Tracking mortality changes from health facility data in Sub-Saharan Africa: how do hospital deaths differ from home deaths in Antananarivo (Madagascar)?

Yempabou Bruno Lankoande [Institut National d'Etudes Démographiques (INED)/Institut Supérieur des Sciences de la Population (ISSP)] Masquelier, Bruno [UCL] et al.

All-cause and cause-specific mortality rates in sub-Saharan Africa remain poorly estimated due to the incompleteness of death registration systems and a high percentage of home deaths. Even though health facility data are increasingly used to monitor changes in disease patterns, they are not representative of the general population. We take advantage of a death notification system that was set up in Antananarivo (Madagascar) to document over time (1976-2015) the hidden causes of deaths that are not usually reported in health facility data. In Antananarivo, age at death has increased and the overall proportion of death occurring in health facilities has declined over time. However, these changes have been uneven across causes of death and the concordance between cause-specific mortality fractions inferred from deaths that occurred in health facilities and those based on all deaths is gradually increasing. Yet some deaths remain consistently under-represented, especially those due to respiratory diseases.