This study reports, for the first time, breastfeeding outcomes in a cohort of women experiencing a PPH. Despite experiencing a significant, and in some cases life threatening blood loss, participants in this study achieved remarkably good rates of both initiation and duration of full breastfeeding. In the first postpartum week, 63% were fully breastfeeding their babies and by four months postpartum 45% were still doing so. Further, maternal ratings of infant health at both two and four months were high, with over 95% rating their infant's health as excellent or good at both time points.
The population from which our sample was drawn has a high rate of initiation of full breastfeeding and indeed over 85% of the women in our sample expected to fully breastfeed their babies from birth. While in this cohort only 63% were able to achieve this goal, in general, women may not always achieve their desired goals in relation to breastfeeding. For example, in a UK study only 75% of women expressing a desire to breastfeed actually initiated breastfeeding after delivery  and in a population of women in the United States, 61% expected to exclusively breastfeed whereas 51% actually did so . Numerous factors may impact on the ability to successfully establish full breastfeeding despite antenatal intentions. For example, women who have caesarean deliveries or other interventions such as epidural anaesthesia, as well as those whose infants are admitted to a special care nursery, are less likely to be fully breastfeeding at discharge [17, 25, 26]. Fetal stress, birth trauma and maternal stress are also known to contribute to delayed onset of copious milk secretion [12, 27]. Thus, for women experiencing a PPH there is the potential for multiple factors/pathways to impact on breastfeeding success: vascular insult to the pituitary gland as a direct consequence of the PPH (in the most extreme cases, Sheehan's syndrome); interventions and stress during labour and delivery; delay in breastfeeding initiation due to separation from the infant; and, further, maternal exhaustion may result in the infant receiving early complementary feeds and thereby subsequent difficulties for the mother in establishing full breastfeeding.
In our cohort of women with significant PPH we have documented delays in breastfeeding initiation. Just under half of the mothers who intended to breastfeed were unable to initiate breastfeeding within the recommended time period of one hour after birth, and delays in initiation were more common among women with higher estimated postpartum blood loss. Delayed initiation of breastfeeding is understandable in the context of a high proportion of mothers possibly requiring transfer to theatre, care in a HDU or ICU, or of their babies requiring special care. Among women for whom initiation of breastfeeding was delayed, we found lower rates of full breastfeeding -- or, conversely, higher rates of partial breastfeeding and of formula feeding. This is of concern, as it is known from other studies that women who commence partial, rather than full breastfeeding, are at risk of early cessation of breastfeeding . In our cohort, partial breastfeeding in the first postpartum week is likely to be indicative of problems in establishing breastfeeding, including delayed lactogenesis Stage II.
We were also able to examine the association between severity of blood loss and breastfeeding in the first postpartum week. We found that estimated blood loss was negatively associated with full breastfeeding in the first postpartum week (p for trend 0.01). After adjustment for method of birth and timing of the first opportunity to suckle, this association almost reached statistical significance using p value of ≤ 0.05. Further exploration of this association in a larger study is warranted.
We also report for the first time data on duration of breastfeeding to four months postpartum among women experiencing a PPH. Many factors are known to influence the duration of breastfeeding  and this may include experiencing a traumatic stressor, such as a PPH . In our cohort of women with significant PPH, full breastfeeding rates at two months postpartum and at four months postpartum were 58% and 45% respectively. Rates of full breastfeeding in the first postpartum week and at both two and four months postpartum were lowest among the subgroup of women with the highest estimated postpartum blood loss. Overall, for women with significant PPH, the rates of full breastfeeding fall well short of the WHO recommendation of 100% full breastfeeding for six months, although current data indicate that this is rarely achieved in any populations [21, 30, 31]. The rates observed do however compare quite favourably with other published full breastfeeding rates for general populations of Australian women (87% in first week and 57% at four months)  and are substantially higher than, for example, US general population data (51% at one week and 22% at three months) . Our observation that among women with 2000-2999 mL estimated blood loss there was some recovery in terms of reversion from partial to full breastfeeding by two months postpartum, is encouraging. This suggests that that even if full breastfeeding cannot be established immediately, there is the prospect of doing so later, and offers potential for interventions to support and encourage women to continue breastfeeding following a significant PPH despite early difficulties.
This study was not designed to examine a possible effect of receiving a transfusion of blood or blood products on breastfeeding success. Overall, lower rates of full breastfeeding were actually observed among women who received transfusion compared with those who did not. This is most likely a reflection of the fact that women receiving a transfusion were less well than those who did not, thereby confounding any association between transfusion and initiation of full breastfeeding.
Consistent with the quantitative results, our qualitative data indicate that difficulties with breastfeeding may be due to delayed lactogenesis Stage II in this population, with women reporting delays in onset of milk secretion. In addition, early separation from their baby, their stressful birth experience, ongoing fatigue and the physical sequelae of PPH were all cited by women as factors influencing their ability to successfully breastfeed. The qualitative data are also consistent with the concept that inability to successfully breastfeed is not benign and has emotional sequelae including disappointment, loss, regret and sense of failure . Women's accounts of their breastfeeding experiences also highlight the crucial role of health care providers in supporting women to breastfeed, in particular, providing them with adequate information, reassurance and practical advice.
This is a descriptive study and as such subject to a number of limitations. Firstly, it was not designed to establish causal associations between the PPH and breastfeeding outcomes. Also, while clinically significant Sheehan syndrome is now very uncommon , we did not assess signs of pituitary failure which might contribute to maternal lactation failure. Nonetheless, in the absence of other studies to date we are able to provide descriptive outcome data which may prompt further studies to examine specific associations and causal pathways. Secondly, the information regarding prenatal intention to breastfeed is subject to recall bias as it was asked in the postnatal period -- albeit in the week immediately postpartum. Thirdly, our outcome data are limited to four months postpartum and it would be important in future studies to extend the follow-up period to at least six months -- the recommended duration of full breastfeeding. In addition, it would be valuable to include more detailed assessment of maternal and infant experiences of breastfeeding and include objective measurements of milk production, proportion of daily fluid intake from breast milk and of infant growth patterns in the first few days and then over time. Fourthly, there are potential issues in relation to selection of our sample. It is possible that we may have recruited a 'healthy' cohort with possible bias towards women who were less severely affected by their birth experiences. Further, recruitment took place across 17 different sites and while ideally we would have recruited in direct proportion to the number of women in each site meeting the eligibility criteria, in fact there was some bias towards those hospitals able to commit time and resources to the study. Lastly, we cannot ignore a possible Hawthorne effect, whereby practices within participating hospitals may have altered as a result of being under study, possibly in the direction of providing greater support and attention to women experiencing a PPH perhaps thereby positively influencing breastfeeding rates.
In relation to women's written accounts of their breastfeeding experiences, we did not specifically set out to examine this in-depth, rather, their comments are serendipitous. Nonetheless, in the absence of other published information we consider them worthy of reporting, with the acknowledgment that of the 206 participants only 39 wrote about breastfeeding-related issues and that responses are most likely from women who had problems or negative experiences. Future qualitative studies might explore breastfeeding experiences in greater depth and focus more specifically on those factors associated with both negative and positive breastfeeding experiences among women following a PPH.