Development and assessment of systems for population-based estimation of neonatal and perinatal mortality in The Gambia

Background: Accurate data on perinatal and neonatal mortality in sub-Saharan Africa are scarce. This thesis assesses methods for their estimation in The Gambia. Methods: Under-five mortality was estimated, comparing the results of the first Demographic Health Survey in 2013 with previous inquiries. Three studies collecting primary data were conducted to assess and develop systems to improve estimation of perinatal and neonatal mortality. First, a pregnancy history survey that employed female interviewers was conducted and the results were compared with those from routine health and demographic surveillance in Basse. A second study linked records of women visiting antenatal clinics (ANC) to the Health and Demographic Surveillance System (HDSS) and used the augmented data to re-estimate neonatal, infant mortality and stillbirth rates. Third, a qualitative study was embedded in the retrospective survey to explore reasons for under-reporting of neonatal deaths and stillbirths. Findings: The DHS estimates of mortality for the recent past were consistent with those from other surveys. Age-disaggregated estimates for 2010-11 reveal that neonatal mortality constitutes 40% of under-five deaths in The Gambia. Women reported significantly more pregnancies in the pregnancy history survey than the HDSS and, in particular, more adverse pregnancy outcomes. Thus, while collecting data prospectively in HDSS generally limits omissions of deaths and age/date reporting errors, it may yield less complete data than pregnancy histories on adverse perinatal outcomes. However, the ANC-HDSS linkage study improved the capture of pregnancies. Women whose pregnancy was only detected in the clinic were 2.5 times more likely to have had a stillbirth than women whose pregnancies were picked up by the HDSS. The qualitative study showed that the cultural practice of pregnancy concealment, reluctance to speak of the deceased and to be counted affect the accuracy of data on pregnancy outcomes and mortality in The Gambia. It also revealed the importance of identification and use of culturally-appropriate metaphors and idioms to discuss sensitive matters compared to training interviewers to ask standardised verbatim questions.