Pregnancy-Related Weight and Postpartum Depressive Symptoms: Do the Relationships Differ by Race/Ethnicity?

Abstract

Background:

There are significant racial/ethnic disparities in the prevalence of postpartum depression. Prior research in the general population suggests that weight status is related to depression and that this relationship varies by race/ethnicity. However, few studies have investigated whether race/ethnicity moderates the relationship between pregnancy-related weight and postpartum depressive symptoms (PPDS). The objective of this study is to examine the relationship between pregnancy-related weight and maternal PPDS overall and by race/ethnicity.

Materials and Methods:

This study used data from the Early Childhood Longitudinal Study-Birth Cohort (n ≈ 6950). Ordinary least-squares and logistic regression was used to examine whether pregnancy-related weight, including preconception weight status and gestational weight gain (GWG), was associated with PPDS measured using the Center for Epidemiologic Studies-Depression Scale (CES-D). Stratified analyses were used to assess whether these relationships varied by race/ethnicity.

Results:

Preconception obesity (body mass index [BMI] ≥30 kg/m²) was associated with higher levels of PPDS (β = 0.601, 95% confidence interval [CI], 0.149–1.053). GWG adequacy was not associated with PPDS. Among non-Hispanic (NH) whites, preconception obesity was positively associated with PPDS (β = 1.016, 95% CI, 0.448–1.584). In contrast, among Hispanics, preconception overweight (25 kg/m² ≤ BMI <30 kg/m²) was associated with lower levels of PPDS (β = −0.887, 95% CI, −1.580 to −0.195). There were no statistically significant relationships between pregnancy-related weight and PPDS among NH black or Asian women, but both groups were significantly more likely than NH whites to report PPDS.

Conclusion:

Whether and how pregnancy-related weight is associated with PPDS varies by race/ethnicity. Addressing preconception weight could help reduce overall levels of PPDS among NH whites but would likely fail to mitigate racial/ethnic disparities in postpartum mental health.

Introduction

Postpartum depression (PPD) is defined as any major depressive disorder that occurs within 30 days after giving birth.¹ PPD has important short- and long-term effects on maternal physical and mental health, maternal/infant bonding, and children's cognitive well-being.^2,3 It can be challenging to generate accurate population-level estimates of PPD despite the existence of well-validated screening tools (e.g., the Edinburgh Postnatal Depression Tool).⁴ This is largely due to the lack of universal screening for perinatal depression in obstetric care, disparities in access to care (necessary to receive a confirmed diagnosis), and societal stigma surrounding mental health disorders.⁵ However, the most recent multistate survey data from the Pregnancy Risk Assessment Monitoring System suggest that approximately 12.5% of U.S. women experience postpartum depressive symptoms (PPDS).⁶ Furthermore, prior studies have found that non-Hispanic (NH) black, Hispanic, and NH Asian women are often more likely than NH whites to exhibit PPDS,^7–12 although there are exceptions to this overall pattern.^13,14 Identifying the key determinants of PPDS is critical to both achieving the Healthy People 2020 goal to reduce the incidence of PPD and narrowing racial/ethnic disparities in maternal mental health.¹⁵

In this study, we focus on one potential determinant of PPDS: body weight. Prior research has shown that obesity (body mass index [BMI] ≥30 kg/m²) is strongly related to depression in the general population, particularly among women.^16–18 There are likely both psychological and biological mechanisms linking obesity to subsequent depression. First, individuals with bodies that do not conform to U.S. societal ideals (i.e., those who are not thin and those who are overweight or obese) may be more likely to experience body dissatisfaction and low self-esteem, both of which are precursors of depression.^19–21

Second, obesity might lead to dysregulation of the hypothalamic/pituitary/adrenal axis (HPA axis), increased insulin resistance, and/or inflammation, any of which can lead to a greater risk of depression.^22–24 One meta-analysis found that individuals classified as obese are significantly more likely to develop depression over time (odds ratio [OR], 1.55, 95% confidence interval [CI], 1.22–1.98).²³ Evidence also suggests that individuals classified as underweight (BMI <18.5 kg/m²) have among the highest rates of depression, while rates are lowest among individuals classified as normal weight (18.5 kg/m²≤BMI <25 kg/m²) or overweight (25 kg/m²≤BMI <30 kg/m²).²⁵ The reverse may also be true. That is, depression might drive subsequent obesity or underweight, through changes in food intake and/or physical activity.²⁵

The relationship between body weight and mental health outcomes may vary significantly by race/ethnicity, although the evidence is somewhat mixed. Several prior studies have found that the obesity/depression relationship is strong among white women but relatively weaker or nonexistent among NH black and Hispanic women.^26–28 In contrast, other research has found no evidence of racial/ethnic differences in the obesity/depression link.^29,30

