Antenatal diet quality and perinatal depression: the Microbiome Understanding in Maternity Study (MUMS) cohort

Abstract

Background

Previous findings from research investigating the role of antenatal nutrition in preventing postpartum depression (PPD) are inconsistent. Our primary aim was to investigate the association between pregnancy diet quality and PPD. Our secondary aim was to investigate associations between (a) diet quality and depression during pregnancy and (b) depression during pregnancy and PPD.

Methods

This analysis represents data from 73 women participating in the Microbiome Understanding in Maternity Study (MUMS) cohort in Sydney, Australia, which followed women from Trimester 1 of pregnancy to 1‐year postpartum (PP). Participants' diet quality was assessed using the Australian Eating Survey at Trimester 1 and 3 to calculate diet quality, known as the Australian Recommended Food Score (lower diet quality defined as score <39; higher diet quality ≥39). Depression was assessed using the Edinburgh Depression Scale at Trimesters 1, 2, 3 and 6 weeks PP (defined as score ≥11).

Results

Depression scores during pregnancy were significantly associated with depression score 6 weeks PP (Trimester 1: r = 0.66, Trimester 2: r = 0.69, Trimester 3: r = 0.67; all p < 0.001). Diet quality during pregnancy was not significantly correlated with 6‐week PPD score. In unadjusted analysis, diet quality during pregnancy was not associated with pregnancy depression scores. When adjusted for age, parity and Trimester 1 body mass index, Trimester 1 physical activity levels and gestational weight gain, higher Trimester 3 diet quality was associated with reduced Trimester 3 depression only.

Conclusions

Depression scores during pregnancy were positively associated with PPD, highlighting the importance of screening for depression during pregnancy and postnatally. Larger longitudinal prospective studies may elucidate the association between diet quality and PPD.

Highlights

• Depression during pregnancy was associated with postpartum depression score, highlighting the importance of adhering to routine screening guidelines throughout the perinatal period.

• Although diet is clearly important for the health of both the mother and the developing foetus during pregnancy, our analysis of data from the Microbiome Understanding in Maternity Study (MUMS) cohort did not find any associations between diet quality and depression.

INTRODUCTION

Postpartum (PP) blues is a transient form of moodiness experienced by ~85% of women 3–4 days PP, which usually dissipates within 1–2 weeks. ¹ It becomes a major depressive disorder when the feeling of despondence prolongs, known as postpartum depression (PPD). ¹ Approximately 12%–16% of women suffer from PPD worldwide, with a prevalence of 6%–12% in Australia. ² , ³ Women experience severe mood swings, extreme sadness and worthlessness that impair their ability to concentrate, leading to deleterious health consequences for the mother and infant. ¹ For the mother this includes increased suicide risk, a leading cause of maternal mortality, ⁴ and impaired mother–infant bonding and interactions, which increase infant risk of growth retardation, social‐emotional delay, attenuated cognitive skills and behavioural problems. ⁵ , ⁶

Although the causes of PPD remain unclear, nutrition has been suggested as one modifiable risk factor. ⁷ Findings from studies conducted to date investigating the association between antenatal nutrition and PPD are conflicting. A higher diet quality pattern indicative of increased intake of micronutrients and unsaturated fats via higher consumption of vegetables, fruit, pulses, nuts, dairy products, fish and olive oil has been associated with reduced PPD symptoms in two cohort studies, ⁸ , ⁹ whereas two other studies reported no association between higher diet quality and PPD. ¹⁰ , ¹¹ Furthermore, Barker and colleagues reported an association between a poorer diet quality pattern indicative of reduced intake of micronutrients and increased intake of saturated fat via higher consumption of meat, potatoes, sugar and sweets, cereals, fats except olive oil, salty snacks, eggs, beverages and sauces and increased PPD symptoms, ⁸ whereas three other studies reported no association between poor diet quality and PPD. ⁹ , ¹⁰ , ¹¹ These inconclusive findings could reflect factors such as varying PPD definitions and the limitation that all studies conducted to date have assessed diet only once, yet diet is known to change throughout pregnancy. ¹² The use of multiple measurement time points to assess antenatal diet would increase the precision of diet assessment, ¹³ suggesting the need for well‐designed longitudinal studies that measure diet at numerous time points.

