髋部骨折老年患者术后1年死亡的危险因素及其预测模型的准确性:基于LASSO-logistic回归  

Risk factors for 1-year death after surgery in elderly patients with hip fractures and accuracy of prediction model: based on LASSO-logistic regression

作　　者：吴红 蔡伟茶 金琪琪 周盈丰 袁开明 李挺 李军 Wu Hong;Cai Weicha;Jin Qiqi;Zhou Yingfeng;Yuan Kaiming;Li Ting;Li Jun(Department of Anesthesiology and Perioperative Medicine,the Second Affiliated Hospital&Yuying Children′s Hospital,Wenzhou Medical University,Wenzhou 325024,China)

机构地区：[1]温州医科大学附属第二医院、育英儿童医院麻醉与围术期医学科,温州325024

出　　处：《中华麻醉学杂志》2024年第1期15-19,共5页Chinese Journal of Anesthesiology

基　　金：浙江省公益技术研究计划(LGF21H250004)。

摘　　要：目的基于LASSO-logistic回归分析,筛选髋部骨折老年患者术后1年死亡的危险因素并评价预测模型的准确性。方法对2019年1月至12月在温州医科大学附属第二医院行髋部骨折手术治疗的老年患者(年龄≥65岁)进行病例对照研究。根据患者术后1年的生存状态分为死亡组和存活组。获取患者一般资料及术前实验室指标。采用LASSO回归筛选出变量后将其纳入多因素logistic回归分析筛选髋部骨折老年患者术后1年死亡的危险因素。根据结果建立预测模型并对模型进行评价。结果死亡组63例,存活组564例。经LASSO回归筛选变量及多因素logistic回归分析,年龄、术前认知功能障碍、查尔森合并症指数(CCI)≥3分和术前血清前白蛋白(PAB)水平是髋部骨折老年患者术后1年死亡的独立危险因素(P<0.05)。预测模型的受试者工作特征曲线下面积为0.788(95%CI 0.731~0.846),灵敏度和特异度分别为76.2%和68.6%。校准曲线的平均绝对误差为0.007。Hosmer-Lemeshow拟合优度检验结果显示模型预测值与实际观测值差异无统计学意义(χ2=5.065,P=0.751)。决策曲线分析显示阈概率范围为0~0.7时患者的净获益率较高。结论年龄、术前认知功能障碍、CCI≥3分和术前PAB水平是髋部骨折老年患者术后1年死亡的独立危险因素,基于LASSO-logistic回归建立的预测模型准确性较高。Objective To identify the risk factors for 1-year death after surgery in elderly patients with hip fractures and evaluate the accuracy of the prediction model based on LASSO-logistic regression analysis.Methods A case-control study was conducted on elderly patients(age≥65 yr)who underwent surgical treatment for hip fractures in the Second Affiliated Hospital of Wenzhou Medical University from January to December 2019.Patients were divided into death group and survival group according to their survival status at 1-year after surgery.General data and preoperative laboratory indicators were obtained.The variables were selected by utilizing LASSO regression and incorporated into multivariate logistic regression analysis to identify the risk factors for 1-year death after surgery in elderly patients with hip fractures.Then a prediction model was established based on the results and evaluated.Results There were 63 patients in death group and 564 in survival group.The results of LASSO regression and multivariate logistic regression analysis showed that age,preoperative cognitive dysfunction,Chalson comorbidity index≥3 points and preoperative serum prealbumin level were the independent risk factors for 1-year death after surgery in elderly patients with hip fractures(P<0.05).The area under the receiver operating characteristic curve of the prediction model was 0.788(95%confidence interval[0.731-0.846]),with the sensitivity and specificity of 76.2%and 68.6%respectively.The average absolute error of the calibration curve was 0.007.The results of Hosmer-Lemeshow goodness-of-fit test showed that there was no significant difference between the predicted value and actual observed value(χ2=5.065,P=0.751).Decision curve analysis showed that patients had a high net benefit rate when the threshold probability range was 0-0.7.Conclusions Age,preoperative cognitive dysfunction,Chalson comorbidity index≥3 points and preoperative serum prealbumin level are the independent risk factors for 1-year death after surgery in elderly p

关 键 词：髋部骨折 老年人 危险因素 预测模型 LASSO-logistic回归 

分 类 号：R687.3[医药卫生—骨科学]