Birth interval effects and healthy families in Brazil

Curtis, S.L., Diamond, I. and McDonald, J. (1991) Birth interval effects and healthy families in Brazil. Demographic and Health Surveys World Conference, Washington, United States. 05 - 07 Aug 1991. pp. 1207-1227 .

Abstract

The interaction between family behavior, infant mortality, and birth spacing are estimated with a model using the binomial approximation of the normal distribution and fit to data from the 1986 Brazil Demographic and Health Survey. Also explored are the determinants of high familial mortality risks. The statistical package for determining maximum likelihood was EGRET (1989). The assumption of dependence between variables is made, in contrast to other analyses using log linear, hazards, or logistic regression modeling which assume independence. There is a short discussion of the justification of the selected model. The effect of preceding birth interval on post neonatal mortality is observed. Birth histories provided data on unweighted single births born 1-10 years prior to the survey who survived to 1 month of age. Explanatory variables were the duration of the preceding birth interval and survival status, maternal age and education, birth order, sex of child, region of residence (northeast is differentiated from the rest because of higher mortality). Analysis is presented in 1) a random effects analysis with a parsimonious logistic model and corresponding logistic binomial model, 2) a distribution of family behavioral and environmental variable analysis of family mortality risks, and 3) a logistic regression model for family mortality. The results indicate that in the random effects logistic model that there is significant variation in the mortality risks of children that can be attributed to the family after controlling for birth interval, maternal education, and region of residence. There was a higher probability of death both for children with a very short preceding birth interval and for those with a preceding birth interval of >2 years. The suggestion is that family effects can cancel out the effects of short birth intervals, even though usually short intervals are associated with greater risk of death. For families experiencing the death of a child since January 1981, the differences between those having a death and those not were small. Those with a death were less likely to have a piped water supply in the kitchen, to own a television, to breastfeed surviving children for shorter duration, and have less prenatal care. The differentials for these groups were large when grouped socioeconomically. The multivariate analysis confirmed average duration of breastfeeding of surviving children and ownership of a television as significant at the 5% level. The television variable may act as an indicator of families wealth and resources, which are more important that use of health care and having a water supply and sewage disposal. Further exploration of these issues is suggested.

This record has no associated files available for download.

Contributors

Download statistics