In this prospective study, we found that women giving birth to a male versus a female infant were more likely to develop depressive symptoms during the postnatal period. These results are consistent with a recent, retrospective study (n=296 women, 651 births) conducted in the United Kingdom that found the odds of postpartum depression (PPD) were 71-79% higher among women carrying a male versus female fetus, after controlling for other risk factors and individual-level characteristics (28). Similarly, a cross-sectional study of 181 women based in France found postnatal depression was significantly associated with having a male infant (35) and a study of 2,267 women in Sweden found male infant sex was associated with an increased odds of experiencing postpartum depressive symptoms within 5 days of delivery, but not at 6 weeks or 6 months postpartum (36). In the present study, we found that women carrying a male fetus also had a higher odds of prenatal depressive symptoms, although the magnitude of this association was small and did not reach statistical significance. The stronger findings observed with depressive symptoms during the postnatal, versus prenatal, period could suggest that fetal-sex linked events during late pregnancy or parturition may specifically contribute to PPD susceptibility. Notably, the majority of prior research has not been designed to examine temporal variation in associations across the pregnancy and postpartum periods.
Several biological mechanisms could underlie our observed findings, including sex-differential shifts in maternal reproductive and other hormones or immune activity. The Reproductive Hormone Model of PPD, which is supported by multiple lines of research, including animal models, clinical trials, and biomarker-based observational studies, purports that PPD arises among a sensitive phenotype of women following hormonal fluctuations in the brain that occur postpartum and during other reproductive phases (i.e., premenstrual and peri-menopausal periods) (26, 27). For example, following parturition, estradiol, which modulates the serotonergic system, and progesterone levels rapidly decrease (37). Research conducted in rats has shown experimental induction of a hormone withdrawal state that mimics that of the postpartum period results in displays of depressive behavior, despair and anhedonia (38-41). Corresponding research in humans has shown pharmacologically neutralizing estrogen withdrawal following delivery reduces the onset of depressive symptoms among women with a history of PPD (42). Notably, recent research has demonstrated that fetal sex modifies the association between reproductive hormone profiles across pregnancy and maternal behavior during the postpartum period (43), providing evidence that the fetus may influence maternal neurobiology with lasting effects. Other research has reported sex-differential levels of reproductive hormones during pregnancy in maternal serum (22, 44-46) and amniotic fluid (47). For example, several studies have shown higher serum estrogen levels in female-bearing pregnancies (22, 44, 45). It is plausible that the lower pregnancy estrogen levels of women carrying males alters their neurobiological response to estrogen withdrawal following delivery; however, to our knowledge this has been formally investigated. Additionally, it is notable that the finding of sex differences in maternal hormone levels is not consistent across studies, with some groups detecting no differences by fetal sex (47). Importantly, we are not aware of research that has examined fetal-sex differences in maternal hormone levels during parturition or the post-partum period, which may be windows of heightened maternal vulnerability based on the findings reported here. Research that repeatedly measures maternal hormone levels across the course of pregnancy and into the post-partum period would provide the means to examine whether fetal sex is associated with altered trajectories of hormone change.
Hormone surges during pregnancy also play a role in maternal immune system shifts that promote tolerance of the fetus (13, 48). These shifts include an upregulation of generalized inflammatory responses (i.e. release of pro-inflammatory cytokines) towards later pregnancy and into the postpartum period (29). Depressive disorders are increasingly associated with inflammation (24, 25), providing a putative link between unusually prolonged or excessive immune activation during pregnancy and perinatal depression (15). Recent research suggests that maternal immune profiles may vary by fetal sex, with women carrying a male fetus having increased levels of pro-inflammatory markers (19, 49). Sex differences are also evident in studies of the placenta, with male placentas associated with a stronger inflammatory response (18). Notably, pro-inflammatory cytokines have been shown to increase activity of serotonin precursors (50), suggesting a role for overlapping biological pathways between the immune and other neurobiological systems. Our finding of a significant association during the postpartum, but not prenatal, period could relate to sex-differential priming of the maternal immune system that only becomes apparent with hormonal fluctuations following delivery. For example, estrogen, which has been shown to vary by fetal sex during pregnancy and withdraws following delivery, binds to receptors expressed on most immune cells (51) and regulates the immunological shift that occurs during pregnancy (29). While it is plausible that estrogen-regulation of immune activity varies by fetal sex, we are not aware of research that has specifically examined interactions between sex, reproductive hormones, immune activity, and PPD susceptibility.
