Systematic review of the association between dietary patterns and perinatal anxiety and depression

The inclusion criteria used were as follows: 1. cross-sectional and cohort studies without language restriction that included pregnant women, women in the 6 months before pregnancy or 6 months after delivery of any age group; 2. studies that utilized factor analysis, principal component analysis, cluster analysis, reduced rank regression, or treelet transform for the derivation of dietary patterns assessed 6 months before pregnancy and up to 6 months postpartum; and 3. studies that included an assessment of depression and/or anxiety during pregnancy or after delivery. The exclusion criteria used were review articles and studies using dietary indices or scores for the derivation of the dietary patterns.

The virtual search was performed using PubMed, Scopus and LILACS databases. In PubMed and LILACS, the search strategy was: (diet OR diet pattern OR diet patterns OR dietary pattern OR dietary patterns OR eating pattern OR eating patterns OR food pattern OR food patterns) AND (posteriori OR factor analysis OR principal component factor analysis OR principal component analysis OR cluster analysis OR reduced rank regression OR treelet transform) AND (mental health OR mental problems OR mental disorder OR depress* OR anxiety) AND (pregnancy OR gestation OR pregnant OR postnatal OR postpartum OR perinatal OR ante-natal OR post-natal OR post-partum OR peri-natal). The search strategy used in the Scopus database was: TITLE-ABS-KEY (diet OR “diet pattern” OR “diet patterns” OR “dietary pattern” OR “dietary patterns” OR “eating pattern” OR “eating patterns” OR “food pattern” OR “food patterns”) AND TITLE-ABS-KEY (posteriori OR “factor analysis” OR “principal component factor analysis” OR “principal component analysis” OR “cluster analysis” OR “reduced rank regression” OR “treelet transform”) AND TITLE-ABS-KEY (“mental health” OR “mental problems” OR “mental disorder” OR depress* OR anxiety) AND TITLE-ABS-KEY (pregnancy OR gestation OR pregnant OR postnatal OR postpartum OR perinatal OR ante-natal OR post-natal OR post-partum OR peri-natal).

Two review authors (DS, SL) independently screened titles and abstracts from the articles identified through the database searches and subsequently assessed full texts articles to see if they met the eligibility criteria. Disagreements were resolved via consensus.

The data extracted from the studies were entered in Table 1. Data included: name of the authors, year and language of publication, country where the study was conducted, study design, publication period of the article, number and age of participants, food surveys used to evaluate habitual consumption and corresponding statistical technique for the derivation of dietary patterns, identified dietary patterns, and association between dietary patterns and depression and anxiety.

Assessment of methodological quality and risk of bias in the primary study with cohort design was performed using NOS [36], which included eight items related to selection, comparison, and outcome. For each item a star is awarded (except for comparison) that can receive up to two stars. The studies with six stars (maximum of nine) were classified as good quality. For cross-sectional studies, we used a modified version of NOS [37] in which studies that received at least five stars (maximum of eight) were classified as good quality.

Summary of evidence was provided for associations between western, healthy and traditional dietary patterns with PAAD. Considering that studies often present different definitions for dietary patterns with a similar food composition, we defined the composition of the Western dietary pattern (independent of the denomination given in the original study) as a diet involving the high intake of meats, processed cereals, high fat foods, and products with high sugar content. In contrast, a healthy dietary pattern consists predominantly of whole foods, fruits, vegetables, fish and seafood, and skimmed milk products. Finally, a traditional dietary patterns was defined as those described by the typical food consumption patterns of the studied regions.

The GRADE framework was used to assess the quality of evidence in regard to the association between dietary patterns with PAAD [38, 39]. GRADE ranks the evidence as: 1. high (there is strong certainty that the association is close to the estimated); 2. moderate (there is moderate certainty in the estimated association); 3. low (certainty in association is limited); and 4. very low (certainty in the association estimate is very limited owing to a significant degree of uncertainty in the findings). The evaluation criteria used for outcomes included the design in which clinical trials were given an A and observational studies a C. The high estimate of the association and overall consistency of the findings also affected the grade. Additionally, the presence of relevant methodological limitations, indirect evidence, and publication bias were all contributing factors that affected the score [38, 39].

The virtual searches recovered a total of 121 studies (90 from PubMed, 31 from Scopus, and 0 from LILACS. There were 23 duplicate records that were excluded, which resulted in 98 abstracts. Following evaluation of the title and abstract, an additional 87 articles were excluded. For the 11 studies that had full-text analysis, 10 met this study’s eligibility criteria and were subsequently included in the review [26‐35]. The flowchart of the selection process is presented in Fig. 1.

The characteristics of included studies are presented in Table 1. The number of participants in each study ranged from 167 [27] to 9530 [30]. The selection period for included studies ranged from 5 weeks of gestation [28, 29, 35] to 3 months postpartum [33].

