A retrospective analysis of pathogen profile, antimicrobial resistance and mortality in neonatal hospital-acquired bloodstream infections from 2009–2018 at Tygerberg Hospital, South Africa

Abstract

Background Analysis of hospital-acquired bloodstream infection (HA-BSI) trends is important to monitor emerging antimicrobial resistance (AMR) threats and guide empiric antibiotic choices. Methods A retrospective 10-year review of neonatal HA-BSI was performed at Tygerberg Hospital’s neonatal unit in Cape Town, South Africa. Neonatal clinical and laboratory data from 2014 to 2018 (Period 2) was compared with published data from 2009 to 2013 (Period 1). Results The neonatal unit’s HA-BSI rate declined between periods from 3.9/1000 inpatient-days in Period 1 to 3.3/1000 inpatient-days in Period 2 (p = 0.002). Pathogen yield and blood culture contamination rate were unchanged (11.0% to 10.4%, p = 0.233; 5.1% to 5.3%, p = 0.636 respectively). Gram-negative pathogens predominated (1047/1636; 64.0%); Klebsiella species, Staphylococcus aureus, Serratia marcescens, Enterococcus species and Acinetobacter baumannii were the most frequent pathogens. Extended spectrum beta-lactamase production was observed in 319/432 (73.8%) of Klebsiella species, methicillin resistance in 171/246 (69.5%) of Staphylococcus aureus and extensive drug resistance in 115/137 (83.9%) of Acinetobacter species (2009–2018). The crude mortality rate of neonatal HA-BSI episodes increased from Period 1 to Period 2 from 139/717 (19.4%) to 179/718 (24.9%) (p = 0.014), but HA-BSI attributable mortality remained unchanged (97/139 [69.8%] vs 118/179 [65.9%], p = 0.542). The in-vitro activity of piperacillin-tazobactam and amikacin declined during Period 2 (74.6% to 61.4%; p<0.001). Conclusion Although HA-BSI rates declined in the neonatal unit, antimicrobial resistance rates in BSI pathogens remained high. Continuous BSI surveillance is a valuable tool to detect changes in pathogen and AMR profiles and inform empiric antibiotic recommendations for neonatal units in resource-limited settings.