Prenatal depression and diet quality during pregnancy

Abstract

Background:

Maternal nutrition during pregnancy has a significant effect on the health of the offspring and mother, highlighting the need for identifying factors that may impact diet during pregnancy. Research in non-pregnant and pregnant populations suggest depression may play a role.

Objective:

Investigate the relationship between prenatal depression and diet quality during pregnancy overall and by race/ethnicity and explore the relationships between prenatal depression and the 12 Healthy Eating Index (HEI-2010) dietary components.

Design:

A cross-sectional secondary analysis of a cohort study of Kaiser Permanente Northern California women entering prenatal care between October 2011 and April 2013.

Participants/setting:

Participants included 1,160 adult pregnant women.

Main outcome measures:

Poor diet quality was defined as a HEI-2010 score in the lowest quartile.

Statistical analyses performed:

Logistic regression was used to assess the relationship between prenatal depression (defined as a depression diagnosis, PHQ-9 score of 10 or greater or antidepressant medication dispensing between the last menstrual period and completion of the FFQ) and poor diet quality overall and by race/ethnicity. T-tests and linear regression analyses were used to assess the relationships between prenatal depression and each of the 12 HEI-2010 dietary components.

Results:

159 (14%) of the participants had prenatal depression. Women with prenatal depression had nearly two times the odds of poor diet quality (OR:1.80;95%CI:1.23,2.60) compared to women without prenatal depression, after adjusting for potential confounders. Differences emerged by race/ethnicity; after adjusting for potential confounders the adjusted odds of poor diet quality were significant only among Hispanic women. Hispanic women with prenatal depression had an increased odds of poor diet quality compared to Hispanic women without prenatal depression (OR:2.66, 95%CI:1.15,6.06). Women with prenatal depression had a higher consumption of empty calories (from solid fats, alcohol, and added sugars; threshold for counting alcohol is greater than 13g/1,000 kcal ) (p=0.01) and lower consumption of greens and beans (p<0.05), total fruit (p<0.01), and whole fruit (p<0.01), compared to women without prenatal depression. Except for empty calories, these findings remained after adjusting for potential confounders.

Conclusions:

Study findings suggest that women with prenatal depression are at a higher risk of poor diet quality compared to women without prenatal depression, and the relationship is stronger among Hispanic women. Nutritional counseling interventions for women with depression should consider the use of culturally sensitive material and target limiting empty calories from solid fats, alcohol, and added sugars and encouraging eating more greens, beans and fruit.

INTRODUCTION

Maternal nutrition during pregnancy has a significant impact on fetal growth and development, as well as the woman’s health status during and after pregnancy. Both under- or over-nutrition during pregnancy is associated with poor long-term health outcomes of the offspring in childhood and adulthood, including metabolic disease and obesity^1-4. A poor quality diet during pregnancy may also compromise maternal health by promoting obesity^5,6, excessive gestational weight gain⁷ and postpartum weight retention⁸.

The assessment of overall diet quality has garnered attention because it considers the complex synergistic effect of food and nutrients rather than individual items. Index scores such as the Healthy Eating Index (HEI) - 2010⁹ are often used to assess overall diet quality. The HEI-2010 measures conformance to the 2010 Dietary Guidelines for Americans (2010 DGAs)¹⁰ jointly published by the United States Department of Agriculture (USDA) and Department of Health and Human Services. Healthy eating patterns that are consistent with the 2010 DGAs emphasize fruits, vegetables, whole grains, fat-free/low fat milk and milk products, seafood, lean meats and poultry, eggs, beans and peas, and nuts and seeds while limiting intake of sodium, solid fats, added sugars, and refined grains¹⁰. Disparities in diet quality by race/ethnicity have been documented in the US among both non-pregnant¹¹ and pregnant women¹²; specifically, suboptimal diet quality is more common among pregnant Hispanics and non-Hispanic Blacks.

Little is known about factors that may impact diet quality during pregnancy, but the nascent literature in both non-pregnant and pregnant populations suggests depression may play a role. Prenatal depression impacts up to 22% of pregnant women^13,14 with higher rates in racial/ethnic minorities. An inverse relationship between prenatal depression and diet quality in low-income pregnant women has been suggested, such that women with prenatal depression reported a low-quality diet during pregnancy¹⁵. However, the research in pregnant populations is limited. Similar findings have also emerged in studies of non-pregnant populations ^16,17.

The impact that maternal nutrition has on the long-term health of both the offspring and the mother highlights the need for a greater understanding of factors that may impact diet quality during pregnancy. If prenatal depression is related to poor diet quality, it will provide information for targeted interventions. Further, understanding differences in the individual HEI-2010 components may provide important information for dietary guidance. Thus, the purpose of this study was to investigate the differences in overall diet quality during pregnancy between women with and without prenatal depression and whether this relationship differs by race/ethnicity. Additionally, this study aimed to examine relationships between prenatal depression and the twelve individual HEI-2010 components in a cohort of pregnant women.

MATERIALS AND METHODS

Study Setting

The study setting was Kaiser Permanente Northern California (KPNC), a large group practice prepaid health plan that provides comprehensive medical services to members living in a 14-county region of Northern California (approximately 30% of the surrounding population). The demographic, racial/ethnic, and socioeconomic characteristics of the KPNC membership are representative of the population residing in the same geographic area except that the very poor and very wealthy are underrepresented^18,19.

