Background: Maternal deaths occur universally and are largely avoidable. Postpartum haemorrhage accounts for a disproportionate amount of maternal deaths. There remains a great need to expeditiously decrease the rate of postpartum haemorrhage to prevent further mortality.

Methods: This study is a cohort analysis of data collected for the pragmatic international multi-centre randomized double blind placebo controlled design WOMAN Trial. It will present a univariate analysis of patient and delivery characteristics (age, type of delivery, placenta fully delivered, primary cause of haemorrhage, severity of haemorrhage), physiologic characteristics (systolic blood pressure, estimated blood loss, clinical signs of haemodynamic instability) and management characteristics (receipt of blood products, uterotonics). Multivariable logistic regression models and likelihood ratio tests will be used to examine the evidence for interaction between death from PPH and region after adjusting for any independent effects of 1) systolic blood pressure 2)age 3) type of delivery 4) receipt of blood products Discussion: This analysis of the WOMAN trial dataset will explore the relationship between geographical location, patient and environment characteristics and outcomes of postpartum haemorrhage. A protocol and statistical analysis plan is presented here.

Global burden of disease Maternal deaths occur universally and are largely avoidable(1). According to the first publication of the Maternal Health Series published in the Lancet October 2016, approximately 300,000 maternal deaths occurred in the past year. Postpartum haemorrhage was the cause of one tenth of these deaths(2,3). While this number has decreased globally since 1990, haemorrhage still accounts for a disproportionate amount of maternal deaths. There remains a great need to expeditiously decrease the rate of postpartum haemorrhage to prevent further mortality.

Why study postpartum haemorrhage? The most severe outcome of postpartum haemorrhage (PPH) is the death of a mother. The World Health Organization defines maternal death as "death of a woman while pregnant or within 42 days of termination of pregnancy" (4). The clinical definition of postpartum haemorrhage refers to more than 500mL of blood loss after a vaginal delivery and more than 1000mL after a caesarean section(5). Haemorrhage is a rapidly fatal condition, most commonly caused by uterine atony, or the inability of the uterus to contract, constricting the network of vessels in the uterine muscle, gradually slowing active bleeding. For this reason, research efforts have been directed at minimizing or preventing uterine atony through preventative or treatment measures targeted at the third stage of labour(6,7). Anti-fibrinolytics have been proposed as an alternative method of slowing bleeding and have been studied both as a prophylactic intervention and a treatment modality(8-11). Risk of death from postpartum haemorrhage increases if the woman is anaemic as she cannot tolerate blood loss to the same extent that her counterparts with normal haemoglobin levels can(3,12). Recent publication of preliminary findings of the woman trial suggest that the anti-fibrinolytic tranexamic acid may reduce maternal mortality from PPH by over one third(13).

Why stratify by region? While the absolute number of maternal deaths around the world is impressive, quoting a global rate does not accurately reflect the burden of illness in different regions of the world. In developed countries like the United States of America, the risk of dying from postpartum haemorrhage after delivering a live baby is 13.4% (pooled, range: 4.7-34.6). This stands in stark contrast to the African experience, where studies have documented maternal death rates from PPH to be as high as 33.9% (pooled, range: 13.3-43.5)(14). There are other reasons why these numbers can vary so broadly. An obvious hypothesis is that more women die from postpartum haemorrhage in developing countries than their counterparts in developed countries because it is more prevalent, more severe or is managed differently than in other parts of the world. It has been proposed, however, that the different rates in fact reflect a paucity of reliable data that is region specific. This is indeed the case when we look at the continent of Africa, where studies done to date capture only the experiences of eight of the 54 countries on the continent (Egypt in the North, Senegal in the West, and the Democratic Republic of Congo, South Africa, Tanzania, Zambia, Zimbabwe in the East and South)(14). A third factor in estimating deaths from postpartum haemorrhage reflects the different birthing environments across regions. In some countries, up to 50% of women deliver at home with support from a traditional birth attendant and never access a medical clinic or hospital. When most women do not deliver in a hospital or clinic, they do not die in a hospital or clinic so their data goes largely unregistered(9). Despite the fact that the "poorest countries have the poorest data"(12) enough is known about postpartum haemorrhage to inform action and spur further investigation, as many organizations have done(15).

While developing countries carry a higher burden of maternal deaths from postpartum haemorrhage, the impact of maternal death from PPH does not spare the developed world(14). Haemorrhage remains a leading cause of maternal death in countries where the majority of deliveries are performed in hospital with highly skilled support(16). This may be, in part, due to the unpredictability of postpartum haemorrhage, even in low-risk women(12). Risk factors for postpartum haemorrhage are commonly accepted to include previous postpartum haemorrhage, pre-eclampsia, disorders of the placenta, induction or augmentation of labour, perineal trauma, high birthweight and retained products(7)(16). Further, reporting occurrence of postpartum haemorrhage varies due to varying definitions of blood loss, the way blood loss is measured, intra-partum management strategies including uterotonics, uterine massage, and cord traction, interventions including method of delivery as well as the underlying characteristics of the population being studied(6)(3). Much time and resources have been invested in investigating the use of uretonics such as oxytocin to treat the most common causes of postpartum haemorrhage(7).

Inclusion Criteria:

• Women aged 16 years and older
• Clinical diagnosis of postpartum haemorrhage (PPH) after vaginal delivery or caesarean section
• "clinical uncertainty" about benefit of tranexamic acid in addition to routine management of PPH

Exclusion Criteria:

- Women aged 16 years and older with a clinical diagnosis of postpartum haemorrhage with clear indication or contra-indication for tranexamic acid

Female

Eligible adult women with a clinical diagnosis of postpartum haemorrhage following vaginal or caesarean section delivery will be included in this cohort analysis. While the diagnosis was clinical, a definition of postpartum haemorrhage could be based on clinically estimated blood loss of more than 500mL after vaginal birth or 1000mL after caesarean section or any blood loss sufficient to compromise haemodynamic stability.