terventions) while a healthy diet was reported to beneficially effect methods were made.
depressive symptoms (effect size -0.14; 95%CI -0.24, -0.04) based on 15
interventions (Thomas-Odenthal et al., 2020). 2.1. Search strategy
Currently there is limited robust evidence of the association between
diet and depression focusing on population-based middle-age and older The Medline Complete, Embase and PsycINFO electronic databases
adults. However, several studies suggest the relationship between were searched to identify relevant studies published from inception to
nutrition and depression changes with age (Chang et al., 2016; Ver December 2020. The final search terms were nutrition, depression, older
meulen et al., 2016). Although the mechanisms behind a potential adults, incidence, cohort and related terms. Full details of search terms
relationship between nutrition and depression are not clear, a popular are included in Supplemental Table A1. The search was limited to peer
view is that inflammation is involved in the aetiology of depression (Liu reviewed original studies conducted in humans and published in En
et al., 2017) with evidence suggesting that diet influences inflammation glish. Reference lists of eligible articles as well as reviews were hand
(Wang et al., 2018). Late-life depression may be due to a number of searched to identify additional articles.
factors associated with advanced age such as age-associated physiolog
ical changes (e.g. increased inflammation), the presence of chronic 2.2. Inclusion and exclusion criteria
disease or the occurrence of stressful life events associated with older
age (e.g. bereavement) (Fiske et al., 2009). As approximately half the To be included in this systematic review, studies had to be an original
older adults with depression experience late-onset depression, i.e. a first study investigating the longitudinal association between diet and inci
episode of depression at or after the age of 60 (Brodaty et al., 2001), the dence of depression (at follow-up) and meet the criteria set out in
potential exists to make lifestyle changes at an earlier age to impact the Table 1.
subsequent risk of depression. Given the increasing proportion of older
adults in the global population (United Nations Department of Eco
2.3. Study selection process
nomics and Social Affairs, 2017) and the higher impact of depression in
this age group, providing dietary recommendations to reduce depression
Two reviewers (SR, AM) independently assessed all articles obtained
incidence in this age range has the potential to reduce global burden-of
from the electronic search against inclusion and exclusion criteria.
disease.
Disagreements between reviewers were resolved through consensus and
To date reviews examining diet and the risk of depression have
a third reviewer (KM) provided final judgement where consensus was
generally examined both cross-sectional and longitudinal studies and
not reached. The retrieved articles were managed using the Covidence
recent reviews have focused on dietary patterns. Depression is known to
software (Veritas Health Innovation, 2021).
affect appetite, therefore cross-sectional evidence of an association be
tween diet and depression may reflect reverse causation. Evidence of a
relationship between diet and depression comes from studies examining 2.4. Data extraction and quality assessment
either dietary patterns, foods/food groups or nutrients. Evidence to help
understand what is driving these relationships at the dietary pattern Data extraction included: participants (country, number, age, sex,
level can be gained from examining diet at the sublevels of food/food ethnicity, education, recruitment source), nutrition (type, assessment
groups and nutrients. tool), depression (assessment tool, threshold, incidence) method of
In this review we will focus on examining the evidence of baseline analysis, follow-up period, confounders, results and industry funding.
dietary intake and the longitudinal risk of new cases of depression. There Data were extracted by AM and verified by SR with differences
is a lack of intervention studies examining the effect of nutrition on resolved by discussion until consensus was reached. Where a study ran
incidence of depression in middle-aged to older adults. Therefore, to multiple analyses, data were extracted from the analysis using the
reduce heterogeneity, this review will focus on longitudinal cohort strictest depression definition, the longest follow-up duration and the
studies. most covariates. If multiple papers examined the same form of nutrition
in the same cohort, results from the largest sample size were used. Where
Table 1
Inclusion and exclusion criteria for systematic review selection.
Include Exclude
Participants • Population-based, community-dwelling adults ≥ 45-years at base • Examining only participants with specific health conditions (including pregnant/post-
line (or study provided sub analysis of participants ≥ 45-years) partum populations and overweight/obese populations) or institutionalised participants
• Included only food insecure or malnourished participants
Exposure • Any form of nutrition intake (e.g. food patterns, food groups, foods, • Eating patterns or dietary habits (e.g. fasting or binge eating)
micronutrients, or dietary supplements) • Examined only biomarkers of diet
Outcome • Incidence of depression • Bipolar depression
• Depression not separately measured (e.g. combined measure of depression and anxiety
or stress)
Study • Prospective observational study • Interventions, cross-sectional studies and case-control studies
design
data was incomplete, the corresponding author was contacted to provide for the lowest (3.7%) and highest (36%) incidence of depression
additional information. extracted from studies included in this systematic review.
The quality of included studies was assessed using a modified Random-effects models were used to account for potential hetero
Newcastle-Ottawa scale for assessing the quality of non-randomised geneity between diet types and study design. Between study heteroge
studies in meta-analysis (Stang, 2010; Wells et al., 2001). Details and neity was assessed using I2, with an I2 of 25% considered low, 50%
results of the quality assessment are included in Supplemental Table A2 moderate and 75% high (Higgins et al., 2003). Potential sources of
and A3. A traffic light system using green, amber and red was used to heterogeneity were assessed by comparing studies for: similarity of
assess study quality with green indicating higher quality and red indi diets, method of assessing diet and depression, participants age and
cating lower quality. Study quality was independently assessed by SR study location. Studies identified as potentially causing heterogeneity
and AM and conflicts were resolved by consensus. were systematically omitted from the meta-analysis to determine if they
improved heterogeneity. Publication bias was assessed by Egger’s test
and visual inspection of a funnel plot for all meta-analyses which con
2.5. Statistical analysis
tained nearly 10 studies.
With the exception of Egger’s test, statistical analysis was performed
Meta-analysis was performed for each dietary type where it was
in Review Manager, version 5.4 (RevMan, 2020). Egger’s test was per
considered that the underlying dietary intake was sufficiently similar to
formed in R (Viechtbauer, 2010). Statistical significance was set as p <
justify combining studies. Combined odds ratios (OR) and 95% confi
0.05. Study results where the dietary type used was not able to be
dence intervals (95%CI) were calculated. Results were based on a
combined by meta-analysis were summarised in a narrative review.
comparison of the highest intake versus lowest intake of each dietary
pattern, food or nutrient. Risk ratios (RR) were treated as OR where the
2.6. Sensitivity analysis
incidence of depression was ≤ 10% as previously recommended (Zhang
and Yu, 1998). If the incidence of depression was > 10% the OR was
Sensitivity analysis was performed based on the results of the quality
estimated using the RR and incidence of depression in the lowest intake
assessment. We performed sensitivity analyses by excluding studies
group (Zhang and Yu, 1998). Hazard ratios (HR) were treated as OR and
rated red for measures of diet and/or depression diagnosis. Sensitivity
sensitivity analysis was performed excluding these studies to assess
analysis was also performed excluding studies only reporting HR.
whether the overall result had been impacted by the inclusion of studies
reporting HR. Regression coefficients were converted to OR and 95%CI
using the β and standard error (SE). Absolute risk reduction/increase
was calculated based on the pooled OR and 95%CI for each form of
nutrition where meta-analysis results were significant and is presented
Fig. 1. Flowchart of study screening and selection process.
3. Results including both females and males (n = 20). Studies were conducted in
North America (11), Australia (9), Europe (8) or Asia (5). Seven cohorts
3.1. Search results were the subject of multiple papers assessing different types of nutrition
(see Supplemental Table A5). One study included two cohorts (Vashum
The database search yielded 1,995 results, with a further 8 studies et al., 2014). Sample size varied from 526 to 263,923 participants. In
identified by handsearching of included articles. After removing dupli total 547,606 participants from 19 different cohorts were included in
cates (635) and studies not meeting inclusion criteria (1,396 based on this review. Eleven studies were conducted in participants aged 65 years
initial screening and 52 based on full-text review), 33 articles, published and older. Study duration varied from 3 to 12 years.
between 01/2008 and 12/2020, were eligible for inclusion in this sys The relationship between the incidence of depression and a variety of
tematic review (Fig. 1). Details of studies subject to full-text review but different types of nutrition were assessed, with the majority of studies
excluded from this systematic review are set out in Supplemental investigating multiple types of nutrition. The most commonly used
Table A4. method to assess dietary intake was a food frequency questionnaire (27
studies), followed by 24-h dietary records (4 studies) and diet history (2
studies). Depression was measured using standardised tools, self-report
3.2. Characteristics of studies
of clinical diagnosis, psychiatric examination, or deduced from Medi
care or Pharmaceutical benefits claims or antidepressant usage.
Thirty-three studies were included in this review. Dietary patterns
Due to the potential for industry funding to bias study results, all
were examined in 17 studies (Table 2), foods/food groups in 13 studies
studies were reviewed for industry funding. Four authors of one study
(Table 3) and nutrients were reported in ten studies (Table 4). Eleven
reported working for an organisation funded by Nestlé Nutrition,
studies included females only, two used males only with the remainder
Fig. 2. Forest plots depicting meta-analyses of studies assessing the longitudinal association between diet and incident depression. Results are odds ratios comparing
the highest category of adherence with the lowest category.
A.P. Matison et al.
Table 2
Characteristics of included studies examining the association of dietary patterns with incident depression.
Author, Year Country Sample % Age Dietary pattern Dietary Diagnosis of Incidence of Follow-up Adjustment for Results
size females range assessment depression depression n period (years) confounders

(number) (mean tool (%) (interim
± SD assessments)
years)
Australian dietary
≥ 75 ADG A, BMI, S, smoking, guideline index OR 1.00
Das et al., 2021 Australia 781 0 (81.2 ± Diet history GDS-15 ≥ 5 13.2% 3 alcohol, self-reported (95%CI 0.97, 1.02)##
4.5) health, PA, C, EI, ADs MedDiet OR 1.02 (95%
MedDiet
CI 0.89, 1.17)##
For a 10% increase in
9
Adjibade et al., Proportion ultra- Three 24-h CES-D-20 ≥17 for A, G, BMI, S, EI, ultra-processed foods in
France 13,620 na 49–86 872 (every 2
2019 (a)* processed foods records ♂; 23 for ♀ smoking, alcohol, PA the diet HR 1.26 (95%CI
years)
1.15, 1.38)
9
Adjibade et al., Three 24-h CES-D-20 ≥17 for A, G, BMI, S, EI, alcohol, Qr4 vs Qr1 HR 1.22
France na na 60–86 ADII na (every 2
2019 (b)* records ♂; 23 for ♀ smoking, PA, C (95%CI 0.91, 1.64)
years)
9 S, smoking, PA,
CES-D-10 ≥ 10 or OR T3 vs T1 0.94 (95%
Lai et al., 2017 Australia 7,877 100 50–55 ADG FFQ 2,841 (36%) (every 3 anxiety/nervous
self-report CI 0.83, 1.00)
years) disorder
For every additional
Psychiatric component of dietary
examination or guidelines adhered to
Voortman et al., The ≥ 45 18.3 A, G, smoking, S, EI, PA,
6,217 58.1 Dutch dietary guidelines FFQ self-report or 1,686 (27%) HR 0.97 (95%CI 0.95,
2017 Netherlands (64.1) (mean 10.9) BMI
medical records 1.00)
and ADs RR Qr1 v Qr4 0.79 (95%
CI 0.66, 0.95)###
5
50-–55 12 EI, S, menopausal

