Dietary Pattern and Depressive Outcomes in Children and Adolescents: Systematic Review and Meta-analysis of Observational Studies

Abstract

Context

Research suggests that dietary pattern plays an important role in mental health and constitutes a modifiable risk factor for depression.

Objective

The aim of this systematic review and meta-analysis was to evaluate the association between dietary patterns and depressive outcomes in adolescents.

Data Sources

Medline/PubMed, Lilacs, EMBASE, PsycINFO, and PsycARTICLES, gray literature up to March 2024 were searched; the reference lists were also verified. Observational studies in participants with a mean age ⩽19 years reporting associations between dietary patterns and clinical depression or depressive symptoms were searched. Overall, 21 studies were included in this systematic review.

Data Extraction

Data from eligible articles were extracted by 2 reviewers.

Data Analysis

Odds ratios (ORs) and confidence intervals (CIs) were determined under a random-effects model. The risk of bias assessment was conducted by 2 independent researchers using the Joanna Briggs Institute Critical Appraisal Checklist tool.

Results

The qualitative results revealed that a higher unhealthy diet score was positively associated with depressive symptoms, while a healthy diet was negatively associated with depressive symptoms. In the study that included adolescents with a clinical diagnosis of depression, the relationship between inflammatory dietary pattern tertiles and depression was attenuated after all covariates were adjusted for. The meta-analysis to evaluate the association between depressive symptoms and a posteriori dietary patterns found that the “healthy” dietary pattern decreased depressive symptoms in adolescents (OR: 0.69; 95% CI: 0.44, 0.95). There was no statistically significant association between depressive symptoms and “unhealthy” and “snacks” patterns (OR: 1.20 [95% CI: 0.95, 1.46]; OR: 1.20 [95% CI: 0.70, 1.48]) dietary patterns.

Conclusion

The results identified that a healthy dietary pattern decreased depressive symptoms in adolescents. However, considering the high heterogeneity and the low level of certainty of the evidence, these results should be interpreted with caution.

Systematic Review Registration

PROSPERO registration no. CRD42020159921.

INTRODUCTION

Depression is a multifactorial disease characterized by the presence of depressed mood, feelings of hopelessness, loss of interest in usual activities, in addition to other cognitive and somatic disorders.¹ Depression affects approximately 350 million people in all social classes and ages, contributing to the global burden of diseases and reduced quality of life.² In adolescents, depression is a common mental health disorder, with a prevalence of 6.2%.³ However, a higher prevalence could be estimated for depressive symptoms, ranging between 8.2% and 28.7% in this phase of life.^4–7

In addition to the psychological and social factors involved in the etiology of depression, studies suggest that food consumption plays an essential role in mental health and constitutes a modifiable risk factor for this mood disorder.^8,9 In this context, researchers indicate the importance of dietary pattern analysis to investigate the relationship between diet and health outcomes, since dietary patterns better express the complexity of eating and the diversity of food components,^10,11 in addition to facilitating nutritional recommendations and providing a broad approach to the prevention of diseases and nutritional disorders.¹² Therefore, when evaluating these associations, all individual and collective determinants influencing dietary practices—for example, perceptions towards a “healthy diet” and the socioeconomic, environmental, and cultural factors—should be considered.¹³

In adolescence, adequate nutrition is essential to meet physiological requirements of rapid growth and development, and for cognitive function.¹⁴ However, the potential of diet in the prevention of depression is a recent area of research, and the results of studies are still inconsistent, especially in this age group.^15,16 For example, in a systematic review conducted by Khalid et al,¹⁵ the increased occurrence of depressive symptoms or the risk of depression among adolescents was associated with unhealthy snacking and Western dietary patterns. An umbrella review that included 19 meta-analyses of observational studies and randomized controlled trials revealed weak or no evidence in 7 studies that assessed the relationship between healthy dietary patterns and depression risk, while 2 meta-analyses found suggestive evidence of an association between healthy dietary patterns and a reduced odds of depression. The strength of evidence was graded as weak for Dietary Approaches to Stop Hypertension (DASH), vegetarian, and Western diets.¹⁷

Considering that diet is a modifiable risk factor for the occurrence of depression and the scarcity of studies in this area, research on this topic can widen the knowledge gap between dietary patterns and depression in adolescents, suggesting the need for consistent evidence on this relationship. In addition, it can contribute to the implementation of actions to promote mental health and the development of prevention strategies for depressive disorders at earlier stages of life.

