Child and adolescent mortality: sociodemographic variation in cause of death in Eastern and Southern Africa

Too many children and adolescents are dying globally, with particularly high numbers of deaths in Eastern and Southern Africa. To ensure policymakers can make informed choices on how to prioritise interventions and optimise health systems to reduce mortality requires data on the causes of death in children and adolescents and an understanding of how the risk of these varies across sociodemographic groups. While in most high-income countries there are robust systems to capture cause of death information on the whole population, these do not generally exist in low-income countries. In the absence of such systems, demographic surveillance sites (DSS) have been established to collect demographic data at regular intervals within geographically defined populations (e.g., on the number of births and deaths in the population) (Herbst et al, 2021). In DSS, reported deaths are typically followed up with a verbal autopsy (VA), an interview where a relative or caregiver provides information on the signs and symptoms, and the social and healthcare circumstances, preceding the death which can be used to assign the most likely cause of death (Chandramohan et al, 2021; Fottrell and Byass, 2010). Unfortunately, the accuracy of cause of death assignment using VA remains particularly poorly understood among children and adolescents. There are two main approaches to assigning cause of death based on VA information. The first method is physician review, where 2-3 physicians review the VA and assign cause of death. More recently, however, algorithms have been developed to automate the cause of death assignment (Byass et al, 2019; McCormick et al, 2016; Serina et al, 2015). Our aim is to provide population-based estimates of the causes of child and adolescent deaths (0-19 years old) in Eastern and Southern Africa, and assess how these vary by sociodemographic characteristics. To do this, we will use a rich VA dataset that has been compiled by the Analysing Longitudinal Population-based HIV/AIDS data on Africa (ALPHA) network, harmonising data from seven DSS across five countries in Eastern and Southern Africa and covering the period 1994 to 2022. The specific objectives are as follows: To assess the quality of VA data for children and adolescents. To compare methods of assigning cause of death for children and adolescents from VA data. To provide population-based estimates of the causes of child and adolescent death, and look at how these vary by key sociodemographic characteristics. To explore the social and health system circumstances surrounding child and adolescent deaths. There are several important applications anticipated. Firstly, we will provide recommendations on using algorithms to assign cause of death in children and adolescents from VA. For example, we will ascertain at what level of detail causes can be reliably assigned (e.g., where we can attribute a death to specific cause of death, such as “malaria”, and where we can we only ascertain cause of death at a broader level, such as “infectious disease”). Secondly, we will provide training to DSS teams on how to conduct these analyses, so there is a strong platform to produce similar estimates at regular intervals in the future. Finally, we will provide much needed empirical estimates of the sociodemographic and healthcare drivers and causes of child and adolescent mortality in a region lacking these data, which can be used by country policymakers as well as international agencies such as the World Health Organization.