基于logistic回归构建住院老年患者衰弱预测模型及验证

Construction and validation of a predictive model for frailty in hospitalized elderly patients based on logistic regression

目的 随着我国快速步入老龄化时代,老年人成为入住医院最主要的群体,本研究旨在调查住院老年患者衰弱的危险因素并构建列线图,为住院患者的衰弱综合征预防提供科学的参考依据。方法 于2020年1月—2021年12月,采用便利抽样法抽取芜湖市2家医院共30个病区进行问卷调查,采用专人一对一采集问卷。其中,一般资料包括性别、年龄、文化水平、户籍、月收入、5年以上慢性病、跌倒、午睡、吸烟、饮酒、睡眠障碍、抑郁情绪。通过单因素分析,研究结果表明,睡眠障碍、抑郁、跌倒、5年以上慢性病、午睡等具有统计学意义(P<0.05)。通过R Studio 4.2.3软件构建风险因素的列线图,并且对模型进行相关验证。结果 住院老年患者在5年以上慢性病、跌倒、午睡、睡眠障碍及抑郁情绪具有统计学差异(P<0.05),5年以上慢性病、有跌倒史、有午睡习惯、有睡眠障碍和抑郁情绪是老年住院患者衰弱的独立危险因素,建立模型并绘制列线图,ROC曲线下面积(Area under curve,AUC)是0.822。结论 本研究通过构建老年住院患者衰弱风险预测模型预测性能较好,为本地区医院的医护工作者提前筛选衰弱发生的高风险人群,并可提前预判住院患者衰弱的发生提供参考依据。

Objective As China rapidly enters the era of aging,the elderly have become the main group of people admitted to hospitals.The purpose of this study is to investigate the risk factors of frailty in inpatients and construct a graph,so as to provide scientific reference for the prevention of frailty syndrome among inpatients.Methods From January 2020 to December 2021,a total of 30 wards of 4 hospitals in Wuhu were selected by convenience sampling method for questionnaire survey,and questionnaires were collected by specially-assigned persons one to one.Among them,general information included sex,age,education level,household registration,monthly income,chronic diseases for more than 5 years,falls,napping,smoking,alcohol consumption,sleep disorders and depression.Through univariate analysis,the results showed that sleep disorders,depression,falls,chronic diseases,and napping were statistically significant.RStudio4.2.3 software was uese to build the risk factor diagram to verify the model.Results There were statistically significant differences among hospitalized elderly patients chronic disease for more than five years,or fall,or nap,or sleep disorder or depressive mood(P<0.05).Chronic disease more than five years,fall history,nap habit,sleep disorder and depressive mood were independent risk factors of frailty in hospitalized elderly patients.By building a model and drawing the nomogram,the area under curve(AUC)was 0.822.Conclusions The frailty risk prediction model of the elderly inpatients established in this study has good predictive performance,which can provide reference for medical workers in local hospitals to screen the high-risk groups of frailty occurrence and predict the occurrence of frailty in inpatients in advance.