Thus, the objective of this systematic review and meta-analysis was to systematically synthesize the results of observational studies with the aim of demonstrating the association between dietary patterns and depression-related outcomes in adolescents.

METHODS

This systematic review and meta-analysis of observational studies was reported according to the recommendations of the Meta-analysis of Observational Studies in Epidemiology (MOOSE).¹⁸ The study protocol followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines¹⁹ and is registered in the Prospective Register of Systematic Reviews (PROSPERO) under number CRD42020159921.

Eligibility Criteria

The study used the PEO (Population, Exposure, and Outcome) criteria to structure the research question and eligibility criteria (Table 1).

Information Sources and Electronic Search

The electronic search was carried out in October 2020 and updated in March 2024, and the selection of articles was conducted in the following databases: Medline/PubMed, Lilacs, Embase, PsycINFO, and PsycARTICLES. The gray literature (Google Scholar) and reference lists of included studies and reviews were searched manually. The terms depression, depressive disorder, and dysthymic disorder were identified, and their corresponding MeSH (Medical Subject Heading) terms and Embase subject headings (Emtree) were selected. The descriptors for “dietary pattern” were searched for in previously published reviews.^5,8,10,15 The Boolean operators “AND” and “OR” were used. There were no date limits or language restriction.

Study Selection

Two independent reviewers (L.E.M.d.S. and K.B.B.d.S.) conducted the study selection process with Rayyan (Copyright© 2022 Rayyan, avaliabe at: https://www.rayyan.ai/), a web and mobile app for systematic reviews,²⁰ which was also used to identify and remove duplicates. In the first stage, the reviewers screened the title and abstract of the retrieved studies. Next, eligible studies were fully reviewed, and those that met the eligibility criteria were included in this review. Disagreements were discussed and resolved after consultation with a third reviewer (M.L.P.d.S.).

Data Extraction

Data extraction was performed by an independent reviewer (C.d.M.C.) using a standardized form elaborated in the Microsoft Office Excel® (Microsoft Corporation, Redmond, WA, USA). Another author (L.E.M.d.S.) confirmed the accuracy of the information collected, and a third researcher (M.L.P.d.S.) resolved any disagreements. In general, the extracted data included participants’ characteristics, sample size, inclusion and exclusion criteria, criteria for the definition of exposure and outcome of interest, data analysis methods, the measure of association, and control for potential confounders. When an eligible study did not provide sufficient data, the corresponding authors and the first authors were contacted.

Risk of Bias Assessment

The risk of bias assessment was conducted by 2 independent researchers (L.E.M.d.S. and W.P.d.O.A.) using the Joanna Briggs Institute (JBI) Critical Appraisal Checklist tool.^21,22 The total number of “yes” responses to the items guided the evaluation of the studies, being judged as “high risk” when these reached up to 49% of the score, “moderate risk” when these were between 50% and 69%, and “low risk” when the total affirmative responses to the items were 70% or higher. Disagreements were discussed and resolved in consensus with the third researcher.

Summary Measure Analysis

The results of the meta-analysis were displayed in a forest plot. Heterogeneity was investigated using Cochran’s Q test, and Higgins’s I² statistic (I² >50%) was used as an indicator of high heterogeneity. The primary studies reported the results of dietary pattern scores and depressive outcomes in terms of quantiles. There was high heterogeneity in the studies, so the random-effects model was adopted to calculate the odds ratio (OR), combining dietary patterns in the higher categories and comparing them with the lower categories. It was not possible to explore the causes of heterogeneity using meta-regression because only 4 studies were included in the quantitative analysis, and a minimum of 10 studies is recommended to perform this analysis.²³ Also, it was not possible to carry out the funnel plot and Egger's regression to evaluate the publication bias considering the small number of studies included in the quantitative analysis. All statistical analyses were conducted using the statistical package STATA version 16.0 (StataCorp, College Station, TX, USA), and P < .05 was considered statistically significant. The characteristics of the dietary patterns of the studies included in the meta-analysis can be found in the Supplementary Material.

