Postpartum Depression: The Role of Gestational Weight and Adiposity, Prenatal Cortisol, Socioeconomic Resources, and Breastfeeding

Abstract

This study examined the mediating role of prenatal cortisol on the relationship between gestational weight or adiposity and postpartum depression (PPD), while considering the moderating roles of breastfeeding (BF) or socioeconomic resources. We hypothesized that women with a higher pre-pregnancy body mass index (PPBMI) or a larger abdominal circumference would have elevated diurnal cortisol levels in late pregnancy, which would predict more PPD symptoms. Additionally, we hypothesized that BF and access to more socioeconomic resources would buffer the positive relationship between prenatal diurnal cortisol and PPD symptoms. We used longitudinal data from the Infant Development and Health Outcomes in Mothers Study, in which women self-reported PPBMI, BF frequency at 6 months, familial education, occupation, and income and completed the Edinburgh Postnatal Depression Scale. The abdominal circumference, cortisol area under the curve with respect to ground (AUC_G), and cortisol awakening response (CAR) were measured. Higher breastfeeding frequency and greater socioeconomic resources were associated with fewer PPD symptoms. There were main and interactive associations of AUC_G with BF frequency with PPD. Notably, higher cortisol levels were linked to more PPD symptoms among women with less frequent BF. Thus, BF may mitigate the relationship between prenatal stress and PPD, highlighting the importance of BF support in PPD prevention.

1. Introduction

The postpartum period includes a unique set of stressors that can be understood via several psychobiological frameworks to explain the relationship between these stressors and maternal health. Two such frameworks include the allostatic load model (ALM) and the adaptive calibration model (ACM). The ALM was created to explain the effect of stress on underlying biological processes that serve as risk factors for the development, maintenance, and/or exacerbation of health-related difficulties, while the ACM helps to understand the functioning of stress-responsive psychobiology including the hypothalamic–pituitary–adrenal (HPA) axis, which adapts and changes in ways that may increase disease susceptibility [1]. For example, stress below or above an optimal level is considered a deviation from the norm, and the body must work to regain homeostasis via allostasis or internal, physiological equilibrium following actual or perceived environmental or psychological stressors. Allostasis is achieved by using mediators that interact with receptors in several tissues and organs throughout the body. Other theories, like the life history theory and resiliency model, propose that each organism allots a unique amount of time and energy into activities that promote fitness and that the ability to be resilient begins with maternal–infant interactions. Therefore, the current study utilized a primary mediator (cortisol) and a secondary mediator (abdominal circumference) as well as the body mass index (BMI) to explore the influence of stress on maternal health, while simultaneously considering subjective stress and potential protective factors (breastfeeding (BF) and socioeconomic resources). A higher BMI is associated with the dysregulation of the HPA axis system, and this type of dysregulation is also associated with an increased risk for mental health disorders [2].

1.1. Cortisol

The HPA axis neuroendocrine system utilizes cortisol to maintain organismal homeostasis. Cortisol mediates various bodily processes like metabolism, cardiovascular respiration, and blood flow [3]. The basal HPA axis follows a clear diurnal rhythm including a peak within 30 min after awakening or the cortisol awakening response (CAR). Due to its consistency, including over the perinatal period, the CAR is useful in understanding HPA axis functioning [4]. When analyzing the repeated measures of cortisol, it is also beneficial to use the area under the curve (AUC) metric, especially in the perinatal period when circulating cortisol levels increase, because it is an index of the total amount of cortisol output while considering the diurnal rhythm [5].

The influence of weight gain on cortisol levels is well documented, including prenatal cortisol and maternal weight. For example, mothers who were obese prior to pregnancy displayed high levels of evening cortisol within the third trimester. These associations are moderated by weight gain, with mothers who were obese prior to pregnancy who experienced the greatest amount of weight gain displaying the highest levels of evening cortisol [6]. Similar to the case for maternal weight, the associations between maternal mental health and cortisol are bidirectional. For instance, higher pregnancy-related anxiety is associated with flatter diurnal cortisol levels postpartum [7]. Likewise, Scheyer and Urizar [8] found higher rates of postpartum depression (PPD) among women with flatter diurnal cortisol curves and a blunted CAR. Studies have also found that pregnant women who have an increased physiological stress response to a lab stressor go on to develop more PPD symptoms [9].