Prior research among childbearing women has found that mothers classified as underweight, overweight, or obese before conception are more likely to experience PPDS and/or PPD relative to those classified as normal weight. However, a thorough review of the literature indicated that to our knowledge, no extant studies have investigated racial/ethnic differences in the relationships between pregnancy-related weight and PPDS. This is an important oversight, given that the significant racial/ethnic disparities in pregnancy-related weight could contribute to racial/ethnic disparities in PPDS. For example, relative to NH whites, NH Asians are more likely to be classified as underweight before pregnancy (8.6% vs. 3.8%) and NH blacks exhibit relatively higher rates of preconception obesity (34.8% vs. 24.7%).³¹ However, most existing research on this topic has used study samples that are largely NH white and/or from a single city or state, potentially limiting generalizability.^32–41 Furthermore, very few studies have examined the association between rapid changes in body weight during pregnancy (i.e., gestational weight gain [GWG]) and postpartum mental health.^37,42

In the present study, we address these gaps in the literature by using a diverse nationally representative data source to examine the relationships between pregnancy-related weight, including preconception weight status and GWG, and PPDS. We focus on whether these relationships vary by race/ethnicity among NH white, NH black, Hispanic, and NH Asian mothers.

Materials and Methods

Data

This study used data from the Early Childhood Longitudinal Study-Birth Cohort (ECLS-B), a longitudinal, nationally representative cohort study of U.S. infants born in 2001.⁴³ The survey, conducted by the National Center on Educational Statistics (NCES) gathered data on children, with interviewers surveying families multiple times, starting when the children were 9 months old and continuing until they entered kindergarten. Asian and Native American infants, multiple births, and infants born at low and very low birth weights were oversampled. Interviewers surveyed mothers about preconception weight, GWG, and PPDS when the children were ∼9 months old. To our knowledge, the ECLS-B is the only nationally representative survey data that allow an investigation of the relationships between pregnancy-related weight and PPDS.

Sample derivation

The original eligible sample consisted of approximately 8800 biological mothers who gave birth to a single child in 2001. (Note: NCES reporting guidelines require us to round all sample sizes to the nearest 50 to protect the privacy of study participants.) We restricted the eligible sample to singleton births because the Institute of Medicine (IOM) has not provided provisional GWG guidelines for multiple pregnancies.⁴⁴ We excluded mothers with missing or implausible information (e.g., BMI <15 kg/m²) on preconception weight or height (n ≈ 250) and GWG (n ≈ 700) as well as mothers with missing information on PPDS (n ≈ 800). Finally, we excluded mothers with any missing information on other important covariates, such as race/ethnicity, cesarean section delivery, parity, and maternal smoking (n ≈ 100). The final sample included approximately 6950 women.

Measures

The main outcome of interest was PPDS among mothers when their children were approximately 9 months old. (The ages of the focal children ranged from 6 to 22 months.) During the 9-month survey (Wave 1), interviewers asked mothers a series of questions about depressive symptoms using a 12-question abbreviated version of the Center for Epidemiologic Studies-Depression Scale (CES-D).⁴⁵ The original CES-D is a 20-item measure that asks respondents to rate how often over the past week they have experienced symptoms associated with depression, such as restless sleep, poor appetite, and feeling lonely.⁴⁶ While neither the abbreviated nor the original versions of the CES-D can be used as a diagnostic tool, they are frequently used as a screening device for depression.^47,48 Furthermore, the CES-D has been validated among perinatal populations both within and outside of the United States.^49–52 Mothers rated the frequency of each symptom on a scale from 0 (rarely or never) to 4 (most or all days of the week). We summed respondents' scores across the 12 items and created two measures of PPDS. First, we developed a continuous measure with total scores ranging from 0 to 36. Second, we created a categorical measure of moderate/severe PPDS, where the outcome = 1 if a mother had moderate (CES-D score 10–14) or severe depressive symptoms (CES-D score greater than 14) and = 0 otherwise. Prior studies have used this categorization to identify moderate/severe PPDS.^47,48,53

The main independent variables were measures of pregnancy-related weight, including preconception weight status and GWG. Using self-reported weight and height, we categorized a mother's preconception weight status as underweight (BMI <18.5 kg/m²), normal weight (18.5 ≤ BMI <25 kg/m²), overweight (25 ≤ BMI <30 kg/m²), or obese (BMI ≥30 kg/m²). We calculated GWG adequacy based on the 2009 IOM guidelines for GWG, which vary by preconception weight status.⁴⁴ Recommended weight gain ranges are highest for women classified as underweight (28–40 pounds) and normal weight (25–35 pounds), and lowest for women classified as overweight (15–25 pounds) and obese (15–20 pounds). To account for differences in pregnancy duration (e.g., preterm birth) that could have affected total GWG, we calculated the ratio of observed weight gain to expected weight gain (i.e., observed GWG/expected GWG).^54–56 If this ratio falls below 85%, GWG is classified as inadequate. Similarly, if the ratio is above 122%, GWG is classified as excessive. Finally, if the ratio is between 85% and 122%, GWG is classified as normal.

