The association between time-to-pregnancy and postpartum depressive symptoms in a North American prospective cohort study.

Compared with participants who conceived in less than 3 cycles, those who conceived in greater than or equal to 12 cycles were more likely to be aged greater than or equal to 35 years (15.0% vs. 9.0%), identify as Black, Indigenous, Latina, Asian, or multiracial (17.8% vs. 10.9%), have a household income <$50,000/year (16.8% vs. 7.4%), have a history of subfertility or infertility (36.1% vs. 6.2%), have gynecological conditions such as polycystic ovarian syndrome (7.7% vs. 4.0%) and endometriosis (3.4% vs. 1.5%), and have less than a college education (24.0% vs. 11.2%), and were slightly more likely to report preconception histories of diagnosed depression (25.7% vs. 23.6%), anxiety (27.4% vs. 23.8%), or PPD in a previous pregnancy (3.9% vs. 3.0%) ( Table 1 ). In the analytic cohort, 10.6% of women reported a EPDS score that was consistent with possible PPD (EPDS ≥ 13). After adjusting for potential confounders, longer TTP was associated with a modest increased risk of PPD ( Table 2 ). Compared with women who took less than 3 cycles to conceive, RRs for those who took 3–5, 6–11, and greater than or equal to 12 were 1.06 (95% confidence intervals [CI]: 0.77, 1.45), 1.24 (95% CI: 0.90, 1.70), and 1.31 (95% CI: 0.87, 1.99), respectively. When TTP was dichotomized at the clinical definition of infertility (i.e., ≥12 vs. <12 cycles), the RR was 1.17 (95% CI: 0.83, 1.64), indicating that infertility was modestly positive associated with risk of PPD. Consistent with the categorical results, restricted cubic splines modeling TTP as a continuous variable demonstrated that PPD risk increased with increasing TTP, with the greatest risk for PPD occurring at TTP greater than or equal to 6 months ( Fig. 1 ). Although imprecise, the TTP-PPD association was stronger among participants with a history of anxiety and/or depression/PPD before study enrollment (≥12 vs. <3 cycles: RR = 1.55, 95% CI: 0.95–2.53; Table 3 ). Notably, risk of PPD was 18.3% among women with a history of anxiety and/or depression/PPD and 5.3% among women without a history of anxiety and/or depression/PPD. In addition, associations between TTP and PPD were somewhat stronger among nulliparous women (i.e., ≥12 vs. <3 cycles, nulliparous women: RR = 1.45, 95% CI: 0.85, 2.48; parous women, RR = 1.20, 95% CI: 0.62, 2.31; Table 4 ). However, estimates are imprecise. In mediation analyses, early pregnancy perceived stress explained 30% of the association between infertility and PPD. The adjusted RRs for the natural direct and indirect effects were 1.13 (95% CI: 0.74, 1.73) and 1.05 (95% CI: 0.96, 1.14), respectively.

Pregnancy Study Online (PRESTO) is an ongoing internet-based preconception cohort study of pregnancy planners residing in the United States and Canada [ 25 ]. Self-identified females attempting to conceive, aged 21–45 years, not using contraception or receiving fertility treatment, and in a relationship with a male partner were invited to participate. Participants completed online questionnaires at baseline and every eight weeks for up to 12 months or until reported pregnancy, initiation of fertility treatment, or loss to follow-up. Participants who conceived completed questionnaires at ~4–12 weeks’ and ~32 weeks’ gestation of pregnancy. Approximately 6 months after their reported due date, participants completed a postpartum questionnaire containing questions about maternal and infant health outcomes. Participants who reported a stillbirth or infant death were not prompted to complete the postpartum questionnaire. Only Pregnancy Study Online participants who conceived, delivered a live born infant, and completed the postpartum questionnaire were included in this analysis. The study was approved by the institutional review board at Boston University Medical Campus, and all participants provided informed consent. From October 2017 through November 2021, we invited 3653 women to complete the postpartum questionnaire, and 2928 women responded (80%). We excluded 44 women who did not complete the EPDS portion of the questionnaire, and 241 women who completed the postpartum questionnaire greater than 1 year after the birth of their infant. The final analytic population comprised 2643 women. On the baseline questionnaire, participants reported their menstrual cycle length, the date of their last menstrual period (LMP), and the number of menstrual periods they had since they began trying to conceive. On each follow-up questionnaire, participants reported the date of their LMP and their current pregnancy status. Total menstrual cycles until conception were calculated using the following formula: menstrual cycles of attempt time at study entry plus [(LMP date of reported pregnancy - LMP date before baseline enrollment)/usual cycle length] plus 1. Participants completed the 10-item Perceived Stress Scale (PSS; [ 26 ]) at baseline and during early pregnancy (i.e., ~4–12 weeks’ gestation). The PSS is a widely used self-report measure of perceived stress that reliably captures stress over the past 4 weeks in community-based samples [ 26 - 28 ]. The PSS has been found to be a better measure of stress than the occurrence of stressful life events because it takes into account the appraisal of and distress related to the stressful event rather than assuming that the stressful event in and of itself causes psychopathology or adverse health-related outcomes [ 26 ]. The total PSS score can