Social and environmental influences on child mortality in Brazil: Logistic regression analysis of data from census files

Census data were used to investigate the influences of socio-economic and environmental variables on child mortality rates in southern Brazil. By multivariate logistic regression analysis the effects of correlated factors were distinguished, after adjustment for maternal age and parity. Low family income and, to a lesser degree, low employment status of the head of the family were associated with high child mortality levels. Place of residence, education of the mother and of the head of the family, availability of piped water in the home, access to a toilet and type of housing were all associated with childhood mortality variation, even after allowing for the effects of income and employment. The contributions of the source of the water supply and type of sanitation facilities, however, were less clear and tended to become unimportant after controlling for the socioeconomic variables. There was also no apparent advantage in being covered by government health insurance.