老年糖尿病患者股骨骨折术后肺部并发症的风险因素分析及其预测模型构建

Risk factors and prediction model construction for postoperative pulmonary complication in elderly diabetic patients with femoral fractures

原文传递

导出

摘要目的探讨老年糖尿病患者股骨骨折术后肺部并发症(PPC)的风险因素及其预测模型的构建。方法回顾性队列研究分析2016年1月至2023年12月陆军军医大学第二附属医院收治的1057例诊断为糖尿病并接受股骨骨折手术患者的临床资料,按术后是否发生PPC分为PPC组(115例)和非PPC组(942例)。将两组差异有统计学意义的因素纳入多因素logistic回归分析,分析老年糖尿病患者股骨骨折发生PPC的风险因素。建立列线图模型,分别采用受试者操作特征(ROC)曲线及曲线下面积(AUC)评估模型区分度,校准曲线评估模型预测风险与实际风险的一致性,Hosmer⁃Lemeshow拟合优度检验评估模型的校准度,决策曲线分析(DCA)评估模型的临床实用性,以Bootstrap法进行内部一致性验证。结果PPC的发生率为10.88%。与非PPC组比较,PPC组患者年龄较大,体重指数(BMI)、术前心室射血分数(EF)、术前白蛋白(ALB)水平较低,美国麻醉医师协会(ASA)分级Ⅲ或Ⅳ级、低蛋白血症、术前血红蛋白(Hb)≤100 g/L、术前肾小球滤过率(GFR)30~90 ml·min^(−1)·1.73 m^(−2)、合并呼吸系统疾病、手术等待时间>7 d的比例较高,全麻、术前肌酐<110μmol/L的比例较低,麻醉时间、手术时间、住院时间较长(均P<0.05)。多因素logistic回归分析结果显示,高龄[比值比(OR)2.34,95%置信区间(CI)1.43~5.65]、低蛋白血症(OR 19.93,95%CI 4.89~81.09)、术前Hb≤100 g/L(OR 3.25,95%CI 1.68~7.35)、术前合并呼吸系统疾病(OR 13.49,95%CI 5.35~34.05)、手术时间(OR 3.31,95%CI 1.53~7.19)、麻醉时间(OR 2.15,95%CI 1.17~3.93)、全麻(OR 3.82,95%CI 1.56~9.32)、手术等待时间>7 d(OR 5.23,95%CI 1.37~19.96)、住院时间(OR 2.11,95%CI 1.12~3.96)是老年糖尿病患者股骨骨折发生PPC的独立风险因素。列线图模型预测PPC发生风险的ROC AUC的OR及95%CI为0.91(0.88~0.94),灵敏度、特异度分别为83.7%、81.2%,表明模型区分度良好;校准曲线为斜率接近1的直线,表明该模型预测老年糖尿病患者股骨骨折PPC的发生风险与实际发生风险一致性良好;Hosmer⁃Lemeshow拟合优度检验χ^(2)=1.78,P=0.314,表明模型有较好的校准度;DAC显示当潜在风险阈值取4.2%时可获得较好的净收益,表明模型具有较好的临床实用性;Bootstrap法内部验证表明模型具有良好的一致性。结论高龄、低蛋白血症、术前Hb≤100 g/L、术前合并呼吸系统疾病、手术时间、麻醉时间、全麻、手术等待时间>7 d、住院时间是老年糖尿病患者股骨骨折发生PPC的风险因素,列线图模型具有较好的临床预测能力及临床实用性。 Objective To explore the risk factors for postoperative pulmonary complication(PPC)in elderly diabetic patients with femoral fractures and construct a prediction model.Methods A retrospective cohort study was conducted on 1057 diabetic patients who underwent femoral fracture surgery at the Second Affiliated Hospital of Army Medical University from January 2016 to December 2023.Based on the occurrence of PPC,the patients were divided into two groups:a PPC group(n=115)and a non⁃PPC group(n=942).Factors showing statistical differences between the groups were included in multivariate logistic regression analysis to identify risk factors for PPC in elderly diabetic patients with femoral fractures.A nomogram prediction model was constructed.The model's discriminative ability was evaluated using the receiver operating characteristic(ROC)curve and the area under the curve(AUC).Calibration curves were plotted to assess the consistency between predicted and actual risks.The model's calibration was evaluated by the Hosmer⁃Lemeshow goodness⁃of⁃fit test.Decision curve analysis(DCA)was conducted to assess clinical utility.Internal validation was performed using the Bootstrap method.Results The incidence of PPC was 10.88%.Patients in the PPC group were older than the non⁃PPC group,and they showed decreases in body mass index(BMI),preoperative cardiac ejection