Background: Early-onset sepsis (EOS) is an important cause of morbidity and mortality in neonates, and its diagnosis remains challenging. The complete blood cell count and differential have been previously evaluated as diagnostic tools for EOS in small, single-center reports. We evaluated the diagnostic accuracy of the complete blood cell count and differential in EOS in a large, multicenter population of neonates admitted to the neonatal intensive care unit.
Methods: Using a cohort of 166,092 neonates with suspected EOS with cultures admitted to 293 neonatal intensive care units, we calculated odds ratios and receiver operating characteristic curves for complete blood cell count indices and prediction of a positive culture. We determined sensitivity, specificity and likelihood ratios for various commonly used cutoff values from the complete blood cell count.
Results: Low white blood cell counts, low absolute neutrophil counts and high immature-to-total neutrophil ratios were associated with increasing odds of infection (highest odds ratios: 5.38, 6.84 and 7.97, respectively). Specificity and negative predictive values were high (73.7%-99.9% and >99.8%). However, sensitivities were low (0.3%-54.5%) for all complete blood cell count indices analyzed.
Conclusion: Low white blood cell count, absolute neutrophil count and high immature-to-total neutrophil ratio were associated with increasing odds of infection, but no complete blood cell count-derived index possesses the sensitivity to rule out reliably EOS in neonates.