Related Experiment Video
Updated: Jul 2, 2026
A Neonatal Imaging Model of Gram-Negative Bacterial Sepsis
Published on: August 12, 2020
Development and validation of a diagnostic model for late-onset neonatal sepsis using the haematological profile: a
Emily Hyde1, Janno Schouten2, Jarob Saker3
1Nuffield Department of Women's and Reproductive Health, University of Oxford, Oxford, OX3 9DU UK.
Purpose:
Neonates admitted to intensive care units are at high risk of hospital-acquired infections. While blood cultures remain the gold standard for sepsis diagnosis, they do not detect infection in every case. Host immune response markers, such as C-reactive protein and procalcitonin, can assist in earlier detection but also have limitations and carry additional costs. We have developed a Neonatal Intensive Care Infection Score (NICIS), which leverages extended complete blood count (CBC+Diff + EIP) parameters, which may require specific analyser configuration or software licensing, to extract additional diagnostic value.
Methods:
We conducted a retrospective cohort study of neonates admitted to the Neonatal Intensive Care Unit at the John Radcliffe Hospital (Oxford, UK) over one year. Eighteen CBC+Diff + EIP parameters were initially considered based on their ability to differentiate culture-positive patients from those without clinical suspicion of infection. NICIS was developed as a weighted score and validated internally using a temporally distinct dataset and externally using a four-year retrospective cohort from Erasmus University Medical Center (Rotterdam, The Netherlands).
Results:
Incorporating eight CBC+Diff + EIP parameters, NICIS values ranged from 0 to 25, with higher scores indicating a greater likelihood of sepsis. NICIS achieved an area under the curve of 0.906 in training, 0.879 in internal validation, and 0.841 in external validation, outperforming C-reactive protein and total white blood cell count in all datasets.
Conclusions:
NICIS provides an interpretable tool for sepsis detection using routinely collected CBC+Diff + EIP data without additional sampling. Its performance across internal and external cohorts suggests sufficient diagnostic accuracy, although prospective studies are warranted to further assess generalisability and clinical utility.
Supplementary Information:
The online version contains supplementary material available at 10.1007/s44253-026-00121-9.
