脓毒症性凝血病患者预后预测模型的建立  

Development of prognostic model for sepsis-induced coagulopathy patients

作　　者：何习斯 邹志业 文莎 陈燕宏 孟新科[1] He Xi-si;Zou Zhi-ye;Wen Sha;Chen Yan-hong;Meng Xin-ke(Department of Critical Care Medicine,Shenzhen Second People′s Hospital,Shenzhen 518035,China)

机构地区：[1]深圳市第二人民医院重症医学科,广东深圳518035

出　　处：《中国急救医学》2022年第8期701-706,共6页Chinese Journal of Critical Care Medicine

基　　金：广东省高水平临床重点专科(深圳市配套建设经费资助)(SZGSP006);深圳市第二人民医院高水平医院医疗质控与提升项目(202104029)。

摘　　要：目的 探讨重症监护病房(ICU)内影响脓毒症性凝血病(SIC)患者预后的危险因素,建立SIC预后预测模型。方法 从美国重症监护医学信息数据库-Ⅲ(MIMIC-Ⅲ)筛选年龄≥18岁、入院诊断SIC并且首次入ICU的患者。根据患者28 d预后分成存活组和死亡组,分析两组患者一般资料、合并症、入ICU首个24 h内的实验室指标、干预措施及病情严重程度等。采用多因素Logistics回归分析确定影响SIC患者28 d预后的危险因素并建立预测模型,用受试者工作特征(ROC)曲线和校准曲线评价列线图模型区分度和校准度,建立决策曲线评估模型的临床实际应用价值。结果 最终筛选出6347例SIC患者,其中存活组5396例,死亡组951例,总体病死率为14.98%。多因素Logistics回归分析结果显示,重度SIC[优势比(OR)=1.341,95%可信区间(CI)1.144~1.572,P<0.001]、年龄(OR=1.028,95%CI 1.023~1.034,P<0.001)、序贯器官衰竭评分(SOFA)(OR=1.228,95%CI 1.199~1.257,P<0.001)、血白细胞计数最大值(WBCmax)(OR=1.009,95%CI 1.002~1.016,P=0.012)是SIC患者28 d死亡的独立危险因素。预测模型方程:-5.031+0.332×重度SIC+0.025×年龄+0.204×SOFA评分+0.009×WBCmax。使用Bootstrap法内部重复抽样1000次进行验证,校正曲线和理想曲线基本拟合,预测值和实际值一致性较好。Hosmer-Lemeshow检验显示,预测模型有较好的校准能力(P=0.075>0.05)。决策曲线显示,列线图模型在高风险阈值范围(0.3~0.8)有一定临床实用性。ROC曲线分析显示,SIC患者预后预测模型的ROC曲线下面积(AUC)为0.703(95%CI 0.685~0.722,P<0.001),当模型截断值为0.170时,敏感度为55.7%,特异度为73.9%。结论 重度SIC患者、年龄、SOFA评分、WBCmax是影响SIC患者预后的独立危险因素,基于MIMIC-Ⅲ建立的预测模型有较好的临床预测价值,对患者预后风险评估和治疗有重要意义。Objective To investigate the risk factors affecting the prognosis of patients with sepsis-induced coagulopathy(SIC) in the intensive care unit(ICU),and to establish a predictive model for the prognosis of SIC.Methods Patients(age≥18 years old) with SIC,admitted to the ICU for the first time were screened from the Medical Information Mart for Intensive Care-Ⅲ(MIMIC-Ⅲ).According to the 28-day outcome,they were divided into survival group and death group.The general data,comorbidities,laboratory examinations within 24 h of ICU admission,intervention measures,and the severity of illness were analyzed.Multivariate Logistic regression analysis was used to determine the risk factors of SIC patients and a prediction model was established.Receiver operating characteristic(ROC) curve and calibration curve were used to evaluate the differentiation and calibration degree of the nomogram model and the clinical decision curve was established to evaluate its clinical application value.Results A total of 6347 SIC patients were screened out,including 5396 in survival group and 951 in death group,with the mortality of 14.98%.Multivariate Logistic regression analysis showed that severe SIC [odds ratio(OR)=1.341,95% confidence interval(95%CI) 1.144-1.572,P0.05).The decision curve showed that the nomogram model had a certain clinical practicability in the high risk threshold range(0.3-0.8).ROC curve analysis showed that the area under ROC curve(AUC) of prediction model for SIC prognosis was 0.703(95%CI 0.685-0.722,P<0.001).When the cut-off value of the model was 0.170,the sensitivity was 55.7% and the specificity was 73.9%.Conclusions Severe SIC,age,SOFA score and WBCmax were independent risk factors affecting the prognosis of patients with SIC.The prediction model based on MIMIC-Ⅲ had a good clinical prediction value and had a great significance in prognostic risk assessment and treatment.

关 键 词：脓毒症性凝血病(SIC) 预后 危险因素 预测模型 美国重症监护医学信息数据库-Ⅲ(MIMIC-Ⅲ) 

分 类 号：R459.7[医药卫生—急诊医学]