急性百草枯中毒患者并发肝损伤预测模型的构建与评估  

Development and validation of a nomogram for predicting the risk of liver injury in the patients with acute paraquat poisoning

作　　者：唐郭 蒋臻[1] 胥伶杰[1] 杨莹[1] 杨莎 姚蓉[1] Tang Guo;Jiang Zhen;Xu Ling-jie;Yang Ying;Yang Sha;Yao Rong(Emergency Medicine Laboratory and Department of Emergency,West China Hospital of Sichuan University,Sichuan University Disaster Medicine Center,Chengdu 610041,China)

机构地区：[1]四川大学华西医院急诊科、急诊医学研究室、四川大学灾难医学中心,四川成都610041

出　　处：《中国急救医学》2023年第4期253-257,共5页Chinese Journal of Critical Care Medicine

基　　金：国家自然科学基金面上项目(82072156);四川省自然科学基金面上项目(2022NSFSC0669);四川省科技厅重点研发项目(2022YFS0273);四川省卫健委重点课题(18ZD002)。

摘　　要：目的构建和评估急性百草枯中毒(APP)患者并发肝损伤的预测模型。方法回顾性分析2010年9月至2022年1月期间就诊于四川大学华西医院急诊科的APP患者,收集患者人口学特征及血清学指标结果,根据住院期间最差的肝功能指标分为肝损伤组和非肝损伤组,采用随机森林进行APP患者并发肝损伤的变量优化,应用多因素Logistic回归分析确定APP患者并发肝损伤的独立危险因素并构建列线图模型,使用C指数、校准曲线和决策曲线分析来评估预测模型的区分性、校准度和临床实用性。结果列线图中的预测因素包括年龄、就诊时血清胱抑素C和尿素氮。基于上述指标构建列线图模型,该模型的准确度与区分性较好,C指数为0.902(95%CI 0.867~0.937),校准C指数为0.894。决策曲线分析表明,该列线图模型具有良好的临床应用潜力。结论年龄、就诊时血清胱抑素C和尿素氮是APP患者发生肝损伤的独立危险因素,基于上述3项指标构建的列线图模型具有较好的预测准确性、区分性和临床实用性。Objective To construct and evaluate a predictive model for liver injury in the patients with acute paraquat poisoning(APP).Methods The study was performed from September 2010 to January 2022 in the emergency department of West China Hospital of Sichuan University.Patients who were diagnosed with APP were included in the study.Demographic data,clinical characteristics,and laboratory findings of APP patients were collected.Based on the worst test result of liver function recorded during hospitalization,the patients were classified into the normal liver function group and liver injury group.The Random Forest was carried out to select optimal predictors.Finally,a nomogram model was established by using the independent risk factors screened out by multivariate Logistic regression.The predictive accuracy of different prognostic models was compared by using a concordance index(C-index),a calibration curve,and decision curve analysis.Results Age,cystatin C and blood urea nitrogen were independent risk factors for the progression to liver injury in the patients with APP.Based on the above indicators,we established a nomogram model.Our nomogram showed good accuracy with C-index of 0.902(95%CI 0.867-0.937)and calibrated C-index of 0.894,which suggested good discriminatory ability of the risk predicting model.The decision curve analysis also showed that the nomogram has good clinical practicality.Conclusions Age,cystatin C and blood urea nitrogen are independent risk factors for liver injury in the patients with APP,and the nomogram constructed by the above three indicators has good prediction accuracy,discrimination,and clinical practicality.

关 键 词：百草枯中毒 肝损伤 预后 危险因素 列线图 年龄 胱抑素C(CysC) 尿素氮(BUN) 

分 类 号：R595.4[医药卫生—内科学] R575[医药卫生—临床医学]