Diagnostic accuracy of complete blood cell count and neutrophil-to-lymphocyte, lymphocyte-to-monocyte, and platelet-to-lymphocyte ratios for neonatal infection

Abstract

Background: Complete blood cell (CBC) counts and neutrophil-to-lymphocyte (NLR), lymphocyte-to-monocyte (LMR), and platelet-to-lymphocyte ratios (PLR) are simple measurements that are conducted as part of routine diagnostic procedures. Objective: To determine the diagnostic importance, specificity, and sensitivity of these measurements for the diagnosis of neonatal infections and in discriminating between neonatal sepsis and various other infections. Methods: We conducted a retrospective study of data from a consecutive series of 232 neonatal patients admitted to Yildirim Beyazit University Yenimahalle Training and Research Hospital in Ankara for 2 years from 2016 to 2018. We included patients with a diagnosis of or clinically suspected infection, and healthy neonates were included as controls. Data included CBC counts, and bacterial culture results, considered the criterion standard for the diagnosis of neonatal sepsis. NLR, LMR, and PLR were calculated. We compared data using independent Student t and Mann-Whitney U tests and determined the sensitivity, specificity, and likelihood ratio (LHOR) of the characteristics for neonatal sepsis using receiver operating characteristic curve analyses. Results: We included data from 155 neonatal patients with a diagnosis or suspicion of infection and 77 healthy neonates. NLR was significantly higher in neonates with sepsis or fever due to dehydration (P < 0.001) than in neonates with other infections or healthy neonates. LMR was significantly higher in neonates with sepsis or viral infection than in those with other infections or healthy controls (P = 0.003). In neonates with early-onset sepsis (EOS), we found cut-off values of ≥4.79 [area under curve (AUC) 0.845, 95% confidence interval (CI) 0.76-0.93, LHOR 11.6, specificity 98.7%, sensitivity 15%] for NLR, ≥1.24 (AUC 0.295; CI 0.18-0.41, LHOR 1.02, specificity 2.6%, sensitivity 100%) for LMR, and ≥37.72 (AUC 0.268; CI 0.15-0.39, LHOR 0.86, specificity 7.8%, sensitivity 80%) for PLR. We found cut-off values of ≥4.94 (AUC 0.667; CI 0.56-0.77, LHOR 4.16, specificity 98.7%, sensitivity 5.4%) for NLR and ≥10.92 (AUC 0.384; CI 0.26-0.51, LHOR 6.24, specificity 98.7%, sensitivity 8.1%) for LMR in those with late-onset sepsis (LOS). Conclusions: CBCs, NLR, LMR, and PLR may be useful for the differential diagnosis of EOS and LOS, and neonates with sepsis from those with other infection. NLR may be a useful diagnostic test to identify neonatal patients with septicemia more quickly than other commonly used diagnostic tests such as blood cultures. NLR has high specificity and LHOR, but low sensitivity.