Study Design

The current analysis is a secondary data analysis using data from the Study of how Pre-pregnancy and Pregnancy Lifestyle Influences the Outcome of Delivery (PEAPOD) study. PEAPOD is a pregnancy cohort study in which participants were recruited from a sample of 12,662 KPNC reproductive-aged women members who completed a health survey between 2007 and 2009 and consented to be included in future research. Participants from this cohort who were pregnant between October 2011 and April 2013, English-speaking and 18 years of age or older were eligible to be recruited for the PEAPOD study. The PEAPOD study enrolled women throughout pregnancy to examine the associations between maternal lifestyle and gestational weight gain and preterm birth and childhood overweight/obesity. A total of 2,136 eligible participants were identified from KPNC’s electronic health records (EHR) and mailed a questionnaire to ascertain information on demographic characteristics including race, ethnicity and lifestyle behaviors during pregnancy, as well as a modified Block Food Frequency Questionnaire (FFQ). The questionnaire was conducted to obtain patient reported information on lifestyle and demographic factors not available in the EHR. Follow-up phone calls were conducted if the survey was not returned within 3 weeks. Return of the survey was considered consent. The PEAPOD study includes a total of 1,810 pregnancies (85% of the 2,136 originally identified as eligible) for which women completed the survey and FFQ. This study was approved by the Kaiser Permanente Northern California Institutional Review Board.

Measures

Prenatal depression —

Kaiser Permanente Northern California began implementing a universal perinatal depression screening program in 2010 in which women are screened two times during pregnancy (at the first prenatal visit and at their 24-week visit)¹⁴ using the Patient Health Questionnaire (PHQ-9). The PHQ-9 has been validated in many studies as an instrument for screening for depression with high sensitivity (> 88%) and specificity (> 88%) in obstetric patients^20-24, as well as a tool to establish depression severity and outcome²⁵. It includes nine questions with potential scores from zero to 27. A score of 10 or greater suggests moderate to severe depression. For this analysis prenatal depression was defined as any of the following between the start of pregnancy and the FFQ completion date: (1) a PHQ-9 score of 10 or greater (indicating moderate to more severe depression severity) (if more than one PHQ-9 score was greater than 10, the highest PHQ-9 score was used), (2) a depression diagnosis (International Classification of Diseases, Ninth Revision (ICD-9) codes²⁶: 296.20-296.25, 296.30 - 296.35, 296.82, 298.0, 300.4, 309.0, 309.1, 309.28, 648.41-648.44, or 311), or (3) an antidepressant medication dispensing (see Table 1 online only for a list of medications).

Table 1.

List of Antidepressant Medications Included in the Definition of Prenatal Depression in a Study of Prenatal Depression and Diet Quality Conducted in Kaiser Permanente Northern California between 2011-2013.

Amitriptyline

Bupropion

Citalopram

Desipramine

Doxepin

Duloxetine

Escitalopram Oxalate

Fluoxetine

Mirtazapine

Nefazodone

Nortriptyline

Paroxetine

Sertraline

Trazodone

Venlafaxine

Diet Quality —

Dietary intake during pregnancy was assessed using a modified Block 2005 Food Frequency Questionnaire (FFQ) tailored to accommodate the diverse dietary habits of the multi-racial/ethnic study cohort and used in previous studies^27,28. The FFQ is designed to collect information on usual dietary intake during the previous three months. The food list for the Block FFQ 2005 was developed from NHANES 1999-2000 dietary recall data²⁹; the nutrient database was developed from the USDA Food and Nutrient Database for Dietary Studies (FNDDS), version 1.0³⁰. A series of “adjustment” questions provide greater accuracy in assessing fat and carbohydrate intake³¹. Individual portion size is asked for each food, and pictures are provided to enhance accuracy of quantification. Dietary supplement intake was not included in the FFQ. Participants reported their usual intake and portion sizes of food and beverages. The Block FFQ has demonstrated adequate reliability and validity in comparison to multiple dietary records³², serving as a useful instrument for estimating energy, food, and nutrient intakes among diverse populations including pregnant women^27,33.

The HEI-2010 score was calculated using the simple HEI scoring algorithm method to assess overall diet quality on the individual level. The HEI-2010 score is a valid and reliable measure and was chosen over other measures of diet quality, such as the alternate Mediterranean diet scores³⁴, because it reflects adherence to the 2010 Dietary Guidelines for Americans⁹. Notably, the HEI-2010 and not the HEI-2015³⁵ was used because the 2010 dietary guidelines covered the study period 2011-2013, whereas the 2015-2020 guidelines³⁶ were not released until 2016. The HEI-2010 consists of 12 components (i.e., total fruit, whole fruit, total vegetables, greens and beans, whole grains, dairy, total protein foods, seafood and plant proteins, fatty acids, refined grains, sodium and empty calories) with a maximum possible score of 100³⁷. The HEI-2010 scores are inversely related to dietary intake for refined grains, sodium and empty calories components and positively related to dietary intake of the other 9 dietary components. Based on the distribution of the HEI-2010 scores, quartiles were calculated. For the primary outcome, participants who scored in the lowest quartile were classified as consuming a poor-quality diet and compared to participants in the second, third and fourth quartiles combined. The secondary outcome was the mean scores for each of the individual 12 components of the HEI-2010.

