A prospective cause of death classification system for maternal deaths in low and middle-income countries: results from the Global Network Maternal Newborn Health Registry
Background
Maternal deaths worldwide have fallen from an estimated 532,000 in 1990 to 303,000 in 2015 representing a maternal mortality rate (MMR) for 2015 of approximately 220/100,000 live births. To reach the World Health Organization’s (WHO) Sustainable Development Goal of 70 deaths per 100,000 live births globally by 2030, low- and middle-income countries (LMIC) will have to reduce their MMR by 7.5% annually, a rate of reduction currently achieved by only Rwanda, Cambodia and Timor-Leste. Reliable cause of death (COD) is essential to strategies to avert maternal mortality. However, misclassification of maternal COD is widespread, even in countries with complete vital registration. In countries with the highest burden of maternal mortality, the lack of COD data is critical. Only one of ten countries with high MMR has published maternal COD data. The standard for assignment of maternal COD is a diagnostic autopsy. However, this procedure is rarely performed where maternal mortality is common. Alternatively, various methods determine maternal COD, ranging from multi-disciplinary investigations such as confidential enquiries in the United Kingdom to verbal autopsies to identify COD outside health facilities. There are numerous challenges in each methodology, particularly in LMIC, including identification of maternal deaths, validity of data collection instruments, time required to gather requisite data, reliability of information, capacity of those gathering data and lack of standardization of classification across methods.11 Common to all is a concern related to the expert assignment of COD, where, given the same information, various experts may select different causes. Attempts to improve attribution of maternal mortality have focused on benchmarks for certification of COD by health-care providers. However, in settings where COD is ascertained by providers untrained in clinical diagnosis, there is little standardization of methods. To improve consistency, inter-observer agreement and comparability, we developed an algorithm16 based on the International Classification of Disease – Maternal Mortality (ICD-MM) system, to assign COD. In this system, the causes include trauma, abortion-related mortality, preeclampsia/eclampsia, hemorrhage, pregnancy-related infection and medical conditions not associated with pregnancy. The purpose was to determine maternal COD within the Maternal Newborn Health (MNH) Registry, a population-based cohort of pregnant women of the National Institute of Child Health and Human Development (NICHD)’s Global Network for Women’s and Children’s Health Research (Global Network). We used an algorithmic classification system to assign COD, and
compared the algorithmic-assigned results to the clinically-assigned COD by the health-care attendant.
By: Pasha O, McClure EM, Saleem S, Patel A, Tikmani SS, Lokangaka A et al.