重症颅脑损伤老年患者围手术期深静脉血栓形成的风险预测模型构建

Construction of risk prediction model for perioperative deep vein thrombosis formation in elderly patients with severe traumatic brain injury

目的构建重症颅脑损伤(sTBI)老年患者围手术期深静脉血栓(DVT)形成的风险预测模型。方法回顾性选取2021年1月—2023年2月阳光融和医院神经外科重症监护室收治的206例sTBI患者,将2021年1月—2022年6月收治的146例患者用于建模(建模组),其中围手术期发生DVT老年患者33例作为DVT组,其余113例患者作为非DVT组。将2022年7月—2023年2月收治的60例患者用于外部验证(验证组)。采用Logistic回归分析法分析DVT形成的高危因素,根据AUC及Hosmer Lemeshow拟合优度检验模型效能。结果两组高血糖、高血压、多发伤、中心静脉置管、机械通气≥1 d、血管活性药、受伤至手术时间、术中输血量、术中失血量、血红蛋白(Hb)、血浆D-二聚体、血小板、凝血酶原时间(PT)、同型半胱氨酸(Hcy)、肺部感染比较,差异有统计学意义(P<0.05)。高血糖、多发伤、血管活性药、血浆D-二聚体、肺部感染、受伤至手术时间是老年sTBI患者发生DVT的独立危险因素(P<0.05)。建模组AUC为0.912,P<0.05,95%CI为0.859~0.962,最大约登指数为0.667,灵敏度为0.738,特异度为0.945,最佳截断值为0.209;验证组中AUC为0.805,本预测模型的灵敏度为0.625,特异度为0.898,准确率为83.6%。结论肺部感染、血管活性药使用、血浆D-二聚体水平>0.5 mg/L、高血糖、多发伤、受伤至手术时间>5 d可增加老年sTBI患者围手术期DVT的发生。

Objective To construct risk prediction model for perioperative deep vein thrombosis(DVT)formation in elderly patients with severe traumatic brain injury(sTBI).Methods Two hundred and six sTBI patients admitted to Neurosurgical Intensive Care Unit of Yangguang Ronghe Hospital from January 2021 to February 2023 were retrospectively selected.One hundred and forty-six patients admitted from January 2021 to June 2022 were used for modeling(modeling group).Among them,33 elderly patients with DVT during perioperative period were included in DVT group,while the remaining 113 patients were included in non DVT group.Sixty patients admitted from July 2022 to February 2023 were used for external validation(validation group).The high-risk factors for DVT formation were analyzed by using Logistic regression analysis,and the model's effectiveness was tested based on AUC and Hosmer Lemeshow goodness of fit.Results Hyperglycemia,hypertension,multiple injuries,central venous catheterization,mechanical ventilation more than one day,vasoactive drugs,time from injury to surgery,intraoperative blood transfusion volume,intraoperative blood loss,hemoglobin,plasma D-dimer,platelets,prothrombin time,homocysteine and pulmonary infection of two groups were compared,and the differences were statistically significant(P<0.05).Hyperglycemia,multiple injuries,vasoactive drugs,plasma D-dimer,pulmonary infection and time from injury to surgery are independent risk factors for DVT in elderly sTBI patients(P<0.05).AUC of modeling group was 0.912,P<0.05,with 95%CI ranging from 0.859 to 0.962.The maximum Jordan index was 0.667,sensitivity was 0.738,specificity was 0.945,and the optimal cutoff value was 0.209.AUC in validation group was 0.805,the sensitivity of this prediction model was 0.625,the specificity was 0.898,and the accuracy rate was 83.6%.Conclusion Pulmonary infection,use of vasoactive drugs,plasma D-dimer levels more than 0.5 mg/L,hyperglycemia,multiple injuries and injury to surgery time more than five days can increase the occurrence of perioperative DVT in elderly sTBI patients.