目的 探讨血糖变异系数（CV）、新生儿急性生理学评分围生期补充Ⅱ（SNAPPE-Ⅱ）与危重新生儿预后的关系。方法 回顾性分析2018年8月至2020年9月新生儿重症监护室（NICU）住院的危重新生儿的临床资料。新生儿根据预后情况分为死亡组和存活组，比较两组间临床特征差异，采用二分类logistic回归模型分析危重新生儿预后不良的危险因素，采用受试者工作特征（ROC）曲线分析评估CV、SNAPPE-Ⅱ对于危重新生儿预后不良的诊断价值。结果 共纳入385例新生儿，男228例、女157例。中位入院年龄1.0（1.0~24.0）h，平均胎龄（35.3±4.1）周，平均体质量（2 456.6±860.3）g。低血糖患儿85例，高血糖115例。死亡组80例，入NICU至死亡中位天数5.0（2.0~10.3）d。二分类logistic回归分析显示，CV，SNAPPE-Ⅱ，胎龄，体质量，主要诊断为肺出血、新生儿败血症与危重新生儿死亡相关（P<0.05）。CV诊断危重新生儿死亡的ROC曲线下面积（AUC）为0.72，SNAPPE-Ⅱ的AUC为0.85。当CV≥0.31时，其诊断危重新生儿死亡的灵敏度为0.75，特异度为0.59；当SNAPPE-Ⅱ≥18.50时，其诊断危重新生儿死亡的灵敏度为0.68，特异度为0.93。结论 CV与SNAPPE-Ⅱ对危重新生儿的不良预后均有一定预测价值。

Objective To explore the association of coefficient of glycemic variation (CV) and score for neonatal acute physiology with perinatal extension-Ⅱ (SNAPPE-Ⅱ) with prognosis in critically ill neonates. Methods The clinical data of critically ill neonates admitted to the neonatal intensive care unit (NICU) from August 2018 to September 2020 were retrospectively analyzed. According to the prognosis, they were divided into death group and survival group, and the difference of clinical features between the two groups was compared. Binary logistic regression model was used to analyze the risk factors of poor prognosis in critically ill neonates. Receiver operating characteristic (ROC) curve analysis was used to evaluate the diagnostic value of CV and SNAPPE-Ⅱ in poor prognosis of critically ill neonates. Results A total of 385 critically ill newborns were included, including 228 boys and 157 girls. The median age of admission was 1.0 (1.0-24.0) hours, the mean gestational age was (35.3±4.1) weeks, and the mean weight was (2456.6±860.3) grams. There were 85 hypoglycemic and 115 hyperglycemic patients. There were 80 patients in the death group, and the median time from NICU admission to death were 5.0 (2.0-10.3) days. Binary logistic regression analysis showed that CV, SNAPPE-Ⅱ, gestational age, weight, the main diagnosis of pulmonary hemorrhage and neonatal sepsis were correlated with death of critically ill neonates (P<0.05). The area under curve (AUC) of CV predicting death of critically ill neonates was 0.72, and the AUC of SNAPPE-Ⅱwas 0.85. When CV≥0.31, the sensitivity and specificity for the diagnosis of critical neonatal death were 0.75 and 0.59. When SNAPPE-Ⅱ≥18.50, the sensitivity and specificity for the diagnosis of critical neonatal death were 0.68 and 0.93. Conclusions Both CV and SNAPPE-Ⅱ have certain predictive values for the poor prognosis of critically ill neonates.