Shivappa et al., 10-item CES-D ≥ RR Qr1 v Qr4 0.79 (95%
Australia 6,438 100 (52.0 ± DII FFQ 1,156 (8.1%) (every 2.5 status, illness/injury,
2016 10 CI 0.66, 0.95) ###
1⋅4) years) smoking, PA, BMI
Western vs not Western
OR 1.48 (95%CI 0.52,
4.20)
≥ 53 8 Traditional vs not
Westerna, traditionalb, A, G, S, smoking,
Tsai et al., 2016 Taiwan 3,778 46.9 (66.7 ± FFQ CES-D-10 ≥ 10 na (every 4 traditional OR 1.60
healthyc alcohol, PA, ADL
8.8) years) (95%CI 1.12, 2.29)
Healthy vs not healthy
OR 0.97 (95%CI 0.53,
1.78)
A, EI, S, BMI, C, HRT, GI OR Qn5 vs Qn1 1.22
PA, alcohol, smoking, (95%CI 1.09, 1.37)
Gangwisch et al. United 50–79 8-item Burnam 4,643
69,954 100 GI, glycaemic load FFQ 3 SLE, fatty acids, fruit, Glycaemic load Qn5 v
2015 States (63.7) scale ≥ 0.06 (6.6%)
vegetables, legumes, Qn1 OR 1.01 (95%CI

nuts/seeds, fibre, HEI 0.82, 1.24)
Varied OR T3 vs T1 0.86
(95%CI 0.54, 1.37)
A, G, S, smoking, EI, PA, Traditional OR T3 vs T1
Gougeon et al., 67–84 Variedd, traditionale, Three 24-h GDS-30 ≥11 or 3
Canada 1,358 50.4 170 (12.5%) BMI, C, autonomy, 0.96 (95%CI 0.61, 1.52)
2015 (74 ± 4) conveniencef (using PCA) recalls ADs (annually)
cognition, SLE Convenience OR T3 vs
T1 0.89 (95%CI 0.59,
1.35)
≥ 55 10
Sanchez-Villegas A, G, BMI, smoking, PA, HR Qn5 v Qn1 2.70
Spain 1,388# 25.1# (62.6 ± DII FFQ Self-report or ADs 88# (every 2
et al., 2015*# M, EI, C (95%CI 1.22, 5.97)
6.1)# years)
Chan et al., 2014 Hong Kong 2,211 40.1 FFQ GDS-15 ≥ 8 81 (3.7%) 4
(continued on next page)
A.P. Matison et al.
Table 2 (continued )
Author, Year Country Sample % Age Dietary pattern Dietary Diagnosis of Incidence of Follow-up Adjustment for Results

± SD assessments)
years)
Fruit-vegetables OR Qr4
vs Qr1 OR 0.94 (95%CI
0.51, 1.74)
≥ 65 Fruit-vegetablesg, snacks- A, G, EI, BMI, PA, Snacks-drinks-milk OR
(71.8 ± drinks-milkh, meat-fishi IADLs, smoking, Qr4 vs Qr1 0.85 (95%CI
4.8) (using FA) alcohol, S, C, cognition 0.43, 1.68)
Meat-fish OR Qr4 vs Qr1
OR 1.04 (95%CI 0.56,
1.90)
Prudent** RR Low v
high 1.18 (95%CI 1.01,
7 G, baseline depressive
Prudentj, Westernk (using 1.39)###
Jacka et al., 2014* Australia 1,437 48.3 60–64 FFQ GDS-9 ≥ 6 na (4 and 7 symptoms, S, PA,
PCA) Western** RR High v
years) smoking, C
low 1.14 (95%CI 0.98,
1.34)
A, BMI, EI, smoking, PA,
50–77 12
United menopausal status, RR Qn5 vs Qn1 1.41
Lucas et al., 2014 43,685 100 (62.6 ± Inflammatory dietary*** FFQ Self-report and ADs 2,594 (every 2–4
States HRT, S, M, C, MHI, (95%CI 1.22, 1.63)
6.9) years)
alcohol, caffeine
A, BMI, EI, smoking, PA, Prudent RR Qn5 vs Qn1
12
Chocano-Bedoya United Prudent l, Westernm menopausal status, 1.05 (95%CI 0.91, 1.20)
50,605 100 50–77 FFQ Self-report and ADs 3,002 (every 2–4
et al., 2013 States (using PCA) HRT, S, caffeine, M, C, Western RR Qn5 vs Qn1
years)
MHI 1.05 (95%CI 0.89, 1.23)
6
12
≥ 65 Year 4 7.0%, A, G, S, EI, BMI,
Skarupski et al., United (every 4 T3 vs T1 β = − 0.03
3,502 59.0 (73.5 ± MedDiet FFQ CES-D-10 ≥ 4 Year 8 10.7%, smoking, alcohol, C,
2013 States years, mean SE=0.01 p<0.001
6.1) Year 12 13.4% cognition
7.2)
Cooked vegetables OR
0.93 (95%CI
0.84,1.02)##
Fruit OR 1.01 (95%CI
0.91, 1.11)##
Cooked vegetablen, fruito, Mediterranean style OR
50–55 Mediterranean stylep, 0.83 (95%CI 0.75,
EI, smoking, PA, S,
Rienks et al., 2013 Australia 6,060 100 (52.5 ± meat/processed meatq, FFQ CES-D-10 ≥ 10 647 (10.7%) 3 0.91)##
stress, BMI
1.5) dairyr, high fat/sugars Meat/processed meat
(using FA) OR 1.09 (95%CI 0.98,
1.21)##
Dairy OR 0.93 (95%CI

0.84, 1.02)##
High fat/sugar OR 1.08
(95%CI 0.96, 1.20)##
Whole food OR T3 vs T1
0.73 (95%CI 0.51, 1.02)
Akbaraly et al. 47–67 Whole foodt, processed A, G, EI, S, PA, smoking,
UK 3,059 26.2 FFQ CES-D-20 > 15 265 5 Processed food OR T3 vs
2009 (55.6) foodu (using FA) C, ADs, cognition
T1 1.69 (95%CI 1.10,
2.60)
Studies listed in reverse date order * Results use sub-analysis of participants meeting our age criteria; ** Includes participants with baseline depression (4.4%) but noted in text that excluding these participants produced
identical results; *** Identified using inflammatory markers; # data obtained direct from author via email; ##uses logistic regression and continuous dietary score; ### Result inverted for meta-analysis to reflect high
versus low adherence. Abbreviations: GDS - Geriatric Depression Scale; CES-D - Center for Epidemiologic Studies Depression Scale; FFQ – food frequency questionnaire; 24-h – 24 hour; MedDiet - Mediterranean diet; DII -
Dietary Inflammatory Index; ADII - Alternate Dietary Inflammatory Index; ADG - Australian Dietary Guidelines; GI - Glycaemic Index; BMI – body mass index; EI – energy intake; ADL – activities of daily living; PA –
Metagenics Inc. and AXA (Voortman et al., 2017). Two authors of
physical activity; A – age; G – gender; S – sociodemographic variables; C – comorbidities; M – multivitamin usage; SLE – stressful life events; ADs – antidepressant usage; IADLs - instrumental activities of daily living; HEI -
week, and fruit and vegetables <10 times/week; b meat/poultry/eggs 2–6 times/week, fish 1–5 times/week, beans/legumes ≥3 times/week, fruit and vegetables >7 times/week; c meat/poultry/eggs <5 times/week, fish
meats, French fries, desserts, high-fat dairy, refined grains; n includes cauliflower, cabbage, brussels sprouts, broccoli, green beans; o includes strawberries, pineapple, melon, apricots, mango; p includes garlic, peppers,
mushrooms, salad greens, pasta, red wine; q includes pork, bacon, sausages, lamb; r includes cream cheese, low-fat cheese, yoghurt, skim milk; s sweet biscuits, cakes, jam, meat pies, chocolate; t high intake of vegetables,
Healthy Eating Index score; MHI – mental health inventory score; HRT – hormone replacement therapy; OR – odds ratio; HR – hazard ratio; RR – relative risk; PCA – principle component analysis, FA – factor analysis; CI –
confidence interval; SD – standard deviation; SE – standard error; T – tertile; Qr – quartile; Qn – quintile;a n – number; na - not available; meat/poultry/eggs ≥7 times/week, fish ≤4 times/week, beans/legumes <2 times/
≥5 times/week, beans/legumes ≥3 times/week, fruit and vegetables ≥10 times/week; d high intake of fruit/fruit juice, vegetable juice, other vegetables, nuts, seeds, refined and whole grain products, pizza, pasta,
chocolate, sweets, snacks, hot beverages, processed meats, high-fat dairy; e high intake of red meat, butter and fats, alcoholic beverages, potatoes, tubers and starchy vegetables, soups, peas, legumes, desserts, processed
meats, high-fat dairy and low intake of low-fat dairy; f high intake of sugar beverages, fried foods, fast foods, mixed dishes and low intake of fish/seafood, margarine, oils, salad dressing, leaf vegetables; g high intake of
poultry, fish/seafood, wine; j fresh vegetables, salad, fruit, grilled fish; k roast meat, sausages, hamburgers, steak, chips, crisps, soft drink; l fruit, vegetables, fish, whole-grain products, low-fat dairy; m red and processed
vegetables, fruits, soy/soy products, legumes; h high intake of condiments, drinks, fast food, French fries, potatoe chips, sweets/desserts nuts, milk products, whole grains; i high intake of dim sum, red and processed meats,
another study reported financial associations with a company planning
to license the development of the dietary inflammatory index (Shivappa
et al., 2018).
3.3. Dietary patterns and incidence of depression
Dietary patterns were assessed in 17 studies, with the majority

assessing more than one dietary pattern. We were able to perform meta-
analysis on studies assessing the Mediterranean diet, dietary inflam
matory index, “healthy” diet, and western diet. Dietary patterns which
could not be combined in meta-analysis were included in a narrative
review.
3.3.1. Mediterranean diet

