Sarah Saleem, Elizabeth M McClure, Shivaprasad S Goudar, Archana Patel, Fabian Esamai, Ana Garces, Elwyn Chomba, Fernando Althabe, Janet Moore, Bhalachandra Kodkany, Omrana Pasha, Jose Belizan, Albert Mayansyan, Richard J Derman, Patricia L Hibberd, Edward A Liechty, Nancy F Krebs, K Michael Hambidge, Pierre Buekens, Waldemar A Carlo, Linda L Wright, Marion Koso-Thomas, Alan H Jobe, Robert L Goldenberg & on behalf of the Global Network Maternal Newborn Health Registry Study Investigators
An estimated 340 000 maternal deaths, 2.7 million stillbirths and 3.1 million neonatal deaths occur worldwide each year – almost all in low-income countries.1–4 In some parts of sub-Saharan Africa, a woman’s lifetime risk of dying in childbirth is as high as one in seven.2 Although women may die at any time during pregnancy and up to six weeks postpartum, the majority die during the last trimester or within the first week following birth.5 Similarly, it has been estimated that, in low-income countries almost half of stillbirths occur during or around the time of delivery and nearly three quarters of neonatal deaths take place within the first few days following birth.6–9Thus, the period around delivery is thought to be the time when the woman and her fetus or infant are at the highest risk of dying.
Although the timing of the mother’s death will, in itself, have a substantial influence on the risk of a fetal or neonatal death, the conditions that cause maternal death will also contribute to the risk.4,10–19 For example, pre-eclampsia and eclampsia are important causes of maternal death and major contributors to fetal and neonatal mortality because of their association with asphyxia and preterm birth.13,18Haemorrhage and obstructed labour increase the risk of both stillbirth and early neonatal death associated with birth asphyxia.19 In addition, intrapartum stillbirth is usually attributed to obstetric conditions.14–17 Rates of maternal death, stillbirth – especially intrapartum stillbirth – and early neonatal death have all declined markedly in high- and some middle-income countries, primarily because of improved obstetric care. However, similar reductions have not yet been achieved in low-income countries, where maternal, fetal and neonatal mortality rates remain high.11,20
In low-income countries, one half to two thirds of births occur either at home or in community health clinics, often without a skilled health-care worker being present.9,21,22 In these situations, it may not always be possible to transfer a woman to an emergency obstetric care facility in time to perform a life-saving procedure should the need arise.23–25 Furthermore, as more emphasis is placed on delivery at health-care facilities and as women become more aware of the benefits, there has been an increase in the workload at referral hospitals in low-resource areas, many of which are underequipped and understaffed.25–27 Thus, even when a referral is made, the quality of care is often inadequate, especially for women who arrive late with a complication.10
However, maternal deaths are relatively rare even in areas with a high maternal mortality rate and, as a result, few studies have investigated maternal deaths and their relationship to fetal and newborn outcomes. Those that have been performed have generally been limited to hospital births and have included only a small number of deaths.27–30 Consequently, it is difficult to generalize their findings. One ecological study, which used demographic and health surveillance data to evaluate the association between childhood and maternal deaths in Bangladesh, found that the probability of survival to the age of 10 years was 24% for children whose mothers died compared with 89% for those whose mothers were still alive at their tenth birthday.28
Since there is a lack of population based studies on maternal and perinatal mortality we wanted to determine the rate and timing of maternal death among a large group of women who gave birth in the community in several low- and middle income countries and to investigate associations between maternal death and stillbirth and neonatal death.
