Antenatal depressive symptoms and early initiation of breastfeeding in association with exclusive breastfeeding six weeks postpartum: a longitudinal population-based study

This study was undertaken as part of the BASIC (Biology, Affect, Stress, Imaging and Cognition) project, a population-based longitudinal study ongoing since September 2009, which included more than 4500 pregnancies and focuses on antenatal and postnatal maternal wellbeing; detailed information may be found elsewhere [16]. All pregnant women attending the routine ultrasound examination, in gestational week 17–20, at the Uppsala University Hospital, receive written information about the project and are invited to participate. Exclusion criteria at recruitment include inability to adequately communicate in Swedish, protected identity, pathologic pregnancies as diagnosed by the routine ultrasound or intrauterine demise and age < 18 years. Written consent was obtained from all the participating women. After obtaining consent, participating women are asked to complete self-administered web-based questionnaires at recruitment, at gestational week 32 and at six weeks postpartum. The BASIC project has a participation rate of 22% and the study protocol has been approved by the Regional Research and Ethics Committee of Uppsala (no. 2009/171). For the whole cohort, 81.3% of those giving consent to participate complete the questionnaire in gestational week 32 and for the whole cohort, 80.9% at 6 weeks postpartum.

In the present study, the main outcome was exclusive breastfeeding at six weeks postpartum, dichotomized into exclusive breastfeeding versus partial breastfeeding or cessation of breastfeeding at six weeks postpartum, as self-reported by the participants. Several background and antenatal sociodemographic, lifestyle and medical variables, such as age (< 25, 25–34, > 35 years), body mass index (BMI) before pregnancy (< 25, 25–29, 30–34, > 35 kg/m²), educational level (high school or lower vs college/university), smoking during pregnancy (yes vs no), medical history of depression (no history of depression vs history of depression), employment status (working/studying vs on maternity leave/sick/unemployed) were included in the first web-questionnaire, answered by the women at gestational week 17–20. Obstetric variables such as parity (primiparas vs multiparas), mode of delivery (vaginal birth, vacuum extraction, planned caesarean section, emergency caesarean section), use of local epidural anesthesia as well as obstetric complications in pregnancy (any of the following, as self-reported or reported in medical records; vaginal bleeding during pregnancy, significant Braxton-Hicks contractions, symphysiolysis, diabetes, hypothyroidism, hypertonia and preeclampsia) and postpartum complications (hemorrhage > 1000 ml, manual placental expulsion, Apgar score at 5 min < 7, admission to the neonatal unit, laceration grade III or IV) were obtained from the medical records. The questions on timing of the first breastfeeding session after birth, experience of the first breastfeeding session (very positive/positive vs negative/very negative) and use of the hands-on approach, i.e. when health care professionals force the baby to the breast by using their hands and touching the woman’s breast and the baby in order to stimulate latch-on and breastfeeding, during the first breastfeeding session (no vs yes) were answered by the women at six weeks postpartum. Depressive symptoms during pregnancy were determined by the Swedish version of the Edinburgh Postnatal Depression Scale (EPDS) at gestational week 17–20 and/or gestational week 32. In line with previous studies, a score of > 13, giving a sensitivity of 77% and a specificity of 94%, was considered indicative of presence of depressive symptoms [17]. EPDS was also used for evaluation of depressive symptoms in the postpartum periods and particularly at six weeks postpartum; a cut-off of ≥12, giving a sensitivity of 96% and a specificity of 49%, was used, as recommended for a Swedish sample [18].