We also included covariates that prior research has shown are associated with PPDS.^33,35,36,53 These included indicators of maternal race/ethnicity, including NH white, NH black, Hispanic, NH Asian, and NH other race. (Throughout the remainder of the article, “white” refers to NH white, “black” refers to NH black, “Asian” refers to NH Asian, and “other race” refers to NH other race. Other race is a heterogeneous group that includes Native Hawaiian, American Indian, and multiracial individuals.) We also included measures of maternal nativity (i.e., foreign born), maternal age at conception (15–19, 20–29, 30–34, and 35 years or older), whether the mother delivered by cesarean section, number of prior live births (0 vs. 1 or more), postpartum smoking status (smoke vs. does not smoke), marital status (married vs. not), education (less than high school, high school, some college, college graduate), poverty status (less than 100% of the federal poverty level [FPL], 100%–129% FPL, 130%–185% FPL, greater than 185% FPL), number of household members, employment status (employed vs. unemployed), and type of residence (urban vs. rural).

Methods

We first compared depressive symptoms (continuous CES-D score and moderate/severe PPDS, CES-D ≥ 10) and other characteristics by preconception weight status using t-tests and χ² tests for continuous and categorical variables, respectively. We then conducted multivariate regression analyses, using ordinary least squares regression for the continuous outcome (total CES-D score) and logistic regression for the binary outcome (being moderately or severely depressed). The results of logistic regressions were expressed as ORs. In regression analyses, we first estimated the associations between preconception weight status, GWG, and PPDS, controlling for all relevant covariates. Subsequently, we estimated separate regression models for each racial/ethnic subgroup (white, black, Hispanic, Asian, and other race). For each set of models, we used appropriate sampling weights to account for the sampling structure of the ECLS-B. In addition, we calculated robust standard errors to address any nonconstant variance in the error terms. Finally, we used STATA 12 (StataCorp, College Station, TX) to perform all statistical analyses. This study was deemed exempt from review by our university's institutional review board.

Results

Table 1 shows that the mean CES-D score was 4.91 and the prevalence of moderate/severe PPDS (CES-D ≥ 10) was 16.10%. When we compare PPDS across preconception weight categories, we show that both the mean CES-D score (5.68 vs. 4.72, p < 0.01) and the prevalence of moderate/severe PPDS (19.70% vs. 15.20%, p < 0.01) were higher among mothers with preconception obesity than among those who were normal weight. Mothers who were classified as underweight before pregnancy also had higher CES-D scores (5.83) and were more likely to report moderate/severe depressive symptoms (20.20%) than normal weight mothers. Similarly, Table 2 shows that mothers who experienced either excessive or inadequate GWG had higher mean CES-D scores (5.35 and 5.40, respectively) and higher rates of moderate/severe depressive symptoms (19.06% and 18.44%, respectively) than mothers with adequate GWG (4.81 and 15.79%, respectively).

Table 1.

Summary Statistics, Full Sample, Stratified by Prepregnancy Weight Status Early Childhood Longitudinal Study, Birth Cohort (n ≈ 6950)