Confidence in Cumulative Evidence

According to the Grading of Recommendations Assessment, Development, and Evaluation (GRADE), the certainty of the evidence was evaluated by narrative and quantitative (meta-analysis) summaries. This approach classifies the quality of evidence into 4 levels: high, moderate, low, and very low.²⁴ Each level reflects the reliability in estimating the effects presented. The initial judgment of the quality of evidence was completed from the study design and the evaluation of each analyzed outcome. The summary of findings tables were produced with the GRADE online software (GRADEpro GDT; GRADE Working Group).

RESULTS

Selection of Studies

The study selection process is shown in Figure 1. An electronic search in the databases retrieved 16 589 articles, of which 2680 were duplicates. After reading the title and abstract, 13 909 studies were excluded. Forty-seven studies were selected for full reading, and 26 were excluded for not meeting the eligibility criteria. A total of 21 were selected for qualitative analysis^25–46 and 4 studies^29,41,46,47 for quantitative analysis. Three publications (with the same outcome) of a single study used different methods to determine the dietary pattern (exposure).^30–32

Study Characteristics

The characteristics of the included studies in this systematic review are described in Table 2. Three studies were conducted in Australia,^25,26,46 8 in Iran,^{28–32,34,35,39,44,47} 3 in the England,^27,42,45 1 in Fiji,⁴³ 1 in Norway,³³ and 4 in China.^36,38,40,41 These studies were published between 2010 and 2024, and the majority were cross-sectional.^{25,28–37,39–41,43,44,46,47} In 5 prospective cohort studies, follow-up of children and adolescents ranged from 1 year to 10 years.^{26,27,33,42,45} The total number of baseline participants recruited in the 21 studies^25–47 was 89 947, and the age of participants ranged from 10 to 19 years. Most studies involved individuals of both sexes,^{25,26,33,36–38,40,41,43,45,46,48} and of those, 2 conducted analyses stratified by gender,^43,46 while other studies conducted their analyses only with female adolescents.^{28–32,34,35,39,44} Three studies included children in their samples.^33,38,45

Evaluation of Depression and Depressive Symptoms

The outcome of depressive symptoms was categorical and evaluated by the following instruments: Short Mood and Feelings Questionnaire (SMFQ),^25,27,33,46 Moods and Feelings Questionnaire (MFQ),⁴² Beck Depression Inventory (BDI),^28,29 the Persian version of the BDI,^34,35,39,44 the 21-item Depression, Anxiety, and Stress Scale (DASS-21),^30–32,37 the Spanish version of the DASS-21, Depression Self-Rating Scale for Children (DSRS),⁴¹ the Chinese short version of the Center for Epidemiological Studies–Depression Scale (CES-D),³⁶ the 10-item Children's Depression Inventory–Short Form (CDI-S),³⁸ and the Chinese version of the 9-item Patient Health Questionnaire (PHQ-9).⁴⁰ Three studies presented the continuous outcome^26,43,47; among them, 2 used the emotional subscale (5 items) of the Pediatric Quality of Life Inventory (PedsQL)^26,43 as a proxy measure for depressive symptomatology and 1 study used the Persian version of the BDI-Second Edition (BDI-II).^47,28 Most instruments used were validated for children and adolescents, except for the DASS-21.^30,32,37,35 One study included adolescents with clinically diagnosed depression.⁴⁵ This study assessed the outcome through the computerized version of the Clinical Interview Schedule–Revised (CIS-R), a validated and standardized tool to assess depression according to the International Classification of Diseases, Tenth Revision (ICD-10)⁴⁵(Table 2).

