RESUMO
BACKGROUND: Infant mortality rates vary substantially among municipalities in the State of Ceará, from 14 to 193 per 1000 live births. Identification of the determinants of these differences can be of particular importance to infant health policy and programmes in Brazil where local governments play a pivotal role in providing primary health care. METHODS: Ecological study across 140 municipalities in the State of Ceará, Brazil. RESULTS: To determine the interrelationships between potential predictors of infant mortality, we classified 11 variables into proximate determinants (adequate weight gain and exclusively breastfeeding), health services variables (prenatal care up-to-date, participation in growth monitoring, immunization up-to-date, and decentralization of health services), and socioeconomic factors (female literacy rate, household income, adequate water supply, adequate sanitation, and per capita gross municipality product), and included the variables in each group simultaneously in linear regression models. In these analyses, only one of the proximate determinants (exclusively breastfeeding (inversely), R2 = 9.3) and one of the health services variables (prenatal care up-to-date (inversely), R2 = 22.8) remained significantly associated with infant mortality. In contrast, female literacy rate (inversely), household income (directly) and per capita GMP (inversely) were independently associated with the infant mortality rate (for the model including the three variables R2 = 25.2). Finally, we considered simultaneously the variables from each group, and selected a model that explained 41% of the variation in infant mortality rates between municipalities. The paradoxical direct association between household income and infant mortality was present only in models including female illiteracy rate, and suggests that among these municipalities, increases in income unaccompanied by improvements in female education may not substantially reduce infant mortality. The lack of independent associations between inadequate sanitation and infant mortality rates may be due to the uniformly poor level of this indicator across municipalities and provides no evidence against its critical role in child survival. CONCLUSIONS: These results suggest that promotion of exclusive breastfeeding and increased prenatal care utilization, as well as investments in female education would have substantial positive effects in further reducing infant mortality rates in the State of Ceará.