1.2. Prenatal Abdominal Circumference

Anthropometry is a straightforward, reliable measure used to assess body size and proportions, with the waist-to-hip ratio, waist circumference, and waist-to-height ratio being used to assess disease risk. The waist circumference is an effective measure of abdominal adiposity associated with cardiovascular risk. It should be used in combination with the BMI to clarify potential health risks [10]. Abdominal circumference and BMI are predictive of maternal–infant health outcomes including maternal gestational diabetes and pre-eclampsia and fetal macrosomia [11]. Additionally, variations in prenatal weight and adiposity may reflect fluctuations in maternal cortisol throughout pregnancy [12].

1.3. Pre-Pregnancy BMI

Prenatal weight gain recommendations depend on multiple factors including pre-pregnancy weight, height, and ethnicity [13]. The National Academy of Medicine defines obesity as a BMI ≥ 30 kg/m² and overweight as BMI ≥ 25 and <30 kg/m². It is estimated that almost 3.5 billion people worldwide are overweight or obese [14]. Genetics, lack of sleep, socioeconomic resources, weight-gaining medication, diet, exercise, and chronic stress contribute to obesity risk [15]. Rates of obesity in reproductive-age women have substantially increased over the past decade, with 37% reportedly being currently obese and 10% morbidly obese (BMI > 40 kg/m²) [16]. Obesity prior to and during pregnancy comes with its own unique set of complications like pre-eclampsia, gestational diabetes, gestational hypertension, and psychological difficulties, including increased rates of depression and anxiety [17]. Obesity is associated with HPA axis dysregulation, which is also associated with an increased risk for mental health disorders including PPD symptoms [2].

1.4. Maternal Adjustment

Experiencing mood changes, including depressive symptoms, is common during and after pregnancy [18]. While there is no formal diagnosis of PPD in the Diagnostic and Statistical Manual of Mental Disorders—Fifth Edition, the specifier “with peripartum onset” captures the experience [19]. The specifier is defined as the most recent major depressive episode occurring during the pregnancy or 4 weeks after delivery [19]. Important protective factors like BF and socioeconomic resources access may serve as buffers between elevated maternal stress and weight and maternal postnatal adjustment.

1.5. Breastfeeding

Women who breastfeed report less depressive symptomology [20], with indices like duration and frequency playing an important role in maternal mental health. One study found a dose–response relationship between greater BF frequency and lower maternal anxiety levels [21]. These relationships are also connected to changes in the HPA axis. Tu et al. [22] found that mothers who bottle-fed had higher rates of cortisol than mothers who breastfed. Therefore, greater BF duration and frequency may reduce the relationship that HPA axis dysregulation has with maternal mental health difficulties.

1.6. Socioeconomic Resources

The effects of perinatal stressors may be particularly salient for those with fewer socioeconomic resources, which increases the risk for mental health difficulties. For example, Mahmoodi et al. [23] conducted a study in Tehran with mothers between 24 and 32 weeks’ gestation and found that parental education and income had the greatest impact on maternal mental health. Namely, women who had partners with less educational attainment received poorer pregnancy care and less social support [23], while families with less income had more maternal mental health difficulties. Likewise, having more resources enhances a mother’s ability to effectively address stress, which may be related to greater access to intervention services [24] and psychophysiological stress regulation [8]. In particular, Scheyer and Urizar [8] found that mothers with less income before and during pregnancy had flatter diurnal cortisol slopes during pregnancy and reported more PPD symptoms after pregnancy.