range from 0–40 with higher scores indicating greater perceived stress. Internal consistency of the PSS scale was high at both baseline (α = 0.87) and early pregnancy (α = 0.87). On the 6-month postpartum questionnaire, participants completed the Edinburgh Postnatal Depression Scale (EPDS; [ 29 ]). The EPDS is the most widely used self-reported measure of depressive symptoms during the perinatal period; it is comprised of 10-items with scores ranging from 0–30. The EPDS demonstrates good reliability across countries and languages [ 30 ]. However, the positive predictive value with depression prevalence using clinician administered instruments varies widely due to differing clinical cut-off recommendations in the literature (i.e., cut-off score recommendations range from 8–15), which have frequently over estimated depression prevalence [ 30 ]. A recent meta-analysis and review on the accuracy of the EPDS compared to clinician-administered interviews found that although a cut-off of greater than or equal to 11 maximized both sensitivity and specificity, a score of greater than or equal to 13 was more specific at identifying women with perinatal depression [ 31 ], and was reported previously as the optimal threshold for possible PPD [ 32 ]. On the baseline questionnaire, participants provided information on age, self-identified race/ethnicity, education, household income, employment status, history of subfertility or infertility, gynecological conditions (e.g., polycystic ovarian syndrome), height, weight, intercourse frequency, physical activity, and smoking. They were also asked about history of diagnosis of depression and/or anxiety via two separated questions (i.e., “Have you ever been diagnosed with depression? Have you ever been diagnosed with anxiety/panic disorder?”). Most covariate data were collected at baseline to minimize the possibility of exposure being influenced by TTP. However, follow-up questionnaires ascertained information regarding the initiation of fertility treatment to conceive the current pregnancy and self-reported history of PPD (i.e., “Did a healthcare provider ever tell you that you have postpartum depression?”). The following covariates were included in all analytic models: age (< 35 vs. ≥ 35 years), a history of diagnosed anxiety or depression/PPD (yes vs. no), parity (parous vs. nulliparous), race/ethnicity (white non-Hispanic, Black non-Hispanic, Hispanic, and other race/ethnicity), education (<16 vs. ≥16 years), household income (<50,000 vs. ≥50,000 United States Dollars per year), currently employed (yes vs. no), number of primary care provider (PCP) visits in the past year (0–1 vs. ≥2), a history of subfertility or infertility prior to the enrollment in the study (yes vs. no), body mass index, and initiation of fertility treatment to conceive the current pregnancy (yes vs. no). Analyses were conducted in SAS version 9.4. First, we evaluated the frequency distributions of TTP overall and according to key baseline characteristics. We then used multivariable modified Poisson regression models with robust error variance to estimate risk ratios (RRs) and 95% confidence intervals for the association between TTP and PPD [ 33 ]. TTP was modeled categorically as follows: <3, 3–5, 6–11, greater than or equal to 12 cycles, with less than 3 cycles serving as the reference group. The highest categories of TTP (6–11 cycles and ≥12 cycles) comprise individuals meeting criteria for subfertility and infertility, respectively [ 10 ]. Possible PPD was modeled as a binary outcome with an EPDS cutoff score of greater than or equal to 13. Potential confounders were chosen a priori based on existing literature [ 6 , 34 ]. As discussed in detail above, models controlled for the following variables: age, history of anxiety, history of depression or PPD, history of subfertility/infertility, parity, preconception primary care physician visits, race/ethnicity, income, employment, use of fertility treatment to conceive, and body mass index. Women who did not complete the EPDS were excluded from the original analytic sample. We had complete data on exposure, but covariate missingness ranged from less than 1% for PCP frequency to 1.7% on early pregnancy perceived stress. Women who were missing covariate data were excluded from adjusted analysis. Given that history of depression and/or anxiety is the strongest predictor of future PPD, we stratified models by history of anxiety and/or depression/PPD prior to enrollment in the study. Models were also stratified by parity, a possible effect modifier of the association. We used causal mediation analyses to calculate the proportion of the association between TTP and PPD that was mediated by early pregnancy perceived stress as a continuous variable [ 35 ]. We estimated the direct effect of TTP on PPD (natural direct effect) and the effect that operates through early pregnancy stress (natural indirect effect). We calculated the proportion mediated as (RR nde x [RR nie −1])/(RR nde x RR nie −1). For the mediation analysis, we dichotomized TTP to correspond with the definition of infertility (≥12 cycles) and modeled early pregnancy PSS score as a continuous variable. The assumptions of causal mediation include: a) no unmeasured confounding between TTP and PPD, b) no unmeasured confounding between the mediator (i.e., PSS) and PPD, c) no unmeasured confounding between TTP and mediator (i.e., PSS), and d) no confounding between PSS and PPD that is affected by TTP. We used a causal diagram to check these assumptions.