Limited research has also shown fetal-sex differences in maternal levels of a range of other biological molecules, including: human chorionic gonadotropin (52), thyroid hormone (53), leptin (54), cortisol (55), and angiogenic factors (19). In turn, dysfunction of the regulatory systems that orchestrate production and homeostasis of these molecules, including the hypothalamic-pituitary-thyroid and hypothalamic-pituitary-adrenal axes, have been implicated in the pathophysiology of depression (56, 57), including during the postpartum period (16, 58-61). However, we acknowledge that clinical and mechanistic research examining fetal sex-differential levels of these hormones and other biological effectors in relation to PPD remains limited. Other research has shown sex-differential expression of imprinted genes in the placenta that in turn have been linked to maternal mood disorders (62). For example, PEG3 is an imprinted gene that plays important roles in controlling fetal growth and nurturing behaviors. Murine models have demonstrated loss of Peg3 expression has sexually-dimorphic consequences for placental function (63), including the expression of hormones, and influences maternal behavior (64).
In contrast to findings from studies conducted in Western cultures, research conducted in Asia (China, India), Africa (Nigeria) and Turkey has reported female infant sex is positively related to PPD (65-67), with qualitative research supporting a role for male gender preference (68). For example, a prospective cohort study based in China found an increased odds of PPD among women who gave birth to a female compared to a male infant; however, the association did not hold following adjustment for social support after childbirth (69). In the United States (U.S.), there is limited research to tie PPD with gender preference; however, we cannot definitively rule out the potential contribution of cultural factors to our findings. For example, econometrics research has identified several differences between U.S. parents of girls versus boys, including a higher divorce rate and differential monetary investment in families (70). Other research suggests subtle gender preference may vary by nativity, with first- and second-generation American immigrants showing greater preference for boys (71). It is thus plausible that cultural factors that track with race/ethnicity or nativity could have contributed to the difference in the magnitude of association that we observed between our diverse cohorts. Moreover, it is well established that in Black and Hispanic communities, boys are often subjected to a range of negative stereotypes and biases related to aggression and violence that place them at greater risk for systemic disadvantages (e.g., school suspensions, incarceration) (72). It is thus plausible that among minority communities, the increased pressure and parenting stress associated with raising boys may lead to greater psychological stress and precipitate the onset of depressive symptoms. Unfortunately, here we were unable to explore interactions with race/ethnicity or nativity owing to our limited sample size across these categorical variables in combination with sex and levels of prenatal and postnatal depressive symptoms.
Across models, we found that women pregnant with a male had lower, albeit not statistically significant, odds of having depressive symptoms during both the prenatal and postnatal periods, which may indicate a more chronic, severe phenotype of depression. Male fetuses have been shown to be more vulnerable to a host of prenatal challenges and adverse intra-uterine conditions, including maternal stress (73), nutrient deprivation (74), exposure to environmental toxicants (75), and maternal chronic health conditions (76). While speculative, it is plausible that this finding reflects a higher incidence of miscarriage among males conceived by women with more severe depression.
Strengths and limitations
A primary strength of this study relates to the prospective nature and general consistency of results across two distinct pregnancy cohorts, although findings were not significant in ACCESS. We also administered the EPDS at two time points, allowing us to examine the importance of timing and understand differential susceptibility across the prenatal and postnatal periods. Despite this, future studies that include additional screening points across the antenatal, perinatal and postpartum periods would allow for the effect of timing to be more comprehensively evaluated. Additionally, both cohorts enrolled a high proportion of lower-income, minority women, a demographic with elevated risk for experiencing depression and other mental health disorders (77). While the diversity of our cohorts is a strength that contributes key research spanning sociodemographic strata, our findings may not generalize to all populations. We also note that Black women enrolled in ACCESS and PRISM were more likely to be excluded from the present analyses due to missing EPDS assessments; this is consistent with other epidemiologic studies that have long cited difficulty retaining African American participants in observational and clinical research (78). While this selection could plausibly be a threat to the generalizability of our results, we do not expect sex to vary by race/ethnicity, and thus the internal validity of our findings is unlikely to be compromised. Additionally, while there were no significant differences in EPDS scores between those included and excluded, it is still possible those with the most severe symptoms were lost to follow-up for reasons relating to depression. Other limitations of this study include the use of the EPDS, which is a screening instrument, rather than a diagnostic tool, to identify women with depressive symptoms. However, we do show strong associations between scores on the EPDS and maternal self-reported history of depressive feelings, prior depression diagnosis and medication use in these samples. Given stigma associated with perinatal depressions, it is also possible that our use of in-person and telephone interviews to administer the EPDS resulted in underreporting of symptom severity. Additionally, among women who exhibited depressive symptoms during the prenatal period, we do not know whether the onset of symptoms occurred prior to or during pregnancy, which may be an important factor relating to the underlying biological mechanism at play. It is thus possible that misclassification of symptom onset could have biased associations between sex and prenatal depressive symptoms towards the null.