Race/Ethnicity.

Self-reported race/ethnicity was categorized as 1) non-Hispanic White, 2) Asian/Pacific Islander, 3) Hispanic, or 4) Other/Unknown. There were very few non-Hispanic Black participants, thus they were combined with American Indian/Alaskan Native, multiple races, other, or unknown race/ethnicity participants in the Other/Unknown category.

Confounders

Potential confounders were ascertained from both the self-report questionnaire and EHR data. Self-report variables were: maternal age at the start of pregnancy in years, marital status (married/partner versus single/divorced/widowed), maternal education (less than college versus college or greater), parity (0 versus 1 or more), any alcohol use during pregnancy (yes/no), any smoking during pregnancy (yes/no), and prenatal vitamin use (yes/no). Pre-pregnancy body mass index (BMI; kg/m²) was calculated using data from the questionnaire and supplemented with EHR data when missing. BMI was categorized as underweight (<18.5 kg/m²), normal weight (18.5 - < 25 kg/m²), overweight (25 - < 30 kg/m²) and obese (≥30 kg/m²)³⁸.

Data Analysis

Differences in the percentage of women with prenatal depression (yes/no) across quartiles of the HEI-2010 total score were assessed with a chi-squared test. Bivariate analyses were conducted to compare characteristics between women with and without prenatal depression; chi-squared analyses were conducted for categorical data and t-tests for continuous data. Unconditional logistic regression analysis was used to obtain odds ratios (ORs) and confidence intervals (CIs) estimating the odds of poor diet quality (vs. good diet quality) given prenatal depression status. Covariates included in the models were chosen based on the literature and included maternal age, marital status, maternal education, parity, pre-pregnancy body mass index (BMI), alcohol use during pregnancy, smoking status during pregnancy, prenatal vitamin use during pregnancy, and race/ethnicity^39-42. A likelihood ratio test was conducted to assess the interaction between prenatal depression and race/ethnicity on diet quality by comparing the full model with all covariates and an interaction term for race/ethnicity and prenatal depression to the nested model that excluded the interaction term. Tests for interaction generally have less power to assess statistical significance and thus a p-value < 0.10 was accepted for the cutoff suggesting statistical significance⁴³. The aORs and 95% CIs from the full model were calculated and reported for non-Hispanic Whites, Hispanic and Asian/Pacific Islander women. A sensitivity analysis was conducted restricting the overall sample to participants who had a PHQ-9 screen and followed the same analytic methods described for the main analysis. T-tests assessed differences in the means of the 12 individual HEI-2010 between women with and without prenatal depression. Linear regression models were conducted to estimate the relationship between prenatal depression and each component of the HEI-2010 after adjusting for covariates. Chi-squared analyses were conducted to compare women in the final analytic sample to women who were excluded. A p-value of <0.05 was considered statistically significant. Data were extracted from the EHR and merged with survey data using SAS 9.4 (Cary, NC)⁴⁴ and analyzed using R version 3.4.4⁴⁵.

RESULTS

Among the 1,810 pregnancies in the PEAPOD study the second pregnancy of any woman in the study period was excluded to avoid non-independent observations (n=13). Additionally, to ascertain depression status during pregnancy and prior to completing the FFQ, women who did not have at least one of the following were excluded: a PHQ-9 screen administered through the perinatal depression screening program, a clinical depression diagnosis, or an antidepressant medication fill after the start of pregnancy and before completing the FFQ (n=578). Finally, women whose response on the FFQ resulted in a food energy intake that was not considered plausible, defined as less than 800 kcal (n=38) or greater than 3500 kcal (n=21) were excluded. The final sample included 1,160 women (Figure 1). The median gestational age at FFQ completion was 25 weeks and more than 75% of the women completed the FFQ at 16 weeks gestation or later. Women excluded from the sample were more likely to have not taken prenatal vitamins during pregnancy compared to women in the sample (3.9% vs. 2.7%, p<0.05). There were no other significant differences between women in the study and women excluded from the study sample (Table 2).

Figure 1.

A Flow Diagram of the Inclusion and Exclusion Criteria for a Study on Prenatal Depression and Diet Quality during Pregnancy

Table 2.

Characteristics of Pregnant Women Included in the Analytic Sample (n=1160) Compared to Women participants of the PEAPOD Study that were Excluded from the Analytic Sample (n=637) in a Study of Prenatal Depression and Diet Quality conducted in Kaiser Permanente Northern California between 2011-2013.