	Full sample	Normal weight	Maternal prepregnancy weight status		Obese
			Underweight	Overweight
Observations	6950	3950	450	1550	1000
Depressive symptoms
CES-D score (mean)	4.91 (0.10)	4.72 (0.11)	5.83*** (0.37)	4.71 (0.16)	5.68*** (0.23)
Moderate/severe PPDS (CES-D ≥ 10)	16.10	15.20	20.20***	15.00	19.70***
Gestational weight gain adequacy
Inadequate	26.40	21.10	32.10***	12.60***	18.80***
Adequate	19.30	32.30	34.70	21.10***	10.50***
Excessive	54.30	46.70	33.20***	66.20***	70.70***
Race/ethnicity
NH white	60.30	62.50	56.20	59.30	54.80***
NH black	13.90	12.40	13.10	13.60	21.00***
Hispanic	19.90	18.60	20.20	22.40***	20.09
NH Asian	3.00	4.00	6.70	1.70***	0.40***
NH other	2.80	2.60	3.90***	3.00	3.00
Maternal nativity
Foreign born	18.40	19.40	22.10	19.10	12.00
Maternal age (in years)
15 to 19 years	7.40	8.70	16.00***	4.80***	3.90***
20 to 29 years	50.80	50.10	57.80***	49.70	52.90
30 to 34 years	24.90	24.60	17.00***	26.20	27.00
35 years or more	16.80	16.60	9.20***	19.40***	16.20
Mode of delivery
C-section birth	15.10	13.90	13.10	14.50	21.20***
Number of prior live births
Zero	41.10	43.50	54.50***	36.60***	33.90***
Maternal smoking 9 months postpartum	20.00	19.40	30.00***	18.10	22.10
Maternal marital status
Married	68.00	69.10	59.20***	70.30	63.20***
Maternal education
Less than high school	17.60	16.30	22.70***	17.90	20.10**
High school	31.40	30.10	39.60***	30.10	33.90**
Some college	25.50	23.80	22.10	26.60***	31.80***
College graduate	25.50	29.80	15.60***	24.50***	14.30***
Household poverty level
<100% of FPL	21.99	19.80	26.50***	22.20***	28.80***
100%–129% FPL	11.30	11.00	12.90	11.10	12.10
130%–185% FPL	12.60	10.40	17.50***	14.70***	15.70***
More than 185% FPL	54.10	58.80	43.10***	52.10***	43.30
Number of household members (mean)	4.24 (0.02)	4.16*** (0.02)	4.22*** (0.09)	4.31*** (0.04)	4.50*** (0.07)
Maternal employment after pregnancy	51.77	51.50	44.60***	53.40	52.80
Rural residence	15.30	14.80	17.40	15.70	15.60

All variables expressed in percentages or means (standard deviations); summary statistics weighted to account for complex sampling structure.

χ² and t-tests separately compare individuals classified as underweight, overweight, and obese to normal weight (reference group).

^** p < 0.05.

^*** p < 0.01.

CES-D, Center for Epidemiologic Studies-Depression Scale; FPL, federal poverty level; PPDS, postpartum depressive symptoms.

Table 2.

Postpartum Depressive Symptoms, Stratified by Gestational Weight Gain Adequacy Early Childhood Longitudinal Study, Birth Cohort (n ≈ 6950)

	Adequate	Maternal gestational weight gain adequacy	Excess
		Inadequate
Observations	2100	1900	3000
Depressive symptoms
CES-D score (mean)	4.81 (5.06)	5.40 (5.87)***	5.35 (5.63)***
Moderate or severe PPDS (CES-D score ≥10)	15.79 (0.37)	18.44 (0.39)***	19.06 (0.39)***

All variables expressed in percentages or means (standard deviations); summary statistics weighted to account for complex sampling structure.

χ² and t-tests separately compare individuals who gained inadequate or excess weight with those with adequate weight gain (reference group).

Combined sample size across categories ≈7000 due to rounding.

^*** p < 0.01.

Figures 1 and 2 display variation in the link between pregnancy-related weight and depressive symptoms by race/ethnicity. As seen in Figure 1, the relationship between preconception weight status and moderate/severe depressive symptoms was generally U-shaped for white, black, and Asian mothers. Specifically, moderate/severe PPDS was highest among mothers classified as obese or underweight before conception and lowest among mothers classified as normal weight or overweight. The results for Hispanics, however, deviate from this pattern. For this group, moderate/severe depressive symptoms were highest among those classified as normal weight or obese before pregnancy and lowest among those classified as underweight or overweight. Figure 2 shows that Hispanic mothers also follow a different pattern than other racial/ethnic groups with respect to GWG. Among Hispanics, mothers with adequate GWG exhibited the highest rates of moderate/severe depressive symptoms, followed by those with excessive GWG, while those with inadequate GWG had the lowest rates of moderate/severe depressive symptoms. In contrast, for whites, blacks, and Asians, mothers with excessive and inadequate GWG had higher levels of depressive symptoms than those with adequate GWG.

FIG. 1.

Moderate/severe depressive symptoms by preconception weight status and race/ethnicity (n ≈ 6950). Tables based on data from the ECLS-B. Figure includes only major racial/ethnic groups (NH white, NH Asian, Hispanic, NH Asian); NH other race mothers are included in regression models. ECLS-B, Early Childhood Longitudinal Study-Birth Cohort; GWG, gestational weight gain; NH, non-Hispanic.

FIG. 2.

Moderate/severe depressive symptoms by gestational weight gain adequacy and race/ethnicity (n ≈ 6950). Tables based on data from the ECLS-B. Figure includes only major racial/ethnic groups (NH white, NH Asian, Hispanic, NH Asian); NH other race mothers are included in regression models.