Dietary Assessment

One study used items from the Simple Dietary Questionnaire (SDQ),⁴⁶ 2 studies adopted a questionnaire with 14 food items based on the Dietary Guidelines for Children and Adolescents in Australia and modified to include additional questions on breakfast consumption, different types of beverages, and takeaway food.^25,26,29,31 Jacka et al²⁷ collected dietary information from the Health and Behavior in Teenagers Study (HABITS), including questions about fruit and vegetable consumption and breakfast frequency. One study obtained food-intake data from the Adolescent Behaviors, Attitudes and Knowledge Questionnaire (ABAKQ) designed to assess nutrition, physical activity, and other health behaviors among Fijian adolescents.⁴³ Fourteen studies used a semi-quantitative food-frequency questionnaire to assess dietary intake.^{28–41,44,45} In the study by Hemmati et al⁴⁷ the dietary intakes of each of the participants were assessed by completing three 24-hour dietary recalls, and in the study by Winpenny et al,⁴² the participants were asked to complete a 4-day diet diary.

Fifteen studies adopted an a priori approach^{25–31,33–37,39,42,44,45} defining the dietary^28–34 patterns through the healthy/unhealthy diet quality score^25–27,29^-31 of the DASH score,²⁸ the Dietary Inflammatory Pattern score,⁴⁵ the Global Dietary Quality score (GDQS),³⁴ the Mediterranean Diet Quality Index in children and adolescents (KIDMED),³⁷ the Dietary Total Antioxidant Capacity (DTAC) index,³⁹ the Lifelines Diet score (LLDS),³⁵ the Nordic Diet (NND) score,³³ and the Mediterranean Diet score (MDS).⁴²

The study that produced 3 publications,^30–32 used the Dietary Inflammatory Index³⁰ and the Adherence to the Mediterranean Dietary Pattern^31,32 to determine the dietary patterns. The study by Sangouni et al⁴⁴ used the food-frequency questionnaire to collect participants' dietary intake and generate the Dietary Phytochemical Index.

Six studies chose the a posteriori method.^{29,40,41,43,46,47} The dietary patterns labelled in the studies were as follows: healthy,^29,43,46,47 unhealthy,^29,43,46,47 traditional,^29,40,41 Western,^29,47 snacks,^41,46 animal food,⁴¹ and modern, snack-aquatic (with high loadings for aquatic products, dairy products, snacks and low loadings for rice flour, vegetables, was labelled the snack-aquatic pattern) and vegetarian.⁴⁰ The study conducted by Sinclair et al⁴³ used exploratory factor analysis and grouped dietary variables into 2 unique factors: “healthy diet quality” and “unhealthy diet quality” (Table 2).

Dietary Patterns and Depressive Symptoms

Most studies reported the exposure as a categorical variable and the dietary patterns in quantiles,^{25–27,29–32,34–43,45,46} and used a logistic regression model to estimate the ORs of the association between diet quality and depressive symptoms in adolescents.^{29,31,35–38,40,41,44–46} Eight studies adopted multivariate linear regression analysis,^{25,26,28,32,39,42,43,47} and the others analyzed the association using both models.^27,30,33,34 All studies considered confounding factors in multivariate analyses to adjust the association of interest.

A Priori Dietary Patterns

Fifteen studies evaluated the association between diet quality and depressive symptoms.^{25–27,29,30,44,45,47} In a cross-sectional study, Jacka et al²⁵ demonstrated that the highest quintile categories of a healthy diet reduced the chances of depressive symptoms between 39% and 50% (95% CI; OR: 0.55; 0.40, 0.77) when compared with adolescents with lower scores. Nevertheless, adolescents in the highest quintile of an unhealthy diet increased the probability of depressive symptoms by almost 80% (95% CI; OR: 1.79; 1.52, 2.11).

In a cohort study,²⁷ researchers explored the cross-sectional and longitudinal relationship between diet quality and mental health. After adjustments for diet and family factors, only the highest quintile of unhealthy diet score was positively associated with baseline depressive symptoms (95% CI; OR: 1.51; 1.04, 2.19); however, no association was identified in the follow-up (95% CI; OR: 0.75; 0.55, 1.04).