1.7. Proposed Model and Hypotheses

This study seeks to investigate the relationships among pre-pregnancy BMI (PPBMI) or abdominal circumference, cortisol level, and maternal adjustment. More research is needed investigating the mental health of women who experience obesity or being overweight during pregnancy. We believe that those with a higher PPBMI or abdominal circumference will experience the greatest changes in cortisol levels and report more PPD symptoms than women with a lower PPBMI or abdominal circumference. We also investigate whether BF or greater socioeconomic resources may serve as protective factors to help stabilize cortisol levels and lessen depression symptoms.

2. Materials and Methods

2.1. Participants

Data were collected as part of the Infant Development and Healthy Outcomes in Mothers (IDAHO Mom) Study in the Perinatal Psychobiology Lab at Idaho State University (ISU; approved Human Subjects Committee protocol 4191), which examined maternal health and infant development. A total of 125 pregnant women completed their prenatal visit in their third trimester. The majority (n = 96) completed a 6-month postnatal follow-up session. The participants included adult women between ages 18 and 35. A summary of the participants’ sociodemographic variables can be seen in

Table 1. Sociodemographic characteristics of participants.

Sample Characteristics	N/%	M	SD
Age (years)		27.29 yrs	4.02 yrs
Ethnicity
White/Caucasian	90/94
Black/African American	2/2
Native Hawaiian or Other Pacific Islander	2/2
American Indian/Alaska Native	1/1
Hispanic/Latino	16/14
Asian	1/1
Other	5/5
Marital Status
Married	80/84
Single/Never Married	10/10
Divorced	1/1
Committed Relationship	3/3
Engaged	1/1
Annual Familial Income
Less than USD 5000	1/1
USD 5000–USD 9000	2/2
USD 10,000–USD 19,999	14/15
USD 20,000–USD 29,999	15/17
USD 30,000–USD 39,999	11/13
USD 40,000–USD 49,999	8/9.8
USD 50,000–USD 74,999	27/28.3
USD 75,000–USD 99,999	6/7.6
USD 100,000 or greater	5/6.5
Postpartum Total EPDS Raw Score		4.54	3.66
PPBMI (kg/m2)		27.5 kg/m2	8.02 kg/m2
Abdominal Circumference (in)		46.46 in	5.39 in
BF Frequency per Day (6 months)		4.67	3.70
Social Prestige		3.35	0.97
AUCg (μg/dL)		181.20 μg/dL	21.87 μg/dL
CAR (μg/dL)		0.1276 μg/dL	0.018 μg/dL

Note: AUCg (area under the curve with respect to ground) and CAR (cortisol awakening response) are both measured in micrograms per deciliter (μg/dL). Participants could select more than one racial/ethnic identity; therefore, the total percentage of people across categories exceeds 100%.

.

2.2. Recruitment and Screening

The participants were recruited from doctor’s offices, businesses, and schools across southeastern Idaho, and this study was advertised on the news, radio, and social media platforms. The exclusion criteria included the following: carrying more than one baby; specific health conditions (e.g., gestational diabetes, pre-eclampsia, and toxemia); specific behavioral or physical health diagnoses/symptoms (e.g., schizophrenia, bipolar disorder, HIV, and AIDS); ongoing use of recreational substances (e.g., marijuana or cocaine), medications from FDA categories D and X (i.e., due to documented detrimental fetal effects), or alcohol use (>40 drinks) during pregnancy [25]. The participants had to be fluent in written and spoken English to provide written informed consent and complete the study measures.

2.3. Procedures

The prenatal session included interviews about previous and current pregnancy-related information, general health history, and sociodemographic characteristics (e.g., socioeconomic resources, ethnicity, race, and age); anthropometric assessment (height, weight, and midsection); and self-report questionnaires of maternal and infant health. The participants were reimbursed USD 30 for completing the prenatal session. The participants were also given a saliva collection kit including materials for a 3-day collection period with four saliva samples per day (12 samples total). The samples were collected immediately after awakening, 30 min after awakening, 45 min after awakening, and prior to going to bed. The participants completed a written log to record the sample day and time, which was facilitated by text message reminders prompting the participants to confirm when they completed samples. Taken together, these data were used to ensure that the samples were completed within the appropriate diurnal time window. The participants were instructed not to eat, drink, smoke, or brush their teeth before providing samples and to store the samples in their freezers until RAs retrieved them. The participants were given USD 5 in gift cards for each day of samples completed. The samples were then stored in a lab freezer at −20 degrees Celsius until they were sent to the ISU Molecular Research Core Facility (MRCF) for analysis. During the 6-month session, mothers completed interview and self-report questions similar to those asked during the prenatal session regarding maternal and infant health, behavior, and psychological symptoms.

