Kordek Agnieszka, Hałasa Maciej, Podraza Wojciech
Pomeranian Medical University, Szczecin, Poland.
Clin Chem Lab Med. 2008;46(8):1143-8. doi: 10.1515/CCLM.2008.214.
The aim of this study was to test the diagnostic model of combining procalcitonin (PCT) and C-reactive protein (CRP) levels in the cord blood and routinely used biochemical parameters and clinical data in the prediction of early onset neonatal infection.
PCT and CRP concentrations were measured in cord blood of neonates with infection (group A, n=46) and compared with uninfected neonates (group B, n=240). Inclusion criteria for group A were based on obstetric history, clinical data and results of laboratory tests. Logistic regression was applied. The receiver operating characteristic (ROC) curves were constructed for PCT, CRP and the diagnostic model.
There was a highly significant (p<0.000001) difference in PCT and CRP concentrations between both groups. The cut-off point for PCT in cord blood was 1.22 ng/mL [sensitivity % (SE%) 80.43, specificity % (SP%) 71.67, positive predictive value % (PPV%) 35.24, negative predictive value % (NPV%) 95.03], and 1.0 mg/L for CRP (SE% 73.91, SP% 77.92, PPV% 39.08, NPV% 93.97). In total, seven variables were included in the model (concentrations of PCT and CRP in cord blood, tocolysis, nutritional status of the newborn, Apgar score, neutrophil ratio and red blood cell count in neonatal venous blood), which proved to offer the highest sensitivity (91.3%; 95% CI: 83-99) and specificity (90%; 95% CI: 86-94) for the detection of early onset neonatal infection. The likelihood ratio for the model was high at 9.13, with PPV% 63.64 (95% CI: 52-75), NPV% 98.18 (95% CI: 96-100) and calculated area under the curve at 0.973.
The diagnostic model based on seven clinical and laboratory parameters, using the concentration of PCT and CRP measurements in the cord blood, could be a useful tool for the prediction of early onset neonatal infection.
本研究旨在测试将脐血中降钙素原（PCT）和C反应蛋白（CRP）水平与常规生化参数及临床数据相结合的诊断模型，以预测早发型新生儿感染。
检测感染新生儿（A组，n = 46）脐血中的PCT和CRP浓度，并与未感染新生儿（B组，n = 240）进行比较。A组的纳入标准基于产科病史、临床数据和实验室检查结果。应用逻辑回归分析。构建PCT、CRP及诊断模型的受试者工作特征（ROC）曲线。
两组间PCT和CRP浓度存在高度显著差异（p < 0.000001）。脐血中PCT的截断点为1.22 ng/mL [敏感度%（SE%）80.43，特异度%（SP%）71.67，阳性预测值%（PPV%）35.24，阴性预测值%（NPV%）95.03]，CRP为1.0 mg/L（SE% 73.91，SP% 77.92，PPV% 39.08，NPV% 93.97）。该模型共纳入七个变量（脐血中PCT和CRP浓度、安胎治疗、新生儿营养状况、阿氏评分、新生儿静脉血中性粒细胞比例和红细胞计数），结果显示该模型对早发型新生儿感染的检测具有最高的敏感度（91.3%；95%可信区间：83 - 99）和特异度（90%；95%可信区间：86 - 94）。该模型的似然比高达9.13，PPV%为63.64（95%可信区间：52 - 75），NPV%为98.18（95%可信区间：96 - 100），计算得到的曲线下面积为0.973。
基于七个临床和实验室参数，利用脐血中PCT和CRP测量浓度构建的诊断模型，可能是预测早发型新生儿感染的有用工具。
