Content area
Full text
An unresolved issue in research on child survival is the extent to which familial mortality risk in infancy is due to biological influences net of sociodemographic and economic factors. We examine the effect of consanguinity on early childhood mortality in an Old Order Amish settlement by using the inbreeding coefficient, an explicit measure of the degree of relatedness in one's ancestry. Inbreeding has a net positive effect on neonatal and postneonatal deaths. We find social, demographic, and population-based sociocultural explanations for this effect among the Amish population, which is known to experience certain genetically transmitted defects associated with mortality.
An unresolved issue in research on child survival is the extent to which familial mortality risk in infancy is due to biological influences net of sociodemographic and economic factors, such as household economic status, household health-related knowledge and attitudes, parental competence in child rearing, and genetic viability (Guo 1993:17; see also Curtis, Diamond, and McDonald 1993; Curtis and Steele 1994; Zenger 1993). One potentially important indicator of genetic viability is degree of consanguinity associated with rare recessive phenotypes (Castilla and Adams 1990; FreireMaia and Elisbao 1984; Thompson 1990) that affect mortality and morbidity.
We extend prior research on familial mortality risk in the first year of life by using the inbreeding coefficient to predict perinatal, neonatal, and postneonatal mortality. The inbreeding coefficient is an explicit measure of genetic viability unavailable in most data sets. We use data for a population with substantial ancestral inbreeding and include selected sociodemographic and economic measures.
UNOBSERVED MEASURES AND RECENT FAMILIAL MORTALITY RESEARCH
Research on familial mortality risk has reached mixed conclusions when examining whether unmeasured variables produce biased estimates. Curtis et al. (1993) find highly significant variation in their study of familial postneonatal mortality in Brazil, with the effects of birth interval and other mortality covariates differing little between random-effects and standard logistic models. Guo (1993), studying child mortality in several Latino communities in Guatemala, compares results from hazard models incorporating random family effects with results from a standard hazard model. He finds that genetic factors (as unobserved measures) play a relatively minor role in child survival. Guo concludes that bias from unobserved mortality determinants is likely to be small, although such factors could be of greater importance in...