Three studies were combined by meta-analysis. Studies used either a
priori measures of the Mediterranean diet (Das et al., 2021; Skarupski
et al., 2013) or factor analysis to identify a “Mediterranean style” dietary
pattern (Rienks et al., 2013). Overall, there was no difference in odds of
incident depression between participants in the highest versus lowest
category of adherence to a Mediterranean diet (Fig. 2a, OR 0.93; 95%CI
0.84, 1.04; P = 0.22). Heterogeneity between studies was high (I2 = 81%
fruits and fish; u high intake of sweetened desserts, chocolate, fried food, processed meats, pies, refined grains, high-fat dairy products, condiments.
P = 0.006). A comparison of studies suggested Rienks et al. was the

major contributor to heterogeneity. Notable differences between Rienks
et al. and the other included studies were the underlying dietary
assessment (a priori versus a posteriori) and age of participants (50− 55
vs ≥ 65 years). Meta-analysis was not re-run excluding the Rienks et al.
study as only two studies remained.
3.3.2. Dietary inflammatory index / alternative dietary inflammatory

index
Three studies used either the original Dietary Inflammatory Index
(DII) (Sanchez-Villegas et al., 2015; Shivappa et al., 2016), or the
Alternate DII (Adjibade et al., 2019b) to assess the relationship between
the level of inflammation in the diet and risk of incident depression.
Overall, the meta-analysis indicated that participants in the highest
category of inflammatory diet had a higher odds of incident depression
compared to those in the lowest (Fig. 2b, OR 1.33; 95%CI 1.04, 1.70; P =
0.02). Heterogeneity between studies was low (I2 = 43%, P = 0.17). It
was estimated that for every 1,000 participants who switch from the
lowest to the highest category of inflammatory diet, depression inci
dence would increase by 12–68 cases (12 cases; 95%CI 1, 24: 68 cases;
95%CI 9, 129).
An additional study assessed the relationship between an inflam
matory dietary pattern and depression incidence. Lucas et al. (2014)
identified a dietary pattern associated with increased levels of bio
markers of inflammation in 43,685 females aged 50− 77 years. This di
etary pattern was used in turn to analyse the relationship of an
inflammatory diet and incident depression. The authors found an asso
ciation with increased risk of depression (RR 1.41; 95%CI 1.22, 1.63; for
the highest versus lowest quintile of inflammatory diet). This study was
not included in the meta-analysis as the underlying dietary score
differed from the DII. However, the results are consistent with those
contained in the meta-analysis.
3.3.3. “Healthy” diet

Eight studies were combined by meta-analysis. Three studies used
adherence to dietary guidelines (two Australian (Das et al., 2021; Lai
et al., 2017); one Dutch (Voortman et al., 2017)). Four studies used a
posteriori methods to identify dietary patterns which they labelled
“prudent” (Chocano-Bedoya et al., 2013; Jacka et al., 2014), “whole
food” (Akbaraly et al., 2009) or “vegetables-fruits” (Chan et al., 2014).
One study used an a priori measure of a “healthy” dietary pattern (Tsai,
2016).
Overall, the risk of incident depression did not differ between par
ticipants in the highest versus lowest category of intake (Fig. 2c, OR
A.P. Matison et al.
Table 3
Characteristics of included studies examining the association of foods/food groups with incident depression.
Author, Year Country Sample % Age Food Dietary Diagnosis of Incidence of Follow-up Adjustment for Results

± SD assessments)
years)
Vegetables (serving/day) HR ≥ 5
A, G, S, BMI, vs none 0.93 (95%CI 0.87, 0.98)
Medical services
Shang et al., Vegetables, fruits, 13,497 psychological distress, C, Fruits (serving/day) HR ≥ 3 vs
Australia 98,958 54.8 45–64 FFQ and 9
2020 chicken (13.7%) family history of none 0.85 (95%CI 0.79, 0.91)
medication**
depression, dementia Chicken (serving/day) HR ≥ 3 vs
none 1.16 (95%CI 1.04, 1.29)
Vegetables (times/week) OR ≥ 6
vs ≤ 5 0.79 (95%CI 0.49, 1.27)
Smoking, alcohol, areca
Chi et al., Fruits (times/week) OR ≥ 6 vs ≤
Taiwan 2,630 46.8 ≥ 53 Vegetables, fruits, tea FFQ CES-D-10 ≥ 10 13.4% 4 nut chewing, S, C,
2016 5 0.82 (95%CI 0.64, 1.07)
cognition
Tea (times/week) OR <3 vs ≥ 3
0.75 (95%CI 0.57, 0.97)
Citrus fruit and juices HR Qn5 vs
Qn1 0.82 (95%CI 0.74, 0.91)
A, S, smoking, BMI, Citrus fruit HR Qn5 vs Qn1 0.87
10 alcohol, PA, coffee, (95%CI 0.75, 1.01)
Chang et al., Citrus fruit and juices, Self-report or (every 2–4 modified alternative Citrus juices HR Qn5 vs Qn1
United States 45,985 100 52–77 FFQ 4,896
2016 tea, onions ADs years mean 8.7 MedDiet, EI, M, 0.90 (95%CI 0.82, 0.98)
± 6.3) menopausal status, HRT, Tea HR Qn5 vs Qn1 0.88 (95%CI
sleep, C 0.78, 1.00)
Onions HR Qn5 vs Qn1 0.99
(95%CI 0.89, 1.09)
8
Added sugar OR Qn5 vs Qn1

1.23 (95%CI 1.07, 1.41)
Total sugar OR Qn5 v Qn1 OR
0.99 (95%CI 0.83, 1.18)
Whole grains OR Qn5 vs Qn1
0.92 (95%CI 0.82, 1.02) P-trend
Added sugar, total A, EI, S, BMI, C, HRT, PA, 0.0166
sugar, whole grains, alcohol, smoking, SLE, Refined grains OR Qn5 vs Qn1
Gangwisch 50–79 8-item Burnam 4,643
United States 69,954 100 refined grains, fruit, FFQ 3 fatty acids, fruit, 1.12 (95%CI 1.01, 1.24)
et al. 2015 (63.7) scale ≥ 0.06 (6.6%)
vegetables, nuts/seeds, vegetables, legumes, Fruit OR Qn5 vs Qn1 0.88 (95%
legumes nuts/seeds, fibre, HEI CI 0.79, 0.99)
Vegetables OR Qn5 vs Qn1 0.88
(95%CI 0.79, 0.99)
Nuts/seeds OR Qn5 vs Qn1 0.92
(95%CI 0.83, 1.02)
Legumes OR Qn5 vs Qn1 0.99

(95%CI 0.86, 1.09)
Fruit OR ≥ 2 pieces/day vs < 2
pieces/day 0.82 (95%CI 0.70,
Mihrshahi 53–58
6 S, BMI, PA, alcohol, fish, 0.96)
et al., Australia 5,117 100 (55.5 ± Fruit, vegetables FFQ CES-D-10 ≥ 10 795 (12.7%)
(every 3 years) EI, C Vegetables OR ≥ 5 serves/day vs
2015 1.5)
0–1 serves/day 0.83 (95%CI
0.62, 1.10)
7
Ritchie A, S, C, BMI, HDL Males HR ≥ 3 cups 0.85 (95%CI
Caffeine (cups of 1,076 (2, 4 and 7
et al., France 5,785 60.4 ≥ 65 FFQ CES-D-20 ≥ 16 cholesterol, triglycerides, 0.66, 1.08); Females HR ≥ 3
coffee/tea) (18.6%) years, mean
2014 mobility cups 0.86 (95%CI 0.74, 1.01)
3.6)
United States 263,923 38.8 FFQ 9
A.P. Matison et al.

± SD assessments)
years)
Caffeinated coffee (cups/day)

OR ≥ 4 vs none 0.90 (95%CI
0.83, 0.98); decaffeinated coffee
(cups/day) OR ≥ 4 vs none 0.88
(95%CI 0.78, 1.00)
Caffeinated soft drink (cups/
day) OR ≥ 4 vs none 1.28 (95%
CI 1.12, 1.46); decaffeinated soft
drink (cups/day) OR ≥ 4 vs none
1.29 (95%CI 1.12, 1.49)
Iced tea (cups/day) OR ≥ 4 vs
none 0.91 (95%CI 0.82, 1.00);
decaffeinated iced tea (cups/
day) OR ≥ 4 vs none 1.37 (95%
CI 1.20, 1.57)
Hot tea (cups/day) OR ≥ 4 vs
Caffeinated drinks, none 1.10 (95%CI 0.92, 1.31);
50–71 Self-reported
Guo et al., decaffeinated drinks, 11,311 A, G, S, smoking, alcohol, decaffeinated hot tea (cups/day)
(61.6 ± clinical
2014 regular drinks, diet (4.3%) PA, BMI, EI OR ≥ 4 vs none 1.27 (95%CI
5.3) diagnosis
drinks 0.98, 1.65)
Regular soft drinks (cups/day)
OR ≥ 4 vs none 1.22 (95%CI
1.03, 1.45); diet soft drinks
9
(cups/day) OR ≥ 4 vs none 1.31

(95%CI 1.16, 1.47)
Regular Fruit drinks (cans/day)
OR ≥ 4/day vs none 1.08 (95%
CI 0.79, 1.46); diet Fruit drinks
(cups/day) OR ≥ 4/day vs none
1.51 (95%CI 1.18, 1.92)
Regular sweetened iced tea
(cans/day) OR ≥ 4/day vs none
0.94 (95%CI 0.83, 1.08); diet
sweetened iced tea (cans/day)
OR ≥ 4/day vs none 1.25 (95%
CI 1.10, 1.41)
Meat RR < 5 times/week vs ≥ 5
times/week 0.90 (95%CI 0.68,
1.19)

Fish RR ≥ 3 times/week vs < 3
times/week 1.03 (95%CI 0.67,
Almeida Meat, fish, full-cream 3–8
1.61)
et al., Australia 4,636 0 65–83 milk: reduced-fat milk FFQ GDS-15 ≥ 7 229 (4.5%) (mean 5.7 ± A, S, medical illnesses
Reduced fat milk used vs
2013 ratio, salt 0.9)
reduced fat not used RR 0.79
(95%CI 0.60, 1.03)
No salt added to meals vs salt
added to meals RR 0.88 (95%CI
0.67, 1.16)
Tsai et al., 8 A, G, S, stress, C, IADL, Alcohol OR Heavy (≥ 1 time/
Taiwan 2,145 46.8 ≥ 53 Alcohol FFQ CES-D-10 ≥ 10 n/a
2013 (4 and 8 years) audio acuity week, ≥ 2 drinks/time) vs <1
A.P. Matison et al.
± SD assessments)
years)
time/month 0.70 (95%CI 0.30,