2.4. Measures

2.4.1. Maternal Pre-Pregnancy BMI

Height was measured using an adult stadiometer (Shorr Productions, Olney MD (accuracy to 0.1 cm)), and weight was measured using a battery-powered digital mother–infant scale (Seca 876 (accuracy to 0.1 kg)). The PPBMI was calculated using a maternal report of pre-pregnancy weight and the height measured in the lab via the following equation: weight (kg)/[height (m)]² [26].

2.4.2. Maternal Prenatal Abdominal Circumference

The maternal abdominal circumference was measured using a non-stretchable ShorrTape© Measuring Tape (accuracy to 0.1 cm) [27]. The standardized procedure for collecting abdominal circumference was based on Lohman’s (1981) research and the CDC National Health and Nutrition Examination Survey [28].

2.4.3. Cortisol Assays

Cortisol was assayed from the saliva samples using Salimetrics enzyme-linked immunosorbent assay (ELISA) kits conducted in duplicate for each sample for each participant (i.e., 24 total). The kits had a sensitivity of <0.007 μg/dL and an assay range of 0.012–3.00 μg/dL. The cortisol concentration values were quantified in μg/dL using CAR and AUC_G indicators.

2.4.4. Breastfeeding

At the 6-month session, the participants completed a 7-question, self-report measure to capture BF experiences and difficulties. It included questions about the infants’ type of feeding (e.g., formula vs. BF) and the onset of BF. To measure BF frequency, the question “Presently, for how many feedings a day do you breastfeed?” was used.

2.4.5. Socioeconomic Resources

The Hollingshead Four-Factor Index of Social Status [29] was used to estimate socioeconomic resources. For this study, the combined social prestige variable was used, which accounts for a mother’s and contributing partners’ education, occupation, and earnings levels.

2.4.6. Depression Symptoms

Depression symptoms were measured using the EPDS [30], which is a 10-item questionnaire asking participants to rate how they have felt over the last week. The responses are measured on a 4-point Likert scale (0–3), with the minimum score being 0 and the maximum being 30. The greater the score, the more severe the depressive symptoms [31]. The current study used the total EPDS score from the 6-month visit. Prior studies suggest that the EPDS has good internal consistency (α = 0.76) and criterion validity, and a factor analysis supported the measure’s internal structure and construct validity [32].

2.4.7. Sociodemographic Descriptors

The maternal age in years was collected at the prenatal session. The annual gross family income level was collected at the prenatal session. Educational attainment was collected at the prenatal session (see Table 1 for categorical distribution). Maternal racial/ethnic identity was also collected at the prenatal session as part of the sociodemographic interview (see Table 1 for categorical distribution).

2.5. Statistical Analyses

In order to control for important factors that may influence relationships between predictors in the model and the outcome variable, covariates that are empirically associated with PPD in the literature (i.e., maternal age, familial income level, maternal education, and maternal reported racial/ethnic identity) were investigated in relation to maternal pre-pregnancy BMI/prenatal abdominal circumference and 6-month EPDS scores and are included as covariates in the moderated mediation models (see Figure 1). Specifically, in this model, the pre-pregnancy BMI or abdominal circumference is X; postpartum depressive symptoms are Y; CAR or AUCG is M; and breastfeeding frequency or socioeconomic resources is W. The relationship between X and Y is the direct path; M is the mediator; and W is the moderator.