Six studies examined usual dietary intake during pregnancy [26, 27, 30‐32, 35]. Two studies [28, 29] evaluated 6 months prior to pregnancy, whereas two other studies evaluated in the postpartum period [33, 34]. The survey methods used included a food frequency questionnaire (n = 7) [26‐30, 32, 33], food history questionnaire (n = 2) [31, 35], and 3-day food record (n = 1) [34]. For statistical analysis, nine studies used factor analysis [27‐35] and one used cluster analysis for evaluation of dietary pattern [26].

The 10 studies included in this review identified food consumption patterns characterized by a Western pattern [26‐35], which was referenced as processed (n = 3) [28‐30], Western (n = 3), [31, 32, 35]^, restricted (n = 1) [26], unhealthy (n = 1) [27], confectionery (n = 1) [30], mixed (n = 1) [33], or Eat-Out diet (n = 1) [34]. Nine studies identified dietary patterns classified as healthy patterns [27‐35], whereas traditional patterns were identified in seven studies [26, 28‐31, 34, 35] conducted in Brazil (n = 3) [26, 28, 29], the United Kingdom (n = 1) [30], Japan (n = 2) [31, 35], and Singapore (n = 1) [34]. The food composition for western, healthy, and traditional dietary patterns is presented in Table 2.

Assessment of prenatal (n = 2) [26, 30], postpartum (n = 1) [34], and perinatal (during pregnancy and postpartum – included 30 to 45 days after delivery; n = 1) [28] anxiety was carried out by four studies. The criteria used for evaluation of anxiety included the State-Trait Anxiety Inventory [28, 34], Primary Care Evaluation of Mental Disorders (PRIME-MD) [26], and Crown-Crisp Experiential Index [30]. Assessment of prenatal (n = 3) [26, 29, 35], postpartum (8 weeks to 9 months; n = 4) [31‐34], and perinatal (n = 1) [27] depression/depressive symptoms was conducted in eight studies [27]. The criteria used for the evaluation of depression included the PRIME-MD (n = 1) [26], and Center for Epidemiologic Studies Depression Scale (n = 1) [35]. The Edinburgh Postnatal Depression Scale (n = 6) [27, 29, 31‐34], was used for evaluation of depressive symptoms.

All cohort studies included in the review showed higher scores than six stars on the NOS scale, which is classified as having good methodological quality [27‐29, 31, 32, 34]. The cross-sectional studies also exhibited good quality as observed scores were greater than or equal to five stars [26, 30, 33, 35].

The summary of the associations between dietary patterns and PAAD is presented in Table 3.

A Western dietary pattern was positively associated with prenatal depressive symptoms in one study [27], whereas three other studies [26, 29, 35] reported no statistically significant association. The summary of evidence demonstrated very low quality based on the GRADE framework. In contrast, a healthy dietary pattern was negatively associated with prenatal depressive symptoms/depression in two studies [29, 35]. No other study reported a statistically significant association with a healthy dietary pattern and prenatal depressive symptoms [27]. The summary of evidence also demonstrated very low quality based on the GRADE framework.

There was no significant association between traditional Brazilian dietary pattern and prenatal depressive symptoms/depression in two studies [26, 29]. Another study reported a negative association between traditional Japanese dietary pattern and prenatal depression [35],. with very low quality of evidence based on the GRADE framework.

The five studies that evaluated the association between a Western dietary pattern and postpartum depression reported no statistically significant association [27, 31‐34]. The summary of evidence demonstrated very low quality based on the GRADE framework. The summary of evidence for the association between a healthy dietary pattern and postpartum depression was evaluated in five studies, but was reported to be inconclusive owing to divergences in the findings [27, 31‐34].

A study by Okubo et al. evaluated the association between a traditional Japanese dietary pattern and postpartum depression. No statistically significant association was observed and the summary of evidence showing very low quality based on the GRADE framework. A traditional Indian-confinement dietary pattern was negatively associated with postpartum depression, whereas a traditional Chinese-confinement dietary pattern reported no statistically significant association with the associations from each traditional dietary pattern exhibiting very low evidence quality, based on the GRADE framework [34].

The summary of evidence for the association between a Western dietary pattern and prenatal anxiety was inconclusive. However, a positive association was found in one study [30], whereas another study found no statistically significant association [26]. A healthy dietary pattern was negatively associated with prenatal anxiety in one study, although the summary of evidence demonstrated very low quality based on the GRADE framework [30]. The relationship between anxiety and a traditional Brazilian dietary pattern showed no significant association [26]. A study that examined the association with a United Kingdom traditional dietary pattern reported a negative association, although the summary of evidence demonstrated very low quality based on the GRADE framework [30].

Among the included studies, two evaluated the association between dietary patterns and postpartum anxiety [28, 34]. A healthy dietary pattern [28, 34] and a traditional Brazilian dietary pattern [28] were both reported to be negatively associated with postpartum anxiety. The summary of evidence for both associations was very low-quality based on the GRADE framework. In contrast, there was no significant association between a Western dietary pattern, a traditional Indian confinement dietary pattern, or traditional-Chinese confinement dietary pattern and postpartum anxiety. The summary of evidence for all three associations was reported as very low quality [34].