Therefore, the primary aim of the present analysis was to assess the association between diet quality during pregnancy and PPD at 6 weeks PP. The secondary aim was to assess associations between diet quality during pregnancy and depression during pregnancy and depression during pregnancy and PPD. We hypothesised that there would be an inverse association between diet quality and PPD.

METHODS

The present study represents secondary data analyses from the Microbiome Understanding in Maternity Study (MUMS), an Australian longitudinal prospective cohort study investigating the maternal microbiome in women with low‐risk (≥18 years, singleton pregnancy, did not meet criteria for high‐risk) and high‐risk (body mass index [BMI] > 30, history of gestational or pre‐pregnancy diabetes mellitus or history of a hypertensive disorder of pregnancy or chronic hypertension) pregnancies. ¹⁴ The cohort recruited 117 mother–infant pairs during 2018 and 2019 who were booked in for pregnancy care at St George Hospital, a socio‐demographically diverse area of metropolitan Sydney, Australia, followed from trimester (T)1 of pregnancy through 1‐year PP. The primary objective of MUMS was to define the maternal microbiome across pregnancy and to 1‐year PP and identify key clinical and environmental variables that shape the female microbiota profile during and after pregnancy. A detailed study protocol has been previously published. ¹⁴ Ethical approval was received from the South Eastern Sydney Local Health District Research Ethics Committee (17/293 (HREC/17/POWH/605)), and the written informed consent was obtained from all participants. The inclusion criteria for the present analysis included women enrolled in MUMS, with dietary intake data at T1 or/and at T3 of pregnancy, together with depression data at 6 weeks PP. This resulted in a final sample size of 73 participants and, with power set at 0.8 and alpha set at 0.05, this powers the present analysis to detect a smallest significant correlation of r < −0.32 or r > 0.32. Depressive symptoms were assessed using the paper‐based Edinburgh Depression Scale (EDS), ¹⁵ a validated and reliable self‐rating scale questionnaire developed to screen depression during pregnancy (T1, T2 and T3) and PP (6 weeks). ¹⁶ , ¹⁷ The questionnaire consists of 10 statements, each scored on a four‐point scale, rating the intensity of depressive symptoms present in the past week. A higher sum of score, described as the Edinburgh Depression Scale Score (EDSS), ¹⁵ reflects elevated severity of depressive symptoms with a range of 0–30. There is no universal consensus on the score used to diagnose PPD. For the present analysis, a score ≥11 was used to indicate the presence of major PPD, as suggested by a 2020 systematic review to maximise sensitivity and specificity. ¹⁸ Analyses at additional cut‐points frequently used to diagnose possible/probable PPD (cut‐points 9, 10 and 13) were also conducted to confirm findings.

Dietary intake was assessed using the online Australian Eating Survey (AES), a validated and reliable self‐administered semi‐quantitative food frequency questionnaire (FFQ), designed for the Australian population. ¹⁹ The AES consists of 120 dietary questions (foods, drinks, food groups, macronutrients, micronutrients), asking about the frequency of consumption over the previous 6 months ranging from ‘never’ to ‘≥7 times per day’ in relation to standard adult portion sizes ¹⁹ and 15 supplementary questions (about vitamin supplements usage, food behaviours and sedentary behaviours). Of the 120 dietary questions, 70 focus on the consumption of eight dietary components consistent with the Australian Dietary Guidelines (ADG): vegetables (20 questions), fruit (12 questions), meat/flesh foods (7 questions), meat/flesh alternatives (6 questions), grains (12 questions), dairy (10 questions), water (1 question) and spreads/sauces (2 questions). ¹⁹ Individual mean daily macro‐ and micronutrient intakes were computed from the FFQ using the AUSNUT 2011–2013 Australian food composition database. ²⁰