Table 3 displays the results of regression analyses examining the relationships between pregnancy-related weight and PPDS for the full sample. All of the regressions that follow control for the full set of covariates discussed previously (e.g., maternal age and educational attainment). We focus primarily on presenting the findings from regression analyses using continuous PPDS as the outcome measure. However, the full results using the binary moderate/severe PPDS are available upon request. Preconception obesity was associated with a 0.601-unit increase in the CES-D score (95% CI, 0.149–1.053) relative to normal weight before pregnancy. However, no other weight variables (preconception overweight, preconception underweight, GWG adequacy) were significantly associated with maternal CES-D score. Blacks and Asians were significantly more likely to have higher CES-D scores (β = 0.715, 95% CI, 0.140–1.290 and β = 1.211, 95% CI, 0.508–1.915, respectively) relative to whites. In contrast, being Hispanic was associated with a lower CES-D score (β = −0.612, 95% CI, −1.126 to −0.097). We observed similar patterns with respect to moderate/severe depressive symptoms (i.e., CES-D score ≥10), but note that preconception obesity was not significantly associated with this outcome at conventional levels.

Table 3.

Ordinary Least Squares Regression, Associations Between Pregnancy-Related Weight and Center for Epidemiologic Depression Scale Score, Overall and Stratified by Race/Ethnicity, Early Childhood Longitudinal Study, Birth Cohort

