摘要
目的探究急性缺血性卒中患者发生卒中后认知障碍(PSCI)的危险因素,建立风险列线图预测模型,帮助临床医生提前对易发生PSCI的人群进行预测和干预,完善早期疾病的识别、干预及预防,从而为PSCI的诊疗提供新思路。方法纳入2021年1月至2023年6月于吉林大学白求恩第一临床医院神经内科住院的急性缺血性卒中患者330例。收集其一般临床资料、实验室检查、影像学检查、神经心理学评估资料。对入组患者均在急性缺血性卒中发病7 d内进行神经心理学量表评估并完成相关检查,以此作为基线值。于发病后6个月对患者进行神经心理学量表随访,根据6个月后蒙特利尔认知评估量表(MoCA)评估结果将随访患者分为PSCI组(143例)和卒中后非认知障碍(PSNCI)组(147例)(6个月后随访脱组40例,最终共纳入研究人数为290例)。首先利用统计学方法对PSCI组与PSNCI组进行一般临床资料的比较;接着使用最小绝对值收敛和选择算子算法(LASSO)回归方法选取更有影响力的预测指标,将其纳入多因素Logistic回归分析,建立风险列线图预测模型。利用bootstrap方法重复1000次抽样,进行内部验证;绘制受试者工作特征曲线,利用曲线下面积(AUC)分析预测模型的区分度;使用校准曲线评估模型的准确度;绘制决策曲线分析(DCA)图评估模型的临床实用性。结果LASSO回归筛选出年龄、受教育水平、关键部位脑梗死、低密度脂蛋白胆固醇、脑白质高信号、脑萎缩作为风险列线图预测模型的预测指标,多因素Logistic回归分析结果显示这些预测指标是急性缺血性卒中患者易发生PSCI的独立危险因素;对建立的风险预测模型进行验证,结果显示,本风险预测模型AUC=0.890,内部验证风险预测模型AUC=0.940,提示模型具有良好区分度;在校准曲线中模型与实际预测效果具有良好的拟合度,说明该模型具有良好的准确度;DCA结果显示,该模型可较好地应用于临床。结论由年龄、受教育水平、关键部位脑梗死、低密度脂蛋白胆固醇、脑白质高信号、脑萎缩组成的风险列线图预测模型具有良好的区分度、准确度和临床实用性,可用于临床实际操作,有助于临床医生筛查易发生PSCI的患者,及时对患者进行干预,以达到更好的临床效果。
Objective To investigate the risk factors of post-stroke cognitive impairment(PSCI)in patients with acute ischemic stroke,to establish a nomogram predictive model to help clinicians predict and intervene in the people who are prone to PSCI in advance,and to improve the recognition,intervention and prevention of the disease at an early stage,so as to provide a new way of thinking for the diagnosis and treatment of PSCI.Methods Totally 330 patients with acute ischemic stroke hospitalized in the Department of Neurology,the First Bethune Hospital of Jilin University from January 2021 to June 2023 were collected.Their general clinical data,laboratory examination,imaging examination,and neuropsychological assessment data were collected.Neuropsychological scales assessment was completed within 7 days of the onset of acute ischemic stroke as a baseline value.The patients were followed up with neuropsychological scales assessment 6 months after the onset of stroke,and according to the results of the Montreal Cognitive Assessment(MoCA)scale assessment 6 months later,the patients were divided into PSCI group(143 patients)and post-stroke non-cognitive impairment(PSNCI)group(147 patients)(40 patients were removed from the study after 6 months,and a total of 290 patients were finally included in the study).Comparisons of general clinical information between the PSCI and PSNCI groups were first performed using statistical methods;then more influential predictors were selected using least absolute shrinkage and selection operator(LASSO)regression method and included in multifactor Logistic regression analyses to create a nomogram predictive model.Internal validation was performed by repeating the sampling 1000 times using the bootstrap method;receiver operating characteristic(ROC)curve and area under the curve(AUC)were plotted to analyze the discrimination of the predictive model;the accuracy of the model was assessed using calibration curves;and a decision curve analysis(DCA)diagram was plotted to assess the clinical utility of the model.Results Age,education level,critical area cerebral infarction,low-density lipoprotein-cholesterol(LDL-C),cerebral white matter hyperintensity(WMH),and cerebral atrophy were selected as the predictors of the nomogram predictive model by LASSO regression,and the results of multifactor Logistic regression analysis showed that these predictors were independent risk factors for PSCI in patients with acute ischemic stroke;the risk predictive model established was validated,and the results showed that the AUC of the present predictive model was 0.890,and the AUC of the internally validated predictive model was 0.940,suggesting that the model had a good degree of differentiation;the good fit between the calibration curve and the actual prediction results indicated that the model had good accuracy;the DCA results showed that the model can be well applied in clinical practice.Conclusion The nomogram predictive model consisting of age,education level,critical area cerebral infarction,LDL-C,WMH,and cerebral atrophy has good differentiation,accuracy,and clinical utility,and can be used in practical clinical practice,which can help clinicians screen patients who are prone to PSCI,and intervene in a timely manner to achieve better clinical outcomes.
作者
王梦贞
杨苗苗
韩哲
梁晔琨
鞠维娜
Wang Mengzhen;Yang Miaomiao;Han Zhe;Liang Yekun;Ju Weina(Stroke Center,Department of Neurology,the First Bethune Hospital of Jilin University,Changchun 130061,China;Department of Neurology,Guangzhou Red Cross Hospital,Guangzhou 510030,China;Department of Cardiology,the First Bethune Hospital of Jilin University,Changchun 130061,China)
出处
《中华神经科杂志》
北大核心
2025年第1期26-35,共10页
Chinese Journal of Neurology
基金
吉林省自然科学基金自由探索重点项目(YDZJ20241452ZYTS)。
关键词
认知障碍
卒中
危险因素
列线图预测模型
Cognition disorders
Stroke
Risk factors
Nomogram predictive model