In this prospective cohort of pregnancy planners, there was a modest dose-response association between longer TTP and PPD risk, after controlling for potential confounders. Early pregnancy PSS scores explained a small percentage (30%) of this association, suggesting that stress during pregnancy is one potential mechanism explaining the association between TTP and PPD. There has been inconsistent evidence in the literature indicating that infertility increases risk for PPD, with previous studies finding that a history of psychiatric illness was a more robust predictor of PPD than infertility [ 19 ]. In the present study, delays in conception were associated with PPD even after controlling for preconception psychiatric history. Stratified analyses examining the association between TTP and PPD among women with and without mental health histories revealed imprecise estimates that are difficult to interpret. However, the risk of PPD was much higher among women with a history of anxiety and/or depression/PPD than among those without a reported mental health history. This is not surprising given that history of mental illness is one of the strongest predictors of PPD [ 6 ] and is an important confounder, which we adjusted for in analyses. Differences in methodology (e.g., prospective design, greater control for confounding) and study populations (e.g., inclusion of women along the full fertility spectrum) could explain the discrepant findings between our study and previous work. Specifically, the extant literature on this topic has typically utilized cross-sectional designs to examine PPD among women who conceived with and without the use of ART [ 19 , 22 ]. In contrast with previous work, the current prospective cohort study followed women from the preconception period through approximately 6 months postpartum, and was, therefore, able to examine the association between longer TTP and PPD in the resultant pregnancy among a sample that primarily conceived spontaneously (i.e., 95% conceived spontaneously) rather than via ART. Population characteristics of women who utilize ART differ in several important ways from women who conceive without fertility treatment. Specifically, women who access ART tend to be older, more educated, and have higher income levels [ 36 , 37 ], all of which are associated with a lower risk of PPD [ 38 ]. In contrast, in the current study, those who took longer to conceive were more likely to identify as a person of color, have a lower household income, and be less educated, which are factors associated with increased risk for PPD [ 39 ]. Therefore, the current study is evaluating not only a different population, but a broader exposure (i.e., the full spectrum of TTP, with or without the use of ART), which may explain differences in results. Perceived stress in early pregnancy explained approximately 30% of the association between TTP and PPD. This is not surprising, as perceived stress during pregnancy has been shown to predict PPD [ 7 ], and is also bidirectionally linked to infertility [ 23 , 40 ]. Specifically, increased levels of perceived stress during preconception are associated with slightly longer TTP [ 40 ] and infertility is also associated with increased levels of perceived stress [ 23 ]. Furthermore, a greater number of in vitro fertilization cycles were associated with PPD in studies of infertile couples [ 15 ]. This provides some support for the notion that increased stress following a period of presumed infertility may be a mechanism by which TTP affects PPD. This is especially likely given that on average, women initiating infertility treatment have similar severity of depressive symptoms compared with the general population [ 41 ]. PSS may increase risk for PPD via biological pathways. For example, stress-related cortisol surges during early pregnancy increases corticotropin-releasing hormone (CRH), which may accelerate increases in CRH throughout the entire pregnancy; accelerated CRH trajectories have been associated with PPD [ 7 ]. The major strength of this study was the prospective preconception design that allowed us to examine how TTP influences subsequent PPD and to elucidate the direction of this relationship. Previous studies, which relied on cross-sectional data, were unable to evaluate temporality of these relationships. This study also assessed PPD using a psychometrically-validated self-reported measure with a well-established clinical cut-off score for possible PPD. The prevalence of PPD based on a cut-off score of greater than or equal to 13 on the EPDS in this sample was 10.6%, which is consistent with population estimates [ 1 ]. There are, however, several limitations of this study. First, we were not able to examine depressive symptoms or other mental health conditions during pregnancy as another potential mediator because these data were not collected. Given that early pregnancy stress only explained 30% of the association, there are likely other important factors involved in the pathway from TTP and PPD. Second, only women who were planning a pregnancy and able to complete a web-based survey are included in this study, which limits the generalizability of the findings. However, one benefit of restricting to pregnancy planners is that it might improve internal validity because we controlled for a strong determinant of PPD (i.e., unplanned pregnancy; [ 42 ]), a potential confounder via the