	Full sample	Maternal race/ethnicity
		NH white	NH black	Hispanic	NH Asian	NH other race
Observations	6950	3150	1150	1150	950	550
	Coeff. (95% CI)	Coeff. (95% CI)	Coeff. (95% CI)	Coeff. (95% CI)	Coeff. (95% CI)	Coeff. (95% CI)
Prepregnancy weight status
Normal weight (reference)
Underweight	0.532 (−0.189 to 1.254)	0.807 (−0.370 to 1.984)	1.223 (−0.597 to 3.042)	−0.239 (−1.853 to 1.375)	0.363 (−0.743 to 1.469)	−0.915 (−3.513 to 1.683)
Overweight	−0.024 (−0.407 to 0.360)	0.512 (−0.027 to 1.050)	−0.778 (−1.744 to 0.188)	−0.887** (−1.580 to −0.195)	−0.416 (−1.375 to 0.543)	−0.645 (−1.820 to 0.530)
Obese	0.601*** (0.149 to 1.053)	1.016*** (0.448 to 1.584)	0.142 (−1.038 to 1.322)	−0.330 (−1.412 to 0.753)	1.655 (−0.727 to 4.037)	0.966 (−0.703 to 2.635)
Gestational weight gain adequacy
Adequate (reference)
Inadequate	0.115 (−0.355 to 0.585)	0.055 (−0.600 to 0.711)	0.555 (−0.727 to 1.838)	−0.425 (−1.505 to 0.656)	−0.302 (−1.346 to 0.742)	1.737 (−0.453 to 3.927)
Excess	0.200 (−0.114 to 0.514)	0.250 (−0.188 to 0.687)	0.494 (−0.523 to 1.511)	−0.062 (−0.707 to 0.582)	−0.148 (−1.054 to 0.758)	−0.038 (−1.512 to 1.436)
Race/ethnicity
NH white (reference)
NH black	0.715** (0.140 to 1.290)
Hispanic	−0.612** (−1.126 to −0.097)
NH Asian	1.211*** (0.508 to 1.915)
NH other	0.378 (−0.305 to 1.060)
Maternal nativity
U.S.-born (reference)
Foreign-born	−0.748*** (−1.312 to −0.184)	−0.179 (−1.232 to 0.873)	−0.868 (−2.111 to 0.375)	−1.180*** (−2.032 to −0.328)	1.229*** (0.328 to 2.129)	−0.418 (−2.861 to 2.026)
Maternal age (in years)
15 to 19 years (reference)
20 to 29 years	−0.585 (−1.332 to 0.162)	−0.967 (−2.335 to 0.401)	0.018 (−1.446 to 1.482)	−0.225 (−1.175 to 0.725)	0.719 (−1.363 to 2.800)	−0.269 (−4.058 to 3.521)
30 to 34 years	−0.774 (−1.564 to 0.015)	−1.029 (−2.473 to 0.414)	−0.484 (−1.989 to 1.022)	−0.217 (−1.555 to 1.120)	−0.126 (−2.348 to 2.097)	−1.114 (−5.087 to 2.858)
35 years or more	−0.565 (−1.383 to 0.253)	−0.714 (−2.254 to 0.826)	−0.092 (−1.720 to 1.536)	−0.656 (−2.185 to 0.873)	0.196 (−2.191 to 2.583)	−0.409 (−4.473 to 3.655)
Mode of delivery
Vaginal birth (reference)
C-section birth	−0.421 (−0.848 to 0.006)	−0.621** (−1.191 to −0.051)	−0.523 (−1.614 to 0.569)	0.063 (−0.950 to 1.076)	−0.462 (−1.470 to 0.546)	0.636 (−1.300 to 2.573)
Number of prior live births
Zero (reference)
One or more	0.035 (−0.321 to 0.390)	0.170 (−0.369 to 0.709)	−0.136 (−1.086 to 0.813)	0.044 (−0.644 to 0.733)	−0.106 (−0.855 to 0.643)	0.607 (−1.330 to 2.544)
Maternal smoking 9 months postpartum
Did not smoke (reference)
Smoked	1.184*** (0.683 to 1.684)	1.061*** (0.476 to 1.645)	1.163** (0.038 to 2.288)	1.179 (−0.236 to 2.594)	1.586 (−0.115 to 3.286)	1.335 (−0.596 to 3.265)
Maternal marital status
Unmarried (reference)
Married	−0.797*** (−1.218 to −0.376)	−1.224*** (−1.785 to −0.663)	−0.156 (−1.172 to 0.861)	−0.684 (−1.390 to 0.021)	−1.936*** (−3.376 to −0.497)	1.148 (−0.596 to 2.892)
Maternal education
Less than high school (reference)
High school	0.043 (−0.553 to 0.638)	0.548 (−0.396 to 1.492)	−0.731 (−2.083 to 0.621)	−0.188 (−0.816 to 0.440)	0.730 (−0.629 to 2.089)	−1.761 (−4.315 to 0.792)
Some college	−0.341 (−0.958 to 0.276)	−0.039 (−0.982 to 0.904)	−1.258 (−2.529 to 0.013)	0.123 (−0.951 to 1.197)	1.587** (0.043 to 3.130)	−2.444 (−5.122 to 0.233)
College graduate	−0.989*** (−1.627 to −0.350)	−0.532 (−1.446 to 0.381)	−1.836** (−3.661 to −0.011)	−1.1 (−2.415 to 0.215)	0.958 (−0.422 to 2.339)	−3.529** (−6.300 to −0.757)
Household poverty level
<100% of FPL (reference)
100%–129% FPL	−0.952*** (−1.590 to −0.314)	−1.720*** (−2.635 to −0.805)	−0.945 (−2.197 to 0.308)	0.021 (−1.019 to 1.062)	2.461*** (0.609 to 4.313)	−1.824 (−4.214 to 0.566)
130%–185% FPL	−0.716** (−1.305 to −0.127)	−0.974 (−2.069 to 0.121)	−0.354 (−1.685 to 0.976)	−1.039** (−1.825 to −0.253)	2.366*** (0.757 to 3.974)	−0.423 (−2.888 to 2.043)
>185% FPL	−1.374*** (−1.961 to −0.786)	−1.879*** (−2.802 to −0.957)	−0.929 (−2.098 to 0.239)	−0.917** (−1.773 to −0.062)	0.806 (−0.613 to 2.226)	−1.543 (−4.128 to 1.042)
Number of household members	−0.138** (−0.268 to −0.008)	−0.252** (−0.464 to −0.040)	−0.090 (−0.415 to 0.234)	0.013 (−0.207 to 0.232)	0.190 (−0.094 to 0.474)	−0.505 (−1.116 to 0.106)
Maternal employment after pregnancy
Unemployed (reference)
Employed	−0.461*** (−0.748 to −0.175)	−0.442** (−0.880 to −0.004)	−1.340*** (−2.288 to −0.392)	0.177 (−0.561 to 0.914)	−0.713** (−1.365 to −0.060)	−1.686*** (−2.894 to −0.478)
County of residence at birth
Not rural (reference)
Rural residence	0.111 (−0.327 to 0.550)	0.102 (−0.462 to 0.667)	0.017 (−0.825 to 0.858)	0.564 (−1.240 to 2.367)	1.351 (−0.767 to 3.468)	−0.181 (−1.903 to 1.541)
Constant	7.795*** (6.591 to 8.998)	8.669*** (6.676 to 10.661)	8.713*** (6.409 to 11.016)	6.405*** (4.464 to 8.345)	2.713** (0.433 to 4.993)	10.357*** (6.009 to 14.706)

^** p < 0.05.

^*** p < 0.01.

All analyses weighted to account for complex sampling structure

CI, confidence interval; GWG, gestational weight gain; NH, non-Hispanic.

