Family composition and infant and child mortality in Sri Lanka.

This study examines the effect of family composition on infant and child mortality in Sri Lanka using logit and proportional hazard models. The sex composition of the surviving siblings and the time to nearest siblings alive are hypothesized to be a significant factor, ceteris paribus, in child survival. The study uses data from the 1975 Sri Lanka World Fertility Survey. The analysis is restricted to a sample of 12,007 individual children who were born during the ten years prior to the survey (1966-1975). Mosley and Chen's scheme is used as a general conceptual framework. The analysis is carried out on the four risk periods of infant and child mortality--neonatal, postneonatal, toddler and child. The results show that the effects of family composition on infant and child mortality vary across the four risk periods. The family composition effect is highly significant in postneonatal mortality when controlling for confounding factors. A living sibling of either sex and a previous sibling death increase the risk of postneonatal mortality. In the child period a younger sibling significantly increases the risk of mortality. The risk of death decreases in the presence of an older female surviving sibling under age ten for the toddler period. This supports the hypothesis that older children play an important role in child survival by providing care for younger siblings. Sibling deaths are persistently and strongly associated with infant and child mortality in Sri Lanka. The time to nearest older surviving siblings is only significant for survival in the neonatal period. Family composition does not significantly affect different probabilities of death of males and females in any risk period except postneonatal. The importance of confounding factors differs in the four risk periods. Several biological factors such as sex of child, parity, mother's age, multiplicity of birth and preceding birth interval, and a few socioeconomic factors such as implied child wantedness status and livestock (as indicator of socioeconomic status) are highly significant in neonatal mortality. Mother's ethnicity and father's education are contributory factors for child survival. While other studies found similar results for these socioeconomic and biological variables, this study in unique in its focus on the effect of family composition in child survival in Sri Lanka.PhDDemographyPublic healthUniversity of Michiganhttp://deepblue.lib.umich.edu/bitstream/2027.42/162362/1/9001710.pd