社区老年人髋部骨折早期风险预测模型的构建与验证  

Develop and validate an early risk prediction model for hip fracture among the elderly in community

作　　者：何春梅[1] 张红英[2] 陈丽娟 熊林竹 田苗[2] 廖天成 蒋红梅[1] 窦艳 He Chunmei;Zhang Hongying;Chen Lijuan;Xiong Linzhu;Tian Miao;Liao Tiancheng;Jiang Hongmei;Dou Yan(Department of Nursing Administration,the Second People’s Hospital of Deyang,Deyang 618000,Sichuan Province,China;Department of Traditional Chinese Medicine and Rehabilitation,the Second People’s Hospital of Deyang,Deyang 618000,Sichuan Province,China;Continuing Care Centre,the Second People’s Hospital of Deyang,Deyang 618000,Sichuan Province,China;Department of Endocrinology and Geriatrics,the Second People’s Hospital of Deyang,Deyang 618000,Sichuan Province,China;Department of Orthopaedics,the Second People’s Hospital of Deyang,Deyang 618000,Sichuan Province,China.)

机构地区：[1]德阳市第二人民医院护理部,四川德阳618000 [2]德阳市第二人民医院中医科康复医学科,四川德阳618000 [3]德阳市第二人民医院延续护理中心,四川德阳618000 [4]德阳市第二人民医院内分泌老年全科医学科,四川德阳618000 [5]德阳市第二人民医院骨科,四川德阳618000

出　　处：《现代临床护理》2025年第3期15-23,共9页Modern Clinical Nursing

基　　金：德阳市科研计划项目,项目编号2022SCZ111;四川省护理科研课题计划,项目编号H23029- 2023年四川护理职业学院自然科学课题资助,项目编号2023ZRY31。

摘　　要：目的 调查社区老年人髋部骨折发生率,分析相关影响因素并构建风险预测模型与验证。方法 采用分层抽样方法收集2023年1月—2024年1月德阳市社区723名老年人的社会人口学资料、生活习惯和髋部骨折危险因素相关资料。通过随机拆分的方法,479名(68.00%)为模型训练集,221名(32.00%)为模型验证集。在模型训练集中,根据是否发生髋部骨折分为骨折组和未骨折组进行比较,应用R软件(4.3.1)构建风险预测模型并验证。结果 最终纳入700例社区老年人,1年内发生髋部骨折共62例,累计发生率为8.86%。社区老年人髋部骨折的风险预测模型共6个预测因子:常吃腌制食品、每天锻炼时间、每天日照时间、骨质疏松、1年内发生跌倒、真牙≥20颗。模型训练集曲线下面积(area under the curve,AUC)值为0.945(95%CI 0.908~0.982),灵敏度为88.89%,特异度为89.40%;校准曲线中预测值与实际值显示较好的吻合性,表明预测模型具有良好的校准度;决策曲线分析(decision curve analysis,DCA)表明该预测模型可以取得正的净效益。模型验证集AUC值为0.892(95%CI 0.784~0.999),灵敏度为82.35%,特异度为93.63%,表明验证集数据与模型拟合良好,预测能力较强;校准曲线中预测值与实际值显示较好的一致性;决策曲线分析表明使用该预测模型可以取得正的净效益。结论 构建的社区老年人髋部骨折预测模型具有较好的预测价值,可为社区工作人员和医护人员筛查社区老年人髋部骨折的风险提供参考和借鉴。Objective To investigate the incidence of hip fracture among the elderly in communities,explore related influencing factors,and develop and validate a risk prediction model.Methods A stratified sampling method was used to collect sociodemographic data,lifestyles and risk factors in hip fracture between January 2023 and January 2024 among the elderly residents in communities in Deyang.With random splitting,479 elderly people(68.00%)were assigned to the model training set,and 221(32.00%)to the model validation set.In the model training set,the participants were divided into a fracture group and a non-fracture group based on hip fracture or not.Data from both groups were compared,and R software(version 4.3.1)was employed to develop and validate the risk prediction model.Results A total of 700 elderly residents in communities were included,62 of them had hip fracture within one year yielding a cumulative incidence rate of 8.86%.The risk prediction model identified six predictors:frequent consumption of preserved foods,daily exercise time,daily sunlight exposure,osteoporosis,times of fall within a year,and with≥20 pieces of natural teeth.In the training set,the model achieved an AUC of 0.945(95%CI:0.908-0.982),with a sensitivity of 88.89%and a specificity of 89.40%.The calibration curve demonstrated a good agreement between predicted and actual values,indicating a strong calibration.Decision curve analysis(DCA)showed a positive net benefit.In the validation set,the AUC was 0.892(95%CI:0.784-0.999),with a sensitivity of 82.35%and a specificity of 93.63%,confirming a good model fit and predictive performance.The calibration curve exhibited a strong consistency,and DCA indicated a positive net benefit.Conclusion The developed risk prediction model for hip fracture in elderly community residents demonstrates a strong predictive value.It provides a practical reference for community workers and healthcare professionals to screen and assess the risk of hip fracture among the elderly residents in communities.

关 键 词：髋部骨折 社区 老年人 风险预测模型 列线图 

分 类 号：R473.6[医药卫生—护理学]