Table 3 displays the results of regression models examining the associations between pregnancy-related weight and CES-D score; each model was stratified by race/ethnicity. (Note: Due to the heterogeneity of the other race category, we felt it was unwarranted to make generalizations about this group in our analyses. Therefore, we include “other race” women in the main analyses, but do not present race-stratified results for this group.) Among whites, preconception obesity (compared with normal weight) was significantly associated with an increase in maternal CES-D score (β = 1.016, 95% CI, 0.448–1.584). Among Hispanics, however, preconception overweight (compared with normal weight) was associated with a decrease in maternal CES-D score (β = −0.887, 95% CI, −1.580 to −0.195). Finally, for the black, Asian, and other race subsamples, there were no statistically significant associations between pregnancy-related weight and depressive symptoms. We found very similar results when using moderate/severe PPDS as the outcome.

Discussion

The goal of the present study was to investigate the links between pregnancy-related weight, postpartum mental health, and maternal race/ethnicity. The findings are threefold: First, the results show that overall, preconception obesity but not preconception overweight/underweight or GWG adequacy was significantly associated with higher levels of PPDS relative to normal weight. Second, the nature of the association between preconception weight status and PPDS differed by race/ethnicity. For white mothers, preconception obesity was significantly associated with higher levels of PPDS. In contrast, among Hispanic mothers, being classified as overweight before pregnancy was associated with lower levels of PPDS. For black and Asian women, pregnancy-related weight was not significantly associated with PPDS. Third, there were systematic differences in PPDS by race/ethnicity. Specifically, black and Asian women were significantly more likely to report PPDS than whites, while Hispanic women were less likely than whites to do so. This study is among the first and is the largest (to our knowledge) to investigate racial/ethnic differences in the associations between pregnancy-related weight and PPDS.

The finding that preconception obesity but not GWG adequacy was related to elevated PPDS is consistent with evidence from prior cross-sectional and longitudinal studies.^32,36,41 Although the current data do not allow a direct examination of underlying mechanisms, we propose that body dissatisfaction is one potential factor that might link obesity to subsequent depression. Obesity is linked to body dissatisfaction, and prior research provides strong evidence that body dissatisfaction is related to the onset of depression during and after pregnancy.^42,57,58 Body dissatisfaction tends to be highest at the beginning of pregnancy, decreases during the second and third trimesters, and increases again during the postpartum period,^59–62 which might explain why there were strong relationships between preconception obesity and PPD, but no links between GWG adequacy and PPDS.⁴¹

Two additional mechanisms that might link preconception weight and PPDS are dysregulation of the HPA axis and insulin resistance.²³ Both underweight and obesity have been linked to the HPA axis dysregulation among the general and pregnant populations,^63,64 although there is conflicting evidence about whether these weight extremes lead to under-regulation or hyper-regulation (e.g., decreased vs. increased cortisol levels). HPA axis dysregulation can, in turn, trigger a cascade of systemic responses (e.g., inflammation) that may lead to fatigue, poor sleep, and depression.^22,24,65 Furthermore, individuals with obesity have an increased risk of developing type 2 diabetes and insulin resistance. Insulin resistance, in turn, may drive brain alterations⁶⁶ and subsequently increase depression risk (although the relationship between insulin resistance and depression is likely bidirectional).^23,67,68 Evidence from prior research also suggests that both prepregnancy diabetes and gestational diabetes are associated with PPD, and that treatment of gestational diabetes can improve women's postpartum mental health outcomes, including PPDS.^69–71 Future research using large and diverse study populations should carefully consider whether these psychological and biological mechanisms explain the links between obesity and PPDS, and how these factors vary by race/ethnicity.

Prior research has found that whites are more likely than nonwhites to experience body dissatisfaction,^59,72 which is consistent with the current finding that preconception obesity was only associated with the PPDS measures for white mothers. One earlier study found that compared with white women, Hispanic women are more likely to view larger body sizes as ideal or a sign of good health, which might explain why overweight was associated with lower PPDS among Hispanic mothers in the present study.²⁶ This finding is consistent with the results of a previous study that showed that preconception overweight was associated with lower odds of elevated depressive symptoms during pregnancy in a sample of Hispanic mothers.³⁷ Given that few studies have examined the link between pregnancy-related weight and PPDS among Hispanics, future research should seek to replicate the current findings in other samples. We found that pregnancy-related weight was not related to PPDS among black or Asian women. This finding is consistent with prior results showing no differences in depression by BMI category among an ethnically diverse sample of low-income women (i.e., primarily African American and Latina/Hispanic women).⁷³