	Excluded from study sample	Current Sample
	n (%)	n (%)
	637	1160	p-valuea
Maternal age (at pregnancy start)
< 35 years age	385 (60.4)	704 (60.7)	0.978
35+ years of age	247 (38.8)	446 (38.4)
Missing	5 (0.8)	10 (0.9)
Marital status
Single/Divorced/Separated	34 (5.3)	40 (3.4)	0.081
Married/Living with partner	601 (94.3)	1119 (96.5)
Missing	2 (0.3)	1 (0.1)
Maternal Education
< College	40 (6.3)	61 (5.3)	0.349
College +	595 (93.4)	1098 (94.7)
Missing	2 (0.3)	1 (0.1)
Race/Ethnicity
Non-Hispanic White	329 (51.6)	631 (54.4)	0.684
Hispanic	119 (18.7)	213 (18.4)
Asian/Pacific Islander	118 (18.5)	195 (16.8)
Other/Unknownb	71 (11.1)	121 (10.4)
Parity
1+	375 (58.9)	732 (63.1)	0.086
0	262 (41.1)	428 (36.9)
Pre-pregnancy BMI
Underweight (<18.5 kg/m2)	19 (3.0)	32 (2.8)	0.099
Normal weight (18.5 - < 25 kg/m2)	356 (55.9)	626 (54.0)
Overweight (25 - < 30 kg/m2)	137 (21.5)	270 (23.3)
Obese (≥30 kg/m2)	115 (18.1)	227 (19.6)
Missing	10 (1.6)	5 (0.4)
Alcohol use during pregnancy
Yes	148 (23.2)	240 (20.7)	0.232
No	489 (76.8)	920 (79.3)
Smoking during pregnancy
Yes	13 (2.0)	21 (1.8)	0.871
No	624 (98.0)	1139 (98.2)
Prenatal Vitamin Use
Yes	612 (96.1)	1129 (97.3)	0.041
No	25 (3.9)	31 (2.7)
Diet Qualityc
Poor quality diet	168 (26.4)	291 (25.1)	0.588
Good quality diet	469 (73.6)	869 (74.9)

^aChi-square test p-value.

^bOther/Unknown represents: Non-Hispanic Black, Native American/Alaskan Native, Multiple races, Other, Unknown or Not Reported.

^cDiet was assessed using the Healthy Eating Index (HEI-2010) score calculated from self-report dietary intake using a modified Block Food Frequency Questionnaire (FFQ). Poor quality diet was defined by scores that ranged in the lowest quartile (≤68.3). Good quality diet was defined by scores that ranged in the top three quartiles.

Characteristics of the Study Population

Descriptive characteristics of women with and without prenatal depression are presented in Table 3. Overall, 159 (14%) of the study participants had prenatal depression. Women with prenatal depression were less likely to have a college education, less likely to already have a child(ren), and more likely to report alcohol use or smoking during pregnancy compared to women without prenatal depression (all p<0.05).

Table 3.

Demographic Characteristics of 1160 Pregnant Women Overall and by Depression Status in a Study of Prenatal Depression and Diet Quality conducted in Kaiser Permanente Northern California between 2011-2013.

	Overall	Depression	No Depression
	n (%)	n (%)	n (%)	p-valuea
Total	1160	159 (13.7%)	1001 (86.3%)
Maternal age (at pregnancy start)
< 35 years age	704 (60.7)	89 (56.0)	615 (61.4)	0.374
35+ years of age	446 (38.4)	69 (43.4)	377 (37.7)
Missing	10 (0.9)	1 (0.6)	9 (0.9)
Marital status
Single/Divorced/Separated	40 (3.4)	10 (6.3)	30 (3.0)	0.099
Married/Living with partner	1119 (96.5)	149 (93.7)	970 (96.9)
Missing	1 (0.1)	0 (0.0)	1 (0.1)
Maternal Education
< College	61 (5.3)	15 (9.4)	46 (4.6)	0.037
College +	1098 (94.7)	144 (90.6)	954 (95.3)
Missing	1 (0.1)	0 (0.0)	1 (0.1)
Race/Ethnicity
Non-Hispanic White	631 (54.4)	89 (56.0)	542 (54.1)	0.030
Hispanic	213 (18.4)	30 (18.9)	183 (18.3)
Asian/Pacific Islander	195 (16.8)	16 (10.1)	179 (17.9)
Other/Unknownb	121 (10.4)	24 (15.1)	97 (9.7)
Parity
1+	732 (63.1)	85 (53.5)	647 (64.6)	0.009
0	428 (36.9)	74 (46.5)	354 (35.4)
Pre-pregnancy Body Mass Index (BMI)
Underweight (<18.5 kg/m2)	32 (2.8)	3 (1.9)	29 (2.9)	0.301
Normal weight (18.5 - < 25 kg/m2)	626 (54.0)	79 (49.7)	547 (54.6)
Overweight (25 - < 30 kg/m2)	270 (23.3)	37 (23.3)	233 (23.3)
Obese (≥30 kg/m2)	227 (19.6)	40 (25.2)	187 (18.7)
Missing	5 (0.4)	0 (0.0)	5 (0.5)
Alcohol use during pregnancy
Yes	240 (20.7)	43 (27.0)	197 (19.7)	0.043
No	920 (79.3)	116 (73.0)	804 (80.3)
Smoking during pregnancy
Yes	21 (1.8)	7 (4.4)	14 (1.4)	0.02
No	1139 (98.2)	152 (95.6)	987 (98.6)
Prenatal Vitamin Use
Yes	1129 (97.3)	154 (96.9)	975 (97.4)	0.894
No	31 (2.7)	5 (3.1)	26 (2.6)
Diet Qualityc
Poor quality diet	291 (25.1)	58 (36.5)	233 (23.3)	0.001
Good quality diet	869 (74.9)	101 (63.5)	768 (76.7)

^aChi-square test p-value.

