Nosocomial infection and death in a neonatal intensive care unit

Based on five years of prospective surveillance in a neonatal intensive care unit (NICU), the association of nosocomial infection with death during hospitalization was studied. Low birth weight and patent ductus arteriosus (PDA) were the variables most strongly associated with nosocomial infection. After stratification for these variables, there was a persistent association between nosocomial infection and increased risk of death (relative risk = 1.96; 95% confidence interval, 1.09-4.44; P = 0.03). The relative risk of mortality with nosocomial infection was significantly modified (P = 0.02) by the presence of PDA (relative risk = 3.42; 95% confidence interval, 1.68-6.95 for infants without PDA; no effect for infants with PDA). Relocation of the NICU to an improved, better-staffed facility was associated with a significant decrease in the adjusted nosocomial infection rate (relative risk [old NICU/new NICU] = 9.73; 95% confidence interval, 4.30-22.0). This improvement was accompanied by a statistically insignificant reduction in the overall mortality because other causes of death, such as low birth weight and serious underlying disease, are much more common in this population and thus are more important determinants of outcome.