1.64)
Meat and poultry (times/week)
OR ≥ 3 vs < 3 1.31 (95%CI 0.90,
1.91)
Dairy (times/week) OR ≥ 3 vs <
3 0.87 (95%CI 0.61, 1.24)
Eggs (times/week) OR ≥ 3 vs <3
Meat and poultry, A, G, S, smoking, alcohol, 0.73 (95%CI 0.50, 1.03)
Tsai et al.,
Taiwan 1,609 42.4 ≥ 65 dairy, eggs, legumes, FFQ CES-D-10 ≥ 10 327 (20%) 4 betel-nut chewing, Legumes (times/week) OR ≥ 3
2012
fish, seafood, cereals function, PA, cognition, C vs <3 1.08 (95%CI 0.76, 1.55)
Fish (times/week) OR ≥ 3 vs < 3
0.91 (95%CI 0.62, 1.34)*
Seafood (times/week) OR ≥ 3 vs
< 3 0.92 (95%CI 0.51, 1.65)
10
Cereal (bowls/d) OR ≥ 3 vs < 3

0.85 (95%CI 0.58, 1.26)
A, menopausal status, S,
Fish RR ≥ 5 (times/week) vs < 1
Lucas et al., 50–75 Self-report and 10 obesity, smoking, PA, C,
United States 54,632 100 Fish FFQ 2,823 (times/month) 1.07 (95%CI
2011 (a) (62.8) ADs (every 2 years) M, EI, protein, fatty acids
0.74, 1.55)
intake, alcohol
Caffeinated coffee RR ≥ 4 cups/
day vs ≤ 1 cup/wk 0.80 (95%CI
A, EI, menopausal status,
Lucas et al., 50–75 Caffeinated coffee, Self-report and 10 0.64, 0.99)
United States 50,739 100 FFQ 2,607 (5%) HRT, smoking, BMI, PA,
2011 (b) (63) decaffeinated coffee ADs (every 2 years) Decaffeinated coffee RR ≥ 2
S, C, MHI
cups/day vs ≤ 1 cup/wk 0.84
(95%CI 0.70, 1.01)
Finland, Italy Zung Self-rating A, C, baseline depressive
Bots et al., 70–89 Alcohol OR >31 g/day vs <1 g/
and the 526 0 Alcohol Diet history Depression Scale 59 (11%) 5 and cognition, S, PA,
2008 (75.2) day 0.64 (95%CI 0.23, 1.80)
Netherlands ≥ 48/80 alcohol, cholesterol
Studies listed in reverse date order. * Obtained by email from author; ** deduced from medical services and medication claimed via Medicare benefits schedule or Pharmaceutical benefits schedule. Abbreviations: GDS -

Geriatric Depression Scale; CES-D - Center for Epidemiologic Studies Depression Scale; n – number; FFQ – food frequency questionnaire; MedDiet - Mediterranean diet; HRT – hormone replacement therapy; HDL – high-
density lipoprotein; BMI – body mass index; HEI - Healthy Eating Index score; EI – energy intake; ADs – antidepressant usage; PA – physical activity; A – age; G – gender; S – sociodemographic variables; C – comorbidities;
M – multivitamin usage; SLE – stressful life events; IADLs – instrumental activities of daily living; MHI – mental health inventory; CI – confidence interval; SD – standard deviation; Qn – quintile; OR – odds ratio; HR –
hazard ratio; RR – relative risk.
A.P. Matison et al.
Table 4
Characteristics of included studies examining the association of nutrients with incident depression.
Author, Year Country Sample % Age Nutrient Dietary Depression Incidence Follow-up Adjustment for Results
size females range assessment diagnosis n (%) period (years) confounders

(mean tool (interim
± SD assessments)
years)
Vitamin A OR Qr1 vs Qr4 1.28 (95%

CI 0.59, 2.78)
A, BMI, S, smoking, Vitamin C OR Qr1 vs Qr4 1.86 (95%
≥ 75
alcohol, self- CI 0.85, 4.08)
Das et al., 2021 Australia 781 0 (81.2 ± Vitamins A, C, E, zinc Diet history GDS-15 ≥ 5 13.2% 3
reported health, PA, Vitamin E OR Qr1 vs Qr4 2.05 (95%
4.5)
C, EI, ADs CI 0.99, 4.28)
Zinc OR Qr1 vs Qr4 2.42 (95%CI
1.17, 5.04)
Total flavonoids HR Qn5 vs Qn1 0.88
(95%CI 0.80, 0.97)
Flavonols HR Qn5 vs Qn1 0.93 (95%
CI 0.85, 1.02)
Flavones HR Qn5 vs Qn1 0.94 (95%
CI 0.86, 1.03)
Total flavonoids, flavonols, A, S, smoking, BMI,
10 Flavanones HR Qn5 vs Qn1 0.91
flavones, flavanones, alcohol, PA, coffee,
Self-report or (every 2–4 (95%CI 0.82, 0.99)
Chang et al., 2016 United States 45,985 100 52–77 flavan-3-ol, anthocyanins, FFQ 4,896 MedDiet score, EI,
ADs years mean Flavan-3-ols HR Qn5 vs Qn1 0.94
polymeric flavonoids, M, menopausal
8.7 ± 6.3) (95%CI 0.86, 1.03)
proanthocyanidins status, HRT, sleep, C
Anthocyanin HR Qn5 vs Qn1 1.00
(95%CI 0.91, 1.10)
Polymeric flavonoids HR Qn5 vs Qn1
11
0.91 (95%CI 0.83, 1.00)

Proanthocyanidins HR Qn5 vs Qn1
HR 0.89 (95%CI 0.81, 0.97)
Glucose OR Qn5 vs Qn1 1.16 (95%CI
0.99, 1.36)
Lactose OR Qn5 vs Qn1 0.81 (95%CI
A, EI, S, BMI, C,
0.72, 0.92)
HRT, PA, alcohol,
Fructose OR Qn5 vs Qn1 1.04 (95%CI
Glucose, sucrose, lactose, 8-item smoking, SLE, fatty
Gangwisch et al. 50–79 4,643 0.89, 1.21)
United States 69,954 100 fructose, starch, FFQ Burnam scale 3 acids, fruit,
2015 (63.7) (6.6%) Starch OR Qn5 vs Qn1 1.01 (95%CI
carbohydrate, fibre ≥ 0.06 vegetables, legumes,
0.85, 1.20)
nuts/seeds, fibre,
Carbohydrate OR Qn5 vs Qn1 0.97
HEI
(95%CI 0.77, 1.22)
Fibre OR Qn5 vs Qn1 0.86 (95%CI
0.76, 0.98)
Folate: Males OR T3 vs T1 1.68 (95%
CI 0.77,3.66); Females OR T3 vs T1

0.84 (95%CI 0.46, 1.55)
67–84 Vitamin B-6 : Males OR T3 vs T1 0.83
Gougeon et al., Three 24-h GDS-30 ≥ 11 170 3 A, PA, autonomy,
Canada 1,368 50.5 (74 ± Folate, vitamins B-6, B-12 (95%CI 0.38, 1.80); Females OR T3
2016 recalls or ADs (12.5%) (annually) SLE, EI
4) vs T1 0.70 (95%CI 0.37, 1.30)
Vitamin B-12: Males OR T3 vs T1
0.42 (95%CI 0.20, 0.90); Females OR
T3 vs T1 0.90 (95%CI 0.51, 1.60)
Total energy OR T3 vs T1 0.55 (95%
A, G, S, smoking, EI,
67–84 CI 0.34, 0.87)
Gougeon et al., Three 24-h GDS-30 ≥11 170 3 PA, BMI, C,
Canada 1,358 50.4 (74 ± Total energy, protein, fat No significant association between
2015 recalls or ADs (12.5%) (annually) autonomy,
4) incidence of depression and protein
cognition, SL
or fat (data not provided)
A.P. Matison et al.
Author, Year Country Sample % Age Nutrient Dietary Depression Incidence Follow-up Adjustment for Results
size females range assessment diagnosis n (%) period (years) confounders
(mean tool (interim

± SD assessments)
years)
2 cohorts: Zinc HCS OR Qn5 vs Q1 0.73 (95%CI

HCS 270
HCS 52 55–85 0.44, 1.19); ALSWH OR Qn5 vs Qn1
(12.9%)
2,092 0.70 (95%CI 0.55, 0.89)
Vashum et al., CES-D-20 ≥ A, G, smoking,
Australia Zinc; zinc: iron ratio FFQ 6 Zinc: iron ratio HCS OR Qn5 vs Qn1
2014 16 or ADs ALSWH alcohol, S, M, C, BMI
ALSWH 1.23 (95%CI 0.57, 2.64); Zinc: iron
100 50–55 1,830
9,738 ratio ALSWH OR Qn5 vs Qn1 0.99
(18.8%)
(95%CI 0.99, 1.00)
Total vitamin D intake OR ≥800 IU/
A, S, BMI, waist-to-
d vs <100 IU/d 0.87 (95%CI 0.74,
hip ratio, smoking,
8-item 1.02); only from food sources OR
Bertone-Johnson alcohol, HRT, EI,
United States 57,288 100 50–79 Vitamin D FFQ Burnam scale 3,524 3 ≥400 IU/d vs <100 IU/d 0.80 (95%
et al., 2011 omega-3, PA,
≥ 0.06 CI 0.67, 0.95); only from
function, C, solar
supplements OR ≥800 IU/d vs none
irradiance, ADs
1.02 95%CI (0.83, 1.25)
ALA (each increase of 0.5g) RR 0.81
(95%CI 0.69, 0.95)
EPA and DHA (each increase of 0.3g)
RR 0.96 (95%CI 0.84, 1.10)
A, menopausal LA (each increase of 5g) RR 1.33
ALA, EPA, DHA, LA, AA, 10 status, S, obesity, (95%CI 1.10, 1.61)
Lucas et al., 2011 50–75 Self-report
United States 54,632 100 ALA:LA, omega-3:omega-6 FFQ 2,823 (every 2 smoking, PA, C, M, AA (each increase of 0.1g) RR 1.06
(a) (62.8) and ADs
ratio years) EI, protein, fatty (95%CI 0.89, 1.26)
acids, alcohol ALA:LA (each increase of 0.05U) RR
12
0.77 (95%CI 0.66, 0.88)

omega-3:omega-6 ratio (each
increase of 0.1U) RR 0.74 (95%CI
0.61, 0.90)
Total folate OR 1.00 (95%CI 1.00,
Year 4 471 1.00); only from food OR 1.00 (95%
(13.7%), CI 1.00, 1.00)*
A, G, S, ADs, alcohol,
≥ 65 Year 8 256 12 Total vitamin B-6 OR 1.00 (95%CI
Skarupski et al., 10-item CES- smoking, lag,
United States 3,503 59.0 (73.5 ± Folate, vitamins B-6, B-12 FFQ (10.7%), (every 4 years, 1.00, 1.00); only from food OR 1.02
2010 D≥4 cognitive function,
6.1) Year 12 mean 7.2) (95%CI 0.98, 1.05)*
physical disability
260 Total vitamin B-12 OR 1.00 (95%CI
(13.4%) 1.00, 1.00); only from food 1.00
(95%CI 0.99, 1.00)*
Total energy OR 1.00 (95%CI 1.00,
1.00)*
Fat OR 1.00 (95%CI 0.99, 1.01)*
Zung Self-