study design. Although the online study design may be prone to selection bias, previous studies suggest that it may not be any more susceptible to selection bias than studies that use traditional recruitment and data collection methods [ 43 ]. Furthermore, the online design relies on self-report measurement, which may be subject to some level of response bias. However, previous research has found good agreement between internet-based surveys of mental health and clinician-administered diagnostic interviews [ 44 ]. Third, the number of women with PPD in each TTP category was small, particularly at the highest TTP category, and our results were imprecise. Although our results consistently indicate a modest positive association of TTP with PPD and a potential dose response relationship with increasing TTP, the 95% CIs were generally consistent with a null association. Therefore, we cannot rule out chance as an explanation for the observed association. Future research is needed to replicate this finding. Fourth, although we adjusted for a wide range of potential confounders, unmeasured or residual confounding (e.g., by depression during pregnancy or infant temperament) could have affected the results; however, we anticipate that the magnitude of confounding is small. Fifth, the PSS measures perceived stress. We did not ascertain information about occurrence of daily stressors (e.g., financial stress, work stress), stressful life events (e.g., divorce), stress resulting from longer TTP, or biomarkers of stress (e.g., cortisol), and it is possible results would differ using these other measures of stress, especially given that different stress measures do not always correlate with one another [ 45 ]. Finally, we only assessed PPD once at ~6 months postpartum, and therefore, could have potentially missed earlier episodes of PPD that had already resolved. In sum, this prospective cohort study found that longer pregnancy attempt times were associated with a modest increased risk for PPD, in a dose-response fashion, and that about 30% of this association was mediated through perceived stress in early pregnancy. Assessment of TTP, as well as perceived stress during early pregnancy, may be an important tool for identifying women who are at increased risk for PPD.

Postpartum depression (PPD) is one of the most common pregnancy-related conditions, affecting 10–15% of all mothers [ 1 ]. PPD is associated with impaired maternal–infant bonding [ 2 ] and an increased risk of postpartum suicide [ 3 , 4 ], and may have long-term detrimental effects on children’s cognitive, emotional, and behavioral development [ 5 ]. Therefore, understanding factors that increase risk for PPD is important to inform early intervention and prevention efforts. Several risk factors for PPD have been identified, including pre-pregnancy psychiatric illnesses, antenatal depression or anxiety, inadequate social support, exposure to traumatic events or domestic violence, perceived stress, unplanned or unwanted pregnancy, pregnancy-related anxiety, complications of pregnancy and delivery, and neonatal morbidity and mortality [ 6 , 7 ]. There has been limited research, however, on the impact of fertility on subsequent risk for PPD. Infertility is clinically defined as the inability to conceive during 12 months of unprotected intercourse for women younger than 35 and 6 months for women 35 years or older and affects up to 15% of couples in the United States [ 8 , 9 ]. Subfertility, the inability to conceive within 6 months of unprotected intercourse [ 10 ], affects an estimated 20% of couples [ 9 ]. Women with diagnosed infertility, particularly those receiving fertility treatments, have increased rates of depression in cross-sectional studies compared with women not receiving fertility treatment and not experiencing infertility [ 11 ]. Furthermore, in a recent meta-analysis, the pooled prevalence of depression among infertile women living in high-income countries (28%) was higher than the general population prevalence [ 12 ]. However, it’s unclear whether this association is due to pre-existing depressive symptoms, given that previous work has found that greater severity of depression during the preconception period is associated with longer time-to-pregnancy (TTP; [ 13 ]). Studies of the association between infertility and subsequent PPD have yielded inconsistent results. Some [ 14 - 17 ], but not all [ 18 - 22 ], studies report that use of fertility treatment, particularly the frequency of fertility treatments [ 15 ], is associated with a higher risk of PPD. While the mechanisms are unclear, perceived stress during pregnancy may be a potential mediator of the association between infertility and PPD. Infertility is associated with increased levels of perceived stress in some [ 23 ] but not all studies [ 24 ], and perceived stress during pregnancy is a strong predictor of PPD [ 7 ]. Most studies of the association between fertility and PPD have focused on women who conceived with the use of assisted reproductive technologies (ART). No study, to our knowledge, has prospectively examined the association between the full spectrum of TTP and subsequent PPD. To this end, in a preconception cohort study of pregnancy planners from North America, we aimed to examine the association between delayed conception and PPD and the extent to which perceived stress in early pregnancy mediates this association.