Crude analyses were performed to assess the possible associations of the study variables with breastfeeding at six weeks postpartum. Odds Ratios (OR) and 95% confidence intervals (95% CI) were calculated through cross-tabulations. A multivariable logistic regression model was fitted to estimate the specific effect of breastfeeding initiation later than two hours after birth and depressive symptoms during pregnancy on exclusive breastfeeding at six weeks postpartum. There was no multicollinearity between the variables included in the model. In order to identify potential mediators and confounders, we created a conceptual directly acycled graph (DAG), based on literature data and available variables (Additional file: Figure S1). According to the DAGs, a direct effect model (model 1) was designed including age, mode of giving birth, depressive symptoms during pregnancy, educational level and parity, as well as an indirect effect model (model 2) investigating also the mediators effect, which included age, mode of giving birth, history of depression, depressive symptoms during pregnancy, educational level, parity, use of the hands-on approach during the first breastfeeding session and experience of the first breastfeeding session. The analysis was thereafter repeated stratified by the mode of giving birth. 1

To additionally explore the interplay between depressive symptoms during pregnancy and breastfeeding initiation later than two hours after birth, a composite variable was created and multivariate associations with the outcome variable were investigated. The composite variable included the following categories: (a) women with no depressive symptoms during pregnancy who initiated breastfeeding within two hours after birth (set as the reference category), (b) women with no depressive symptoms during pregnancy who initiated breastfeeding more than two hours after birth, (c) women with depressive symptoms during pregnancy who initiated breastfeeding within two hours after birth, (d) women with depressive symptoms during pregnancy who initiated breastfeeding > two hours after birth.

SPSS version 24 was used for the statistical analyses.

The current study identified a cumulative negative effect of the presence of depressive symptoms during pregnancy and a postponed first breastfeeding session on the duration of exclusive breastfeeding as assessed at six weeks postpartum. This indicates that women with depressive symptoms during pregnancy might be more vulnerable to the consequences of a postponed first breastfeeding session on breastfeeding later in the postpartum period.

Among the main strengths of this study are the large sample size and the number of studied variables on an individual level, which gave the possibility of controlling for multiple confounders in the analyses. On the negative side, it could be argued that women answered questions on breastfeeding initiation via self-report at six weeks postpartum, which poses an eventual problem of recall bias. Nevertheless, it has been shown that women a long time after giving birth, are capable of successfully recalling what happened during the birth process and the early hours postpartum [19]. Additionally, intention to breastfeed was not assessed in the questionnaires, as we assume that nearly all mothers in this Swedish setting had planned to breastfeed; indeed, all women included in the study actually initiated breastfeeding. Nevertheless, one could speculate that some women who delayed the breastfeeding initiation might have a lower commitment to breastfeeding, which could be reflected in not exclusively breastfeeding at six weeks postpartum. To be considered is also the fact that, women participating in the present study have a higher exclusive breastfeeding rate compared to the Swedish national and county-level average. Overall, the BASIC study has a participation rate of 22% pregnant women, probably due to extensive questionnaires and collection of biological material. The participating women have a slightly higher educational level than the background population in Uppsala and are slightly more often primiparas. These differences could affect the generalizability of the findings but they are not expected to greatly affect effect estimates. Further, the prevalence of perinatal depression in the BASIC study is very close to that of other studies, hence the lack of an attrition analysis should not introduce significant bias in our analyses. Lastly, the Edinburgh Postnatal Depression Scale, although widely recognized in medical research and also translated and validated in Swedish, [17] only detects depressive symptoms and does not provide a diagnosis of clinical depression.