The cross-sectional analyses of the study conducted by Jacka et al²⁶ showed a positive relationship between healthy diet scores and depressive symptoms before and after adjustment (β: 0.42; 0.31, 0.53; P < .001). In contrast, unhealthy diet scores were inversely associated with depressive symptoms at baseline (β: -0.29; -0.38, -0.20; P < .001). After the 2-year follow-up period, the increase in healthy diet scores remained associated with lower scores of depressive symptoms (β: 0.14; 0.02, 0.27; P = .03). Still, there was no prospective association between higher scores of unhealthy diet quality and lower PedsQL (higher depressive symptoms) (β: -0.07; -0.18, 0.03; P = .18).

Greater adherence to the DASH-style diet was associated with lower chances of depression among girls in Iran, even after adjustment for confounding variables (OR: 0.47; 0.23, 0.92).²⁸ Also, data from the study by Shivappa et al³⁰ conducted with Iranian adolescents suggested a significant association between the highest dietary inflammatory potential and moderate depressive symptoms (OR: 3.96; 1.12, 13.97). In the study by Cong et al⁴⁵ adolescents from the United Kingdom in the highest tertile of the inflammatory dietary pattern (IDP) were more likely to develop depression at 18 years of age compared with participants in the lower tertile, but this relationship was attenuated after adjustment (OR: 1.21; 0.96, 1.51). In the study by Sangouni et al⁴⁴ the participants in the fourth quartile of the Dietary Phytochemical Index compared with the first quartile had a 50% lower odds of depression (OR: 0.50; 95% CI: 0.30–0.84; P = .009). A cohort study in participants with low vs high NND scores at 7 years showed a slight association with depressive symptoms at 8 years (95% CI; OR: 1.0; 1.0, 1.2).³³ In the same direction, 6 cross-sectional studies showed that high diet quality was associated with lower odds of depression (95% CI; OR: 0.5; 0.28, 0.92), (95% CI; OR: 0.59; 0.39, 0.90), (95% CI; OR: 0.58; 0.35, 0.97), (95% CI; OR: 0.41; 0.24, 0.72), (95% CI; OR: 0.92; 0.87, 0.98), (95% CI; OR: 0.59; 0.39, 0.90).^33–37,39 In the study by Zhang et al,³⁸ poor diet quality (95% CI; OR: 1.89;1.62, 2.21) was associated with a higher odds of depression in children and adolescents. A cohort study found no association between diet quality and mid-adolescent depression symptoms at baseline and 3-year follow-up (β: −0.03 [−0.31, 0.26]; β: 0.35 [−0.04, 0.74]).⁴²

A Posteriori Dietary Patterns

Six studies used the a posteriori approach to evaluate the association of interest.^{29,40,41,43,46,47} In Australia, a study identified that greater adherence to the “unhealthy” dietary pattern was associated with a higher chance of depressive symptoms in boys (OR: 1.18; 95% CI: 1.07, 1.32) but not in girls (OR: 1.09; 95% CI: 0.98, 1.21).⁴⁶ In Chinese adolescents, higher tertile scores in the dietary pattern of “snacks” were positively associated with higher chances of depressive symptoms (OR: 1.64; 1.30, 2.06), while the traditional pattern reduced the occurrence of these symptoms by 62% (OR: 0.38; 0.30, 0.49).⁴¹ Results of the study by Khayyatzadeh et al²⁹ showed that adherence to the highest quartile of the “healthy” dietary pattern was significantly associated with fewer mild to severe depressive symptoms (OR: 0.56; 0.35, 0.92). In the study by Sinclair et al⁴³ it was observed that higher intake of a healthy diet was associated with lower depressive symptomatology scores at baseline (β: 6.0 [4.0, 8.0], P < .00; β: 7.9 [5.8, 9.9], P < .00) and in the follow-up for both sexes (β: 4.8 [2.9, 6.7], P < .00; β: 5.0 [3.0, 6.9], P < .00).

