Neonatal sepsis of nosocomial origin: an epidemiological study from the "Grupo de Hospitales Castrillo"

Abstract

A prospective multicenter study was designed to assess the frequency, etiology, and mortality of nosocomial neonatal sepsis diagnosed between 1996 and 1997 in the neonatology services of 27 acute-care hospitals in Spain ("Grupo de Hospitales Castrillo"). Nosocomial sepsis is defined in the literature using chronological criteria (> 3-7 days of life at the onset of symptoms); accordingly, there is the possibility of including late-onset maternally acquired sepsis or of excluding early-onset nosocomial sepsis (< 3-7 days of life). For these reasons, in this study, cases of nosocomial sepsis that developed at < or = 3-7 days after birth (early onset) were also recorded and maternally acquired sepsis diagnosed beyond 3-7 days of life were excluded. Using these criteria in a total of 30,993 admissions to the neonatal units of the participating hospitals, the nosocomial sepsis rate was 2.1% with an incidence density of 0.89 per 1000 patient days. Sepsis rate was significantly more frequent among very low birth weight (VLBW) infants (15.6%) than among those weighing > or = 1500 g (1.16%) (P < 0.001). Fifty-eight percent of all isolates were Gram-positive organisms, mainly Staphylococcus epidermidis (42%). Gram-negative organisms were isolated in 29.5% of cases (Escherichia coli and Klebsiella spp. were the most commonly isolated pathogens) and fungal infections in 12%, with absolute predominance of Candida spp. The overall mortality rate was 11.8% and the following subgroups had significantly higher (P < 0.001) mortality rates: sepsis caused by Gram-negative organisms (19% vs. 5.1% in Gram-positive pathogens) and sepsis caused by Pseudomonas aeruginosa (33.3% vs. 9.4% for the total number of sepsis caused by the remaining causative pathogens). Sepsis caused by S. epidermidis showed a significantly lower mortality rate (5.5%) compared with overall sepsis for the remaining etiologies (14.2%) (P < 0.001). In VLBW infants, the mortality rate was significantly higher than in infants weighing > 1500 g (17.3% vs. 6.5%, P < 0.001).