肋骨骨折患者发生心肌挫伤预测模型的构建及其临床应用价值  

Establishment of a predictive model for myocardial contusion in patients with rib fractures and its clinical application value

作　　者：余长永 宋月坤 朱康宇 程祥 朱天皓 刘武新 Yu Changyong;Song Yuekun;Zhu Kangyu;Cheng Xiang;Zhu Tianhao;Liu Wuxin(Department of Thoracic Surgery,Affiliated Jiangsu Shengze Hospital of Nanjing Medical University,Suzhou 215228,China;Department of Emergency,Affiliated Jiangsu Shengze Hospital of Nanjing Medical University,Suzhou 215228,China)

机构地区：[1]南京医科大学附属江苏盛泽医院胸外科,苏州215228 [2]南京医科大学附属江苏盛泽医院急诊科,苏州215228

出　　处：《中华创伤杂志》2024年第8期715-726,共12页Chinese Journal of Trauma

基　　金：苏州市科技发展计划项目(SKYD2023022);苏州市吴江区“科教兴卫”项目(WWK202320)。

摘　　要：目的构建肋骨骨折患者发生心肌挫伤(MC)的预测模型并评价其临床应用价值。方法采用回顾性病例对照研究分析2017年1月至2019年12月南京医科大学附属江苏盛泽医院收治的370例肋骨骨折患者的临床资料,其中男257例,女113例;年龄18~95岁[(56.5±14.0)岁]。患者入院24 h内均接受心电图检查和心肌标志物检测,其中159例诊断为MC,211例无MC(NMC)。采用完全随机法将370例患者按7∶3的比例分为训练集264例(MC 106例、NMC 158例)和验证集106例(MC 53例、NMC 53例)。在训练集中,比较MC组和NMC组患者的人口学特征、入院时生命体征、肋骨骨折类型、肋骨骨折数、肋骨骨折部位、关联的胸部损伤、创伤评分、实验室检查指标。通过Spearman相关分析筛选肋骨骨折患者发生MC的正相关变量,且采用单因素二元Logistic回归分析MC的正相关变量以确定肋骨骨折患者发生MC的危险因素。采用LASSO回归和多因素Logistic回归分析筛选肋骨骨折患者发生MC的独立危险因素并构建回归方程,且利用R软件构建基于回归方程的列线图预测模型。绘制受试者工作特征(ROC)曲线评价模型的区分度。采用Hosmer‑Lemeshow(H‑L)拟合优度检验及Bootstrap法重复抽样1000次的校准曲线评价模型的校准度。采用决策曲线分析(DCA)和临床影响曲线分析(CIC)评价模型的临床应用价值。根据独立危险因素的β系数赋值进行风险评分,将入选的370例肋骨骨折患者分为低危组202例、中危组108例、高危组50例和极高危组10例,比较不同亚组患者MC发生率和院内死亡率,进一步验证模型的临床应用价值。结果在训练集中,MC组和NMC组双侧肋骨骨折、连枷胸、肋骨骨折数、上胸部近胸骨侧段、上胸部前外侧段、上胸部近脊柱侧段、中胸部前外侧段、中胸部近脊柱侧段、下胸部前外侧段、气胸、纵隔气肿、血胸、胸骨骨折、胸部简明损伤定级(c‑AIS)Objective To establish a predictive model for myocardial contusion(MC)in patients with rib fractures and evaluate its clinical application value.Methods A retrospective case-control study was conducted to analyze the clinical data of 370 patients with rib fractures admitted to the Affiliated Jiangsu Shengze Hospital of Nanjing Medical University from January 2017 to December 2019,including 257 males and 113 females,aged 18-95 years[(56.5±14.0)years].All the patients underwent electrocardiogram examination and myocardial biomarker test within 24 hours on admission,of whom 159 were diagnosed with MC,and 211 with non-MC(NMC).The 370 patients were divided into a training set of 264 patients(106 with MC,158 with NMC)and a validation set of 106 patients(53 with MC,53 with NMC)at a ratio of 7∶3 through the completely randomized method.In the training set,the MC group and NMC group were compared in terms of their demographic characteristics,vital signs on admission,types of rib fractures,number of rib fractures,locations of rib fractures,associated thoracic injuries,trauma scores,and laboratory indices.Variables of positive correlation with MC in patients with rib fractures were screened by Spearman correlation analysis,followed by univariate binary Logistic regression analysis for these variables to determine the risk factors for MC in patients with rib fractures.LASSO regression analysis and multivariate Logistic regression analysis were applied to identify the independent risk factors for MC in patients with rib fractures,and the regression equation was constructed.A nomogram prediction model was plotted based on the regression equation with R software.The receiver operating characteristic(ROC)curve was plotted to evaluate the model′s discriminability.Hosmer-Lemeshow(H-L)goodness-of-fit test and calibration curves of 1000 repeated samplings by the Bootstrap method were used to evaluate the calibration of the model.The decision curve analysis(DCA)and clinical impact curve analysis(CIC)were plotted to evaluate its

关 键 词：肋骨骨折 预后 危险因素 病例对照研究 

分 类 号：R683.1[医药卫生—骨科学] R655[医药卫生—外科学]