Evaluation of the Neonatal Sequential Organ Failure Assessment and Mortality Risk in Preterm Infants With Late-Onset Infection

IMPORTANCE: Infection in neonates remains a substantial problem. Advances for this population are hindered by the absence of a consensus definition for sepsis. In adults, the Sequential Organ Failure Assessment (SOFA) operationalizes mortality risk with infection and defines sepsis. The generalizability of the neonatal SOFA (nSOFA) for neonatal late-onset infection-related mortality remains unknown. OBJECTIVE: To determine the generalizability of the nSOFA for neonatal late-onset infection-related mortality across multiple sites. DESIGN, SETTING, AND PARTICIPANTS: A multicenter retrospective cohort study was conducted at 7 academic neonatal intensive care units between January 1, 2010, and December 31, 2019. Participants included 653 preterm (<33 weeks) very low-birth-weight infants. EXPOSURES: Late-onset (>72 hours of life) infection including bacteremia, fungemia, or surgical peritonitis. MAIN OUTCOMES AND MEASURES: The primary outcome was late-onset infection episode mortality. The nSOFA scores from survivors and nonsurvivors with confirmed late-onset infection were compared at 9 time points (T) preceding and following event onset. RESULTS: In the 653 infants who met inclusion criteria, median gestational age was 25.5 weeks (interquartile range, 24-27 weeks) and median birth weight was 780 g (interquartile range, 638-960 g). A total of 366 infants (56%) were male. Late-onset infection episode mortality occurred in 97 infants (15%). Area under the receiver operating characteristic curves for mortality in the total cohort ranged across study centers from 0.71 to 0.95 (T0 hours), 0.77 to 0.96 (T6 hours), and 0.78 to 0.96 (T12 hours), with utility noted at all centers and in aggregate. Using the maximum nSOFA score at T0 or T6, the area under the receiver operating characteristic curve for mortality was 0.88 (95% CI, 0.84-0.91). Analyses stratified by sex or Gram-stain identification of pathogen class or restricted to infants born at less than 25 weeks’ completed gestation did not reduce the association of the nSOFA score with infection-related mortality. CONCLUSIONS AND RELEVANCE: The nSOFA score was associated with late-onset infection mortality in preterm infants at the time of evaluation both in aggregate and in each center. These findings suggest that the nSOFA may serve as the foundation for a consensus definition of sepsis in this population.