Diagnostic performance of initial and serial C-reactive protein in late preterm infants with suspected late-onset sepsis in a tertiary NICU in Pakistan

Résumé fourni par la source

C-reactive protein (CRP) is widely used as adjunct biomarker in neonatal sepsis evaluations. Its reported diagnostic performance varies considerably across settings, and performance in infants born preterm remains inconsistent. We conducted a retrospective diagnostic accuracy study in the Neonatal Intensive Care Unit (NICU) of the Aga Khan University Hospital. Medical records of preterm neonates with clinical suspicion of bacterial late-onset sepsis (LOS) who underwent blood culture and CRP testing were reviewed. Only records with blood culture results and initial CRP measurements (obtained at presentation to the NICU during evaluation for suspected LOS) and serial CRP measurements (taken at 24–48 h of admission) were considered for this analysis. Sensitivity, specificity, positive/negative predictive values (PPV/NPV), likelihood ratios, and area under the curve (AUC) for CRP ≥ 10 mg/L at both timepoints were calculated. Of the 300 preterm infant records screened between January 2020 and December 2022 with clinical suspicion of LOS, 293 had blood culture, initial CRP and serial CRP recordings. Two blood culture results were positive for fungal organisms and were excluded from the analysis set. Out of 291, 105 (36.1%) infants had bacterial culture-proven sepsis. Initial CRP showed sensitivity of 97.1%(91.9%-99.4%), specificity 10.2% (6.3%-15.5%), PPV 37.9% (32.1%-44%) and NPV 86.4% (65.1%-97.1%). Serial CRP showed sensitivity 94.3% (88.0%-97.9%), specificity 23.1% (17.3%-29.8%), PPV 40.9% (34.7%-47.4%) and NPV 87.8% (75.2%-95.4%). The area under the receiver operating characteristic curve (AUROC) was 0.68 (0.61–0.74) for initial CRP and 0.67 (0.61–0.73) for serial CRP measurements. Initial and serial CRP demonstrated high sensitivity but poor specificity for bacterial culture-confirmed LOS. CRP should therefore be interpreted as an adjunct to, rather than a substitute for, clinical assessment and microbiological investigations in late preterm infants with suspected bacterial LOS.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.