Determinants of morbidity and mortality related to health care-associated primary bloodstream infections in neonatal intensive care units: a prospective cohort study from the SEPREVEN trial

BackgroundHealth care-associated primary bloodstream infections (BSIs), defined as not secondary to an infection at another body site, including central line-associated BSI, are a leading cause of morbidity and mortality in patients in neonatal intensive care units (NICUs). Our objective was to identify factors associated with severe morbidity and mortality after these infections in neonates in NICUs. MethodsThis ancillary study of the SEPREVEN trial included neonates hospitalized >= 2 days in one of 12 French NICUs and with >= 1 BSI during the 20-month study period. BSIs (all primary and health care-associated) were diagnosed in infants with symptoms suggestive of infection and classified prospectively as possible (one coagulase-negative staphylococci (CoNS)-growing blood culture) or proven (two same CoNS, or >= 1 recognized pathogen-growing blood culture). BSI consequences were collected prospectively as moderate morbidity (antibiotic treatment alone) or severe morbidity/mortality (life-saving procedure, permanent damage, prolonged hospitalization, and/or death). ResultsOf 557 BSIs identified in 494 patients, CoNS accounted for 378/557 (67.8%) and recognized bacterial or fungal pathogens for 179/557 (32.1%). Severe morbidity/mortality was reported in 148/557 (26.6%) BSIs. Independent factors associated with severe morbidity/mortality were corrected gestational age <28 weeks (CGA) at infection (P < .01), fetal growth restriction (FGR) (P = .04), and proven pathogen-related BSI vs. CoNS-related BSI (P < .01). There were no differences in severe morbidity and mortality between proven and possible CoNS BSIs. In possible BSI, S. epidermidis was associated with a lower risk of severe morbidity than other CoNS (P < .01), notably S. capitis and S. haemolyticus. ConclusionsIn BSIs in the NICU, severe morbidity/mortality was associated with low CGA at infection, FGR, and proven pathogen-related BSIs. When only one blood culture was positive, severe morbidity/mortality were less frequent if it grew with S. epidermidis compared to other CoNS. Further studies to help distinguish real CoNS BSIs from contaminations are needed. Study registrationClinicalTrials.gov (NCT02598609).