Responses to the AES FFQ were also used to calculate the diet quality score, described as the Australian Recommended Food Score (ARFS). ²¹ One ARFS point is awarded for a reported frequency consumption aligned with the ADG. The overall ARFS equates to the sum of ARFS points from the eight dietary components with a possible score ranging from 0 to 73 points. The ARFS can then be categorised into four ranks: needs work (<33), getting there (33–38), excellent (39–46), outstanding (47+). ²¹ Due to the small sample size in the present study, for our analysis we combined diet quality ranks as follows: needs work–getting there (<39) indicating lower diet quality and excellent–outstanding (≥39) indicating higher diet quality.

Other variables studied in the present analysis included age, gravidity, parity, rate of high‐risk pregnancy, presence of pregnancy complications (i.e., gestational diabetes, preeclampsia, gestational hypertension), mode of birth, anthropometry (body mass index [BMI], waist circumference, hip circumference), body composition (bone mass %, fat mass %, total body water %) assessed using multichannel bioimpedance analysis (Bodystat 1500: Bodystat Ltd.) ²² and level of physical activity (reported as metabolic equivalent of task [MET] min/week) assessed using the validated self‐administered International Physical Activity Questionnaire – Short Form survey. ²³ Time points of data collection are summarised in Figure 1.

Figure 1.

Timeline of data collection relevant to the present secondary data analysis of the MUMS cohort study. AES, Australian Eating Survey; BMI, body mass index; BMR, basal metabolic rate; EDS, Edinburgh Depression Scale; MET, metabolic equivalent of task; MUMS, Microbiome Understanding in Maternity Study; PP, postpartum; T1, trimester 1; T2, trimester 2; T3, trimester 3; TBW, total body weight; wks, weeks.

Statistical analyses were performed using IBM SPSS Statistics, v27. Participants included and excluded in our analysis were (a) characterised using descriptive statistics and (b) compared using independent sample t‐tests (continuous variables) and chi‐square tests (categorical variables) as appropriate.

Bivariate Pearson's correlations were performed as appropriate to assess the associations between (a) diet quality (ARFS) at T1, T3 and depression (EDSS) at T1, T2, T3 and 6 weeks PP, (b) dietary intake of foods and nutrients of interest (i.e., core foods, non‐core foods, the five food groups, the three macronutrients [including types of fats, sugar and fibre] and seven micronutrients of interest that have previously been suggested to play a role in the development of depression [thiamine, riboflavin, niacin, folate, sodium, magnesium, zinc]) at T1, T3 and EDSS at T1, T2, T3 and 6 weeks PP, and (c) EDSS at T1, T2, T3 and EDSS at 6 weeks PP. Multiple linear regression was also used to adjust for age, parity, T1 BMI, T1 physical activity levels and gestational weight gain (T1–T3) due to their previously reported association with depression and/or diet quality in previous antenatal studies. ²⁴ , ²⁵ , ²⁶

Independent sample t‐tests were performed to compare the mean in ARFS (T1, T3) between EDSS <11 and ≥11 groups and Mann–Whitney U tests were used to compare the mean in EDSS (T1, T2, T3 and 6 weeks PP) between the two predefined ARFS (T1, T3) ranks (needs work–getting there [<39]) and (excellent–outstanding [≥39]).

RESULTS

Of the 117 participants recruited to the MUMS cohort study, 1 was lost to follow‐up, 11 withdrew and 6 were excluded from the primary study. For the present study, a further 14 participants were excluded due to missing dietary data at T1 and T3, and 12 were excluded due to missing depression data at 6 weeks PP, leaving 73 participants in this analysis. Participants included in the present analysis were more likely to be Caucasian, had their babies at a slighter later gestational age and were on average 2 years older than those excluded (Tables S1 and S2). The EDSS at T1 was significantly higher in those excluded from analysis compared with those included in analysis (Table S2).