Finland, Italy Total energy, fat, saturated Saturated fat OR 1.01 (95%CI 0.99,
70–89 rating
Bots et al., 2008 and the 526 0 fat, polyunsaturated fat, Diet history 59 (11%) 5 A 1.02)*
(75.2) Depression
Netherlands cholesterol Polyunsaturated fat OR 0.96 (95%CI
Scale ≥ 48/80
0.92, 1.01)*
Cholesterol OR 1.00 (95%CI 1.00,
1.00)*
Studies listed in reverse date order. * using logistic regression and continuous intake Abbreviations: GDS - Geriatric Depression Scale; CES-D - Center for Epidemiologic Studies Depression Scale; FFQ – food frequency
questionnaire; MedDiet - Mediterranean diet; HRT – hormone replacement therapy; HDL – high-density lipoprotein ; IU/d – international units per day; BMI – body mass; PA – physical activity; HEI - Healthy Eating Index
score; ADs – antidepressant usage; EI – energy intake; 24-h – 24 hour; HCS - Hunter Community Study; ALSWH - Australian Longitudinal Study on Women’s Health; A – age; G – gender; S – sociodemographic variables; C –
comorbidities; M – multivitamin usage; SLE – stressful life events; CI – confidence interval; n – number; SD – standard deviation; T – tertile; Qr – quartile; Qn – quintile; OR – odds ratio; HR – hazard ratio; RR – relative risk;
ALA – α-linolenic acid; EPA – eicosapenic acid; DHA – docosahexaenoic acid; LA – linoleic acid, AA – arachidonic acid.
0.97; 95%CI 0.95, 1.00; P = 0.09). Heterogentity between studies was (Rienks et al., 2013) were not associated with longitudinal risk of inci
low (I2 = 30%, P = 0.19). dent depression.
3.3.4. Western diet 3.4. Foods and food groups and incidence of depression
Seven studies were combined by meta-analysis. Five studies used a
posteriori methods and identified dietary patterns which they labelled as 3.4.1. Vegetables
“western” (Chocano-Bedoya et al., 2013; Jacka et al., 2014), “conve Four studies investigated the relationship between intake of vege
nience” (Gougeon et al., 2015), “high fat/sugar” (Rienks et al., 2013) or tables and incidence of depression. Participants intake of vegetable
“processed food” (Akbaraly et al., 2009). One study used the proportion servings were compared between two categories (≤ 5/week and ≥ 6/
of ultra-processed foods in the diet (Adjibade et al., 2019a) and one week; Chi et al., 2016), three categories (0− 1/day, 2− 4/day and ≥
employed an a priori measure of a western dietary pattern (Tsai, 2016). 5/day; Mihrshahi et al., 2015)), five categories (0− 1/day, 2/day, 3/day,
Overall, participants in the highest category of consumption had a 4/day and ≥ 5/day; Shang et al., 2020), or quintiles (Gangwisch et al.,
higher risk of incident depression than those in the lowest category 2015)).
(Fig. 2d OR 1.15 95%CI 1.04, 1.26; P = 0.005). Heterogeneity between Overall, participants in the highest category of intake had a lower
studies was low (I2 = 46%, P = 0.08). It was estimated that for every odds of incident depression than participants in the lowest category
1,000 participants who switch from the lowest to the highest category of (Fig. 3a, OR 0.91; 95%CI 0.87, 0.96; P<.001). There was no evidence of
western diet, depression incidence would increase by 5–33 cases (5 heterogeneity (I2 = 0%; P = 0.69). It was estimated that for every 1,000
cases; 95%CI 1, 9: 33 cases; 95%CI 9, 55). participants who switch from the lowest to the highest category of
vegetable intake, depression incidence would decrease by 3–21 cases (3
3.3.5. Other dietary patterns cases; 95%CI 1, 5: 21 cases; 95%CI 9, 31).
Five studies contained dietary patterns, which could not be com
bined by meta-analysis. A traditional diet was assessed in two Taiwanese 3.4.2. Fruit
studies. Participants (53+ years) consuming a traditional diet reported a Four studies investigated the relationship between fruit intake and
higher odds of incident depression versus those who did not (OR 1.60; incidence of depression. Participants intakes of fruit were compared
95%CI 1.12, 2.29; Tsai, 2016), whereas higher consumption of a tradi between two categories (≤ 5/week and ≥ 6/week (Chi et al., 2016) or <
tional diet was not associated with incident depression in a Canadian 2/day and ≥ 2/day (Mihrshahi et al., 2015)), four categories (none,
cohort aged 67–84 years (Gougeon et al., 2015). In a cohort of older 1/day, 2/day and ≥ 3/day; Shang et al., 2020) or quintiles (Gangwisch
women (50− 79 years), participants consuming carbohydrates with a et al., 2015).
higher glycaemic index, but not overall glycaemic load were found to Overall, participants in the highest category of intake had a lower
have higher odds of depressive outcomes (Gangwisch et al., 2015). odds of incident depression than those in the lowest category (Fig. 3b,
Higher consumption of a “varied” (Gougeon et al., 2015), a “snacks- OR 0.85; 95%CI 0.81, 0.90; P < 0.001). There was no evidence of
drinks-milk” (Chan et al., 2014), “dairy” (Rienks et al., 2013), “meat- heterogeneity (I2 = 0%; P = 0.87). It was estimated that for every 1,000
fish” (Chan et al., 2014), “meat-processed/meat” (Rienks et al., 2013), participants who switch from the lowest to the highest category of fruit
“cooked vegetable” (Rienks et al., 2013) and “fruit” dietary pattern intake, depression incidence would decrease by 5–47 cases (5 cases; 95%
Fig. 3. Forest plot depicting meta-analyses of studies assessing the longitudinal association between intake of foods/food groups and incident depression. Results are
odds ratios comparing the highest category of adherence with the lowest category.
13
Fig. 4. Funnel plots for the analysis of “Healthy” and Western diets and incident depression. Results compare participants in the highest category versus lowest
category of adherence.
CI 4, 7: 37 cases; 95%CI 24, 47). intake and decreased odds of depression (p = 0.0166), however, there
Although not combined in meta-analysis due to potential underlying was no significant difference between the first and fifth quintiles of
differences, a beneficial relationship was also reported between higher intake (Gangwisch et al., 2015). The ratio of non-whole grains to whole
consumption of citrus fruit and juices and the incidence of depression grains was not associated with depressive risk (Gangwisch et al., 2015).
(HR 0.82; 95%CI 0.74, 0.91; Chang et al., 2016) in 45,984 females aged Higher intake of added sugar, but not total sugar was associated with
52− 77 years. increased risk of depression (OR 1.23; 95%CI 1.07, 1.41; Gangwisch
et al., 2015). A similar result was reported for higher intake of regular
3.4.3. Fish soft drink (OR 1.22; 95%CI 1.03, 1.45; Guo et al., 2014). Higher intakes
Three studies investigated the relationship between intake of fish of diet soft drinks, diet fruit drinks and diet sweetened iced tea were all
and incidence of depression. Fish consumption frequency was either associated with higher odds of incident depression (Guo et al., 2014).
compared between two categories (≥ 3/week and < 3/week; Almeida Results were mixed for studies that investigated meat and chicken.
et al., 2013; Tsai et al., 2012) or five categories (< 1/month, 1–3/month, Higher chicken consumption was associated with increased odds of
1/week, 2− 4/week and ≥ 5/week; Lucas et al., 2011a). depression (HR 1.16; 95%CI 1.04, 1.29; Shang et al., 2020), whereas
Overall, there was no association between fish intake and incident meat (Almeida et al., 2013), and combined meat and poultry (Tsai et al.,
depression (Fig. 3c, OR 1.00; 95%CI 0.80, 1.26; P = 0.99). There was no 2012) were not associated with depressive risk.
evidence of heterogeneity (I2 = 0%; P = 0.83). Intake of dairy (Almeida et al., 2013; Tsai et al., 2012), legumes (Tsai
Although not combined in meta-analysis due to potential underlying et al., 2012), eggs (Tsai et al., 2012), nuts and seeds (Gangwisch et al.,
differences, higher consumption of seafood was not associated with the 2015), onions (Chang et al., 2016), cereal (Tsai et al., 2012) and alcohol
risk of depression in 1,609 Taiwanese adults aged 65 years and older (Bots et al., 2008; Tsai et al., 2013) were not associated with the inci
(Tsai et al., 2012). This finding adds weight to the results of the dence of depression.
meta-analysis.
3.5. Nutrients and incidence of depression
3.4.4. Other food types
Results from studies investigating a wide range of other foods and
Ten studies examined the associations between various nutrients and
beverages could not be analysed by meta-analysis due to underlying
longitudinal risk of depression, however there were insufficient studies
differences in the types of foods. Below is a summary of findings from
examining each of the nutrients to combine in meta-analysis. The results
these studies.
of these studies are discussed in a narrative review.
Three studies investigated the association between tea consumption
and incidence of depression. Two studies were conducted in the United
3.5.1. Vitamins, minerals and flavonoids
States (Chang et al., 2016; Guo et al., 2014) and one in Taiwan (Chi
Studies have investigated the associations between B-group vita
et al., 2016). Due to differences in the types of tea consumed between
mins, vitamin C, vitamin A, vitamin D and vitamin E with incident
these populations, a meta-analysis was not performed. The Taiwanese
depression. In an ethnically diverse cohort of American adults (65+
study reported a beneficial association between higher tea consumption
years), higher intake of vitamin B12 was associated with lower incident
(mainly Oolong tea; Chi et al., 2016) and reduced incidence of depres
depression (Skarupski et al., 2010), while in a Canadian cohort aged
sion (OR 0.75; 95%CI 0.57, 0.97), whereas the studies conducted in the
67− 84 years, males in the highest tertile of vitamin B12 intake had a
United States found no association for hot or iced tea (Chang et al., 2016;
decreased odds of incident depression compared with males in the
Guo et al., 2014) and a detrimental association for decaffeinated iced tea
lowest tertile (Gougeon et al., 2016). This association was not reported
(OR 1.37; 95%CI 1.20, 1.57; Guo et al., 2014).
in females (Gougeon et al., 2016), or in the results for vitamin B6 and
Coffee intake was assessed in three studies, however, as one of these
folate in either study (Gougeon et al., 2016; Skarupski et al., 2010). In a
studies used a combined intake of coffee and tea, the three studies were
cohort of American females aged 50− 79 years, higher vitamin D intake
not combined in meta-analysis. Results were mixed. Two studies re
from food sources, but not total dietary vitamin D intake was associated
ported a beneficial association between higher caffeinated coffee intake
with a lower risk of incident depression (Bertone-Johnson et al., 2011).
and lower incidence of depression (OR 0.90; 95%CI 0.83, 0.98 (Guo
No associations were reported between intakes of vitamins A, C or E and