Finally, the analysis showed that black and Asian women were more likely to exhibit PPDS than white women, which is consistent with prior research.^7,8,74,75 Black women are more likely than women in other racial/ethnic groups to experience racial/ethnic discrimination, financial worries, a lower level of social support, and other stressors during the pregnancy and postpartum periods, which may predispose them to higher rates of PPD.^8,76–78 Asian women are at higher risk of poor mental health outcomes in general due to potential cultural barriers to seeking help (e.g., stigma surrounding mental illness), multiple and conflicting gender stereotypes (e.g., eager to please, shy, erotic, and submissive, but at other times unattractive, impersonal, and efficient worker), and misinterpretation of expressed symptoms due to cultural differences.^79–81 Both black and Asian women may also face negative cultural beliefs about mental illness, which could compound the stress they experience related to mental health, leading to higher rates of depressive symptoms.⁸² Finally, the finding that Hispanics were less likely than whites to report PPDS was generally consistent with the existing literature.^9,33,83–85 Like black women, Hispanic women, on average, experience lower socioeconomic status, lower levels of postpartum social support, and higher levels of social stressors than whites.^8,84 However, some of these obstacles may differ by nativity: Prior research using the ECLS-B data showed that relative to U.S.-born Hispanic women, foreign-born Hispanic women were less likely to have moderate/severe PPDS, more likely to have a partner in the household, and less likely to be a teenage mother.⁵³

Limitations and Strengths

The present study has several important limitations. First, reports of PPDS are different than a clinical diagnosis of PPD. However, subclinical depressive symptoms like the ones we measure here are associated with adverse perinatal outcomes and thus are important to understand.⁸⁶ Furthermore, analyzing the prevalence of diagnosed PPD across racial/ethnic subgroups may introduce bias because factors such as physician screening and maternal willingness to disclose symptoms to a provider likely vary by race/ethnicity.^87,88 Second, like most prior studies, the present analysis relied on self-reported preconception weight and GWG, which may be subject to reporting bias and/or memory bias.⁸⁹ However, existing research suggests that there is generally a strong correlation between maternal self-reported weight and actual weight measures.^89,90 A related issue is that the ECLS-B data were collected in 2001, and national preconception obesity rates have likely significantly increased over time.⁹¹ Third, although prior research suggests that there are important differences in PPD within racial/ethnic groups by country of origin (e.g., Filipino Americans differ from Japanese Americans) and maternal nativity,⁵³ sample size limitations prevented us from examining within-race/ethnicity differences in the relationships between pregnancy-related weight and depression. We also lacked information on other biological and psychosocial mechanisms that may link prepregnancy weight and depressive symptoms, including maternal eating-related attitudes,³⁴ traumatic and emotional stress,^13,63 physical health (e.g., gestational diabetes and/or other pregnancy complications),^86,92,93 and maternal history of depression.³³ The latter may be particularly critical, given that mothers with a history of depression may have been prescribed symptom-relieving medications before the time of the survey. In turn, this may have resulted in an overall underestimate of PPDS in our sample. However, we have no evidence that women are more likely to be assigned medication for PPDS differentially by weight status. Finally, the ECLS-B survey used the CES-D to capture PPDS, rather than the Edinburgh Postnatal Depression Tool (widely considered the gold standard for use for PPD screening). However, evidence from several studies suggests that the CES-D is a validated measure that captures important information on depressive symptoms in postpartum populations.

Importantly, the present study also had several significant strengths, including the use of a large and racially diverse nationally representative data source, the ability to capture two key measures of pregnancy-related weight (i.e., preconception weight and adequacy of GWG), and the ability to control for differences in socioeconomic characteristics related to depression (i.e., education and poverty level) that might confound the relationships between pregnancy-related weight and PPDS.

Conclusion

Due to the short- and long-term implications of maternal depression for maternal and child health outcomes, national health policy goals aim to decrease the proportion of women who experience PPDS.¹⁵ This study provides robust evidence that the relationship between pregnancy-related weight and PPD is complex and varies across population subgroups. These findings are relevant to efforts to promote better mental health among expectant and new mothers. Most notably, the results indicate that preconception obesity is a key risk factor for PPDS among white mothers, while preconception overweight is a protective factor against depression among Hispanic mothers. Furthermore, there is no link between pregnancy-related weight and depressive symptoms among black and Asian mothers. While there are well-documented challenges with identifying effective preconception obesity interventions,⁹⁴ these patterns suggest that if successful, such efforts might reduce overall levels of PPDS (by reducing the level among white women). However, doing so would likely not mitigate black-white or Asian-white disparities in this outcome and could aggravate Hispanic-white disparities. Thus, reducing racial/ethnic disparities in PPDS may be more effectively accomplished by addressing other more salient social determinants of mental health such as socioeconomic status and social support.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

None of the authors have received financial support to conduct the research, writing or publication of this article.