Intercenter variability and time of onset: two crucial issues in the analysis of risk factors for nosocomial sepsis.

BACKGROUND: Nosocomial sepsis is one of the most important causes of morbidity and mortality in neonatal intensive care units (NICUs).
OBJECTIVE: To assess the impact of clinical conditions, exposure to invasive procedures and NICU characteristics on late (3 to 10 days) nosocomial sepsis (LNS) and very late (>10 days) nosocomial sepsis (VLNS) and to describe the variability among NICUs.
METHODS: Multicenter prospective study in 21 NICUs including all newborns admitted in a 18-month period, weighing </=1750 g at birth or suffering major illness. Infections were diagnosed according to Centers for Disease Control and Prevention clinical criteria. Kaplan-Meyer curves and Cox models were used to identify major determinants of LNS and VLNS and its intercenter variability.
RESULTS: Of the 2160 neonates admitted to the 21 NICUs, 196 neonates developed a LNS, and 137 developed a VLNS. Some selected neonatal characteristics were independently associated with LNS, but not with VLNS (i.e. respiratory distress syndrome or patent ductus arteriosus), and the opposite was true for other conditions (i.e., necrotizing enterocolitis or high maximum base excess at admission). Exposure to mechanical ventilation and central venous catheter increased the risk of LNS and VLNS. Being admitted to specific NICUs independently increased the risk of both LNS and VLNS.
CONCLUSIONS: To analyze risk factors correctly for hospital-acquired infections in multicenter studies in NICUs, Cox modeling and stratification by time of infection onset should be used, and the existing intercenter variability should be taken into account.