老年重症颅脑损伤患者发生急性创伤性凝血病的影响因素分析及列线图模型的建立

Analysis of influencing factors and establishment of a nomogram model for acute traumatic coagulopathy in elderly patients with severe traumatic brain injury

中文关键词: 老年  重型颅脑损伤  急性创伤性凝血病  影响因素  列线图模型

英文关键词: elderly  severe traumatic brain injury  acute traumatic coagulopathy  influencing factors  nomogram model

中文摘要:

      目的 探究老年重型颅脑损伤(severe traumatic brain injury,STBI)患者发生急性创伤性凝血病的影响因素及列线图模型的建立.方法 回顾性选取2020 年1 月—2023 年6 月在联勤保障部队第九〇九医院接受治疗的93 例老年ST-BI患者作为研究对象,依照7∶3的比例使用R软件将其随机分为建模队列(n =65)和模型验证队列(n =28);同时根据患者是否发生急性创伤性凝血病(ATC),将建模列队分为ATC组(n =39)和非ATC组(n =26).比较建模队列中 2 组的年龄、性别构成、BMI、ISS评分、收缩压、输液量、体温、动脉血PH、PT、APTT、FIB、D-DT、PLT值;采用二元多因素Lo-gistic回归分析方法分析老年STBI患者发生ATC的影响因素,建立路线图预测模型并验证,ROC评价模型预测效能,校准曲线评估预测事件与实际事件的一致性,DCA曲线评价模型的有效性.结果 建模队列ATC组ISS评分、输液量、PT、APTT及D-DT值显著高于非ATC组,而动脉血pH、GCS评分、PLT值显著低于非ATC组,差异均有统计学意义(P＜0.05).二元多因素Logistics回归分析结果显示,ISS评分、输液量、动脉血pH、GCS评分是老年STBI患者发生ATC的独立预测因子;模型公式:Logistic =-5.373+0.221 ISS评分+0.001 输液量+0.469 动脉血pH-0.232 GCS评分.采用R语言软件计算列线图模型的C统计量为:0.831,95%CI:0.778～0.879,标准误为0.025(P＜0.001),10 000 Bootstrap计算得出C统计量为0.822.所生成的列线图校准曲线斜率接近1,拟和度检验P值＞0.05,预测事件与实际事件的一致性较高;该模型的ROC曲线下面积为0.826(95%CI:0.720～0.933),决策分析曲线显示收益率较高,也进一步证实列线图预测模型的有效性.基于模型验证队列,对ATC风险列线图以ROC曲线进行外部验证,ROC曲线下面积为:0.829(95%CI:0.675～0.983),所生成的列线图校准曲线斜率接近1,Hosmer-Lemeshow检验结果:χ2 =9.362,P =0.303,且决策曲线显示,该模型的净收益较高,提示列线图模型在验证组中的有效性.结论 ISS评分、输液量、动脉血pH、GCS评分均可能是老年STBI患者发生ATC的独立预测因子;且进一步构建的老年STBI患者发生ATC的列线图预测模型表现出良好的预测能力,为临床识别老年STBI患者发生ATC的高风险人群提供了一定的帮助,有利于及时预防.

英文摘要:

      Objective To explore the influencing factors of acute traumatic coagulopathy (ATC) in elderly patients with severe traumatic brain injury (STBI) and establish a nomogram model. Methods 93 elderly patients with STBI who re? ceived treatment in the 909th Hospital of Joint Logistics Support Force from January 2020 to June 2023 were retrospectively se? lected as the study subjects. According to a ratio of 7∶ 3, they were randomly divided into modeling cohort (n =65) and model validation cohort (n = 28) by R software. At the same time, the patients in the model cohort were divided into ATC group (n =39) and non?ATC group (n =26) according to whether they had ATC. The age, gender composition, BMI, ISS score, systolic blood pressure, infusion volume, body temperature, arterial blood pH, PT, APTT, FIB, D?DT and PLT values were compared between the two groups in the modeling cohort. The influencing factors of ATC in elderly STBI patients were ana? lyzed by binary multivariate logistic regression analysis, and a road map prediction model was established and verified. The ROC was used to evaluate the predictive efficiency of the model, calibration curve was used to evaluate the consistency of the predicted event with the actual event, and DCA curve was used to evaluate the effectiveness of the model. Results In the modeling cohort, ISS score, infusion volume, PT, APTT and D?DT values of the ATC group were significantly higher than those of the non?ATC group, while arterial blood pH, GCS score and PLT values were significantly lower than those of the non?ATC group, with statistical significance (P <0. 05). The results of binary multivariate Logistic regression analysis showed that ISS score, infusion volume, arterial blood pH and GCS score were independent predictive factors for the occurrence of ATC in elderly STBI patients. The model equation was: Logistic = - 5. 373 + 0. 221 ISS score + 0. 001 infusion volume + 0?? 469 arterial blood pH?0. 232 GCS score. The C statistic of the nomogram model calculated by R language software was 0. 831 (95% CI: 0. 7788 -0. 879), with a standard error of 0. 025 (P <0. 001). The C statistic calculated by 10 000 Bootstrap was 0. 822. The slope of the calibration curve of the generated nomogram was close to 1, and the P value of the fitting degree test was >0. 05, which indicated that the predicted event has a high consistency with the actual event. The area under ROC curve of the model was 0. 826 (95% CI: 0. 720 -0. 933), and the decision analysis curve showed a high yield, which further confirmed the effectiveness of the nomogram prediction model. Based on the model validation cohort, the nomogram risk prediction model for ATC was externally validated by ROC curve. The area under ROC curve was 0. 829 (95% CI: 0. 675 - 0?? 983), and the slope of the calibration curve of the generated nomogram was close to 1. Hosmer?Limeshow test results: c2 = 9. 362, P = 0?? 303. The decision curve shows that the model had a higher net yield, indicating the effectiveness of the nomogram model in the validation group. Conclusion ISS score, infusion volume, arterial blood pH and GCS score may be independent predictors of ATC in elderly patients with STBI. The constructed nomogram prediction model for occurrence of ATC in elderly STBI pa? tients show good prediction ability. It can provide certain help for clinical identification of high?risk populations of elderly pa? tients with STBI who may develop ATC, and was conducive to prevention in time.