71: Impact of a Novel Predictive Model for Early-Onset Neonatal Sepsis Evaluation

Abstract

Background: The multivariate risk model proposed by Puopolo et al (Pediatrics 2011;128(5):e11655-1163) has been shown to identify 35% fewer asymptomatic infants born at ≥ 36 weeks in a single maternity center as high risk for early onset neonatal sepsis (EOS) when compared with standard algorithms. (Mukhopadhyay et al E-PAS2013:3355.6) Objectives: To retrospectively examine the number of infants assessed and treated for EOS using local guidelines in a Canadian neonatal unit in comparison with the numbers that would have been assessed by using the quantitative multivariate risk model of Puopolo et al. Design/Methods: Retrospective cohort study of infants born at ≥34 weeks gestational age (GA) admitted to a Canadian NICU, between July 2012 and June 2013. The cohort, which includes infants admitted to the NICU for a variety of clinical and child welfare reasons, was first stratified into high- and low-risk groups according to the risk calculated by the model of Puopolo et al. This model uses GA highest maternal intrapartum temperature, duration of rupture of membranes, maternal group B Strep status as well as timing and type of intrapartum antibiotics to calculate a risk. A value of ≥0.5 per 1000 live births was used to define high risk and therefore the threshold for assessment and treatment. The numbers who were investigated (CBC, blood culture, or CSF culture) and/or treated for EOS within each group were then identified. Results: 89.5% (239 of 267) of infants were calculated as being low risk while 10.5% (28 of 267) were high risk, with the mean risk scores being 0.12±0.01 (95% CI 0.0 to 0.47) and 1.68±0.91 (95% CI 0.52 to 13.19), respectively. Within the low risk group, 130 of 239 (54%) were investigated for EOS, while 31 of 239 (13%) received antibiotics. Within the high risk group, 18 of 28 (64%) were investigated with only two of 28 (7%) also being treated for EOS. There were no differences in the mean risk scores in the high risk group between those who were investigated and those who were not. The scores of the two patients in the high risk group who received antibiotics were 13.19 and 0.94, and the mean score of those who did not was 1.26±0.39 (95% CI 0.52 to 4.47). In total, 148 of 267 (55.4%) of infants were investigated for EOS in this cohort based on clinical judgement. No cases of EOS as defined by a positive blood or CSF culture were identified during this period. Conclusions: Use of the Puopolo et al. neonatal sepsis predictive model for evaluation of EOS would have resulted in 81% fewer infants investigated. Costs associated with the investigations and treatment and potential cost savings of using this model are currently being evaluated. Prospective study is needed to further evaluate the practical impact of this predictive instrument on EOS identification and associated cost savings in Canadian NICUs.