Characteristics of the cohort are presented in Table 1. Participants had a mean ± SD age of 34.0 ± 4.4 years, and the majority were of White or Asian ethnicity. The mean EDSS was <11 at all time points, and the proportion of participants with an EDSS ≥11 ranged from 7% to 14% at all time points. The cohort's diet consisted of 69% and 67% of energy from core foods (31% and 33% from non‐core foods) at T1 and T3, respectively. Further details are found in Table S3.

Table 1.

Characteristics of the MUMS cohort eligible for this sub‐study

Participant characteristics	Descriptive	n
Age in years, M ± SD	34.0 ± 4.4	73
Ethnicity, n (%)	‐	73
Asian	13 (18)	‐
Middle Eastern	2 (3)	‐
Latino/Hispanic	3 (4)	‐
White	51 (70)	‐
Mixed	2 (3)	‐
Other	2 (3)	‐
Gravidity, n (%)a	‐	73
Nulligravida (0)	1 (1)	‐
Primigravida (1)	24 (33)	‐
Multigravida (>1)	48 (66)	‐
Parity, n (%)a	‐	73
Nullipara (0)	27 (37)	‐
Primipara (1)	29 (40)	‐
Multipara (>1)	17 (23)	‐
High risk pregnancy, n (%)a	32 (44)	73
Complications, n (%)
Gestational diabetes	11 (16)	70
Preeclampsia	4 (6)	72
Gestational hypertension	6 (8)	72
EDSS, median (IQR)
T1	3.0 (0.5–5.5)	68
T2	4.0 (1.5–6.5)	70
T3	3.0 (0.0–6.0)	70
PPD 6 weeks	4.0 (1.0–7.0)	73
No. of participants with EDSS ≥11
T1 n (%)	10 (14)	70
T2 n (%)	9 (13)	70
T3 n (%)	5 (7)	70
PP 6 weeks n (%)	7 (10)	73
Gestational weight gain (T1 to T3), kg M ± SD	9.8 ± 4.0	72

Participant characteristics	T1	T3	T1	T3
BMI, M ± SD	25.9 ± 5.4	30.0 ± 5.5	73	72
<18.5 kg/m2 n (%)	1.00 (1)	0.00 (0)	‐	‐
18.5–24.9 kg/m2 n (%)	39.0 (53)	14 (19)	‐	‐
≥25 kg/m2 n (%)	33.0 (45)	58 (81)	‐	‐
Fat mass %, M ± SD	33.3 ± 8.0	37.3 ± 6.4	71	72
ARFS M ± SD	35.5 ± 10	36.9 ± 10	71	64
Outstanding 47 + n (%)	11 (15)	13 (20)	‐	‐
Excellent 39–46 n (%)	16 (23)	18 (28)	‐	‐
Getting there 33–38 n (%)	15 (21)	8 (13)	‐	‐
Needs work <33 n (%)	29 (41)	25 (39)	‐	‐

Abbreviations: ARFS, Australian Recommended Food Score; EDSS, Edinburgh Depression Scale Score; IQR, interquartile range; M, mean; MUMS, Microbiome Understanding in Maternity Study; n, number of participants from cohort of 73; PP, postpartum; SD, standard deviation; T1, Trimester 1; T2, Trimester 2; T3, Trimester 3.

^a At the start of pregnancy.

ARFS during pregnancy at T1 and T3 was not associated with EDSS at 6 weeks PP (Table 2), including after adjustment for age, parity, T1 BMI, T1 physical activity levels and gestational weight gain. Similarly, unadjusted analysis indicated that ARFS during pregnancy at T1 and T3 was not associated with EDSS during pregnancy at T1, T2 or T3 (Table 2). In adjusted analysis, T3 ARFS was associated with T3 depression (regression coefficient B [95% confidence intervals]: 0.16 [0.03–0.29]; p = 0.017) only. When comparing participants with a lower diet quality with those who had a higher diet quality, no statistically significant difference in 6‐week PP EDSS was observed at T1 or T3 (Table 3). Similarly, when categorised into the two EDSS groups (i.e., depression score: <11 vs. ≥11) no difference in diet quality (ARFS) at T1 or T3 between the two EDSS groups at 6 weeks PP was observed (Table 4). Similarly, there were no significant differences in diet quality between those above or below the EDSS cut‐points during pregnancy (Tables 3 and 4). There were no differences in diet quality or depression score at any time point among women who took supplements at T1 (90% of women) and/or T3 (88% of women) compared with those who did not.