A positive association was observed between the Western dietary pattern and depression (β: 0.510; P < .001), while the healthy dietary pattern was inversely related to depression (β: -0.508; P < .001) in the study of Hemmati et al.⁴⁷ A cross-sectional study identified by factor analysis 4 dietary patterns: the modern pattern, the snack-aquatic (with high loadings for aquatic products, dairy products, snacks and low loadings for rice flour, vegetables, was labelled the snack-aquatic pattern) pattern, the traditional pattern, and the vegetarian pattern. In this study, vegetarian patterns were associated with a reduced risk of depression and modern and snack-aquatic patterns were associated with an increased risk of depression in Chinese adolescents.⁴⁰

Assessment of Risk of Bias

Most studies were classified as low risk of bias,^{27,28,30–32,34,36–40,42,46,47} and 5 studies presented a moderate risk of bias.^{29,41,43–45} The additional studies were judged to have a high risk of bias according to JBI's critical assessment for cohort studies.^26,27,33 (Figure 2).

Results of Meta-analysis

The meta-analysis included 4 cross-sectional studies and 9395 participants.^29,41,46,47 It was not possible to perform quantitative analysis of other studies because the characteristics and methods for determining dietary patterns were not directly comparable (cross-sectional studies [n = 11]^{25,28,30–32,34–36,38–40,43,44} and cohort studies [n = 5]^{26,27,33,42,45}). One study that met all eligibility criteria could not be part of the meta-analysis because the dietary pattern was presented as a continuous variable. For this study, the authors were contacted, but no response was received.⁴³

For better comparability of data, the dietary patterns were classified as “healthy,” “unhealthy,” and “snacks” to evaluate the associations with depressive symptoms, as shown in Figure 3. The “mixed” pattern, which had a high loading factor for unhealthy and healthy foods, was not included in the quantitative analysis.

“Healthy” Dietary Pattern and Depressive Symptoms

The results of the meta-analysis for the “healthy” dietary pattern are shown in Figure 3A.^29,41,46,47 After calculating the values of combined effects, the highest adherence to the “healthy” dietary pattern decreased by 31% (95% CI; OR: 0.69; 0.44, 0.95) the chance of adolescents having depressive symptoms (I² = 96.18%).

“Unhealthy” Dietary Pattern and Depressive Symptoms

The summary measure of the studies evaluating the “unhealthy” dietary pattern did not identify a significant association between the consumption of this dietary pattern and depressive symptoms among adolescents (OR: 1.20; 95% CI: 0.95, 1.46; I² = 88.53%) (Figure 3B).^29,41,46,47

“Snacks” Dietary Pattern and Depressive Symptoms

Quantitative synthesis did not identify a significant association between the “snacks” dietary pattern and depressive symptoms in adolescents (OR: 1.09; 95% CI: 0.70, 1.48; I² = 93.84%). For this pattern, only 2 studies showed information that allowed comparability and synthesis of the data (Figure 3C).^41,46

Certainty of the Evidence

According to the GRADE system, the overall level of certainty of the evidence was very low for both outcomes (Table 3).

DISCUSSION

This systematic review aimed to synthesize the results of studies that evaluated the association between dietary patterns and depressive outcomes in adolescents. Twenty-one studies met the eligibility criteria and were included in this review.^25–47 Most studies that used indices for diet evaluation revealed associations between higher healthy dietary patterns and lower risk of depressive symptoms.^{25–28,28,34–37,39,43,44} In contrast, an unhealthy (low-quality) diet was associated with greater depressive symptoms.^{25–27,30,33,38} After all covariates were adjusted, in the study that included adolescents with a clinical diagnosis of depression, a higher IDP in childhood seems not to be associated with higher depression risk in late adolescence.⁴⁵ The summary measure of studies using an a posteriori approach showed that the high intake of foods in the “healthy” dietary pattern decreased depressive symptoms in adolescents.^29,41,46,47 No association between depressive symptoms with “unhealthy”^29,41,46,47 and “snacks” dietary patterns^41,46 was found. However, this systematic review did not find enough evidence from high-quality studies to completely support an association between depressive symptoms and specific dietary patterns.

