摘要
目的:探讨老年脑梗死患者跌倒恐惧(FoF)危险因素,建立Logistic回归与随机森林模型,为患者FoF干预管理提供借鉴。方法:纳入本院2022年10月至2024年10月152例老年脑梗死患者,应用一般资料调查表、修订版跌倒效能量表(MFES)、简易精神状态检查量表(MMSE)、下肢运动功能评定量表(FMA)、Berg平衡量表(BBS)、医院焦虑抑郁量表(HADS)进行调查,采用多因素Logistic回归模型确定老年脑梗死患者FoF独立影响因素(MFES条目均分≥8分判定为存在FoF,反之为无FoF),通过R语言软件建立随机森林模型,进行变量重要性分析。通过受试者工作特征(ROC)曲线评价随机森林模型与多因素Logistic回归模型对老年脑梗死患者FoF的预测效能。结果:152例老年脑梗死患者中存在FoF共99例(65.13%)。FoF组年龄、近1年内跌倒史、久坐生活方式、共存疾病数量≥3种、多重用药占比、HADS抑郁分量表(HADS-D)评分均大或高于非FoF组,MMSE评分、FMA评分、BBS评分低于非FoF组(均P<0.05)。多因素Logistic回归分析显示,年龄(OR=1.078,95%CI:1.008~1.153)、近1年内跌倒史(OR=2.826,95%CI:1.149~6.952)为老年脑梗死患者FoF的独立危险因素,MMSE评分(OR=0.768,95%CI:0.664~0.888)、FMA评分(OR=0.845,95%CI:0.761~0.937)、BBS评分(OR=0.847,95%CI:0.779~0.921)为独立保护因素(均P<0.05)。随机森林模型显示,老年脑梗死患者FoF的独立影响因素重要性排序依次为FMA评分、BBS评分、MMSE评分、年龄、近1年内跌倒史。ROC曲线分析显示,随机森林模型对老年脑梗死患者FoF的预测曲线下面积(area under the curve,AUC)为0.885,略高于多因素Logistic回归模型预测AUC的0.838。结论:老年脑梗死患者FoF受下肢运动功能、平衡能力、认知功能、年龄及跌倒史的影响,建立的随机森林模型对FoF预测效能更好,有助于识别高危人群,指导临床干预。
Objective:To explore the risk factors for fear of falling(FoF)in elderly patients with cerebral infarction,establish Logistic regression and random forest models,and provide reference for FoF intervention management in patients.Methods:A total of 152 elderly patients with cerebral infarction from our hospital from October 2022 to October 2024 were included.General information questionnaire,Modified Falls Efficacy Scale(MFES),Mini-Mental State Examination(MMSE),Fugl-Meyer Assessment of Lower Extremity(FMA),Berg Balance Scale(BBS),and Hospital Anxiety and Depression Scale(HADS)were used for the investigation.Multivariate Logistic regression model was used to determine independent influencing factors of FoF in elderly patients with cerebral infarction(FoF was defined by MFES items average score≥8 points,otherwise as no FoF).Random forest model was established through R software for variable importance analysis.The predictive efficacy of random forest model and multivariate Logistic regression model for FoF in elderly patients with cerebral infarction was evaluated through receiver operating characteristic(ROC)curve.Results:Among 152 elderly patients with cerebral infarction,99 cases(65.13%)had FoF.The FoF group exhibited a higher mean age,a greater history of falls within the past year,a more sedentary lifestyle,a larger number of coexisting diseases(≥3),a higher proportion of polypharmacy use,and elevated scores on the HADS Depression Subscale(HADS-D)compared to the non-FoF group.,while MMSE scores,FMA scores,and BBS scores were lower than the non-FoF group(all P<0.05).Multivariate Logistic regression analysis showed that age(OR=1.078,95%CI:1.008-1.153)and falls history within the past year(OR=2.826,95%CI:1.149-6.952)were independent risk factors for FoF in elderly patients with cerebral infarction,while MMSE score(OR=0.768,95%CI:0.664-0.888),FMA score(OR=0.845,95%CI:0.761-0.937),and BBS score(OR=0.847,95%CI:0.779-0.921)were independent protective factors(all P<0.05).The random forest model ranked the importance of independent factors for FoF as follows:FMA score,BBS score,MMSE score,age,and fall history within the past year.ROC curve analysis showed that the random forest model had a slightly higher predictive area under the curve(AUC)of 0.885 as compared to 0.838 for the multivariate Logistic regression model.Conclusion:FoF in elderly patients with cerebral infarction is influenced by lower extremity motor function,balance ability,cognitive function,age,and history of falls.The established random forest model has better predictive efficacy for FoF,which helps identify high-risk populations and guide clinical intervention.
作者
夏丹玲
刘娟
XIA Danling;LIU Juan(Second Department of Wound Repair Extravascular Surgery,Liyuan Hospital of Tongji Medical College of Huazhong University of Science and Technology,Wuhan 430077,China;Geriatrics,Liyuan Hospital of Tongji Medical College of Huazhong University of Science and Technology,Wuhan 430077,China)
出处
《东南大学学报(医学版)》
2025年第4期654-660,共7页
Journal of Southeast University(Medical Science Edition)