Table 2.

Correlations between diet quality during pregnancy and antenatal and PP depression

ARFS and EDSS	Pearson's correlation (r)	p‐Value	n
T1 ARFS and EDSS PP 6 weeks	−0.16	0.18	70
T3 ARFS and EDSS PP 6 weeks	0.03	0.81	64
T1 ARFS and EDSS T1	−0.14	0.25	67
T1 ARFS and EDSS T2	−0.05	0.68	68
T1 ARFS and EDSS T3	−0.09	0.48	68
T3 ARFS and EDSS T3	0.19	0.14	62

Abbreviations: ARFS, Australian Recommended Food Score; EDSS, Edinburgh Depression Scale Score; n, number of participants from cohort of 73; PP, postpartum; (r), bivariate Pearson's correlation analysis; T1, Trimester 1; T2, Trimester 2; T3, Trimester 3.

Table 3.

Antenatal and postnatal depression scores of women with low versus high diet quality during pregnancy, median (IQR)

	PP 6 weeks			T1 depression			T2 depression			T3 depression
	EDSS	n	p‐Value	EDSS	n	p‐Value	EDSS	n	p‐Value	EDSS	n	p‐Value
T1 lower diet quality	3.5 (0.5–6.5)	44	‐	3.0 (0.5–5.5)	40	‐	3.5 (1.5–5.5)	42	‐	3.0 (0.0–6.0)	42	‐
T1 higher diet quality	5.0 (2.0–8.0)	27	0.52	3.5 (0.0–7.0)	26	0.98	4.0 (1.5–6.5)	27	0.59	4.0 (0.5–7.5)	27	0.78
T3 lower diet quality	4.0 (1.0–7.0)	33	‐	‐	‐	‐	‐	‐	‐	3.0 (0.5–5.5)	32	‐
T3 higher diet quality	4.0 (2.5–5.5)	31	0.54	‐	‐	‐	‐	‐	‐	3.5 (0.0–7.0)	30	0.29

Abbreviations: EDSS, Edinburgh Depression Scale Score; IQR, interquartile range; n, number of participants from cohort of 73; T1, Trimester 1; T2, Trimester 2; T3, Trimester 3.

Table 4.

Diet quality during pregnancy in women with versus without depressive symptoms during pregnancy and PP

	T1 diet			T3 diet
	ARFS M ± SD	n	p‐Value	ARFS M ± SD	n	p‐Value
PP 6 weeks <11 EDSS	35.9 ± 9.5	64	0.28	36.9 ± 9.3	59	0.96
PP 6 weeks ≥11 EDSS	31.6 ± 14.2	7		37.4 ± 19.4	5
T1 < 11 EDSS	36.4 ± 9.5	57	0.19	37.6 ± 9.5	53	0.36
T1 ≥ 11 EDSS	31.6 ± 13.8	9		33.9 ± 15.0	7
T2 < 11 EDSS	36.2 ± 9.1	63	0.79	37.4 ± 9.3	56	0.53
T2 ≥ 11 EDSS	34.5 ± 14.4	6		32.3 ± 18.0	6
T3 < 11 EDSS	36.0 ± 9.4	64	0.69	36.4 ± 10.0	58	0.11
T3 ≥ 11 EDSS	32.6 ± 17.5	5		45.0 ± 12.5	4

Note: Lower diet quality = ARFS < 39; higher diet quality = ARFS ≥ 39; EDSS cut‐off point used to indicate major depression ≥11.

Abbreviations: ARFS, Australian Recommended Food Score; EDSS, Edinburgh Depression Scale Score; IQR, interquartile range; M, mean; n, number of participants from cohort of 73; PP, postpartum; SD, standard deviation; T1, Trimester 1; T2, Trimester 2; T3, Trimester 3.

