With the aging population in China, the incidence of vascular cognitive impairment (VCI) will increase year by year. Non-dementia vascular cognitive impairment (VCIND) is the most common form of VCI. At present, the research shows that glycolipid metabolic diseases will accelerate the process of VCI, and the treatment of VCI focuses on controlling risk factors, but there is a lack of relevant research on VCIND caused by glycolipid metabolic diseases.

To analyze the factors influencing the occurrence of VCIND with glycolipid metabolic disease, construct a regression model, and conduct risk prediction.

A total of 410 patients with glycolipid metabolic diseases who were hospitalized in the encephalopathy center of Guangdong Provincial Hospital of Traditional Chinese Medicine from March to December 2023 were selected. Patients were divided into a cognitive normal group (MMSE>26 points) and a VCIND group (MMSE≤26 points) according to the Mini-mental State Examination Scale (MMSE). Multivariate Logistic regression was used to evaluate the influencing factors of VCIND in middle-aged and elderly patients with glycolipid metabolic diseases, and the risk prediction model of VCIND in glycolipid metabolic diseases was constructed. The predictive value of the model was evaluated via the receiver's operating characteristic (ROC) curve, and the area under the ROC curve (AUC) was calculated.

Among the 410 patients, there were 209 cases in the cognitively normal group and 201 cases in VCIND. The results of multivariate Logistic regression analysis showed that low education level [below primary school (OR=25.989, 95%CI=5.656-119.427), primary school (OR=6.839, 95%CI=3.919-11.933) ], Fazekas grade (OR=1.700, 95%CI=1.124-2.570) were independent influencing factors for the occurrence of VCIND in patients with glycolipid metabolism (P<0.05). Based on the results of multivariate Logistic regression analysis, the prediction model was logit (P) =-1.608+ primary school×1.923+ below primary school×3.285+Fazekas grading×0.531. The AUC of this risk prediction regression model was 0.767 (95%CI=0.721-0.813, P<0.001). Hosmer-Lemeshow goodness-of-fit test showed that the model has a good fitting effect (χ²=13.404, P=0.099) .

Low literacy and Fazekas classification are independent influencing factors for the development of VCIND in a population of patients with glycolipid metabolism. Establishing a risk prediction regression model based on the above risk factors has a good predictive value and helps to identify the high-risk group of developing VCIND in patients with glycolipid metabolism disease at an early stage.

随着我国人口老龄化程度的不断加剧，血管性认知障碍（VCI）的发病率逐年增加。非痴呆型血管性认知障碍（VCIND）是VCI最常见的形式。目前研究表明，糖脂代谢病会加速VCI进程且VCI的治疗侧重于控制危险因素，缺少糖脂代谢病发生VCIND的相关研究。

分析糖脂代谢病患者发生VCIND的危险因素，构建回归模型并进行风险预测。

选取2023年3—12月在广东省中医院脑病中心住院的糖脂代谢病患者410例，根据简易精神状态检查量表（MMSE）将患者分为认知正常组（MMSE>26分）和VCIND组（MMSE≤26分）。采用多因素Logistic回归评估中老年糖脂代谢病患者发生VCIND的影响因素，并构建糖脂代谢病发生VCIND的风险预测模型。采用受试者工作特征（ROC）曲线评估模型的预测价值，计算ROC曲线下面积（AUC）。

410例患者中认知正常组有209例，发生VCIND 201例。多因素Logistic回归分析结果显示，低文化程度[小学以下（OR=25.989，95%CI=5.656~119.427）、小学（OR=6.839，95%CI=3.919~11.933）]、Fazekas分级（OR=1.700，95%CI=1.124~2.570）是糖脂代谢病人群发生VCIND的独立影响因素（P<0.05）。根据多因素Logistic回归分析结果建立预测模型为logit（P）=-1.608+小学×1.923+小学以下×3.285+Fazekas分级×0.531，该模型的AUC为0.767（95%CI=0.721~0.813，P<0.001），灵敏度为0.726，特异度为0.756，约登指数为0.482；Hosmer-Lemeshow拟合优度检验显示，模型拟合效果较好（χ²=13.404，P=0.099）。

低文化程度、Fazekas分级是糖脂代谢病人群发生VCIND的独立影响因素。基于此建立的风险预测回归模型，预测价值较好，有助于早期识别糖脂代谢病患者发生VCI的高危人群。