Abstract:

Objective To analyze the influencing factors of frailty in elderly patients with traumatic brain injury （TBI），construct and validate the predictive model of its risk，and provide a reference for the prevention of frailty in elderly TBI patients. Methods The clinical data of 337 elderly patients with TBI who were admitted to the emergency department of a tertiary general hospital in Zhejiang Province from October 2024 to March 2025 were retrospectively collected. The modeling group included 237 cases，while the validation group comprised 100 cases. Univariate analysis，binary Logistic regression analysis，and Lasso regression were used to explore the risk factors of frailty in elderly TBI patients，and collinearity tests were conducted. The nomogram prediction model for frailty in elderly TBI patients was drawn using R software. Internal validation and spatial external validation were performed on the validation set using the Bootstrap method，and the predictive performance of the model was evaluated using the Hosmer-Lemeshow test，the area under the receiver operating characteristic（ROC） curve，calibration curves，and decision curves. Results The model variables included polypharmacy status，Glasgow Coma Score，and Revised Trauma Score. The area under the ROC curve of the modeling set was 0.854（95%CI：0.808~0.900），with the optimal cut-off value of 0.454，a sensitivity of 74.20%，and a specificity of 78.10%. The Hosmer-Lemeshow test（χ²=4.188，P= 0.899） indicated that the model had good goodness of fit. Conclusion The constructed predictive model for frailty in elderly TBI patients has good discrimination，calibration，and clinical practicability，and can provide a reference for the early screening and prevention of frailty in high-risk elderly TBI patients.

Key words: Elderly Patients, Traumatic Brain Injury, Frailty, Prediction Model, Nursing Care

摘要：

目的 分析老年创伤性脑损伤（traumatic brain injury，TBI）患者衰弱的影响因素，构建并验证其风险预测模型的效果，为老年TBI患者的衰弱防范提供参考。 方法 回顾性收集2024年10月—2025年3月，在浙江省某三级甲等综合医院2个院区因TBI急诊入院的337例老年患者的临床资料。其中，建模组纳入237例，验证组纳入100例。通过单因素分析、二元Logistic回归分析和Lasso回归探讨老年TBI患者衰弱的风险因素，并进行共线性检验，运用R软件绘制老年TBI患者衰弱列线图预测模型；采用Bootstrap法进行内部验证和空间外部验证，采用Hosmer-Lemeshow检验、受试者操作特征曲线下面积、校准曲线、决策曲线评价模型的预测性能。 结果 模型变量包括多重用药、格拉斯哥昏迷评分、修正创伤评分。建模组受试者操作特征曲线下面积为0.854（95%CI：0.808~0.900），最佳临界值0.454，灵敏度74.20%，特异度78.10%，内部验证Hosmer-Lemeshow检验（χ²=4.188，P=0.899）提示模型拟合优度较好。 结论 构建的老年TBI患者衰弱预测模型具有较好的区分度、校准度和临床实用性，可为早期筛查、预防衰弱高危的老年TBI患者提供参考。

关键词: 老年患者, 创伤性脑损伤, 衰弱, 预测模型, 护理