Abstract: Background　As sepsis in children has no typical clinical manifestations in its early stage，the optimal timing of treatment is often delayed，resulting in poor prognosis and high mortality.In addition，some of its clinical symptoms are similar to those of other diseases，which undoubtedly increases the rate of misdiagnosis.Therefore，early diagnosis is a most effective way for preventing the incidence of death caused by sepsis-related complications.We aimed to explore the indicators that can predict death in children with sepsis with high accuracy.Objective　To evaluate the value of serological indicators combined with pediatric critical illness（PCIS）and Brighton Pediatric Early Warning Score（PEWS）for predicting the risk of death in children with sepsis.Methods　This nested case-control study was performed from December 2015 to December 2016.The participants were 205 children with sepsis who were diagnosed in the First Affiliated Hospital of Hebei North University.Their baseline data，serological indicators，PCIS and Brighton PEWS were collected.The definite diagnosis of sepsis/severe sepsis was defined as the starting point for participating in the study，and death or discharge was defined as the end point for terminating participation.Those who died during the study period were defined as the experimental group.In addition，whenever there was a death in the cohort，sex-，and primary site of infection-matched children who were alive in the cohort with a similar age（less than a 3-year age gap）were selected as the control at a ratio of 1∶1.The factors influencing the death of children were analyzed by multivariate logistic regression analysis.ROC curve analysis was used to analyze the value of PCT，Alb，PCIS and Brighton PEWS in predicting the death of children with sepsis.Results　According to the study design，94 children with sepsis were selected as the final participants，including 47 in the experimental group and 47 in the control group.The fatality rate was 22.9%.The experimental group had higher rates of use of antibiotics within 1 hour after admission，electrolyte disturbance，and shock and average number of MODS involved organs，average PCT，and Brighton PEWS，longer duration of mechanical ventilation，and lower average Alb and PCIS compared with the control group（P<0.05）.The results of multivariate logistic regression analysis showed that application of antibiotics within 1 hour after admission〔OR=1.795，95%CI（1.723，1.870）〕，shock〔OR=1.673，95%CI（1.299，2.154）〕，number of organs involved by MODS〔OR=2.132，95%CI（1.507，3.016）〕，PCT〔OR=1.392，95%CI（1.336，1.450）〕，PCIS〔OR=0.553，95%CI（0.477，0.641）〕and Brighton PEWS〔OR=2.536，95%CI（2.268，2.836）〕were the influencing factor of death in sepsis children（P<0.05）.The AUC of Alb combined with PCIS and Brighton PEWS in predicting the death of children with sepsis was higher than that of each of the three indicators alone（P<0.05）.Conclusion　Alb combined with PCIS and Brighton PEWS can evaluate the death in children with sepsis with high accuracy，which is expected to be used as an auxiliary index for early clinical prediction of sepsis-related death in such a population.

Key words: Sepsis, Death, Forecasting, Serological markers, Pediatric critical illness score, Brighton pediatric early warning score, Nested case-control study

摘要： 背景　脓毒症患儿在发病初期因临床表现缺乏特异性，常延误最佳治疗时机导致其预后不佳，病死率较高。此外，脓毒症患儿的一些临床症状常与其他疾病的症状相似，这无疑又增加了脓毒症的误诊率。早期确诊是改善脓毒症患儿并发症导致死亡的最有效方法，因此，本研究旨在探索一种更精确的评估脓毒症患儿死亡的指标。目的　评估血清学指标联合小儿危重病例评分（PCIS）、Brighton儿童早期预警评分（PEWS）预测脓毒症患儿死亡的价值。方法　选取2015年12月—2016年12月在河北北方学院附属第一医院确诊的脓毒症患儿205例为研究对象。收集患儿一般资料、血清学指标及PCIS、Brighton PEWS。以入院确诊脓毒症/严重脓毒症为起点，死亡或出院为终点，采用巢式病例对照研究方法，将研究期间发生死亡的患儿定义为试验组。另外，每当队列内出现1例患儿死亡时，在该队列尚未死亡的患儿中按1∶1匹配同步选取性别、年龄（<3岁）、原发感染部位等条件相近或相同的患儿作为对照组。患儿死亡影响因素分析采用多因素Logistic回归分析；采用ROC曲线分析降钙素原（PCT）、清蛋白（Alb）、PCIS、Brighton PEWS预测脓毒症患儿死亡的价值。结果　根据巢式病例对照研究设计最终共选出94例脓毒症患儿进行分析，试验组47例，对照组47例，病死率为22.9%。试验组1 h内应用抗菌药率、电解质紊乱发生率、休克发生率、MODS受累器官个数、PCT、Brighton PEWS大于对照组，机械通气时间长于对照组，Alb、PCIS小于对照组（P<0.05）。多因素Logistic回归分析结果显示，1 h内应用抗菌药〔OR=1.795，95%CI（1.723，1.870）〕、休克〔OR=1.673，95%CI（1.299，2.154）〕、MODS受累器官个数〔OR=2.132，95%CI（1.507，3.016）〕、PCT〔OR=1.392，95%CI（1.336，1.450）〕、PCIS〔OR=0.553，95%CI（0.477，0.641）〕、Brighton PEWS〔OR=2.536，95%CI（2.268，2.836）〕是脓毒症患儿死亡的影响因素（P<0.05）。Alb联合PCIS、Brighton PEWS预测脓毒症患儿死亡的AUC分别高于其单独预测脓毒症患儿死亡的AUC（P<0.05）。结论　Alb联合PCIS、Brighton PEWS对脓毒症患儿死亡有较好的预测能力，有望作为临床早期预测脓毒症患儿死亡的辅助指标之一。

关键词: 脓毒症, 死亡, 预测, 血清学指标, 小儿危重病例评分, Brighton儿童早期预警评分, 巢式病例对照研究