This thesis has three main objectives: first, to estimate age-specific levels, age patterns and period trends in early mortality at national and sub-national levels in Eritrea; second, to investigate the major possible determinants of infant and child mortality; and third to discuss implications of the main findings for health policy and future research directions.
The data used for this study come from the 1995 Eritrea Demographic and Health Survey of 5,469 households and 5,054 women in the reproductive age span.
An assessment of fertility and mortality data quality using different internal consistency checks indicated some rounding of ages at death, but no substantial omission of births or deaths, mis-dating of births, or sex differentials in reporting births or deaths. Over the ten years to 1995, national levels of infant (1q0), childhood (4q1), and under-five (5q0) mortality are estimated to be 80, 67, and 146 per 1000, respectively. Early mortality has declined substantially during the post-war period.
The influences of bio-medical, socio-economic and ecological determinants of mortality were analyzed separately for the neonatal, post-neonatal and childhood periods using logistic regression and proportional hazard models. Among the demographic features of the child and mother, multiplicity, first births of uneducated mothers, previous birth interval, and survival status of the previous child are the most powerful determinants of mortality during the neonatal period, but their effects decline as the child ages. Twin children have a significantly higher risk of death than single children during the first month of life, but the gap narrows as the child gets older. First-born children of uneducated mothers suffer higher risks of neonatal death than other children. The risk of death during the neonatal and post-neonatal periods is high among children born no more than 18 months after the previous birth, but drops after 19 months. During childhood, the risk of death drops when the previous birth interval is at least two years. Children are at a considerably higher risk of death if the previous sibling died than if it survived, but the differential decreases with age of child. Post-neonatal mortality is appreciably higher among children while their mothers are pregnant or after they give birth again than before their mothers become pregnant again. Childhood mortality rises when mothers bear the next child.
Households without piped water and toilet facilities record significantly higher mortality, particularly during the post-neonatal period when supplementary foods are introduced. Socio-economic and community (ecological) factors are important predictors of mortality differentials after the first month of life. Children from poor families and those living in the lowlands have appreciably higher risks of death than other children. Birth year of child is a significant predictor of mortality during all age periods, the risk of death being considerably lower for children born during the post-war period than those born during the war-crisis period.
In order to reduce early mortality in Eritrea, efforts should focus on health programs that target first-time mothers. Health education and family planning programs that promote longer birth intervals should be given special emphasis. The provision of clean water and toilet facilities must be given greater attention in order to increase child survival in Eritrea. Resources should be directed towards lowland communities.
Finally, our findings indicate the need for further research to identify the precise mechanisms by which altitude influences child mortality in Eritrea. Other important areas of study include breast feeding, weaning foods, feeding practises and low birthweight.
Stockholm: Stockholm University , 1999. , 277 p.