^bOther/Unknown represents: Non-Hispanic Black, Native American/Alaskan Native, Multiple races, Other, Unknown or Not Reported.

^cDiet was assessed using the Healthy Eating Index (HEI-2010) score calculated from self-report dietary intake using a modified Block Food Frequency Questionnaire (FFQ). Poor quality diet was defined by scores that ranged in the lowest quartile (≤68.3). Good quality diet was defined by scores that ranged in the top three quartiles.

Healthy Eating Index-2010 Total Scores

The overall mean score for the HEI-2010 was 74.6 SD=10.0; the minimum score was 37.1 and the maximum score 95.8. The cutoff for the first quartile (poor diet quality) was 68.3. Several characteristics were significantly associated with poor diet quality (Table 4). Women with poor diet quality were more likely to be younger (<35 years), have less than a college education and to be obese compared to women with good diet quality.

Table 4.

Demographic Characteristics of 1160 Pregnant Women by Diet Quality in a Study of Prenatal Depression and Diet Quality Conducted in Kaiser Permanente Northern California between 2011-2013.

	Poor Diet Qualitya	Good Diet Qualitya
	n (%)	n (%)	p-valueb
	291	869
Maternal age (at pregnancy start)
< 35 years age	194 (66.7)	510 (58.7)	0.044
35+ years of age	94 (32.3)	352 (40.5)
Missing	3 (1.0)	7 (0.8)
Marital status
Single/Divorced/Separated	14 (4.8)	26 (3.0)	0.288
Married/Living with partner	277 (95.2)	842 (96.9)
Missing	0 (0.0)	1 (0.1)
Maternal Education
< College	25 (8.6)	36 (4.1)	0.003
College +	265 (91.1)	833 (95.9)
Missing	1 (0.3)	0 (0.0)
Race/Ethnicity
Non-Hispanic White	156 (53.6)	475 (54.7)	0.900
Hispanic	53 (18.2)	160 (18.4)
Asian/Pacific Islander	53 (18.2)	142 (16.3)
Other/Unknownc	29 (10.0)	92 (10.6)
Parity
1+	182 (62.5)	550 (63.3)	0.874
0	109 (37.5)	319 (36.7)
Pre-pregnancy BMI
Underweight (<18.5 kg/m2)	7 (2.4)	25 (2.9)	0.006
Normal weight (18.5 - < 25 kg/m2)	138 (47.4)	488 (56.2)
Overweight (25 - < 30 kg/m2)	69 (23.7)	201 (23.1)
Obese (≥30 kg/m2)	77 (26.5)	150 (17.3)
Missing	0 (0.0)	5 (0.6)
Alcohol use during pregnancy
Yes	59 (20.3)	181 (20.8)	0.906
No	232 (79.7)	688 (79.2)
Smoking during pregnancy
Yes	11 (3.8)	10 (1.2)	0.008
No	280 (96.2)	859 (98.8)
Prenatal Vitamin Use
Yes	277 (95.2)	852 (98.0)	0.016
No	14 (4.8)	17 (2.0)

^aDiet quality was assessed using the Healthy Eating Index (HEI-2010) score calculated from self-report dietary intake using a modified Block Food Frequency Questionnaire (FFQ). Poor quality diet was defined by scores that ranged in the lowest quartile (≤68.3). Good quality diet was defined by scores that ranged in the top three quartiles.

^bChi-square test p-value.

^cOther/Unknown represents: Non-Hispanic Black, Native American/Alaskan Native, Multiple races, Other, Unknown or Not Reported.

Prenatal Depression and Diet Quality

Overall, women with prenatal depression had nearly twice the odds of poor diet quality (crude odds ratio (cOR): 1.89; 95% CI:1.32, 2.69) (Table 5) compared to women without prenatal depression. This relationship held after adjusting for potential confounders (adjusted OR: 1.80; 95% CI: 1.23, 2.60).

Table 5.

Crude (cOR) and Adjusted Odds Ratios (aOR) for the Relationship between Prenatal Depression and Poor Diet Quality, Overall and by Race/Ethnicity in a Sample of 1160 Pregnant Women Recruited from Kaiser Permanente Northern California between 2011-2013.

	Poor Diet Qualitya
	n (%)	cORb	95% CI	aORc	95% CI
Full sample (n=1160)d
Depression (n=159)	58 (36.5)	1.89	(1.32, 2.69)	1.80	(1.23, 2.60)
No depression (n=1001)	233 (23.3)	1		1
Non-Hispanic White (n=631)e
Depression (n=89)	28 (31.5)	1.48	(0.90, 2.40)	1.29	(0.76, 2.15)
No depression (n=542)	128 (23.6)	1		1
Hispanic (n=213)e
Depression (n=30)	13 (43.3)	2.73	(1.21,6.09)	2.66	(1.15, 6.06)
No depression (n=183)	40 (21.9)	1		1
Asian/Pacific Islander (n=195)e
Depression (n=17)	7 (41.2)	2.25	(0.76, 6.38)	2.32	(0.77, 6.70)
No depression (n=179)	46 (25.7)			1

^aDiet was assessed using the Healthy Eating Index (HEI-2010) score calculated from self-report dietary intake using a modified Block Food Frequency Questionnaire (FFQ). Poor quality diet was defined by scores that ranged in the lowest quartile. Good quality diet was defined by scores that ranged in the top three quartiles.