et al., 2014) and RR 0.80 95%CI 0.64, 0.99 (Lucas et al., 2011b)),
the incidence of depression (Das et al., 2021).
whereas neither reported an association for decaffeinated coffee. A
The only mineral assessed was zinc. Two studies reported that higher
combined intake of caffeinated coffee and tea was not associated with
intake was associated with reduced odds of incident depression (Das
risk of depression (Ritchie et al., 2014).
et al., 2021; Vashum et al., 2014).
One study assessed whole grain intake and reported a trend of higher
Flavonoids were also assessed. Higher intake of total flavonoids was
14
associated with a decreased risk of incident depression in a cohort of conclusions on the associations between the Mediterranean diet, fish
45,985 participants aged 52− 77 years (Chang et al., 2016). intake and incident depression as both were only examined in three
studies (total participants: Mediterranean diet 10,343; fish 60,877).
3.5.2. Sugars, starch and dietary fibre Moreover, it is likely that the types of fish consumed varied by study as
One study assessed the association between sugars, starch, dietary they were conducted in different countries.
fibre and incident depression. Higher intake of both lactose and fibre This review’s finding that higher intake of fruits and vegetables are
was associated with decreased odds of depression, whereas no associa associated with lower rates of incident depression, while higher adher
tion was reported between intake of either glucose, sucrose, fructose or ence to the DII (i.e. pro-inflammatory diets) and a western diet are
starch and incident depression (Gangwisch et al., 2015) associated with increased risk of incident depression are generally in-
line with existing meta-analyses in the general population (Lassale
3.5.3. Macronutrients, total energy and fatty acids et al., 2019; Li et al., 2017; Liu et al., 2016; Molendijk et al., 2018;
Three studies examined macronutrients. No association was reported Saghafian et al., 2018; Wang et al., 2018).
for carbohydrate (Gangwisch et al., 2015), fat (Bots et al., 2008; Gou Our finding that “healthy” and Mediterranean dietary patterns were
geon et al., 2015) or protein (Gougeon et al., 2015). Results for total not associated with a reduced risk of depression differs from a recent
energy were mixed with a study in Canadians aged 68–84 years finding meta-analysis conducted in older adults (Wu et al., 2020). However, it is
higher total energy intakes were associated with lower odds of incident difficult to compare these findings with our review due to different in
depression (Gougeon et al., 2015), while in a cohort of males aged clusion criteria (cross sectional, cohort and interventions vs only cohort
70− 89 years no association was detected (Bots et al., 2008). in the current study), participants age (65+ vs 45+) and examined
One study examined fatty acids. A beneficial association between outcomes (incident depression, depressive symptoms, combined anxiety
higher intake of the omega-3 polyunsaturated fatty acid, α-linolenic and depression vs incident depression). Nevertheless, it should be noted
acid, and decreased risk of depression was reported, whereas higher that several previous meta-analyses in general populations have re
intake of the omega-6 polyunsaturated fatty acid, linoleic acid, was ported reduced depression risk over time associated with “healthy” di
associated with increased incident depression (Lucas et al., 2011a). etary patterns (Lassale et al., 2019; Molendijk et al., 2018). Our finding
that “healthy” and Mediterranean diets are not associated with reduced
3.6. Sensitivity analysis risk of incident depression may be due to our focus on adults aged 45+
years. Further research is required to understand divergent results be
Sensitivity analysis excluding studies which received a red rating for tween different aged participants.
assessment of diet and/or depression did not change the results of the It is surprising that there was no association for Mediterranean and
main analyses. However, we were unable to perform sensitivity analysis “healthy” diets, while higher intake of fruits and vegetables (key com
for DII, vegetables, fruit and fish as only one study per category did not ponents of Mediterranean and “healthy” diets) were associated with
receive a red rating. lower incident depression, and pro-inflammatory diets were associated
Sensitivity analysis excluding studies which only reported HRs did with increased incident depression (Mediterranean diet foods are
not change the results from the main analysis. generally anti-inflammatory; Shivappa et al., 2014). It is possible that as
a wide variety of different methods were used to define Mediterranean
3.7. Risk of publication bias and “healthy” diets underlying differences in diets combined by
meta-analysis has prevented associations being detected. In contrast,
Publication bias was assessed for the two meta-analyses containing studies assessing the DII used similar validated indices. Further research
nearly 10 studies. There was no indication of publication bias based on is required into the relationship between healthy dietary patterns and
Egger’s test (“healthy diet” p = 0.146, western diet p = 0.691) and visual the foods they contain to better understand the relationships detected.
inspection of funnel plots for asymmetry (Fig. 4). There are likely to be a number of biological mechanisms responsible
for the associations observed. Higher levels of oxidative stress markers
4. Discussion and lower levels of anti-oxidant markers have been observed when
comparing people with and without depression (Liu et al., 2015). An
This systematic review and meta-analysis analysed the evidence of tioxidants contained in high levels in fruits and vegetables may help
longitudinal observational studies assessing associations between diet protect the brain from oxidative damage (Smaga et al., 2015). Depres
and incident depression in adults aged 45 years and older. Findings sion has also been associated with increased levels of pro-inflammatory
support a beneficial relationship between higher intake of both vege cytokines (Goldsmith et al., 2016). Emerging evidence suggests the
tables and fruits with lower odds of incident depression, whereas higher importance of the gut microbiota in protecting against inflammation
inflammatory diets and a western dietary pattern were associated with (Ghosh et al., 2020). The consumption of wholegrains, resistant starch
increased risk of incident depression. Our results do not support a and, potentially, vegetables appear to beneficially impact the composi
relationship between a Mediterranean dietary pattern, a “healthy” di tion of the gut microbiome (Graf et al., 2015). It is also possible that diet
etary pattern or fish intake and incident depression. To our knowledge, impacts incidence of depression indirectly via the role of diet in chronic
our meta-analysis is unique in including dietary patterns (western diet diseases such as cardiovascular disease. A strong relationship has been
and DII) and food groups (vegetables, fruit and fish) not previously demonstrated between myocardial infarction and subsequent risk of
performed in older adults. depression (White and Groh, 2007).
The magnitude of the association was greatest for the detrimental A number of limitations should be considered when interpreting the
association between the DII and incident depression (33% increased results of our meta-analyses. Our meta-analyses generally included a
risk), however only three studies with a total of 7,826 participants small number of studies limiting the robustness of our results and
(Adjibade et al., 2019b did not provide sample size for sub analysis of making it difficult to assess publication bias. Although we did not detect
participants meeting inclusion criteria) were included. The evidence publication bias in the two meta-analyses which combined nearly 10
supporting a beneficial relationship between vegetables and fruits and studies, we were unable assess publication bias in the other five meta-
incident depression appears to be particularly strong with a total sample analyses due to the low number of studies (Sterne et al., 2011). Addi
size of 176,649 and low heterogeneity (I2 = 0%) between the four tionally, due to the small number of studies focusing on participants
studies. Similarly, evidence of a detrimental relationship between a aged 45–65 years or 65+ years, we were unable to perform the planned
western diet and incident depression is supported by seven studies, with sub-analysis by age group. As there are no generally accepted measures
a total sample size of 79,917 (I2 = 46%). It is difficult to draw of western or “healthy” diets, we have combined studies that appear to
15
contain similar underlying foods, which may have led to inconsistent associations detected to be due to reverse causation. We also chose to
results when compared to previous meta-analyses. Additionally, studies focus on incident of depression rather than change in symptoms as this
generally took place in high-income countries, hence results may not be outcome is likely to have more clinical significance. Unlike previous
generalisable to low-income countries. reviews, we have excluded studies where the outcome was psychological
As for all observational studies, it is also possible that the associa distress rather than depression to reduce heterogeneity between studies.
tions we detected reflect residual confounding. A review of the literature Including all dietary types has allowed a better overall understanding of
suggests that sex, socio-economic status, education, physical activity, the effect of diet on depression and helped highlight seemingly incon
living arrangements, functional or cognitive impairment and chronic sistent results (e.g. results suggesting a benefit of fruit and vegetables but
health conditions are associated with both dietary intake and depressive not a “healthy” or Mediterranean diet).
risk (Cebrino and Portero de la Cruz, 2020; Djernes, 2006; Jacka et al.,
2015; Kim et al., 2015). Adjustment of confounders was inconsistent 5. Conclusion and future direction
between studies. For example, only two studies adjusted for
socio-economic status (Chang et al., 2016; Jacka et al., 2014), although The findings from this systematic review and meta-analysis suggest
24 of the remaining 31 studies adjusted for related measures such as that in adults aged 45+ years, a beneficial relationship exists between
education and income. Baseline depression score was only adjusted for higher intakes of vegetables and fruits and lower incident depression
by one study (Jacka et al., 2014), with three other studies adjusting for while high adherence to western and pro-inflammatory diets were