Between 2010 to 2012, as part of a prospective, maternal newborn health study of all pregnancies, we documented maternal, fetal and neonatal deaths that occurred up to six weeks postpartum. The study was done in 106 communities at six sites in five low-income countries (Chimaltenango, Guatemala; Nagpur District and Karnataka District, India; Western Provence, Kenya; Thatta District, Pakistan; and Lusaka, Zambia) and at one site in a middle-income country (Corrientes, Argentina).9 These seven sites were selected by the Eunice Kennedy Shriver National Institute of Child Health and Human Development in the United States of America (USA), which supports the Global Network for Women’s and Children’s Health Research, to represent rural or semi-urban geographical areas served by government health services. Each site included between six and 24 distinct communities. In general, each community represented the catchment area of a primary health-care centre and, in each, 300 to 500 births took place annually. Beginning in 2009, the study investigators at each site initiated an ongoing, prospective maternal and newborn health registry of pregnant women for each community. The objective was to enrol pregnant women by 20 weeks’ gestation and to obtain data on pregnancy outcomes for all deliveries that took place in the community. Each community employed a registry administrator who identified and tracked pregnancies and their outcomes in coordination with community elders, birth attendants and other health-care workers.
All pregnant women resident in study communities were eligible for inclusion. Women were enrolled during pregnancy and data on pregnancy outcomes were collected by the trained registry administrators – usually nurses or health workers – who were supervised by study site investigators. At each site, efforts were made to verify that all pregnant women residing in the study communities were included in the registry and that data on all outcomes had been obtained. The study coordinators, who were generally nurses or physicians, monitored enrolment and follow-up to ensure that the data collected were consistent, complete and of a high quality. For hospital births, registry administrators reviewed hospital birth records routinely to identify deliveries to women from the study communities. In addition, culturally appropriate strategies were used at each site to ensure that all outcomes were reported. For example, elders or chiefs in one village used mobile phones to send text messages when women enrolled in the study gave birth. Such strategies increased the likelihood that we were able to identify all pregnancies and maternal and fetal outcomes.
Demographic and medical data were obtained for each woman by either the registry administrator or the study coordinator. All deaths that occurred during pregnancy or in the six weeks postpartum were reported using World Health Organization (WHO) classifications.31 The cause of each maternal death was assigned by the registry administrator on the basis of clinical and other information provided by the birth attendant and the woman’s family. All death reports were reviewed by the supervising physician at the study site. For deaths for which a definite cause could not be established, we undertook a secondary investigation to identify contributing factors, such as haemorrhage, pre-eclampsia, eclampsia or obstructed labour, and classified the cause accordingly. Details of this procedure have been described elsewhere.9
Study data were entered onto Microsoft Access computer files (Microsoft, Redmond, USA) at each study site and data were edited before transmission to the central data centre at RTI International in the United States, where additional data edits were performed and the data were analysed using SAS version 9.2 (SAS Institute, Cary, USA). The study findings were reported using descriptive statistics and risk ratios were calculated for maternal, fetal and neonatal outcomes. Generalized estimation equations were used to adjust for the characteristics of each site and for clustering.
The study was approved by university review boards at each local site, by partner universities in the United States and by RTI International and was registered as trial NCT01073475 at the ClinicalTrials.gov registry (United States National Library of Medicine, Bethesda, USA). All women provided informed consent before enrolment.
Between 2010 and 2012, 224 234 pregnant women were considered for enrolment in the study and 220 365 were regarded as eligible. Of these, 220 235 were enrolled since 130 refused to give consent. Data on outcomes up to six weeks after delivery were available for 214 070 of the 220 235 women (97.2%). The proportion of deliveries conducted by a physician ranged from 1.9% (503/27 072) in Kenya to 72.2% (6381/8861) in Argentina (Table 1). Deliveries by traditional birth attendants were rarely reported in Argentina, whereas 62.2% (12 276/19 724) of births in Guatemala were conducted by such attendants. The proportion of deliveries that took place in a hospital ranged from 7.7% (1609/21 008) in Zambia to 98.9% (8757/8861) in Argentina. Across all study sites, 44.0% (94 094/214 070) of deliveries took place in a hospital, compared with 30.4% (65 152/214 070) at home and 24.9% (53 233/214 070) in a health clinic. The caesarean section rate ranged from 34.4% (3048/8861) in Argentina to 1.1% (231/21 008) in Zambia