Aim
To evaluate and compare the performance of the early‐onset sepsis (EOS) risk calculator with procalcitonin (PCT), complete blood count (CBC) and C‐reactive protein (CRP) for predicting neonatal EOS.
Methods
This was a retrospective case–control study of neonates who were ≥34 weeks of gestation and ≤12 h of age at admission to our hospital between January 2017 and December 2018. Neonates with strictly defined EOS and those without evidence of infection were included in this study. We reviewed and collected the laboratory data and medical charts of the included neonates. The EOS risk scores for all neonates were calculated using the EOS risk calculator, and the results were analysed and compared with blood biomarkers.
Results
A total of 501 neonates, including 353 infected and 148 uninfected infants, met the inclusion criteria for the study. Comparing these predictors, PCT had the best predictive value (sensitivity: 87.5%, specificity: 95.5%), closely followed by the EOS risk calculator (sensitivity: 81.16%, specificity: 93.92%). Multivariate logistic regression found that risk scores calculated by the EOS risk calculator had strong associations with EOS as an independent risk factor (odds ratio: 57.37, P < 0.05). The combination of the EOS risk calculator, PCT, CBC and CRP could increase the predictive value of the model and reach an area under the receiver operating characteristic curve of 0.987 for predicting EOS.
Conclusions
In this pilot study, applying the EOS calculator in China, the EOS risk calculator and PCT showed good predictive value compared to CBC and CRP. Risk scores from the EOS risk calculator strongly correlated with EOS, and the EOS risk calculator offered increased predictive value when used in combination with blood biomarkers.