fraction(EF),albumin(ALB)levels,compared with the non⁃PPC group.The PPC group presented higher proportions of American Society of Anesthesiologists(ASA)gradesⅢandⅣ,hypoalbuminemia,preoperative hemoglobin(Hb)≤100 g/L,preoperative glomerular filtration rate(GFR)of 30‒90 ml·min^(−1)·1.73 m(−2),respiratory diseases,and surgery waiting time>7 days.In contrast,lower proportions of general anesthesia and preoperative creatinine<110μmol/L,and longer anesthesia time,surgical time and hospitalization stay were observed in the PPC group(all P<0.05).Multivariate logistic regression analysis indicated that advanced age[odds ratio(OR)2.34,95%(confidence interval(CI)1.43,5.65)],hypoalbuminemia[OR 19.93,(95%CI 4.89,81.09)],preoperative Hb≤100 g/L[OR 3.25,(95%CI 1.68,7.35)],preoper⁃ative respiratory diseases[OR 13.49,(95%CI 5.35,34.05)],surgery time[OR 3.31,(95%CI 1.53,7.19)],anesthesia time[(OR 2.15),(95%CI 1.17,3.93)],general anesthesia[OR 3.82,(95%CI 1.56,9.32)],surgery waiting time>7 days[OR 5.23,(95%CI 1.37,19.96)],hospitalization stay[OR 2.11,(95%CI 1.12,3.96)]were independent risk factors for PPC in elderly diabetic patients with femoral fractures.The nomogram model for predicting the risk of PPC showed an area under the ROC AUC of 0.91(95%CI 0.88,0.94)with a sensitivity of 83.7%and specificity of 81.2%,which indicated that the model had a good discriminatory degree.The calibration curve was a straight line with a slope of nearly 1,which indicated that the model predicted the risk of femur fracture PPC in elderly diabetic patients in good agreement with the actual risk.The Hosmer⁃Lemeshow test yieldedχ^(2)=1.78,P=0.314,indicated that the model had a good calibration.DCA showed a net benefit when the potential risk threshold was set at 4.2%,which indicated that the model had clinical utility.Internal validation demonstrated good consistency of the model using the Bootstrap method.Conclusions Advanced age,hypoalbuminemia,preoperative Hb≤100 g/L,preoperative respiratory diseases,surgery time,anesthesia time,general anesthesia,surgery waiting time>7 days,and hospitalization stay are risk factors for PPC in elderly diabetic patients with femoral fractures.The nomogram model exhibits good clinical predictive capacity and clinical utility.

作者方亮张皓琳吴卓熙李洪白福海 Fang Liang;Zhang Haolin;Wu Zhuoxi;Li Hong;Bai Fuhai(Department of Anesthesiology,Second Affiliated Hospital,Army Medical University,Chongqing 400037,China)

机构地区陆军军医大学第二附属医院麻醉科

出处《国际麻醉学与复苏杂志》 2025年第6期572-579,共8页 International Journal of Anesthesiology and Resuscitation

基金国家卫生健康委科学技术研究所科研项目(2024KYS01810) 重庆英才计划“包干制项目”(cstc2022ycjh⁃bgzxm0061)。

关键词糖尿病股骨骨折术后肺部并发症老年患者风险因素列线图预测模型 Diabetes mellitus Femur Fracture Postoperative pulmonary complication Elderly patient Risk fac⁃tor Nomogram Prediction model

分类号 R687.3 [医药卫生—骨科学]

作者简介通信作者:白福海,Email:bfh@tmmu.edu.cn。