In our sample, one fifth of the mother-infant dyads did not have the opportunity to accomplish the first breastfeeding session within two hours after birth, despite the recommendations to initiate breastfeeding during the newborns’ initial alert period during the first hours after birth. This could be due to separation of mother and newborn after a caesarean section or another complication. At the Uppsala University Hospital, all infants being born by caesarean section are to be removed from the operating theatre and separated from the mother, for a short time period. If a midwife or nurse is not available to take responsibility for the infant in the recovery ward, the infant will continue to be separated from the mother and thus the first breastfeeding session will be delayed. Women experiencing larger vaginal tears or retained placenta might also be subjected to the same clinical routine. In the univariate analyses, delayed breastfeeding (> 2 h) after birth was also associated with primiparity and the use of epidural anesthesia during labour. Given that primiparas tend to also breastfeed for a shorter period compared to multiparas [20], women with no previous breastfeeding experience might be in greater need for targeted breastfeeding support. Regarding the use of anesthetics during labour, it has been suggested to affect the newborns reflexes, making it more problematic for them to latch on, and therefore possibly complicating and postponing the first breastfeeding session [21]. Further, we noted an increased frequency of the hands-on approach when the first breastfeeding session takes place more than two hours postpartum, as has been previously described [22]. The higher rate of the hands-on approach among those women could indicate a wish to help and support mother-infant dyads to establish breastfeeding, but this model of breastfeeding support has been shown to increase negative experience of breastfeeding among women [22, 23]. Accordingly, women with a postponed breastfeeding initiation, were more likely to reported a negative first breastfeeding experience.

Despite the recommendations on exclusive breastfeeding during the first 6 months, only 77% of the participating women were breastfeeding exclusively at 6 weeks postpartum. Potential risk factors for an early discontinuation of breastfeeding or partial breastfeeding at 6 weeks postpartum were lower age, primiparity, low maternal education, obesity and depressive symptoms during pregnancy which have all been investigated and pointed out in previous research [12, 13, 20, 24, 25]. Likewise, caesarean section and other obstetric complications during labour, as well as a postponed first breastfeeding session were associated with not exclusively breastfeeding at six weeks postpartum. Women undergoing a caesarean section might experience more pain postpartum affecting breastfeeding negatively [26]. Also, women undergoing a planned caesarean section for psychosocial reasons, such as for example fear of childbirth, might be more vulnerable, possibly also having a lower breastfeeding self-efficacy [27]. Further, our results indicate that a negative first breastfeeding session and the use of the hands-on approach were associated with exclusive breastfeeding lasting less than six weeks, which is also likely linked to a low breastfeeding self-efficacy.

As shown in earlier research [12, 13, 15], women breastfeeding exclusively at six weeks had lower odds for significant depressive symptoms than women who did not. Breastfeeding seems to be associated with decreased odds of postpartum depression, whereas early breastfeeding cessation or negative early breastfeeding experience, as indicated by breastfeeding aversion or severe breastfeeding pain have been associated with a higher risk [28, 29]. This potentially protective effect of breastfeeding against depressive symptoms has been suggested to be exerted via attenuation of the cortisol response to stress, oxytocin release, improvement of the mother’s breastfeeding self-efficacy, emotional involvement and interaction with the infant [13]. Conversely, mothers with a history of depression or those experiencing postpartum depression also more often report shorter breastfeeding duration [29, 30].

Therefore, this study adds to the available evidence investigating the cumulative effect of antenatal depression and that postponed breastfeeding initiation, which also has a negative impact on breastfeeding duration [6–8]. In our study, parallelly, depression during pregnancy and late initiation of breastfeeding interacted in increasing the odds for non-exclusive breastfeeding at six weeks postpartum. Our results show that at six weeks postpartum, women with depressive symptoms during pregnancy and breastfeeding for the first time later than two hours postpartum had almost a 4-fold increase in the odds of not breastfeeding exclusively at six weeks postpartum. Women with depressive symptoms during pregnancy are a vulnerable group, and previous studies have concluded that these women tend to breastfeed for a shorter period [29, 30]. Depressive symptoms as well as the timing of the first breastfeeding session are also linked to a lower breastfeeding self-efficacy, pointing to that these women are in need of targeted encouraging breastfeeding support to enhance the chances of a longer exclusive breastfeeding duration [9, 31].

Additional file 1:Figure S1. Graphical representation as directed acycled graph (DAG) of the conceptual model designed as to determine mediators and confounders in the association of interest between breastfeeding initiation at the first 2 h and exclusive breastfeeding at 6 weeks postpartum. (DOCX 334 kb)