OBJECTIVE:To estimate the risk factors for mortality during hospitalization in a Neonatal Intensive Care Unit (NICU) using a hierarchical logistic model. METHODS:This longitudinal, observational and analytical study enrolled newborns admitted to the NICU from January/2001 to December/2003. The outcome analyzed was in-hospital death and the independent variables were prenatal, perinatal and postnatal factors. A hierarchical model with three levels was built. The variables with p<0.20 by bivariated analysis were included in the model and, after adjustment at the same level, variables with p<0.05 were maintained in the logistic model. Statistic analysis was performed by SPSS.10, software that estimated the accuracy of the model, adopting significance as p<0.05. RESULTS: Among 367 newborns included, 69 (18.8%) died during hospital stay. The following risk factors comprised the hierarchical model of logistic regression and were significantly associated with death among the studied neonates: previous stillbirth, being first or second child, Apgar at five minutes below 7, preterm newborn and use of mechanical ventilation. This model had 86.9% of accuracy. CONCLUSIONS: The model obtained in this study has variables of the three hierarchical levels and might be used in Neonatal Intensive Care Units that share the same characteristic of the unit herein studied.
infant, premature; Intensive Care Units; hospital mortality; risk factors; logistic model
