[Individual and contextual determinants of infant mortality in Brazilian state capitals: a multilevel approach]

The study sought to identify individual and contextual risk factors in healthcare and their interactions and regional differences in the determination of infant mortality in Brazilian state capitals. This was a case-control study that analyzed 7,470 infant deaths in 2012 in the 27 state capitals, recorded in the Brazilian Mortality Information System (SIM) and matched with the Brazilian Information System on Live Births (SINASC) through linkage and 24,285 controls obtained by sampling the surviving liveborn infants from 2011 to 2012 from the total of 1,424,691 births. The individual explanatory variables corresponded to information available in the SINASC database, and the contextual variable consisted of a quality index for hospital care in the 702 healthcare services where the births occurred. A multilevel logistic model was used to analyze interaction. The principal determinants of infant mortality were biological factors (low birthweight, prematurity, congenital malformations, severe/moderate asphyxia, and race/color), mediated by maternal socioeconomic factors (schooling, marital status, and occupation) and insufficiency of prenatal care. Low number of prenatal visits was a risk factor for infant mortality, independently of the service's quality, except in the state capitals in the South of Brazil. In the interaction between income and prenatal care, few prenatal visits and birth in high-income state capitals showed a higher risk when compared to births in low-income state capitals (OR = 0.68). Multilevel analysis evidenced regional inequalities in the risk models and reiterated the importance of biological determinants in the mediation of socioeconomic and healthcare factors in infant mortality.