Objective: Several studies of neonatal bacteremia have identified an increase in bacteremia due to commensal species with a concomitant decrease in bacteremia due to gram-negative rods (GNR). We observed a recent increase in the incidence of GNR bacteremia in the Newborn Special Care Unit (NBSCU) at Yale-New Haven Children's Hospital (YNHCH). This study was aimed at defining the changing incidence and epidemiology of GNR bacteremia in NBSCU patients.

Methods: We retrospectively reviewed the course, treatment, and outcome of all infants with GNR bacteremia in the NBSCU at YNHCH over a 10-year period from Jan. 1, 1987, to Dec. 31, 1996.

Results: There were 111 GNR isolates from 104 episodes of GNR bacteremia. The mean number of episodes increased significantly when comparing the periods 1994-96 and 1987-93 (10.4 vs.5.3 cases per 1000 admissions; p=0.03). There were significant increases in bacteremia due to P. aeruginosa, Enterobacter spp., and S. marescens. The mothers of infants from 1994-96 were more likely to have received antibiotics in labor than the mothers of infants from 1987-93 (31.1%vs. 11.9%; p=0.016); however, there was no difference in the incidence of clinical choramnionitis. There were no significant differences between 1994-96 and 1987-93 in mean gestational age (29.3 weeks vs. 30.2 weeks) or mean birth weight (1421g vs. 1584g). Infants from 1994-96 had a later mean age at onset of GNR bacteremia (24.8vs. 21.1 days; p=0.05) and a longer mean hospital stay(87.3 days vs. 62.4 days; p=0.003). Infants in 1994-96 were also more likely to be receiving mechanical ventilation (73.3vs. 52.5%; p=0.03) or corticosteroid therapy (28.9vs. 6.8%; p=0.006) at or in the 7 days prior to diagnosis. Infants from 1994-96 were more likely to have a central venous catheter at diagnosis (60.0 vs. 25.4%; p=0.0004). Infants from 1994-96 were more likely to have an indwelling chest tube at diagnosis (8.9%vs. 0%; p=0.03) or to have received dopamine at any time prior to diagnosis (48.9% vs. 20.3%; p=0.002) than infants from 1987-93. There was no difference in the incidence of necrotizing enterocolitis between the two study periods. The difference in deaths related to GNR bacteremia between 1994-96 and 1987-93 was not statistically significant (28.9% vs. 16.9%; p=NS).

Conclusion: Recent advances in neonatal intensive care practices may have contributed to the changing incidence and epidemiology of neonatal GNR bacteremia. Despite technological improvements in neonatal care, mortality due to GNR bacteremia is unchanged over the past decade.