We estimated the direct association of X with Y (c’ path), while controlling for the possible association of M and taking into consideration the moderation of W. The a and b paths in the model were used to measure the indirect association of X with Y, while controlling for the possible association of M [33] (p. 92). The formula c = c′ + ab was used to measure the total association of X on Y [33] (p. 93). The direct association in the model was assessed using a significance value of a two-tailed p-value < 0.05 and a 95% confidence interval. Bias-corrected bootstrapping with replacement was used to measure the indirect association in the model. This created a z-score that was normally distributed and used to compare the original distribution and determine the upper and lower bounds of the confidence interval [33] (p. 112). This study used 5000 iterations for the bootstrapping process [33]. To estimate how W affects the relationship between M and Y in the model, a conditional process model was used, which quantified how changes in M mapped onto changes in Y as a function of W [33] (p. 374).

An a priori power analysis using G*Power (version 28.0) was conducted with the following parameters based on the prior literature and studies using IDAHO Mom Study data: a linear multiple regression with a fixed model, R² increase with three tested predictors (X, M, M × W) and up to five covariates, a medium effect size (ƒ² = 0.15), low Type I error probability (α = 0.05), and power of 0.80. The results suggest that a sample size of 77 is sufficient [34]. All the data analyses were conducted using IBM SPSS Statistics for Windows, Version 28.0 [35]. All the models were tested using Hayes Process macro-Model 7 [33].

3. Results

3.1. Descriptive Statistics

See Table 1 for a complete summary of the descriptive statistics for participant sociodemographic and primary study variables. The majority of the participants earned an annual income of USD 50,000–74,999 (28%), had earned a college or university degree (39%), and identified as European American (88%), though 14% identified as Hispanic or Latina. The mean participant age was 27.29 years (SD = 3.85 years). None of the covariates correlated with both predictor and outcome variables, so they were not used in the moderated mediation analyses.

3.2. Primary Analyses

See

Table 2. Summary of moderated mediation models predicting postpartum depression symptoms.

Model	X (Predictor)	M (Mediator)	W (Moderator)	Significant Associations
1	PPBMI	CAR	BF Frequency	None
2	PPBMI	AUCG	BF Frequency	Main Association—BF AUCG × BF
3	PPBMI	CAR	Socioeconomic Resources	Socioeconomic Resources
4	PPBMI	AUCG	Socioeconomic Resources	None
5	Abdominal Circumference	CAR	BF Frequency	None
6	Abdominal Circumference	AUCG	BF Frequency	AUCG Main Association—BF AUCG × BF
7	Abdominal Circumference	CAR	Socioeconomic Resources	Socioeconomic Resources
8	Abdominal Circumference	AUCG	Socioeconomic Resources	None

Note. AUCg (area under the curve with respect to ground), CAR (cortisol awakening response). PPBMI (pre-pregnancy body mass index), and BF (breastfeeding frequency).

for a summary of the moderated mediation models predicting postpartum depression symptoms.

3.2.1. Moderated Mediation Model 1: PPBMI, CAR, BF Frequency, and Depression Symptoms

A statistically significant amount of variance in PPD symptoms was not explained in the model (R² = 0.0242, F[1, 74] = 0.4333, p = 0.7841) and no direct, indirect, or interactive associations were statistically significant.

3.2.2. Moderated Mediation Model 2: PPBMI, AUC_G, BF Frequency, and Depression Symptoms

A statistically significant amount of variance in PPD symptoms was not explained in the model (R² = 0.1266, F[1, 54] = 1.8127, p = 0.141). PPBMI was not a statistically significant predictor of PPD symptoms in the model (b = 8.250, t(55) = 0.7647, SE = 10.7874, p = 0.4480). AUC_G (M) did explain a statistically significant amount of variance in PPD symptoms (b path; b = 3.323, t(55) = 2.050, SE = 1.620, p = 0.0456). The main (b = 4.1423 t(55) = 2.2824, SE = 1.8149, p = 0.0268) and interaction associations (with AUC_G; b = −1.6433, t(55) = −2.2492, SE = 0.7306, p = 0.0289) of BF frequency on PPD symptoms were both statistically significant. Likewise, the significant variance in the EPDS total scores was explained by the addition of the interaction between AUC_G and BF frequency while considering the main associations (R²_CHANGE = 0.0884).