Studies included in a previous systematic review also demonstrated a significant cross-sectional relationship between unhealthy dietary patterns and poorer mental health in children and adolescents; however, the association between healthy dietary patterns and better mental health was less consistent.⁸ Another systematic review showed evidence on the association between high diet quality and lower levels of depression or better mental health in adolescents, and a significant and inverse relationship between worse diet quality and the outcome of interest.¹⁵ A recent meta-analysis of children and adolescents found that a healthy dietary pattern was negatively associated with depressive symptoms (k = 19, r = –0.13, P < .001, 95% CI [–0.18, –0.08]). In contrast, unhealthy dietary patterns were positively associated with depressive symptoms (k = 13, r = 0.11, P = .001, 95% CI [0.05, 0.17]).⁴⁹ It is necessary to emphasize that some eligibility criteria of studies included in previous reviews differ from those established in this review and studies that reported on specific nutrients or food groups were not included in our study.

Considering that adolescence is a life stage marked by intense physical, behavioral, and psychosocial changes,⁵⁰ adolescents may find themselves in conditions of greater vulnerability to situations that negatively affect their physical and mental health.⁵¹ This transition period between childhood and adulthood also involves increased nutritional requirements for rapid growth and development,⁵² in addition to the formation of eating habits that can be maintained until adulthood.^50,53 At this stage, dietary habits are characterized mainly by the high consumption of fried-food preparations, ultra-processed foods, and low intake of fresh foods, which can increase the risk of depressive symptoms since many of these foods make up food patterns considered “unhealthy.”^15,25,27,45 However, this meta-analysis did not identify an association between unhealthy and snacks dietary patterns with depressive symptoms.

Although not fully established, some potential pathways may explain how the dietary pattern influences adolescents' mental health. The components of a healthy diet, such as intake of fruits and vegetables, have a high content of antioxidant compounds (vitamins C and E, β-carotene, and folic acid) that can have protective effects against depression by reducing oxidative stress.^5,54 It is known that neuronal cells in the brain are susceptible to oxidative stress, which can cause neuronal damage and favor the occurrence of depression.^55,56

Other foods of a healthy dietary pattern, such as whole grains, lean red meats, fish, olive oil, and wine, are also important sources of nutrients such as magnesium, selenium, zinc, B-complex vitamins, mono- and polyunsaturated fatty acids, polyphenols, and fiber. Some of these nutrients can mediate and play an essential role in the relationship between dietary patterns and depression.^5,48,57–61

A growing body of evidence suggests that inflammation and endothelial dysfunction may underlie the development of depression and mediate the association with diet.^{30,32,61–65} An increase in the dietary inflammatory potential and Western dietary pattern (high red meat intake, takeaway food, refined foods, and confectionery) may be associated with a higher risk of depressive symptoms.^30–32,61 In contrast, a greater adherence to the Mediterranean and healthy dietary patterns (composed of fruit, vegetables, fish and other seafood, whole grains, wine, olive oil) was inversely associated with inflammation and depressive symptoms in adolescents ^{30–32,62,66}

Elevated levels of proinflammatory cytokines, such as interleukins (IL-1, IL-6), C-reactive protein, and tumor necrosis factor-alpha (TNF-α), may reduce biogenic amines in the brain and induce neuroendocrine changes with impairments in synaptic plasticity.⁶⁷ In this context, cytokines can reduce tryptophan availability for serotonin synthesis by activating the indoleamine-2,3-dioxygenase enzyme, which converts tryptophan to neurotoxic metabolites promoting glutamate excitotoxicity via N-methyl-d-aspartate receptors.⁶⁸ Cytokines may also be responsible for reducing neurotrophins, as the brain-derived neurotrophic factor, which plays an essential role in neurogenesis, memory, and hippocampal plasticity.^62,69,70