^bcrude odds ratio (cOR)

^cadjusted odds ratio (aOR)

^dA total of 1143 women were included in the adjusted analysis due to missing data. The analyses were adjusted for maternal age, marital status, maternal education, parity, pre-pregnancy body mass index (BMI), alcohol use during pregnancy, smoking status during pregnancy, prenatal vitamin use during pregnancy, and race/ethnicity.

^eDue to some missing covariate data, the total sample size for the adjusted analyses for each of the race/ethnicity categories were as follows: n= 623 for Non-Hispanic White, n= 210 for Hispanic, and n=191 for Asian. The analyses were adjusted for maternal age, marital status, maternal education, parity, pre-pregnancy body mass index (BMI), alcohol use during pregnancy, smoking status during pregnancy, prenatal vitamin use during pregnancy and an interaction term for race/ethnicity and prenatal depression status.

For Hispanic women, prenatal depression was significantly associated with an increase in the odds of poor diet quality. After adjusting for potential confounding, Hispanic women with prenatal depression had over two and a half times the odds of poor diet quality compared to those without prenatal depression (aOR: 2.66, 95% CI: 1.15, 6.06). The p-value for the likelihood ratio test of the interaction was p=0.11.

Prenatal Depression and HEI-2010 Component Scores

The mean scores for each component are listed in Table 6 for the total sample. Women with prenatal depression had a significantly lower mean HEI-2010 component score for empty calories compared to women without depression indicating a higher consumption of foods with solid fats, alcohol and added sugars (Mean:14.38, SE: 3.95 vs. Mean: 15.27, SE: 3.71, p<0.05). Additionally, women with prenatal depression had a significantly lower mean HEI-2010 component score for greens and beans (Mean: 3.49, SE: 1.60 vs. Mean: 3.79, SE: 1.48, p<0.05), total fruit (Mean: 4.16, SE: 1.19 vs. Mean: 4.43, SE: 1.01, p<0.05), and whole fruit (Mean: 4.45, SE: 1.05 vs. Mean: 4.70, SE: 0.76, p<0.01), indicating a lower dietary intake compared to women without prenatal depression. After adjusting for potential confounders the relationship between prenatal depression and solid fats, alcohol and added sugars approached statistical significance (p= 0.06), while significant relationships remained with prenatal depression and greens and beans, total fruit and whole fruit. No other significant relationships emerged between prenatal depression and any of the other eight HEI-2010 component scores even after adjusting for potential confounders.

Table 6.

Mean Score and Standard Error (SE) for each of the 12 HEI-2010^a Components, Overall and by Depression Status and β coefficient for the Multivariable Linear Regression Model Assessing the Association between Prenatal Depression and Each of the 12 HEI-2010 Components in a Sample of 1160 Pregnant Women Recruited from Kaiser Permanente Northern California between 2011-2013.

	Overall		Depression		No Depression			Multivariable Modelc
	n=1160		n=159		n=1001			n=1160
HEI-2010 component	Mean	SE	Mean	SE	Mean	SE	p-valueb	β	p-value
Total vegetablesd	3.62	1.19	3.56	1.18	3.63	1.19	0.47	−0.06	0.55
Greens and beansd	3.74	1.50	3.49	1.60	3.79	1.48	0.03	−0.28	0.03
Total fruitd	4.39	1.04	4.16	1.19	4.43	1.01	<0.01	−0.24	<0.01
Whole fruitd	4.66	0.81	4.45	1.05	4.70	0.76	<0.01	−0.24	<0.01
Whole grainse	7.23	2.75	6.92	2.77	7.28	2.74	0.13	−0.40	0.09
Dairye	7.30	2.51	7.30	2.51	7.30	2.52	0.99	−0.11	0.62
Total protein foodsd	4.62	0.67	4.56	0.75	4.63	0.66	0.27	−0.08	0.20
Seafood and plant proteinsd	4.54	0.93	4.45	1.05	4.55	0.91	0.27	−0.08	0.32
Fatty acidse	6.34	2.77	6.15	2.82	6.38	2.77	0.35	−0.05	0.81
Sodiume	4.01	2.49	4.34	2.59	3.96	2.47	0.09	0.19	0.35
Refined grainse	9.01	1.78	8.82	2.07	9.04	1.73	0.20	−0.29	0.06
Empty caloriesf	15.15	3.75	14.38	3.95	15.27	3.71	0.01	−0.59	0.06

^aHealthy Eating Index- 2010.

^bT-test p-value.

^cA total of 1143 women were included in the adjusted analysis due to missing data. The analysis adjusted for maternal age, marital status, maternal education, parity, pre-pregnancy body mass index (BMI), alcohol use during pregnancy, smoking status during pregnancy, prenatal vitamin use during pregnancy, and race/ethnicity.