3.2.3. Moderated Mediation Model 3: PPBMI, CAR, Socioeconomic Resources, and Depression Symptoms

A statistically significant amount of variance in PPD symptoms was not explained in the model (R² = 0.0794, F[1, 74] = 1.5095, p = 0.2088). The main association of socioeconomic resources with PPD symptoms was statistically significant (b = −0.0209, t(55) = −2.0992, SE = 0.0100, p = 0.0394), such that those with more resources reported experiencing fewer PPD symptoms. No other direct, indirect, or interactive associations were statistically significant.

3.2.4. Moderated Mediation Model 4: PPBMI, AUC_G, Socioeconomic Resources, and Depression Symptoms

A statistically significant amount of variance in PPD symptoms was not explained in the model (R² = 0.1277, F[1, 54] = 1.8304, p = 0.1377), and no direct, indirect, or interactive associations were statistically significant.

3.2.5. Moderated Mediation Model 5: Abdominal Waist Circumference, CAR, BF Frequency, and Depression Symptoms

A statistically significant amount of variance in PPD symptoms was not explained in the model (R² = 0.0462, F[1, 74] = 0.8475, p = 0.4999), and no direct, indirect, or interactive associations were statistically significant.

3.2.6. Moderated Mediation Model 6: Abdominal Waist Circumference, AUC_G, BF Frequency, and Depression Symptoms

A statistically significant amount of variance in PPD symptoms was not explained in the model (R² = 0.1165, F[4, 50] = 1.6485, p = 0.1768). Abdominal circumference was not a statistically significant predictor of PPD symptoms in the model (b = 0.1348, t(55) = 0.0678, SE = 1.9891, p = 0.9462). AUC_G (M) explained a statistically significant amount of variance in PPD symptoms (b path; b = 3.5557, t(55) = 2.2138, SE = 1.6062, p = 0.0314). The main (b = 4.4004, t(55) = 2.4435, SE = 1.8009, p = 0.0181) and interaction associations (with AUC_G; b = −1.7462, t(55) = −2.4063, SE = 0.7257, p = 0.0199) of BF frequency with PPD symptoms were both statistically significant. Likewise, the significant variance in the EPDS total scores was explained by the addition of the interaction between AUC_G and BF frequency while considering the main associations (R²_CHANGE = 0.1023). Follow-up analyses suggested that there were significant positive associations between AUC_G and PPD symptoms only when BF frequency was 1 standard deviation below the mean value or about one BF per day (as opposed to at (4.67 times/day) or 1 standard deviation above (8.37 times/day) the mean value; b = 2.5274, t(55) = 2.0357, SE = 1.2415, p = 0.0471, 95% CI [0.0337, 5.0212).

3.2.7. Moderated Mediation Model 7: Abdominal Waist Circumference, CAR, Socioeconomic Resources, and Depression Symptoms

A statistically significant amount of variance in PPD symptoms was not explained in the model (R² = 0.1072, F[1, 74] = 2.1011, p = 0.0898). The main association of socioeconomic resources with PPD symptoms was statistically significant (b = −0.0214, t(55) = −2.1851, SE = 0.0098, p = 0.0322), such that those with more resources reported experiencing fewer PPD symptoms. No other direct, indirect, or interactive associations were statistically significant.

3.2.8. Moderated Mediation Model 8: Abdominal Waist Circumference, AUC_G, Socioeconomic Resources, and Depression Symptoms

A statistically significant amount of variance in PPD symptoms was not explained in the model (R² = 0.0996, F[1, 54] = 1.3829, p = 0.2533), and no direct, indirect, or interactive associations were statistically significant.

