BACKGROUND: Bloodstream infection (BSI) is a significant cause of morbidity and death encountered in the NICU. The rates of BSIs vary significantly in NICUs across the nation. However, no attempt has been made to correlate this variation with specific infection-control practices and policies. We experienced a significant increase in BSIs in the NICU at the George Washington University Hospital and were seeking additional precautionary measures to reduce BSI rates. Our objective was to review policies and practices associated with lower infection rates nationally and to test their reproducibility in our unit. DESIGN AND METHODS: Data on BSI rates in 16 NICUs were reviewed. The BSI rate at Connecticut Children's Medical Center (CCMC) was the lowest among those reviewed. A team from George Washington University Hospital conducted a site visit to CCMC to examine their practices. Differences in the aseptic precautions used for intravenous line management were noted at CCMC, where a closed medication system is used. This system was applied at George Washington University Hospital starting January 1, 2001. Infection rates among low birth weight infants (<2500 g) at George Washington University Hospital in the period from January 1998 to December 2000 (group 1) were compared with those in the period from January 2001 to December 2003 (group 2). Comparisons between the 2 cohorts were made with Fisher's exact test, the Kruskal-Wallis test, and Student's t test. Multivariate analysis was used to control for differences in birth weight, gestational age, central line days, and ventilator days. Analyses were repeated for the subgroup of very low birth weight infants (<1500 g). RESULTS: A total of 536 inborn low birth weight infants were included in this retrospective study (group 1, N = 169 infants; group 2, N = 367). The incidence of sepsis decreased significantly from group 1 to group 2 (25.4% and 2.2%, respectively). The reduction of sepsis observed in association with the new practice was statistically significant after controlling for birth weight, central line days, and ventilator days in a multiple regression model (regression coefficient: 0.95 +/- 0.29). The odds ratio of reduction in sepsis after implementation of the new practice was 2.6 (95% confidence interval: 1.5-4.5). The central line-related BSI rate decreased from 15.17 infections per 1000 line days to 2.1 infections per 1000 line days. The study included 233 very low birth weight infants, ie, 90 in group 1 and 143 in group 2. The rate of BSIs decreased significantly from group 1 to group 2 (46.7% and 5.6%, respectively). The decrease in sepsis rate remained significant in a multiple regression model (regression coefficient: 1.42 +/- 0.35). The odds ratio of decreased sepsis in relation to the new policy application among the very low birth weight infants was 4.15 (95% confidence interval: 2.1-8.3). CONCLUSION: Applying the closed medication system was associated with reduced BSI rates in our unit. This protocol was easily reproducible in our environment and showed immediate results. Serious attempts to share data can potentially optimize outcomes and standardize policies and practices among NICUs.