Bloodstream infections in a neonatal intensive-care unit: 12 years' experience with an antibiotic control program

OBJECTIVE: To assess the prevalence of gram-positive coccal (GPC), gram-neg ative bacillary (GNB), and fungal bloodstream infections (BSIs) during a 12 -year period in which a consistent antibiotic treatment protocol was in pla ce; to evaluate the efficacy of these antibiotic policies in relation to tr eatment, to the emergence of bacterial or fungal resistance, and to the occ urrence of infection outbreaks or epidemics. STUDY DESIGN: Case series. METHODS: Demographic, clinical, and bacteriological information from 363 in fants born during 1986 through 1991 and 1992 through 1997 who developed 433 blood-culture-proven BSIs was analyzed. Early-onset BSIs were defined as t hose infections discovered within 48 hours of birth, and late-onset BSIs as those that occurred thereafter. Suspected early-onset BSIs were treated wi th ampicillin and gentamicin, and suspected late-onset BSIs with vancomycin and gentamicin. Antibiotics were changed on the basis of organism antimicr obial susceptibility. RESULTS: Early-onset BSIs were noted in 52 of 21,336 live births and 40 of 20,402 live births during 1986 through 1991 and 1992 through 1997, respecti vely. GPC (83% due to group B streptococcus [GBS]) accounted for approximat ely one half of early-onset BSI cases and GNB (68% Enterobacteriaceac) for the remainder. Early-onset GBS declined from 24 to 11 cases (P=.04) and lat e-onset BSI increased from 111 to 230 cases (P<.01) from the first to the l ast study period. Sixty-eight percent of late-onset BSIs were due to GPC (p rimarily coagulase-negative Staphylococcus), 18% to GNB, and 14% to fungus. Over the study period, Escherichia coli, Klebsiella pneumoniae, Enterobact er cloacae, and Pseudomonas aeruginosa isolated from the newborn intensive- care unit (unlike those strains from other hospital units) remained fully s usceptible to ceftazidime and gentamicin. Although the hospitalwide prevale nce of methicillin-resistant Staphylococcus aureus increased, all 17 newbor n BSI cases were due to methicillin-sensitive strains. Prevalence of methic illin-resistant coagulase-negative Staphylococcus increased, although all s trains remained vancomycin-susceptible, as did the 16 Enterococcus faecalis isolates. All fungi recovered (from 48 patients) were susceptible to ampho tericin. CONCLUSION: We observed a decrease in the prevalence of early-onset BSIs du e to GBS and an increase in late-onset BSIs due to GPC, GNB, and fungi. The combination of ampicillin and gentamicin for suspected early-onset BSIs an d vancomycin and gentamicin for late-onset BSIs has been successful for tre atment of individual patients without the occurrence of infection outbreaks or the emergence of resistance. Controlled antibiotic programs and periodi c evaluations based on individual unit and not on hospitalwide antibiograms are advisable (Infect Control Hosp Epidemiol 1999;20:242-246).