Epidemiology and complications of late-onset sepsis: an Italian area-based study.

BACKGROUND: Most studies regarding late-onset sepsis (LOS) address selected populations (i.e., neonates with low birth weight or extremely preterm neonates). Studying all age groups is more suitable to assess the burden of single pathogens and their clinical relevance. METHODS: This is a retrospective regional study involving paediatric departments and NICUs in Emilia-Romagna (Italy). Regional laboratory databases were searched from 2009 to 2012. Records of infants (aged 4 to 90 days) with a positive blood or cerebrospinal fluid (CSF) culture were retrospectively reviewed and analysed according to acquisition mode (whether hospital- or community-acquired). RESULTS: During the study period, there were 146,682 live births (LBs), with 296 patients experiencing 331 episodes of LOS (incidence rate: 2.3/1000 LBs). Brain lesions upon discharge from the hospital were found in 12.3% (40/296) of cases, with death occurring in 7.1% (23/296; 0.14/1000 LBs). With respect to full-term neonates, extremely preterm or extremely low birth weight neonates had very high risk of LOS and related mortality (> 100- and > 800-fold higher respectively). Hospital-acquired LOS (n = 209) was significantly associated with very low birth weight, extremely preterm birth, pneumonia, mechanical ventilation, and death (p< 0.01). At multivariate logistic regression analysis, catecholamine support (OR = 3.2), central venous line before LOS (OR = 14.9), and meningitis (OR = 44.7) were associated with brain lesions or death in hospital-acquired LOS (area under the ROC curve 0.81, H-L p = 0.41). Commonly identified pathogens included coagulase-negative staphylococci (CoNS n = 71, 21.4%), Escherichia coli (n = 50, 15.1%), Staphylococcus aureus (n = 41, 12.4%) and Enterobacteriaceae (n = 41, 12.4%). Group B streptococcus was the predominant cause of meningitis (16 of 38 cases, 42%). Most pathogens were sensitive to first line antibiotics. CONCLUSIONS: This study provides the first Italian data regarding late-onset sepsis (LOS) in all gestational age groups. Compared to full-term neonates, very high rates of LOS and mortality occurred in neonates with a lower birth weight and gestational age. Group B streptococcus was the leading cause of meningitis. Excluding CoNS, the predominant pathogens were Escherichia coli and Staphylococcus aureus. Neonates with hospital-acquired LOS had a worse outcome. Antibiotic associations, recommended for empirical treatment of hospital- or community-acquired LOS, were adequate.

The burden of early-onset sepsis in Emilia-Romagna (Italy): a 4-year, population-based study.

OBJECTIVE: To provide the first Italian data on pathogens causing early-onset sepsis (EOS) and their antimicrobial susceptibility, after the successfully prevention of Group B streptococcus (GBS) EOS. METHODS: Retrospective area-based cohort study from Emilia-Romagna (Italy). Cases of EOS registered (from 2009 to 2012) in all gestational age neonates were reviewed. RESULTS: Live births (LB) numbered 146 682. Ninety neonates had EOS and 12 died (incidence rates of 0.61 and 0.08/1000 LB, respectively). EOS and mortality were the highest among neonates with a birth weight <1000 g (20.37/1000 LB and 8.49/1000 LB, respectively). The most common pathogens were GBS (n = 27, 0.18/1000 LB) and Escherichia coli (n = 19, 0.13/1000 LB). Most infants affected by E. coli EOS were born preterm (n = 13), had complications (n = 4) or died (n = 7). Among 90 isolates tested, only 3 were resistant to both first line empirical antibiotics. Multivariate logistic regression analysis showed that low gestational age, caesarean section and low platelet count at presentation were significantly associated with death or brain lesions (area under ROC curve = 0.939, H-L = 0.944, sensitivity 76.0%, specificity 90.7%). CONCLUSIONS: GBS slightly exceeds E. coli as a cause of EOS. However, E. coli is the prominent cause of death, complications and in most cases affects preterm neonates. Empirical antimicrobial therapy of EOS seems appropriate.