4. Discussion

The purpose of the current project was to explore the associations between cortisol levels and women’s PPBMI or abdominal circumference and how this influences maternal adjustment, specifically PPD symptoms. While none of the moderated mediation models were statistically significant, the results indicated that the main association of BF frequency was statistically significant. These findings highlight the importance of considering BF practices in the assessment and management of PPD. Healthcare providers should be aware that BF frequency may be a relevant protective factor for women at risk for PPD. In addition, BF frequency impacted the relationship between AUC_G and PPD symptoms. Specifically, greater cortisol output was positively associated with more depression symptoms only when the BF frequency was low. Thus, more frequent BF may act as a buffer between psychophysiological stress and the development of PPD symptoms. Research has found that BF has protective characteristics in that it reduces the likelihood of developing PPD symptoms [36]. Hahn-Holbrook et al. [37] found that women who breastfed more frequently at 3 months postpartum showed decreased PPD symptomology. Mothers who practiced non-exclusive BF were seven times more likely to experience PPD than exclusively BF mothers [38].

Additionally, there was a significant relationship between more socioeconomic resources and the experience of fewer PPD symptoms in the current sample. This replicates some studies that have found that income, education, and SES may provide more resources to better handle the hardships that may occur after giving birth [8].

Contrary to the study hypotheses, the CAR and the AUC_G measurements did not predict PPD symptoms. Within the context of pregnancy, there are several associations between cortisol and maternal mental health [7,8]. Seth et al. [39] conducted a systematic literature review and found that some studies reported significantly higher levels of cortisol shortly after giving birth. Future research should attempt to measure cortisol levels shortly after birth to better understand these unique shifts. Conversely, several studies have found non-significant associations between cortisol during pregnancy and PPD symptoms [40]. Some studies even reported lower levels of cortisol are associated with mothers experiencing depressive symptoms after birth [39].

Therefore, the current study findings may be because dysregulation of the HPA axis can lead to higher or lower levels of salivary cortisol [41]. Prolonged or repeated exposure to stressors can lead to the overactivation of the HPA axis, causing the sustained elevation of cortisol levels [42]. Paradoxically, prolonged exposure to stressors can lead to HPA axis hypoactivity and blunted cortisol responses. This is often observed in conditions of chronic stress in which the HPA axis becomes dysregulated and fails to create an appropriate cortisol response [42]. Thus, having a mix of opposite directions of associations may have contributed to the non-significant findings for the models that included AUC_G. However, this may also be attributed to the use of a relatively low-risk community sample with few reported psychological symptoms.

Neither PPBMI nor abdominal circumference predicted PPD symptoms or was associated with higher levels of CAR or AUC_G. Adults with greater abdominal adiposity have been shown to have increased cortisol reactivity in response to physical and psychosocial stressors [43]; however, the relationship in pregnancy is unclear. Aubuchon-Endsley et al. [6] found that there was a relationship between PPBMI and diurnal cortisol indicators, but this relationship was moderated by gestational weight gain, suggesting that future studies should consider this important moderator when testing similar models. Based on the dearth of research, the relationship between prenatal abdominal waist circumference and PPD symptoms is not clear. The body undergoes physiological adaptations during pregnancy to support fetal growth and development, including changes in metabolism and body composition. These adaptations may impact the way abdominal adiposity influences mood and depressive symptoms in pregnant women compared with the general population [44]. Similarly, there are few studies that examine the relationship between PPBMI and PPD symptoms. Scott and Aubuchon-Endsley [45] demonstrated a relationship between greater PPBMI and an increased number of PPD symptoms reported at 6 months using a subsample of the IDAHO Mom Study data, which means that, when considering the other variables in the current model (i.e., diurnal cortisol and either BF frequency or socioeconomic resources), this direct relationship is no longer statistically significant. This may mean that cortisol regulation, breastfeeding, and/or psychosocial resources may mediate the relationship between PPBMI and depressive symptoms in some capacity or just serve as more robust predictors of PPD symptoms.

