乙肝肝硬化住院患者衰弱风险预测模型的构建及验证  

Construction and validation of a frailty risk prediction model for inpatients with hepatitis B cirrhosis

作　　者：李莉莉 何娜 邓淑敏 淦伟强[2] 谢日华 高志良[2] Li Lili;He Na;Deng Shumin;Gan Weiqiang;Xie Rihua;Gao Zhiliang(School of Nursing,Southern Medical University,Guangzhou,510515;Department of Infectious Diseases,The Third Affiliated Hospital of Sun Yat-sen University,Guangzhou,510630;Big Data Artificial Intelligence Center,The Third Affiliated Hospital of Sun Yat-sen University,Guangzhou,510630;Affiliated Foshan Maternity&Child Healthcare Hospital,Southern Medical University,Foshan,528000,China)

机构地区：[1]南方医科大学护理学院,广东广州510515 [2]中山大学附属第三医院感染科,广东广州510630 [3]中山大学附属第三医院大数据人工智能中心,广东广州510630 [4]南方医科大学附属佛山妇幼保健院,广东佛山528000

出　　处：《现代临床护理》2023年第6期1-9,共9页Modern Clinical Nursing

基　　金：国家自然基金项目,项目编号为82170612;广东省医学基金课题,项目编号为A2021300。

摘　　要：目的调查乙肝肝硬化住院患者发生衰弱的现状,分析相关影响因素并构建风险预测模型。方法采用便利抽样方法,选取2021年8月至2022年11月广州市某三级甲等综合医院感染科收治的410例乙肝肝硬化住院患者作为建模集,根据是否发生衰弱分为衰弱组和非衰弱组,对两组相关资料进行比较,应用R Studia(4.1.1)软件构建风险预测模型,应用受试者操作特征曲线(receiver operating characteristic curve,ROC)下面积检验模型拟合效果。选取同一所医院2022年11月至2023年3月符合标准的139例患者进行模型预测效果验证。结果乙肝肝硬化住院患者衰弱发生率为19.9%。患者衰弱的风险预测模型共纳入4个危险因素:年龄(OR=1.042)、血红蛋白值(OR=0.982)、NRS2002营养风险筛查量表得分(OR=1.293)、体力活动水平(OR=0.482)。内部验证:ROC曲线下面积是0.747,95%CI(0.686,0.807)。Hosmer-Lemeshow检验,χ^(2)=5.669,P=0.684,灵敏度为64.0%,特异度为76.5%,准确率为80.2%。外部验证:ROC曲线下面积是0.775,95%CI(0.668,0.883)。Hosmer-Lemeshow检验,χ^(2)=15.077,P=0.058,灵敏度为91.3%,特异度为56.9%,准确率为84.2%。结论构建的乙肝肝硬化住院患者衰弱风险预测模型具有较好的预测价值,可为临床医护人员有效识别和筛选衰弱高风险人群提供参考及借鉴。Objective This study was to investigate the status and risk factors of frailty in inpatients with hepatitis B cirrhosis,and to develop a risk prediction model.Methods The convenient sampling method was used to enroll 410 inpatients with hepatitis B cirrhosis admitted to the Department of Infection Diseases of a tertiary general hospital in Guangzhou from August 2021 to November 2022 as the modeling cohort.They were divided into frailty group and non-frailty group,and the relevant data between the two groups were compared.Logistic regression analysis was used to establish a risk prediction model,and the area under the operating characteristic curve of the subjects was used to assess the prediction effect of the model.A total of 139 inpatients in the same hospital who met the criteria from November 2022 to March 2023 were enrolled for model prediction effect verification.Results The prevalences of frailty in inpatients with hepatitis B cirrhosis was 19.9%.Four risk factors were included in the risk prediction model for frailty:age(OR=1.042),hemoglobin(OR=0.982),Nutritional Risk Screening2002(OR=1.293),and physical activity level(OR=0.482).Internal verification:area under ROC curve 0.747,95%CI(0.686,0.807).Hosmer-Lemeshow test,χ^(2)=5.669,P=0.684,sensitivity of 64.0%,specificity of 76.5%,accuracy of 80.2%.External verification:area under ROC curve 0.775,95%CI(0.668,0.883).Hosmer-Lemeshow test,χ^(2)=15.077,P=0.058,sensitivity of 91.3%,specificity of 56.9%,accuracy of 84.2%.Conclusion The nomogram prediction model constructed has good risk prediction value,which can provide reference for clinical staff to effectively identify and screen inpatients with hepatitis B cirrhosis who are at high risk of frailty.

关 键 词：乙肝病毒感染 肝硬化 衰弱 风险预测模型 列线图 

分 类 号：R473.5[医药卫生—护理学]