^dComponent score ranges from 0 to 5.

^eComponent score ranges from 0 to 10.

^fCalories from solid fats, alcohol and added sugar; threshold for counting alcohol is greater than 13 g/1000 kcal. Component score ranges from 0 to 20.

Sensitivity Analysis

Similar results emerged in the sensitivity analysis assessing the relationship between prenatal depression and diet quality restricting the sample to women with a PHQ-9 screening. 1,140 women received a PHQ-9 screening during routine prenatal care. Women with prenatal depression had greater odds of poor diet quality (cOR: 1.91; 95% CI: 1.31, 2.77) compared to women without prenatal depression. Although slightly attenuated, this relationship remained significant after adjusting for potential confounders (aOR: 1.76; 95% CI: 1.19, 2.59). Similar patterns emerged in the sensitivity analysis assessing the interaction between race/ethnicity and prenatal depression on diet quality when compared to the main analysis. Hispanic women with a positive PHQ-9 score (10 or greater) had a significant increased odds of poor diet quality compared to women without prenatal depression (aOR: 2.66, 95% CI: 1.12, 6.30) (Table 7).

Table 7.

Sensitivity Analysis - Crude (cOR) and Adjusted Odds Ratios (aOR) for the Relationship between Prenatal Depression and Poor Diet Quality, Overall and by Race/Ethnicity, Excluding Women who did not have a PHQ-9 Screening Score in a Sample of 1140 Pregnant Women Recruited from Kaiser Permanente Northern California between 2011-2013.

	cOR	95% CI	aOR	95% CI
Full sample (n=1140)a
Depression (n=139)	1.91	(1.31, 2.77)	1.76	(1.19, 2.59)
No depression (n=1001)	1		1
Non-Hispanic White (n=616)b
Depression (n=74)	1.55	(0.90, 2.60)	1.36	(0.77, 2.34)
No depression (n=542)	1		1
Hispanic (n=210)b
Depression (n=27)	2.86	(1.22, 6.60)	2.66	(1.12, 6.30)
No depression (n=183)	1		1
Asian/Pacific Islander (n=195)b
Depression (n=16)	2.25	(0.76, 6.38)	2.35	(0.78, 6.80)
No depression (n=179)	1		1

^aAdjusted for maternal age, marital status, maternal education, parity, pre-pregnancy body mass index (BMI), alcohol use during pregnancy, smoking status during pregnancy, prenatal vitamin use during pregnancy, and race/ethnicity

^bAdjusted for maternal age, marital status, maternal education, parity, pre-pregnancy body mass index (BMI), alcohol use during pregnancy, smoking during pregnancy, prenatal vitamin use during pregnancy and interaction term for race/ethnicity and prenatal depression status.

DISCUSSION/COMMENT

Findings from this study suggest women with prenatal depression or high prenatal depressive symptoms may have a higher risk of poor diet quality compared to women without prenatal depression, and that the relationship may be stronger in Hispanic women. The findings remained significant after restricting the sample to women with a PHQ-9 screening score, suggesting that depressive symptoms at or above a threshold of moderate depression may be as important as a clinical depression diagnosis in the relationship with diet quality. While an inverse association between depression and diet quality has been found previously in pregnant women, the limited research to date has been conducted in small samples (fewer than 120) of low-income women^15,42. This study adds to the literature findings from a large, racially/ethnically diverse sample of pregnant women.

Findings from this study are consistent with the majority of the limited studies in non-pregnant populations evaluating the relationship between depression and diet quality. In a study of breast cancer survivors, depressive symptoms and HEI scores were inversely related⁴⁶. Similarly, Appelhans et al⁴⁷ observed an association between higher depressive severity and lower diet quality in a sample of obese participants with major depressive disorder. An inverse association between depressive symptoms and diet quality was also noted in a representative sample of US adults⁴⁸. In a sample of overweight and obese rural-dwelling adults, depressive symptoms predicted poor overall diet quality among the overweight participants, however findings did not replicate in the obese participants⁴⁹. Three other studies documented mixed findings as well. For example, Rahe et al,⁴⁰ did not find a significant relationship between depression as a homogenous entity, but did find a relationship between distinct subtypes of depression such that participants with atypical depression (depression with improved mood in response to positive events) reported the lowest diet quality scores. Two additional studies, one in a sample of obese and overweight women from disadvantaged neighborhoods⁵⁰ and another in sample of middle-aged adults⁵¹ did not find a relationship between depressive symptoms and overall dietary quality. However, similar to the current study findings, the study of obese and overweight women from disadvantaged neighborhoods documented a significant relationship between depressive symptoms and greater saturated fats and total sugar intake⁵⁰.

