多发伤患者静脉血栓栓塞症列线图预测模型建立的研究

A prognostic nomogram model to predict venous thromboembolism in multiple trauma patients

探究多发伤患者静脉血栓栓塞症(VTE)的危险因素并建立VTE列线图预测模型。方法回顾性收集2019年1月至2020年6月上海交通大学附属第六人民医院急诊医学科重症监护室收治的多发伤患者临床资料。通过Logistic回归分析筛选VTE发生的危险因素,使用R语言软件建立列线图预测模型,并评价该模型的预测能力。结果本研究共纳入患者138例,其中男103例,女35例,平均创伤严重程度(ISS)评分(24.14±8.7)分,VTE发病率40.6%(56例)。Logistic回归分析结果显示,下肢制动、存在药物抗凝禁忌证、机械通气、胸部简明损伤评分(AIS)>3分、四肢和骨盆AIS评分>3分是VTE发生的独立危险因素(P<0.05)。将上述危险因素纳入列线图模型,ROC曲线下面积和自举法内部验证后模型预测结果与实际结果一致。结论基于Logistic回归模型建立的列线图预测模型对多发伤患者VTE发生具有较好的预测效能,可实现个体化、可视化、图形化预测,为临床决策多发伤患者抗凝治疗方案提供参考,从而减少VTE发生。

To assess the risk and prognostic factors of venous thromboembolism(VTE)in multiple trauma patients and to establish a prognostic nomogram model to predict the risk of VTE.Methods Data related to the demographics,nature of the injury,and clinical features of adult patients with multiple trauma(injury severity score≥9)who were treated in the Intensive Care Unit of the Shanghai Sixth People′s Hospital were collected and compared.Binary logistic regression and a nomogram model were used to identify independent risk factors for VTE.Results The study cohort included 138 multiple trauma patients.Logistic regression analysis showed that immobilization of the lower limbs,contra-indication for chemoprophylaxis,mechanical ventilation,abbreviated injury scale(AIS)of the chest>3,and AIS of the pelvisand lower limbs>3 were independent risk factors for VTE.Apredictive nomogrammodel was established with the use of the identified risk factors.A calibration chart of the model showed that the predictive probability was consistent with the actual occurrence probability and decision curve analysis showed that the ability of the nomogram to predict the risk of VTE for multiple trauma patients was excellent.Conclusion Immobilization of the lower limbs,contra-indication for chemoprophylaxis,mechanical ventilation,AIS of the chest>3,and AIS of the pelvis and lower limbs>3 were independent risk factors for VTE in multiple trauma patients.The proposed prognostic nomogram is a useful visual and graphical predictive tool for triage and management of multiple trauma patients.