Specific food groups and nutrients associated with higher depression scores throughout pregnancy and at 6 weeks PP are outlined in Table 5. Correlations of all food groups/nutrients with depression scores throughout pregnancy and at 6 weeks PP are shown in Table S4.

Table 5.

Specific food groups and nutrients associated with higher score on the Edinburgh Depression Scale at Trimester 1 and Trimester 3 of pregnancy

Higher depression scores	Trimester 1 diet factors	Trimester 3 diet factors
Trimester 1	Less dairy intake	‐
Trimester 2	Greater sweetened drink intake	‐
Trimester 3	‐	Greater sugar, packaged snack and magnesium intakes
Postpartum depression	Less daily fruit intake	Smaller % energy from core foods

Higher EDSS at each time point in pregnancy (T1, T2 and T3) was significantly associated with a higher EDSS at 6 weeks PP in unadjusted analysis (Table 6). These associations did not change when adjusted for age, parity, T1 BMI, T1 physical activity levels and gestational weight gain.

Table 6.

Correlations between antenatal depression and postnatal depression

	Pearson's correlation (r)	p‐Value	n
T1 EDSS and EDSS PP 6 weeks	0.66	<0.001	70
T2 EDSS and EDSS PP 6 weeks	0.69	<0.001	70
T3 EDSS and EDSS PP 6 weeks	0.67	<0.001	67

Abbreviations: EDSS, Edinburgh Depression Scale Score; n, number of participants from cohort of 73; PP, postpartum; (r), bivariate Pearson's correlation analysis; T1, Trimester 1; T2, Trimester 2; T3, Trimester 3.

DISCUSSION

To the best of our knowledge, this is the first study to examine associations between multiple assessments of antenatal diet quality and PPD. Overall, our results suggest that diet quality, during early or late pregnancy, does not influence the development of depression at 6 weeks PP, nor does antenatal diet quality influence the presence of depressive symptoms during pregnancy. However, an EDSS indicating higher risk for antenatal depression was associated with an increased depression score at 6 weeks PP.

An association between antenatal diet and perinatal depression is biologically possible. ²⁷ Nutrition plays a role in modulating hormonal, immunological and biochemical processes, all of which are associated with the development of depression, suggesting that nutrition plays a plausible role in the development of this multifactorial illness. ²⁷ The significant physical, physiological and immunological changes that a woman's body undergoes during pregnancy and PP mean that women are particularly susceptible to nutrient deficiencies resulting from suboptimal dietary intake during this life stage. ²⁷ This in turn may influence the mechanisms that underlie depression. ²⁸ Bolton and colleagues have proposed a detailed mechanism of how perinatal consumption of a Western diet may increase the risk of PPD whereby high plasma level of branched‐chain amino acids, typical of a Western diet, competes with mood‐altering neurotransmitter precursors crossing the blood brain barrier, reducing the production of neurotransmitters including dopamine, histamine and serotonin, consequently, increasing PPD risk. ²⁸

Despite the plausible mechanism, we did not find an association between antenatal diet quality and depression at any time point, including our primary time point at 6 weeks PP. This finding is in line with three studies that found healthy ¹⁰ , ¹¹ and unhealthy ⁹ , ¹⁰ , ¹¹ diets during pregnancy were not associated with PPD symptoms at 8 weeks, ⁹ 10 weeks, ⁹ 2 months, ¹⁰ 3 months ¹¹ and 9 months ¹⁰ PP, respectively. However, our findings are in contrast with Barker et al. and Chatzi et al. that healthy ⁸ , ⁹ and unhealthy ⁸ antenatal diets were significantly associated with decreased ⁸ , ⁹ and increased ⁸ PPD symptoms, respectively.

