卒中患者跌倒风险预测模型的构建及验证

Development and validation of a risk prediction model for fall in stroke patients

目的分析卒中患者发生跌倒的影响因素,构建并验证风险预测模型,为预防卒中患者跌倒提供科学依据。方法选取2021年6—12月锦州医科大学附属第一医院就诊的355例卒中住院患者作为研究对象,采用随机抽样法抽取70%(248例)患者作为训练集,剩余30%(107例)作为验证集。回顾性收集其人口学特征,以及基于文献回顾筛选出的相关资料,采用单因素分析和多因素logistic回归筛选出跌倒的影响因素,基于主要影响因素构建列线图模型;采用受试者工作特征(receiver operating characteristic,ROC)曲线、校准曲线评估模型预测效能。结果高年龄(OR=3.975)、高血压(OR=4.557)、使用辅助器具(OR=5.103)、头晕眩晕(OR=6.951)、跌倒史(OR=7.663)是卒中患者跌倒的独立危险因素,日常生活能力得分高(OR=0.286)为保护因素;训练集和验证集的ROC曲线下面积分别为0.892、0.916,训练集和验证集的校准曲线显示预测曲线和实际曲线贴合紧密。结论该研究构建的预测模型具有良好的区分度与校准度,可用于预测卒中患者跌倒的发生。

Objective To analyze the risk factors for falls in stroke patients,and develop and verify a prediction model that provides a scientific basis for preventing falls in stroke patients.Methods Three hundred fifty-five stroke patients who were admitted to the First Affiliated Hospital of Jinzhou Medical University from June 2021 to December 2021 were selected.Of the patients,70%were selected by random sampling as the training set and the remaining 30%served as the validation set.Demographic characteristics,as well as the relevant data selected based on the literature review,were collected retrospectively.Risk factors for falls were identified using a chi-square test and multivariate logistic regression,then a nomogram based on the risk factors was constructed and validated.Receiver operating characteristic(ROC)and calibration curves were used to verify the accuracy and effectiveness of the prediction model.Results High age(OR=3.975),hypertension(OR=4.557),use of assistive devices(OR=5.103),dizziness(OR=6.951),and fall history(OR=7.663)were independent risk factors for falls in stroke patients,while high daily living ability score(OR=0.286)was a protective factor.The area under the ROC curve for the training and validation sets was 0.892 and 0.916,respectively,and the calibration curves for the training and validation sets showed that the predicted and actual curves fit closely.Conclusions The prediction model constructed in this study has good discrimination and accuracy,and can be used to predict the occurrence of falls in stroke patients.