baseline Mental Health Inventory score (Chocano-Bedoya et al., 2013; associated with an increased risk of depression. Mediterranean diets,
Lucas et al., 2014) or psychological distress (Shang et al., 2020). There “healthy” diets and fish intake were not associated with incident
are arguments for and against adjusting for baseline depression. By not depression. Results of this meta-analysis should however be viewed with
adjusting, associations detected may have been due to subclinical caution due to the observational nature and limited number of included
depression at baseline. Countering this argument, as dietary patterns are studies. The varied measures of diet and depression also make it difficult
generally stable in older adults (Thorpe et al., 2019), baseline depression to combine and compare studies.
may reflect dietary intake prior to baseline assessment and therefore Our results provide a promising direction for further investigation.
including as a confounder adjusts for some of the impact of diet. More high-quality intervention and cohort studies using middle-aged
Moreover, there was no consistency between the treatment of comor and older adults (including participants in low-income countries) and
bidities between studies. It appears likely that comorbidities contribute using clinical diagnosis of depression are required to increase the
to the development of depression. Poor diet may contribute to these robustness of our findings. This should allow greater analysis by age
chronic conditions, therefore studies which adjusted for these comor group and may allow more detailed dietary recommendations to be
bidities may have adjusted out this indirect influence of diet on provided.
depression. As studies generally adjusted for baseline confounders, it is
also possible that cofounders as well as diet (despite the observed sta Author contributions
bility of diet in older adults; Thorpe et al., 2019) have changed over the
study duration (3–12 years) impacting results. It is also possible that AM designed the search strategy, AM and SR screened studies,
unknown confounders exist. extracted data and performed quality assessment, AM performed data
The issue of confounding could be mitigated by the use of inter analysis and risk of bias. SR and KM oversaw the planning and execution
vention trials, however to our knowledge, there are currently no such of the research. AM wrote the first draft of the article. All authors read,
studies which assess the association between diet and incident of provided feedback, and approved the final manuscript.
depression in population-based community dwelling adults aged 45+
years. To date interventions studies have generally been of short dura Funding
tion (≤ 12-months) and based on participants with specific health con
ditions, hence results may not be applicable to population-based This research did not receive any specific grant from funding
cohorts. The lack of intervention studies in this area may be due to issues agencies in the public, commercial, or not-for-profit sectors. AM is
such as blinding, expectation bias and adherence associated with long- supported through an Australian Government Research Training Pro
term dietary interventions. Therefore, it is necessary to rely on obser gram Scholarship.
vational studies for evidence.
No study received a green rating for all categories of quality assess Declaration of Competing Interest
ment. Firstly, no study used clinical diagnosis as the sole method for
diagnosing depression. Depression was diagnosed using a variety of The authors declare that they have no known competing financial
depression scales, self-report of diagnosis or antidepressant usage. This interests or personal relationships that could have appeared to influence
may have caused participants depressive status to be inconsistently the work reported in this paper.
classified. Additionally, this has meant we were unable to perform the
planned subgroup analysis of studies with and without clinical diag
Appendix A. Supplementary data
nosis. Secondly, ten studies used unvalidated tools to assess diet. Brief
dietary assessment tools used in large observational studies are known to
Supplementary material related to this article can be found, in the
contain a level of error in capturing dietary intake which makes the lack
online version, at doi:https://doi.org/10.1016/j.arr.2021.101403.
of validation to be of particular concern. We did, however, perform
sensitivity analysis excluding studies rated “red” for assessment of
References
depression and/or diet, which did not significantly change the results.
The third issue identified by quality assessment was high drop-out rates Adjibade, M., Julia, C., Alles, B., Touvier, M., Lemogne, C., Srour, B., Hercberg, S.,
(24 studies had drop-out rates >30%) which may have biased results. Galan, P., Assmann, K.E., Kesse-Guyot, E., 2019a. Prospective association between
Participants with depression may be at increased risk of withdrawing ultra-processed food consumption and incident depressive symptoms in the French
NutriNet-Sante cohort. BMC Med. 17 (1).
from a study. Each of these quality assessment issues are likely to make Adjibade, M., Lemogne, C., Touvier, M., Hercberg, S., Galan, P., Assmann, K.E., Julia, C.,
associations between diet and depression more difficult to detect. Kesse-Guyot, E., 2019b. The inflammatory potential of the diet is directly associated
A strength of this systematic review and meta-analysis includes its with incident depressive symptoms among French adults. J. Nutr. 149, 1198–1207.
Akbaraly, T.N., Brunner, E.J., Ferrie, J.E., Marmot, M.G., Kivimaki, M., Singh-
restriction to longitudinal studies examining incident depression. By Manoux, A., 2009. Dietary pattern and depressive symptoms in middle age. Br. J.
excluding cross-sectional studies, we have reduced the potential for Psychiatry 195, 408–413.
16
Almeida, O.P., Hankey, G.J., Yeap, B.B., Golledge, J., McCaul, K., Flicker, L., 2013. A risk Lassale, C., Batty, G.D., Baghdadli, A., Jacka, F., Sanchez-Villegas, A., Kivimaki, M.,
table to assist health practitioners assess and prevent the onset of depression in later Akbaraly, T., 2019. Healthy dietary indices and risk of depressive outcomes: a
life. Prev. Med. 57, 878–882. systematic review and meta-analysis of observational studies. Mol. Psychiatry 24,
Bertone-Johnson, E.R., Powers, S.I., Spangler, L., Brunner, R.L., Michael, Y.L., Larson, J. 965–986.
C., Millen, A.E., Bueche, M.N., Salmoirago-Blotcher, E., Liu, S., Wassertheil- Li, Y., Lv, M.R., Wei, Y.J., Sun, L., Zhang, J.X., Zhang, H.G., Li, B., 2017. Dietary patterns
Smoller, S., Ockene, J.K., Ockene, I., Manson, J.E., 2011. Vitamin D intake from and depression risk: a meta-analysis. Psychiatry Res. 253, 373–382.
foods and supplements and depressive symptoms in a diverse population of older Liu, T., Zhong, S., Liao, X., Chen, J., He, T., Lai, S., Jia, Y., 2015. A meta-analysis of
women. Am. J. Clin. Nutr. 94, 1104–1112. oxidative stress markers in depression. PLoS One 10, e0138904.
Blazer, D.G., 2003. Depression in late life: review and commentary. J. Gerontol. A Biol. Liu, X., Yan, Y., Li, F., Zhang, D., 2016. Fruit and vegetable consumption and the risk of
Sci. Med. Sci. 58, 249–265. depression: a meta-analysis. Nutrition 32, 296–302.
Bots, S., Tijhuis, M., Giampaoli, S., Kromhout, D., Nissinen, A., 2008. Lifestyle- and diet- Liu, C.S., Adibfar, A., Herrmann, N., Gallagher, D., Lanctot, K.L., 2017. Evidence for
related factors in late-life depression–a 5-year follow-up of elderly European men: inflammation-associated depression. Curr. Top. Behav. Neurosci. 31, 3–30.
the FINE study. Int. J. Geriatr. Psychiatry 23, 478–484. Lucas, M., Mirzaei, F., O’Reilly, E.J., Pan, A., Willett, W.C., Kawachi, I., Koenen, K.,
Brodaty, H., Luscombe, G., Parker, G., Wilhelm, K., Hickie, I., Austin, M.P., Mitchell, P., Ascherio, A., 2011a. Dietary intake of n-3 and n-6 fatty acids and the risk of clinical
2001. Early and late onset depression in old age: different aetiologies, same depression in women: a 10-y prospective follow-up study. Am. J. Clin. Nutr. 93,
phenomenology. J. Affect. Disord. 66, 225–236. 1337–1343.
Cebrino, J., Portero de la Cruz, S., 2020. Diet quality and sociodemographic, lifestyle, Lucas, M., Mirzaei, F., Pan, A., Okereke, O.I., Willett, W.C., O’Reilly, E.J., Koenen, K.,
and health-related determinants among people with depression in Spain: new Ascherio, A., 2011b. Coffee, caffeine, and risk of depression among women. Arch.
evidence from a cross-sectional population-based study (2011-2017). Nutrients 13. Intern. Med. 171, 1571–1578.
Chan, R., Chan, D., Woo, J., 2014. A prospective cohort study to examine the association Lucas, M., Chocano-Bedoya, P., Schulze, M.B., Mirzaei, F., O’Reilly, E.J., Okereke, O.I.,
between dietary patterns and depressive symptoms in older Chinese people in Hong Hu, F.B., Willett, W.C., Ascherio, A., 2014. Inflammatory dietary pattern and risk of
Kong. PLoS One 9, e105760. depression among women. Brain Behav. Immun. 36, 46–53.
Chang, S.C., Cassidy, A., Willett, W.C., Rimm, E.B., O’Reilly, E.J., Okereke, O.I., 2016. Mihrshahi, S., Dobson, A.J., Mishra, G.D., 2015. Fruit and vegetable consumption and
Dietary flavonoid intake and risk of incident depression in midlife and older women. prevalence and incidence of depressive symptoms in mid-age women: results from
Am. J. Clin. Nutr. 104, 704–714. the Australian longitudinal study on women’s health. Eur. J. Clin. Nutr. 69,
Chi, S.H., Wang, J.Y., Tsai, A.C., 2016. Combined association of leisure-time physical 585–591.
activity and fruit and vegetable consumption with depressive symptoms in older Moher, D., Liberati, A., Tetzlaff, J., Altman, D.G., Group, P., 2009. Preferred reporting
Taiwanese: results of a national cohort study. Geriatr. Gerontol. Int. 16, 244–251. items for systematic reviews and meta-analyses: the PRISMA statement. J. Clin.
Chocano-Bedoya, P.O., O’Reilly, E.J., Lucas, M., Mirzaei, F., Okereke, O.I., Fung, T.T., Epidemiol. 62, 1006–1012.
Hu, F.B., Ascherio, A., 2013. Prospective study on long-term dietary patterns and Molendijk, M., Molero, P., Ortuno Sanchez-Pedreno, F., Van der Does, W., Angel
incident depression in middle-aged and older women. Am. J. Clin. Nutr. 98, Martinez-Gonzalez, M., 2018. Diet quality and depression risk: a systematic review
813–820. and dose-response meta-analysis of prospective studies. J. Affect. Disord. 226,
Das, A., Cumming, R.G., Naganathan, V., Ribeiro, R.V., Le Couteur, D.G., Handelsman, D. 346–354.
J., Waite, L.M., Hirani, V., 2021. The association between antioxidant intake, dietary RevMan, 2020. Review Manager(RevMan) Computer Program. The Cochrane
pattern and depressive symptoms in older Australian men: the Concord health and Collaboration.
