By preventing and treating PPH, we hope to avoid maternal deaths from PPH. To achieve this, the WHO has developed a set of recommendations for the prevention and treatment of PPH, focusing on medical and surgical interventions, and their implementation. Based on the WOMAN trial results , TXA was added to these recommendations .
With 20,060 patients and 483 maternal deaths, ours is one of the largest studies of women with PPH ever conducted. Because we conducted the WOMAN trial in countries where the burden of maternal death from PPH is highest, our findings are directly relevant to these countries. A potential limitation to our study is that we obtained narratives on 52% of the maternal deaths in the WOMAN trial. We selected the narratives shown because they provided context for the quantitative data about the women who died. The full set of narratives is available for review in the supplement (Table S1). Another limitation is that nearly half of the narratives (49%) came from one African country, which may be due to the high number of women recruited to the trial in that country. As a result, we cannot make clear inferences on the state of the health care in that country compared to others because the number of women recruited to the trial varied greatly among countries. Similarly, because only one high income country participated to the trial and the trial was not meant to compare health facilities in high income countries vs. low- and middle-income ones, we cannot draw conclusions on the effect of quality of care on mortality.
Although the risk of death was higher for women who delivered outside of hospital, we cannot draw inferences about the relative safety of home versus hospital delivery on the basis of this study. Another problem is the lack of records on the amount of blood that patients lost before the hospital admission, and that may have caused life-threatening consequences. The women who were transferred to hospital for urgent management would be the most severely bleeding patients and would not be a representative sample of all women who develop PPH. As a result of this selection bias we should not make any causal inference. In some low and middle-income countries, about 40% of women deliver at home  with only rudimentary transport in the event of an obstetric emergency. Although health workers attend most births, most cannot give intravenous drugs. Transport to hospital can take hours, and many women exsanguinate before arrival. Our data show that many women arrived in a critical condition and died soon after arrival. Early intravenous TXA reduces death from bleeding, but this is not an option for many thousands of women who deliver outside healthcare facilities. Intramuscular TXA has the potential to increase timely access to this life saving drug. Most health workers can give IM oxytocin and could give IM TXA if shown to be effective. The World Health Organization (WHO) has recommended that research into other routes of administration of TXA is a priority , and this research is in progress.
Many women do not deliver in hospitals because of costs, education or the perception or accessibility of health services [4, 11,12,13,14,15,16]. Studies on the factors that affect maternal mortality extensively used the three delays model examining barriers on the demand side (phases 1 and 2 delay) and on the supply side (phase 3 delay)  to understand and intervene on the barriers to prevent maternal death [4, 11,12,13,14,15,16].
According to the records, bleeding was the leading cause of death. Most women in our study received treatments aimed to prevent and stop blood loss, and, importantly, they also received blood transfusions (92.1%), which should help preventing death in case of major PPH . However, some of the narratives revealed that sometimes the amount of blood transfused might not have been sufficient, or the required blood type was not available, or that the patient or her family could not afford to buy more blood units. Hence, in developing countries availability of blood for transfusion may be a factor that leads to maternal deaths. This agrees with other studies showing that poverty, lack of donors and logistic problems contribute to scarcity of safe blood in low- and middle-income countries [19,20,21,22,23,24,25]. However, our study cannot confirm that deficit of blood transfusion was a cause of death, nor that the transfusion was administered too late. The actual impact of safe blood availability on patients’ health, and how to address blood shortage should be the subject of further research in developing countries . Research and interventions should target diverse areas such as increasing information to donors on safety of donations, the possibility of blood donations before delivery rather than after, financing steady blood availability within countries.
A little more than a quarter (28.4%) of the 483 women died of causes other than bleeding. Most of the women who died more than 24 h after giving birth (60.2%) died of causes other than bleeding, whereas among the women who died within 24 h from giving birth 85% of them died because of bleeding. This suggests a temporal shift in the causes of death, with bleeding being more important early on. The narratives show that these women had concurrent health complications that often lead to organ or respiratory failures, or death due to sepsis. In one narrative, sepsis, which caused the death of 5% of the 483 women, lead to the death of the patient because she could not buy the necessary antibiotics. In another case, the patient died because of she received a transfusion with the wrong blood type. To prevent transfusion errors, which result in harmful consequences to patients, waste of blood and money, new low-cost methods should be researched and developed to improve areas like patient identification, sample labelling or obtaining the correct blood type [24, 26,27,28].
The number of stillbirths was higher for the women who died than the women who lived. Based on our data, it is not possible to conclude the causal direction between stillbirths and the mothers’ complications and deaths. The issue around the causes of stillbirths is a problem at a worldwide level because of the poor quality of information, and the classification of causes is inconsistent between low-, middle-, and high-income countries [29, 30].
Several narratives highlighted the contribution of anaemia to maternal death. Although we did not measure haemoglobin on all women, a sub-study conducted at University College Hospital, Ibadan (Nigeria) [31,32,33], showed that most (88%) women were anaemic (haemoglobin < 110 g/L) and 40% were severely anaemic (haemoglobin < 70 g/L) at the time of PPH onset . Women with such low haemoglobin levels have little physiological reserve and even mild to moderate bleeding can have serious consequences. Anaemia is caused by iron deficiency, micronutrients deficiency, infections, and disorders of haemoglobin synthesis, and is highly prevalent in Africa and South-East Asia . Maternal anaemia increases the risk of PPH and the likelihood of severe morbidity or death should PPH occur [35, 36]. Blood shortages would exacerbate the risk from maternal anaemia. Several of the narratives highlighted the absence of blood for transfusion due to availability or cost. Poverty, lack of donors and logistic problems contribute importantly to blood shortages [19,20,21,22,23,24,25]. For women with severe anaemia, treatment of an established PPH is often too late to prevent death and effective prevention is needed. The WOMAN-2 trial will evaluate the effect of TXA for the prevention of PPH in high-risk anaemic women .
Although women who delivered outside of the hospital had a substantially higher case fatality, the case fatality of women who delivered in hospital was 1.9%. The data presented here highlight some of the factors that may account for this high hospital case fatality rate including the lack of timely intervention, inadequate infrastructure, maternal anaemia with limited access to safe blood transfusion and other important maternal co-morbidities that increase the risk of death. Whilst it seems unlikely that we will achieve maternal death rates comparable to those in high-income settings without addressing these structural issues, it is important to note that early administration of TXA reduced the risk of death due to bleeding by almost one third despite these constraints. Indeed, further analyses of the WOMAN trial data suggest that many women were so critically ill at the time of randomisation that their death was imminent and inevitable regardless of treatment . This raises the possibility that if women are treated very soon after bleeding onset, as recommended by the WHO, the effect of TXA on reducing the risk of death due to bleeding may exceed that observed in the WOMAN trial.