Background Neonatal sepsis with bacteraemia causes high morbidity and mortality. Early empirical antibiotics are essential to reduce this. Antibiotic policies are based on local epidemiological data of pathogens and resistance profiles.

Objective Identify bacteraemia pathogens and antibiotic resistance rates including incidence of Extended Spectrum Beta Lactamase producing Enterobacteriacae, thus informing policy.

Compare pathogen profile in Medical versus Surgical admissions.

Methods Positive blood cultures from January 2006 to December 2010 were analysed. False positives were excluded using concurrent laboratory results (from date of positive culture to 10 days after) if CRP<10, WCC<22 & Platelets>100, along with clinical data from electronic database (SEND) for secondary outcomes.

Results The average positive blood cultures for all 5 years was 15.1%, of which Cogulase negative Staphylococci (clinically significant only) was 48.8%, Enterobacteriaceae was 21.8%, group B streptococcus 2.9% and the rest were S Aureus, Enterobacter and others.

The Cefotoxime resistant Enterobacteriaceae were 18.9%. Tazocin resistance was 36.7%, and Gentamycin resistance (excluding Pseudomonas) was 2.1%.

Commonest pathogens in early onset sepsis (<48 hours age) n=28 (6.6%), were GBS (35%) & coliforms (23%), and in late onset sepsis (n=392) CoNS (48%), Enterococcus (9.1%) and Klebsiella (6.8%). No candidaemia in 2009-10, despite surgical antibiotics.

There was no significant differences in pathogens between surgical (n=232) and medical (n=188) infants. In 5 years, 11 babies died during episodes of sepsis and 2 term babies had cerebral bleed, reflecting the bane of sepsis.

Conclusion Pathogens and resistance patterns were stable with a decreasing trend in Tazocin resistance. Tazocin with Gentamicin (resistance 2.1%) is a reasonable 2nd line for late onset sepsis. ESBL rate was 18.9% compared with BSPI National surveillance data (26%), probably attributable to low Cephalosporin use. Since pathogen profiles change, it is important to review epidemiology periodically.