ageing in men project. Eur. J. Nutr. 60, 443–454. Rienks, J., Dobson, A.J., Mishra, G.D., 2013. Mediterranean dietary pattern and
Djernes, J.K., 2006. Prevalence and predictors of depression in populations of elderly: a prevalence and incidence of depressive symptoms in mid-aged women: results from a
review. Acta Psychiatr. Scand. 113, 372–387. large community-based prospective study. Eur. J. Clin. Nutr. 67, 75–82.
Fiske, A., Wetherell, J.L., Gatz, M., 2009. Depression in older adults. Annu. Rev. Clin. Ritchie, K., Ancelin, M.L., Amieva, H., Rouaud, O., Carrière, I., 2014. The association
Psychol. 5, 363–389. between caffeine and cognitive decline: examining alternative causal hypotheses.
Gangwisch, J.E., Hale, L., Garcia, L., Malaspina, D., Opler, M.G., Payne, M.E., Rossom, R. Int. Psychogeriatr. 26, 581–590.
C., Lane, D., 2015. High glycemic index diet as a risk factor for depression: analyses Saghafian, F., Malmir, H., Saneei, P., Milajerdi, A., Larijani, B., Esmaillzadeh, A., 2018.
from the Women’s Health Initiative. Am. J. Clin. Nutr. 102, 454–463. Fruit and vegetable consumption and risk of depression: accumulative evidence from
GBD, 2017. Disease and injury incidence and prevalence collaborators, 2018. Global, an updated systematic review and meta-analysis of epidemiological studies. Br. J.
regional, and national incidence, prevalence, and years lived with disability for 354 Nutr. 119, 1087–1101.
diseases and injuries for 195 countries and territories, 1990-2017: a systematic Sanchez-Villegas, A., Ruiz-Canela, M., de la Fuente-Arrillaga, C., Gea, A., Shivappa, N.,
analysis for the global burden of disease study 2017. Lancet 392, 1789–1858. Hebert, J.R., Martinez-Gonzalez, M.A., 2015. Dietary inflammatory index,
Ghosh, T.S., Rampelli, S., Jeffery, I.B., Santoro, A., Neto, M., Capri, M., Giampieri, E., cardiometabolic conditions and depression in the Seguimiento Universidad de
Jennings, A., Candela, M., Turroni, S., Zoetendal, E.G., Hermes, G.D.A., Elodie, C., Navarra cohort study. Br. J. Nutr. 114, 1471–1479.
Meunier, N., Brugere, C.M., Pujos-Guillot, E., Berendsen, A.M., De Groot, L., Shang, X., Hodge, A.M., Peng, W., He, M., Zhang, L., 2020. Are leading risk factors for
Feskins, E.J.M., Kaluza, J., Pietruszka, B., Bielak, M.J., Comte, B., Maijo-Ferre, M., cancer and mental disorders multimorbidity shared by these two individual
Nicoletti, C., De Vos, W.M., Fairweather-Tait, S., Cassidy, A., Brigidi, P., conditions in community-dwelling middle-aged adults? Cancers 12, 1–15.
Franceschi, C., O’Toole, P.W., 2020. Mediterranean diet intervention alters the gut Shivappa, N., Steck, S.E., Hurley, T.G., Hussey, J.R., Hebert, J.R., 2014. Designing and
microbiome in older people reducing frailty and improving health status: the NU- developing a literature-derived, population-based dietary inflammatory index.
AGE 1-year dietary intervention across five European countries. Gut 69, 1218–1228. Public Health Nutr. 17, 1689–1696.
Goldsmith, D.R., Rapaport, M.H., Miller, B.J., 2016. A meta-analysis of blood cytokine Shivappa, N., Schoenaker, D.A.J.M., Hebert, J.R., Mishra, G.D., 2016. Association
network alterations in psychiatric patients: comparisons between schizophrenia, between inflammatory potential of diet and risk of depression in middle-aged
bipolar disorder and depression. Mol. Psychiatry 21, 1696–1709. women: the Australian Longitudinal Study on Women’s Health. Br. J. Nutr. 116,
Gougeon, L., Payette, H., Morais, J., Gaudreau, P., Shatenstein, B., Gray-Donald, K., 1077–1086.
2015. Dietary patterns and incidence of depression in a cohort of community- Shivappa, N., Hebert, J.R., Veronese, N., Caruso, M.G., Notarnicola, M., Maggi, S.,
dwelling older Canadians. J. Nutr. Health Aging 19, 431–436. Stubbs, B., Firth, J., Fornaro, M., Solmi, M., 2018. The relationship between the
Gougeon, L., Payette, H., Morais, J.A., Gaudreau, P., Shatenstein, B., Gray-Donald, K., dietary inflammatory index (DII) and incident depressive symptoms: a longitudinal
2016. Intakes of folate, vitamin B6 and B12 and risk of depression in community- cohort study. J. Affect. Disord. 235, 39–44.
dwelling older adults: the Quebec longitudinal study on nutrition and aging. Eur. J. Skarupski, K.A., Tangney, C., Li, H., Ouyang, B., Evans, D.A., Morris, M.C., 2010.
Clin. Nutr. 70, 380–385. Longitudinal association of vitamin B-6, folate, and vitamin B-12 with depressive
Graf, D., Di Cagno, R., Fak, F., Flint, H.J., Nyman, M., Saarela, M., Watzl, B., 2015. symptoms among older adults over time. Am. J. Clin. Nutr. 92, 330–335.
Contribution of diet to the composition of the human gut microbiota. Microb. Ecol. Skarupski, K.A., Tangney, C.C., Li, H., Evans, D.A., Morris, M.C., 2013. Mediterranean
Health Dis. 26, 26164. diet and depressive symptoms among older adults over time. J. Nutr. Health Aging
Guo, X., Park, Y., Freedman, N.D., Sinha, R., Hollenbeck, A.R., Blair, A., Chen, H., 2014. 17, 441–445.
Sweetened beverages, coffee, and tea and depression risk among older US adults. Smaga, I., Niedzielska, E., Gawlik, M., Moniczewski, A., Krzek, J., Przegalinski, E.,
PLoS One 9, e94715. Pera, J., Filip, M., 2015. Oxidative stress as an etiological factor and a potential
Higgins, J.P., Thompson, S.G., Deeks, J.J., Altman, D.G., 2003. Measuring inconsistency treatment target of psychiatric disorders. Part 2. Depression, anxiety, schizophrenia
in meta-analyses. BMJ 327, 557–560. and autism. Pharmacol. Rep. 67, 569–580.
Jacka, F.N., Cherbuin, N., Anstey, K.J., Butterworth, P., 2014. Dietary patterns and Stang, A., 2010. Critical evaluation of the Newcastle-Ottawa scale for the assessment of
depressive symptoms over time: examining the relationships with socioeconomic the quality of nonrandomized studies in meta-analyses. Eur. J. Epidemiol. 25,
position, health behaviours and cardiovascular risk. PLoS One 9, e87657. 603–605.
Jacka, F.N., Cherbuin, N., Anstey, K.J., Butterworth, P., 2015. Does reverse causality Sterne, J.A., Sutton, A.J., Ioannidis, J.P., Terrin, N., Jones, D.R., Lau, J., Carpenter, J.,
explain the relationship between diet and depression? J. Affect. Disord. 175, Rucker, G., Harbord, R.M., Schmid, C.H., Tetzlaff, J., Deeks, J.J., Peters, J.,
248–250. Macaskill, P., Schwarzer, G., Duval, S., Altman, D.G., Moher, D., Higgins, J.P., 2011.
Kim, M.T., Kim, K.B., Han, H.R., Huh, B., Nguyen, T., Lee, H.B., 2015. Prevalence and Recommendations for examining and interpreting funnel plot asymmetry in meta-
predictors of depression in Korean American elderly: findings from the memory and analyses of randomised controlled trials. BMJ 343, d4002.
aging study of Koreans (MASK). Am. J. Geriatr. Psychiatry 23, 671–683. Thomas-Odenthal, F., Molero, P., van der Does, W., Molendijk, M., 2020. Impact of
Lai, J.S., Hure, A.J., Oldmeadow, C., McEvoy, M., Byles, J., Attia, J., 2017. Prospective review method on the conclusions of clinical reviews: a systematic review on dietary
study on the association between diet quality and depression in mid-aged women interventions in depression as a case in point. PLoS One 15, e0238131.
over 9 years. Eur. J. Nutr. 56, 273–281.
17
Thorpe, M.G., Milte, C.M., Crawford, D., McNaughton, S.A., 2019. Education and Viechtbauer, W., 2010. Conducting meta-analyses in R with the metafor package. J. Stat.
lifestyle predict change in dietary patterns and diet quality of adults 55 years and Softw. 36 (3), 1–48.
over. Nutr. J. 18, 67. Voortman, T., Kiefte-de Jong, J.C., Ikram, M.A., Stricker, B.H., van Rooij, F.J.A.,
Tsai, H.J., 2016. Dietary patterns and depressive symptoms in a Taiwanese population Lahousse, L., Tiemeier, H., Brusselle, G.G., Franco, O.H., Schoufour, J.D., 2017.
aged 53 years and over: results from the Taiwan longitudinal study of aging. Geriatr. Adherence to the 2015 Dutch dietary guidelines and risk of non-communicable
Gerontol. Int. 16, 1289–1295. diseases and mortality in the Rotterdam Study. Eur. J. Epidemiol. 32, 993–1005.
Tsai, A.C., Chang, T.L., Chi, S.H., 2012. Frequent consumption of vegetables predicts Wang, J., Zhou, Y., Chen, K., Jing, Y., He, J., Sun, H., Hu, X., 2018. Dietary inflammatory
lower risk of depression in older Taiwanese - results of a prospective population- index and depression: a meta-analysis. Public Health Nutr. 1–7.
based study. Public Health Nutr. 15, 1087–1092. Wells, G.A., Shea, B., O’Connell, D., Peterson, J., Welch, V., Losos, M., Tugwell, P., 2001.
Tsai, A.C., Chi, S.H., Wang, J.Y., 2013. Cross-sectional and longitudinal associations of The Newcastle-ottawa Scale (NOS) for Assessing the Quality of Nonrandomised
lifestyle factors with depressive symptoms in >/= 53-year old Taiwanese - results of Studies in Meta-analyses. http://www.ohri.ca/programs/clinical_epidemiology/o
an 8-year cohort study. Prev. Med. 57, 92–97. xford.asp.
United Nations Department of Economics and Social Affairs, 2017. In: Nations, U. (Ed.), White, M.L., Groh, C.J., 2007. Depression and quality of life in women after a myocardial
World Population Prospects: The 2017 Revision. UN, Geneva. infarction. J. Cardiovasc. Nurs. 22, 138–144.
Vashum, K.P., McEvoy, M., Milton, A.H., McElduff, P., Hure, A., Byles, J., Attia, J., 2014. Worrall, C., Jongenelis, M., Pettigrew, S., 2020. Modifiable protective and risk factors for
Dietary zinc is associated with a lower incidence of depression: findings from two depressive symptoms among older community-dwelling adults: a systematic review.
Australian cohorts. J. Affect. Disord. 166, 249–257. J. Affect. Disord. 272, 305–317.
Veritas Health Innovation, Covidence systematic review software, Melbourne, Australia, Wu, P.Y., Chen, K.M., Belcastro, F., 2020. Dietary patterns and depression risk in older
p. Available at www.covidence.org. adults: systematic review and meta-analysis. Nutr. Rev. Epub ahead of print. PMID:
Vermeulen, E., Stronks, K., Visser, M., Brouwer, I.A., Schene, A.H., Mocking, R.J., 33236111.
Colpo, M., Bandinelli, S., Ferrucci, L., Nicolaou, M., 2016. The association between Zhang, J., Yu, K.F., 1998. What’s the relative risk? A method of correcting the odds ratio
dietary patterns derived by reduced rank regression and depressive symptoms over in cohort studies of common outcomes. JAMA 280, 1690–1691.
time: the Invecchiare in Chianti (InCHIANTI) study. Br. J. Nutr. 115, 2145–2153.
18

Ageing Research Reviews: Annabel P. Matison, Karen A. Mather, Victoria M. Flood, Simone Reppermund

Uploaded by

Copyright:

Available Formats

Ageing Research Reviews: Annabel P. Matison, Karen A. Mather, Victoria M. Flood, Simone Reppermund

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ageing Research Reviews: Annabel P. Matison, Karen A. Mather, Victoria M. Flood, Simone Reppermund

Uploaded by

Copyright:

Available Formats

Ageing Research Reviews 70 (2021) 101403

Contents lists available at ScienceDirect