In the risk of bias assessment, most cross-sectional studies were at low risk of bias^{25,28,30–32,34–40,42,47} and 4 studies were at moderate risk.^29,41,43,44 Three cohort studies were classified as high risk of bias.^26,27,33 Also, the confidence in the cumulative evidence was considered very low for the 2 outcomes. Therefore, caution is needed in comparing the studies and interpreting the measure of the summary effect of this meta-analysis. According to the Higgins inconsistency test (I²), the studies showed moderate to high heterogeneity, which may influence the measure of association. However, this result is expected in meta-analyses involving observational studies because they have biases inherent to the type of design and the number of confounding factors controlled for.⁷¹ Another critical issue concerns that it was not possible to perform meta-regression and subgroup analysis because too few studies were included in the meta-analysis.²³ Moreover, the limitations of the analysis of food items should be considered. The various patterns and the complexity of the methods to determine dietary patterns may be additional sources of heterogeneity.⁷²

Limitations identified in the included studies also included the difficulty in interpreting and comparing the results due to methodological variation in assessing food intake.⁷² Behaviors, socioeconomic and environmental factors, culture, beliefs, and religion can influence food choices and produce dietary patterns with diverse content, contributing to the difficulty in a direct comparison of patterns.^46,73–76 The food-frequency questionnaire also has some limitations because it relies on the participant’s memory and cooperation and their communication ability, and ample time to collect the information.^77,78

It was not possible to perform the meta-analysis of 12 cross-sectional studies^{25,28,30–32,34–40,43,44} and 5 cohort studies^{26,27,33,42,45} because the patterns were not directly comparable. One study that met all eligibility criteria was not included in the analysis since the variable was presented by continuous data.⁴³ Considering that only 4 studies^29,41,46,47 were included in this meta-analysis, it was not possible to perform the funnel plot and Egger’s regression test to identify the presence of publication bias. Moreover, some effect estimates (OR or mean differences) produced using the funnel plot are naturally correlated with their standard errors and may have asymmetry in the graph and erroneous conclusions of the publication bias.⁷¹

However, it is important to highlight that, in the review process, the search was performed using 5 important scientific databases and the gray literature. The reference lists of the retained studies and reviews were manually searched to identify additional studies not indexed in the databases. Additionally, there was no limitation of time or language restriction, and the process was carried out by 2 independent researchers, ensuring the systematic review's methodological rigor.

CONCLUSION

Most studies that determined the a priori dietary pattern showed that a higher unhealthy diet score was positively associated with depressive symptoms. In contrast, a healthy diet (high-quality) was negatively associated with depressive symptoms. In the study that included adolescents with a clinical diagnosis of depression, the relationship between IDP tertiles and depression was attenuated after all covariates were adjusted for. Meta-analysis of studies with an a posteriori dietary pattern found an association between healthy dietary patterns and lower chances of depressive symptoms in adolescents. In contrast, the “unhealthy” and “snacks” dietary patterns were not related to depressive symptoms. Considering the small number of studies included in the meta-analysis, the high heterogeneity, and the low level of certainty of the evidence, these results should be interpreted with caution. Future studies are also required to establish the direction and mechanism of the association between dietary pattern and depression in adolescents, contributing to the development of public policies and strategies that improve the eating habits of younger populations and reduce the prevalence of risk factors throughout life.

Acknowledgments

The authors thank the FAPESB (Foundation for Research Support in Bahia) for providing a scholarship to this study.

Author Contributions

L.E.M.d.S. performed the selection of articles, extracted and analyzed the data, assessed the risk of bias, wrote and revised the manuscript. M.L.P.d.S. assessed the global certainty of the evidence, made the risk of bias assessment, and contributed to the writing and revision of the manuscript. P.R.F.C. performed the statistical analysis of the data. K.B.B.d.S. and L.E.M.d.S. conducted the selection of the studies. L.E.M.d.S. and W.P.d.O.A. determined the risk of bias. C.d.M.C. and L.E.M.d.S. performed data extraction; and M.L.P.d.S., E.M.P., and L.E.M.d.S. assessed the overall certainty of the evidence. All of the authors contributed to the revision of the manuscript and read and approved the final version.

Supplementary Material

Supplementary Material is available at Nutrition Reviews online.

Funding

This study was supported, in part, by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior—Brasil (CAPES), Finance Code 001.

Conflicts of Interest

None declared.

REFERENCES