ARCOⅡ期非创伤性股骨头坏死塌陷风险预测模型的构建与验证  

Development and verification of collapse risks prediction model for ARCO stageⅡnontraumatic osteonecrosis of femoral head

作　　者：韦建杰 曾平[2] 刘金富 农焦 梁娟 陶红成 陈立民 陆智豪 李晁辉 王颖赞 WEI Jianjie;ZENG Ping;LIU Jinfu;NONG Jiao;LIANG Juan;TAO Hongcheng;CHEN Limin;LU Zhihao;LI Chaohui;WANG Yingzan(Graduate school,Guangxi University of Chinese Medicine,Nanning 530200,Guangxi,China;NO.2 Orthopedics Department(Osteoarthropathy,Traumatic Orthopedics,and Femoral Head Necrosis Specialty),the First Affiliated Hospital of Guangxi University of Chinese Medicine,Nanning 530023,Guangxi,China)

机构地区：[1]广西中医药大学研究生院,广西南宁530200 [2]广西中医药大学第一附属医院骨二科(骨病创伤骨科、股骨头坏死专科),广西南宁530023

出　　处：《暨南大学学报（自然科学与医学版）》2024年第4期378-389,共12页Journal of Jinan University(Natural Science & Medicine Edition)

基　　金：国家自然科学基金项目(82160913,81960876);广西壮族自治区教育厅项目(2022KY0298)。

摘　　要：目的:分析ARCOⅡ期(未塌陷)非创伤性股骨头坏死(NONFH)发生塌陷的危险因素,并构建列线图风险预测模型,为识别ARCOⅡ期NONFH高塌陷风险患者提供简便工具和理论依据。方法:以2017年7月至2024年2月诊治的ARCOⅡ期和ⅢA期(塌陷早期)NONFH患者为研究对象。根据6∶4将患者随机分为训练组和验证组,以筛选塌陷的独立危险因素、构建预测模型并检验模型效能。首先利用医院电子病历系统提取患者资料并用R统计软件对训练组患者的各项指标进行LASSO回归分析以筛选最优特征值,再用多因素Logistic回归分析筛选塌陷的危险因素并构建预测模型。在训练组及验证组中比较模型对ARCOⅡ期塌陷风险的预测性能和区分度,评价指标包括一致性指数(C-index)、校准曲线,采用列线图对模型可视量化并通过决策曲线评估其临床效用。结果:共纳入符合标准的ARCOⅡ期NONFH患者78例(95髋)和ARCOⅢA期NONFH患者80例(102髋),共197髋。多因素Logistic回归分析显示,疼痛时间长(OR=1.176,95%CI:1.090~1.313)、股骨头受累百分比(PFHI)>30%(OR=25.257,95%CI:2.771~436.321)、骨髓水肿(BME)3级(OR=1.963,95%CI:0.017~22.424)、联合保留角(CPA)<118.7°(OR=6.814,95%CI:1.658~37.099)、红细胞体积分布宽度高(OR=1.097,95%CI:0.611~1.570)、血小板体积分布宽度高(OR=1.174,95%CI:0.119~13.149)为股骨头塌陷的独立危险因素(P<0.05)。本研究构建的ARCOⅡ期NONFH患者塌陷风险预测模型具有良好区分度,其在训练组和验证组中的C-index分别为0.844、0.878。在训练组和验证组中,校准曲线显示列线图预测结果与实际观测结果具有良好的一致性。决策曲线分析结果表明,预测模型可在较大概率阈值范围内带给患者更多净获益。结论:疼痛时间长、PFHI>30%、BME 3级、CPA<118.7°、红细胞和血小板体积分布宽度高可作为ARCOⅡ期股骨头塌陷发生的预测因素,有助于临床医生早期发现并干预�Objective:The risk factors for collapse in association research circulation osseous(ARCO)stageⅡ(non-collapse)nontraumatic osteonecrosis of the femoral head(NONFH)were analyzed,and a nomogram risk prediction model was developed to provide a simple tool and theoretical basis for identifying ARCO stageⅡpatients with a high collapse risk.Methods:Patients with ARCO stageⅡand stageⅢA(early collapse)NONFH diagnosed and treated from July 2017 to February 2024 were retrospectively selected as subjects.The patients were randomly divided into a training group and a verification group at a ratio of 6∶4 to screen the independent risk factors of collapse and develop a prediction model and test the model′s effectiveness.The optimal characteristic values were screened using LASSO regression analysis of the indexes of the patients in the training group with R statistical software.The risk factors of collapse were screened using a multifactor Logistic regression analysis.A prediction model was developed,and thus,the prediction and differentiation performance of the model for the risk of collapse in ARCO stageⅡpatients were compared between the training and verification groups.The evaluation indexes included the C-index and calibration curve.The model was visually quantified by a nomogram and its clinical effectiveness was evaluated by a decision curve.Results:A total of 78 patients(95 hips)with ARCO stageⅡNONFH and 80 patients(102 hips)with ARCO stageⅢA NONFH who met the criteria were included,totaling 197 hips.Multifactor Logistic regression analysis showed a long pain time(OR=1.176,95%CI:1.090-1.313),the percentage of femoral head involvement(PFHI)>30%(OR=25.257,95%CI:2.771-436.321),bone marrow edema(BME)grade 3(OR=1.963,95%CI:0.017-22.424),combined preserved angle(CPA)<118.7°(OR=6.814,95%CI:1.658-37.099),high red blood cell volume distribution width(OR=1.097,95%CI:0.611-1.570)and high platelet volume distribution width(OR=1.174,95%CI:0.119-13.149)were independent risk factors for femoral head collapse(P<0.05)

关 键 词：非创伤性股骨头坏死 股骨头塌陷 回归分析 

分 类 号：R274.9[医药卫生—中西医结合]