We also did not find any significant associations between antenatal diet and antenatal depression. This is in contrast to findings from a narrative review of 27 studies which concluded that antenatal depression was a barrier to good antenatal diet quality with 22 of 27 studies finding an inverse association between these two outcomes. ²⁹ Despite the lack of associations between diet quality and depression observed in the present study, diet clearly remains an important factor for the health of both the mother and the developing foetus during pregnancy for a variety of reasons, including foetal/child growth and risk of subsequent cardiometabolic disease. ³⁰ , ³¹

To provide a comprehensive investigation of the associations between antenatal diet and depression, we also conducted correlation analysis between specific food groups/nutrients and depression. In these analyses we demonstrated significant associations between depression and particular dietary items. For example, increased depression scores during pregnancy or at 6 weeks PP were associated with reduced dairy, increased sweetened drinks, sugar, packaged snacks, magnesium intakes and increased non‐core foods, reduced core foods and fruit intake. However, these associations did not track throughout the whole pregnancy and PP time course and, although findings are in accordance with some other literature, ¹² it is possible that these represent chance findings given that we did not correct for type 1 error which may have existed given the number of correlation analyses run as part of this study. ³² Therefore, these findings should be interpreted with caution.

Our study also identified a significant association between higher depression scores during pregnancy and 6 weeks PP. This is in keeping with multiple prior longitudinal cohort studies, which have shown that about 40% of women with elevated depressive symptoms PP also had elevated depression scores during pregnancy. ³¹ , ³³ , ³⁴ , ³⁵ , ³⁶ Furthermore, antenatal depression has consistently been identified as a risk factor for PPD. ³⁷ , ³⁸ , ³⁹ This highlights the importance of screening for depression throughout the antenatal period, to not only treat antenatal depression, but also to prevent postnatal depression. This is in line with the 2017 Australian Clinical Practice Guideline Mental Health Care in the Perinatal Period, which recommends routine, universal antenatal and postnatal mental health screening including monitoring and repeating in 2–4 weeks for women with an initial EPDS score between 10 and 12, and arranging further assessment for woman with an EPDS score of 13 or more. ⁴⁰

Strengths of this study include the longitudinal assessment of both diet quality and depression throughout the perinatal period. We also conducted our analysis at multiple EDSS cut‐points, with results unchanged, adding further strength to our findings. Furthermore, our assessment of depression at 6 weeks PP aligns with the 2017 Australian Clinical Practice Guideline Mental Health Care in the Perinatal Period recommendation to first screen for postnatal depression at 6–12 weeks PP. ⁴⁰ This time point at 6 weeks PP also allowed for the association between antenatal diet and PPD to be assessed with minimal influence from postnatal confounders. Although diet assessments are imperfect by nature, the use of the AES to assess diet quality was a strength of this study as it is well validated, provided a detailed assessment of the maternal diet and accounts for increased nutritional needs during pregnancy. However, we could not account for the influence of supplement use on nutrient intake as the AES does not record the brand, type or dosage which is a limitation. Although both the AES and EDS are self‐reported questionnaires, both are validated and reliable. However, participants reported energy intake of 7855 ± 2706 kJ/day (46% energy from carbohydrates, 19% energy from protein and 35% energy from fat) in T1 and 8319 ± 2416 kJ/day (45% energy from carbohydrates, 19% energy from protein and 36% energy from fat) in T3. This is below the recommended dietary intake for pregnant women ⁴¹ suggesting some level of underreporting which is typical of self‐report surveys. Other methods of dietary assessment, including doubly labelled water and weighed food records, were considered to be not practical or acceptable in this population, and assessment of plasma carotenoids, although useful for indicating fruit and vegetable intake, does not address the main problem of underreporting of calories observed in our cohort. Of note, the AES has been validated against doubly labelled water, ⁴² plasma carotenoids ⁴³ and fatty acids ⁴⁴ in adult comparative studies. Participants in our study received ongoing care fortnightly throughout their pregnancy from an obstetrician who delivered both health and social support. This high level of support could have significantly improved participants' depressive feelings and eating habits, which may have influenced our findings. The development of PPD is a multifactorial process of which nutrition may be but one modifiable risk factor. ⁷ Although we were able to adjust for age, parity, T1 BMI, T1 physical activity levels and gestational weight gain, our present analysis did not adjust for other factors known to influence depression and/or diet quality including socio‐economic status ⁴⁵ and history of depression. ²⁴ Although not significant, at both T1 and T3, diet quality was reduced in women who scored ≥11 on the EDS compared with those who did not, suggesting that our study may not have been powered to find a significant difference which is a common limitation of secondary analyses. In addition, there were some socio‐demographic differences between the included and excluded cohort, so findings may be less applicable to a non‐Caucasian, younger cohort. Excluded participants also had higher depression scores in T1 which may have affected our findings and requires further investigation. Finally, our sample size of 73 participants had limited power to detect significant associations between outcomes of interest.