The underlying mechanisms that link depression and eating behavior may include psychological as well as sensory and physiological pathways. For example, several studies have demonstrated an association between the presences of depressive symptoms and emotional eating ^50,52,53. Emotional eating is the propensity to choose to eat energy-dense sweet and high-fat foods in response to negative affects and stress⁵². Findings from the current study suggest that women with prenatal depression had a significantly lower HEI-2010 empty calories component score compared to women without prenatal depression. The HEI-2010 empty calorie component score is inversely related to dietary intake of empty calories; thus, in this study women with prenatal depression consumed a diet higher in solid fats, alcohol and added sugars (threshold for counting alcohol is greater than 13g/1,000 kcal) compared to women without prenatal depression. These findings suggest that women with prenatal depression may benefit from nutritional counseling targeted at limiting dietary intake from empty calories from solid fats, alcohol and added sugars. This study also found that women with prenatal depression consumed less overall greens and beans, total fruit and whole fruit compared to women without prenatal depression suggesting that nutritional counseling for women with prenatal depression may also target increasing consumption of fruit, greens and beans. Given that poor maternal diet quality during pregnancy has been linked to poor outcomes of both the offspring into adulthood^1-3 and to the woman’s health^5-8,54-57, further research is needed to understand the extent to which depression may contribute to these outcomes by influencing diet quality. This study is the first to our knowledge to assess the relationship between prenatal depression and individual HEI components.

Strengths and limitations

Dietary intake was assessed using a validated instrument which allowed us to construct a reliable and internally valid index of diet quality (as assessed with the HEI-2010) that represents a comprehensive assessment of a woman’s diet during pregnancy. Further, the HEI-2010 was designed to reflect the Dietary Guidelines for Americans⁹ and the use of this a priori diet score would allow comparability across studies in the United States, in contrast to the use of other data-driven study specific dietary scores. Additionally, prenatal depression was defined using a combination of clinical diagnoses, antidepressant medications and PHQ-9 scores all ascertained from EHR data. The sensitivity analyses restricting the sample to women uniformly screened for depression with the PHQ-9 during the study period demonstrates that symptom severity may be as important as a clinical diagnosis of prenatal depression in the relationship with diet quality. Finally, the authors acknowledge that the study was limited to English-speaking women and thus the associations observed among the Hispanic women may not be representative of non-English speaking Hispanic women.

This study offers some unique insights into the relationship between depression including depressive symptoms and diet quality in pregnant women; however, the findings and implications of this study should be interpreted with consideration of the limitations. Maternal dietary intake was self-reported using the FFQ, with potential recall bias and subsequent exposure misclassification⁵⁸. There is no evidence that women with prenatal depression may recall their dietary intake differentially from women without prenatal depression. The mean HEI-2010 total score in the current study sample (74.6) was a little higher in comparison to some previous studies of pregnant women which have ranged from 54.2 – 64^12,59, yet similar to others (mean HEI-2010 score of 67⁶⁰) including a study conducted in pregnant women in Northern California (mean HEI-2010 score of 71.3²⁸). Causality cannot be inferred given the cross-sectional study design. Despite defining prenatal depression prior to the completion of the FFQ, the FFQ asks for dietary intake in the previous three months and thus there may be overlap of the timing of prenatal depression and dietary intake, thus the ability to determine the direction of the relationship is limited. It is plausible that the relationship between depression and diet is bidirectional. Observational studies have found that adherence to higher quality diets is associated with reduced future onset of depressive symptoms⁶¹. Additionally, recent clinical trials^62,63 suggest dietary interventions may have a modest impact on reducing depressive symptoms. It is possible that depressive symptoms may be related to dietary misreporting, although research assessing this is insufficient⁶⁴. The use of a validated and reliable instrument to assess dietary intake may diminish this possibility, however we note that previous validation studies in women did not report information on inclusion or exclusion of pregnant women with depression^32,33. The study sample excluded over a quarter of the participants recruited into the PEAPOD study because they did not have any evidence of being assessed for depression. There were very few significant differences between the final study sample and those who were excluded. Finally, the relationship between prenatal depression and diet quality among non-Hispanic Black women was not able to be assessed given the sample size.

Conclusions

Findings from this study suggest that women with prenatal depression and or prenatal depressive symptoms are at a higher risk of poor diet quality compared to women without prenatal depression, and the relationship is stronger among Hispanic women. Nutritional counseling interventions for women with depression may consider the use of culturally sensitive material, target limiting empty calories from solid fats, alcohol and added sugars and encourage eating more greens, beans and fruit. Additional prospective studies are needed to clarify the direction of the relationship.

Research Snapshot.

Research Question:

Is prenatal depression associated with diet quality during pregnancy as measured with the Healthy Eating Index-2010 (HEI-2010) and does this relationship differ by race/ethnicity? Are there differences in the 12 HEI-2010 dietary components for women with prenatal depression compared to women without prenatal depression?

Key Findings:

In this cross-sectional examination of 1160 pregnant participants in the PEAPOD study, findings suggest that women with prenatal depression (defined as either a depression diagnosis, positive screen on the Patient Health Questionnaire (PHQ-9), or antidepressant medication dispensing) may have a higher risk of poor diet quality (defined as falling in the lowest quartile of the HEI-2010 scores) compared to women without prenatal depression, and that the relationship may be stronger in Hispanic women. Women with prenatal depression had a significantly lower mean HEI-2010 component score for empty calories compared to women without depression indicating a higher percent energy intake from solid fats and added sugars. Additionally, women with prenatal depression had a significantly lower mean HEI-2010 component score for greens and beans, total fruit and whole fruit indicating a lower dietary intake per the density-based approach used by the HEI-2010 compared to women without prenatal depression.