欢迎访问《中国循证儿科杂志》官方网站，今天是
English Version
中国循证儿科杂志

中国循证儿科杂志 ›› 2017, Vol. 12 ›› Issue (5): 337-341.

• 论著 • 上一篇    下一篇

危重症监护病房死亡与存活患儿血糖波动的病例对照研究

杜语慧，李静，刘成军，许峰   

  1. 重庆医科大学附属儿童医院重症医学科，儿童发育疾病研究教育部重点实验室，儿科学重庆市重点实验室，儿童发育重大疾病国家国际科技合作基地重庆，400014
  • 收稿日期:2017-10-13 修回日期:2017-10-25 出版日期:2017-10-25 发布日期:2017-10-25
  • 通讯作者: 李静

Blood glucose variability and outcomes in critically ill children

DU Yu-hui, LI Jing, LIU Cheng-jun, XU Feng   

  1. Department of Critical Care Medicine,Children's Hospital,Chongqing Medical University,Ministry of Education Key Laboratory of Child Development and Disorders,Chongqing 4000l4,China
  • Received:2017-10-13 Revised:2017-10-25 Online:2017-10-25 Published:2017-10-25
  • Contact: LI Jing

PDF

919

摘要： 摘要目的：分析不同血糖波动指标对危重患儿预后的预测价值。方法回顾性采集2016年1月1日至2016年12月31日在PICU滞留时间≥72 h患儿的血糖水平和儿童死亡风险评分Ⅲ（PRISMⅢ）中的临床指标，以入PICU后28 d转归情况分为生存组和死亡组。单因素分析血糖水平和PRISMⅢ评分临床指标与入PICU后28 d转归情况的相关性，通过ROC曲线确定入科时首次血糖、72 h平均血糖、血糖标准差、血糖变异系数、高血糖发生率、低血糖发生率、平均血糖波动幅度和血糖不稳定指数中对入PICU后28 d 转归情况预测价值最大的指标，连同单因素分析结果中有意义的指标行多因素分析。结果：生存组与死亡组24 h有创机械通气、PRISMⅢ评分、72 h平均血糖、血糖标准差、血糖变异系数、平均血糖波动幅度及72 h血糖不稳定指数差异有统计学意义（P＜0.05）。 双变量相关分析显示，72 h平均血糖、血糖标准差、平均血糖波动幅度及72 h血糖不稳定指数与PRISMⅢ评分均正相关（P＜0.01），相关系数分别为0.281、0.202、0.204和0.335。ROC曲线分析显示，PRISMⅢ评分的曲线下面积（AUC）为0.733（95%CI：0.665~0.801，P＜0.001），平均血糖、血糖标准差、血糖变异系数、平均血糖波动幅度及血糖不稳定指数AUC分别为0.624（95%CI：0.539~0.708，P=0.003）、0.654（95%CI：0.577~0.730，P＜0.001）、0.630（95%CI：0.552~0.708，P= 0.002）、0.605（95%CI：0.525~0.686，P=0.012）、0.687（95%CI：0.611~0.764，P＜0.001）。多因素Logistic回归分析显示，PRISMⅢ评分和血糖不稳定指数均是危重患儿死亡的独立危险因素。四分位数分层比较，随着血糖不稳定指数的升高，血糖不稳定指数≥3.13 mmol2·h-1·d-1的患儿病死率和PRISMⅢ评分差异有统计学意义（P＜0.05）。结论：危重患儿血糖波动与病死率密切相关，72 h血糖不稳定指数比他血糖波动指标对预后更有预测价值，将其控制在3.13 mmol2·h-1·d-1以下可能是一个相对合理范围。

Abstract: AbstractObjective：The value of different blood glycaemic variability(GV) indices to predict outcomes was studied in critically ill Children. Methods：The retrospective study was conducted in Critical Care Medicine,Children's Hospital of Chongqing Medical University from January 2016 to December 2016.The included patients were divided into death group and survival group,by comparing the baseline characteristics of study and blood glucose concentration at admission(GLUAdm), average blood glucose(GLUAve), incidence of hyperglycemia, incidence of hypoglycemia,standard deviation(GLUSD), coefficient of variation(GLUCV),mean amplitude of glycaemic excursion(MAGE), and the glycaemic lability index(GLI). It was sought whether there were statistical differences among the 2 groups, and the correlation between different blood glucose indices and PRISMⅢ score was analyzed. ROC(Receiver operating curve) was drawn to evaluate predictive values of different indicators.The best indicator to assess GV was looked for and then the logistic regression analyses were performed to assess the association between GV and ICU mortality.The cases were stratified according to the GLI. Results：① The length of stay,invasive mechanical ventilation, PRISMⅢ score, SD, CV, MAGE, GLI and AVE showed a significant statistical difference between survival group and death group(P＜0.05). ② 72 h SD, MAGE, GLI, AVE and PRISMⅢ score had significantly positive correlation(correlation coefficient= 0.202, 0.204, 0.335, 0.281, P＜0.05). ③ The area under the curve(AUC) of GLI(0.687, 95%CI: 0.6110.708,P=0.001) was superior to AVE(0.624, 95%CI: 0.5390.708,P=0.003),SD(0.654, 95%CI: 0.5770.730,P＜0.001), CV(0.630, 95%CI: 0.5520.708, P=0.002) and MAGE(0.605, 95%CI: 0.5250.686,P=0.012) under the determination of ROC respectivel. ④ Multivariate logistic regression analysis showed that length of stay,invasive mechanical ventilation,PRISMⅢ score and GLI were independently associated with mortality. ⑤ With the GLI increased, the 28day mortality and PRISMⅢ score between subgroup Q3(GLI=1.003.13 mmol2·h-1·d-1) and subgroup Q4(GLI=3.13~12.96 mmol2·h-1·d-1) showed statistical difference(P＜0.05), therefore, the GLI below subgroup Q3 should be controlled. Conclusion：GV was significantly associated with ICU mortality, the GLI had the greater superiority in predicting outcome ,to control the GLI below3.13 mmol2·h-1·d-1may be acceptable in critically ill children.