OBJECTIVE: To describe the causes of maternal death in a population-based cohort in six low- and middle-income countries using a standardised, hierarchical, algorithmic cause of death (COD) methodology.
DESIGN: A population-based, prospective observational study.
SETTING: Seven sites in six low- to middle-income countries including the Democratic Republic of the Congo (DRC), Guatemala, India (two sites), Kenya, Pakistan and Zambia.
POPULATION: All deaths among pregnant women resident in the study sites from 2014 to December 2016.
METHODS: For women who died, we used a standardised questionnaire to collect clinical data regarding maternal conditions present during pregnancy and delivery. These data were analysed using a computer-based algorithm to assign cause of maternal death based on the International Classification of Disease-Maternal Mortality system (trauma, termination of pregnancy-related, eclampsia, haemorrhage, pregnancy-related infection and medical conditions). We also compared the COD results to healthcare-provider-assigned maternal COD.
MAIN OUTCOME MEASURES: Assigned causes of maternal mortality.
RESULTS: Among 158 205 women, there were 221 maternal deaths. The most common algorithm-assigned maternal COD were obstetric haemorrhage (38.6%), pregnancy-related infection (26.4%) and pre-eclampsia/eclampsia (18.2%). Agreement between algorithm-assigned COD and COD assigned by healthcare providers ranged from 75% for haemorrhage to 25% for medical causes coincident to pregnancy.
CONCLUSIONS: The major maternal COD in the Global Network sites were haemorrhage, pregnancy-related infection and pre-eclampsia/eclampsia. This system could allow public health programmes in low- and middle-income countries to generate transparent and comparable data for maternal COD across time or regions.
TWEETABLE ABSTRACT: An algorithmic system for determining maternal cause of death in low-resource settings is described.