Zhongguo cuzhong zazhi (Oct 2024)
脑小血管病患者认知障碍影响因素分析及列线图模型的构建与验证(Analysis of Influencing Factors of Cognitive Impairment in Patients with Cerebral Small Vessel Disease and Construction and Validation of a Nomogram Mode)
Abstract
目的 分析影响脑小血管病患者发生认知障碍的因素。 方法 回顾性选取（系统抽样）沧州市中心医院2020年1月—2022年5月收治的脑小血管病患者作为建模组，根据是否发生认知障碍分为无认知障碍组和认知障碍组。多因素logistic回归模型分析脑小血管病患者发生认知障碍的因素，于R3.6.3中构建预测脑小血管病认知障碍的列线图模型。另以建模组∶验证组=7∶3的方案收集2022年5月—2023年5月收治的脑小血管病患者为验证组，对构建的模型进行外部验证；绘制ROC曲线评价列线图模型在脑小血管病认知障碍风险评估中的区分度。 结果 本研究共纳入247例患者，建模组173例，验证组74例；建模组中61例（35.26%）存在认知障碍，验证组中24例（32.43%）存在认知障碍。多因素分析显示，CRP水平较高（OR 1.527，95%CI 1.271～1.834，P＜0.001）、高血压（OR 2.106，95%CI 1.044～4.248，P=0.037）、颈动脉粥样硬化（OR 3.917，95%CI 1.416～10.833，P=0.009）、高同型半胱氨酸血症（OR 3.220，95%CI 1.261～8.226，P=0.015）、脑白质损伤病变程度（OR 2.862，95%CI 1.496～5.475，P=0.001）是脑小血管病患者发生认知障碍的危险因素。ROC的AUC为0.834（95%CI 0.791～0.902，P＜0.001），提示列线图预测模型的区分度较好；校准曲线斜率≈1（χ2=4.388、P=0.820），提示列线图模型拟合效度较好。外部验证结果显示，ROC的AUC为0.845（95%CI 0.802～0.911，P＜0.001），校准曲线斜率接近1，拟合效度较好（χ2=6.042，P=0.302）。 结论 CRP、高血压、颈动脉粥样硬化、高同型半胱氨酸血症、脑白质损伤病变程度均可能导致脑小血管病患者发生认知障碍，基于以上指标构建的预测脑小血管病患者认知障碍发生风险的列线图模型具有较好的区分度和一致性。 Abstract: Objective To analyze the factors influencing cognitive impairment in patients with cerebral small vessel disease. Methods Patients with cerebral small vessel disease admitted to Cangzhou Central Hospital from January 2020 to May 2022 were retrospectively selected (systematic sampling) as the modeling group, and divided into the non-cognitive impairment group and the cognitive impairment group according to the occurrence of cognitive impairment. Multivariate logistic regression model was used to analyze the influencing factors of cognitive impairment in patients with cerebral small vessel disease. A nomogram model was constructed in R3.6.3 to predict cognitive impairment in cerebral small vessel disease. In addition, patients with cerebral small vessel disease admitted from May 2022 to May 2023 were collected as the validation group. The radio of the modeling group to the validation group is 7∶3. This validation group was used for external validation of the constructed model. The ROC curve was developed to evaluate the discrimination of the nomogram model in assessing cognitive impairment in cerebral small vessel disease. Results A total of 247 patients were included in this study, including 173 in the modeling group and 74 in the validation group. In the modeling group, 61 cases (35.26%) had cognitive impairment, while in the validation group, 24 cases (32.43%) had cognitive impairment. Multivariate analysis showed that high levels of CRP (OR 1.527, 95%CI 1.271-1.834, P＜0.001), hypertension (OR 2.106, 95%CI 1.044-4.248, P=0.037), carotid atherosclerosis (OR 3.917, 95%CI 1.416-10.833, P=0.009), hyperhomocysteinemia (OR 3.220, 95%CI 1.261-8.226, P=0.015), and the degree of white matter lesion (OR 2.862, 95%CI 1.496-5.475, P=0.001) were the risk factors for cognitive impairment in patients with cerebral small vessel disease. The AUC of ROC was 0.834 (95%CI 0.791-0.902, P<0.001), indicating that the discrimination of the nomogram model was good. The calibration curve slope was approximately 1 (χ2=4.388, P=0.820), indicating that the fitting degree of the nomogram model was good. External validation results showed that the AUC of ROC was 0.845 (95%CI 0.802-0.911, P<0.001), the slope of calibration curve was close to 1, and the fitting degree was good (χ2=6.042, P=0.302). Conclusions CRP, hypertension, carotid atherosclerosis, hyperhomocysteinemia, and the degree of white matter lesion may all lead to cognitive impairment in patients with cerebral small vessel disease. The nomogram model constructed with these indicators to predict the risk of cognitive impairment in patients with cerebral small vessel disease has good discrimination and consistency.
Keywords