Future studies assessing the association between pregnancy diet quality and PP or antenatal depression should ensure supplement usage is suitably accounted for. Furthermore, standardisation of a cut‐off point for the detection of depression would allow more direct comparisons to be made between studies.

In conclusion, antenatal diet quality was not associated with perinatal depression. More studies are warranted to further investigate this multifaceted relationship, including longitudinal studies with standardised methodology, a larger sample size and detailed supplement usage. Antenatal depression scores were predictive of higher‐risk PPD scores, highlighting the need to adhere to routine screening guidelines throughout the perinatal period.

AUTHOR CONTRIBUTIONS

Amanda Henry, Daniella Susic and Maria E. Craig contributed to conception of the MUMS cohort study. Daniella Susic collected data. Megan L. Gow and Hiba Jebeile designed this sub‐study. Megan L. Gow and Yei W. I. Lam conducted statistical analyses, data interpretation and prepared the manuscript. All other authors provided intellectual input in revisions of the manuscript and contributed to finalising and approving the final version of the manuscript. The content of the present study has not been published elsewhere.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

TRANSPARENCY DECLARATION

Dr. Gow affirms that this manuscript is an honest, accurate, and transparent account of the study being reported. The reporting of this work is compliant with STROBE guidelines. Dr. Gow affirms that no important aspects of the study have been omitted and that any discrepancies from the study as planned (prospectively registered with Australian New Zealand Clinical Trials Registry: ACTRN12618000471280; https://www.anzctr.org.au/Trial/Registration/TrialReview.aspx?id=374454%26isReview=true) have been explained.

PEER REVIEW

The peer review history for this article is available at https://publons.com/publon/10.1111/jhn.13081

Supporting information

Supporting information.

Biographies

Megan Gow, doctor, is an NHMRC Early Career Research Fellow and dietitian at the University of Sydney with a research programme focused on optimising maternal and infant health postpartum.

Yei W. I. Lam, Miss, is a new graduate dietitian who completed a Masters of Nutrition and Dietetics at the University of Sydney in 2020–21.

Hiba Jebeile, doctor, is an early career researcher and dietitian at the University of Sydney. She is the co‐lead and programme manager of the Eating Disorders In weight‐related Therapy (EDIT) Collaboration, assessing individual eating disorder risk during obesity treatment in adolescents and adults.

Maria Craig is professor of paediatric endocrinology at the Children's Hospital in Westmead, a principal investigator at Westmead Applied Research Centre, an academic co‐director at Charles Perkins Centre Westmead and medical director of the Australasian Diabetes Data Network (ADDN).

Daniella Susic, doctor, is an obstetrics and gynaecology professional and senior lecturer and PhD student at the University of New South Wales. Dr. Susic's PhD included facilitating and leading the Microbiome Understanding in Maternity Study (MUMS) cohort study at St George Hospital.

Amanda Henry is associate professor in obstetrics and gynaecology, an obstetrician at St George Public Hospital and the Royal Hospital for Women, Sydney and honorary senior research fellow, Global Women's Health at the George Institute for Global Health.

Gow ML, Lam YWI, Jebeile H, Craig ME, Susic D, Henry A. Antenatal diet quality and perinatal depression: the Microbiome Understanding in Maternity Study (MUMS) cohort. J Hum Nutr Diet. 2023;36:754–762. 10.1111/jhn.13081