Contrary to the study hypotheses, CAR and AUC_G did not predict PPD symptoms. Prior research has shown mixed results regarding associations between cortisol and maternal mental health. Some studies report that elevated cortisol levels shortly after birth are associated with increased PPD symptoms [46], suggesting PPD symptoms may be more detectable immediately postpartum [39]. Notably, recent findings indicate that higher perceived prenatal stress is associated with greater PPD symptoms reported both as early as 3 days and again at 12 weeks postpartum, highlighting critical windows for symptom emergence and detection [47]. Additionally, recent research using hair cortisol from a large population-based sample found that elevated third-trimester cortisol levels were associated with increased risk of PPD symptoms at 2 months postpartum [48]. However, other studies have found no significant association between cortisol during pregnancy and PPD symptoms [40], with some even reporting lower cortisol levels in mothers with depressive symptoms [39]. In summary, cortisol levels fluctuate across the perinatal period, and their relationship with depressive symptoms may differ depending on whether they are assessed during pregnancy, immediately postpartum, or several months later. Indicators of psychological and physiological stress do not fully overlap based on other important psychological and biological factors, so it is possible that physiological stress alone is not always robustly predictive of PPD without considering these other factors. Further research is needed to clarify the nature of this relationship.

This study has several limitations. First, due to the homogenous nature of the sample collected, the generalizability of the current findings should be carefully considered among a broader, more diverse population. Specifically, this sample was predominantly White, middle-class, and lived in southeastern Idaho. Second, while the data were collected at two time points, the analytic approach used is structurally correlational and does not allow for causal inference. The observed associations between prenatal physiological measures and postpartum depressive symptoms should be interpreted with caution, as the temporal sequence of changes in cortisol and depressive symptoms cannot be definitively established. Future research should seek to investigate these findings with a more diverse sample using experimental designs.

Implications

Based on the current study findings, healthcare providers and policymakers should prioritize initiatives that support and promote BF initiation and maintenance in line with the American Academy of Pediatrics guidelines. By encouraging more frequent BF, especially in mothers at risk of PPD, healthcare systems can leverage the protective effects of BF against psychophysiological stress and depressive symptoms. In addition, integrating BF support and mental health services within perinatal care settings can optimize maternal mental health outcomes. By addressing both BF challenges and depressive symptoms concurrently, healthcare providers can provide comprehensive support to mothers and improve overall well-being. Further, continued prevention efforts aimed at decreasing PPD are important based on health implications to the mothers and children. Research on PPD should consider cultural, social, and contextual factors that influence its prevalence, expression, and help-seeking behaviors across diverse populations based on differences noted within groups from the literature review. In addition, research on PPD can help raise awareness and reduce stigma surrounding maternal mental health issues. By highlighting the prevalence and impact of PPD, efforts can be made to address misconceptions, promote discussion, and encourage help-seeking behaviors among those affected.

5. Conclusions

Due to patterns of increasing weight gain in pregnant populations, this study sought to investigate associations between cortisol levels and women’s PPBMI or abdominal circumference and how this influences maternal adjustment, specifically in terms of PPD symptoms via important evidence-based mediators (i.e., diurnal cortisol) and moderators (i.e., BF and access to socioeconomic resources). The current study supported statistically significant associations of BF frequency and socioeconomic resources with PPD as well as the influence of BF frequency on the association of cortisol levels with the development of PPD symptoms. Specifically, findings in the prior literature were replicated in that having greater BF frequency and more socioeconomic resources are predictive of experiencing fewer PPD symptoms. Additionally, greater cortisol output was positively associated with more depression symptoms only when the BF frequency was low. Thus, BF may serve as a buffer to associations between elevated prenatal psychophysiological stress and PPD symptoms, suggesting that PPD prevention and intervention strategies should consider including BF promotion along with other mental health screening, management, and treatment options. However, more studies are needed to replicate the findings in larger samples with participants who have greater sociodemographic diversity. The current study failed to support statistically significant mediation of PPBMI or abdominal circumference and postpartum symptoms by cortisol levels and an interaction of socioeconomic resources with cortisol levels in relation to PPD symptoms. Thus, future studies are needed to minimize the current study’s limitations and to continue addressing important gaps in the literature.