Xu Yikang, Ma Jingru, Yang Yang, Liu Limin, Zhao Xinran, Wang Yu, Mijiti Alimu, Cheng Qiangru, Ma Jun
Department of Cardiovascular Medicine, The Second Affiliated Hospital of Shenyang Medical College, Shenyang, China.
School of Public Heath, Shenyang Medical College, Shenyang, China.
Front Cardiovasc Med. 2024 Dec 12;11:1514103. doi: 10.3389/fcvm.2024.1514103. eCollection 2024.
To analyze the risk factors for coronary heart disease (CHD) in patients hospitalized in general hospitals in the Tacheng Prefecture, Xinjiang, and to construct and verify the nomogram prediction model for the risk of CHD.
From June 2022 to June 2023, 489 CHD patients (CHD group) and 520 non-CHD individuals (control group) in Tacheng, Xinjiang, were retrospectively selected. Using a 7:3 ratio, patients were divided into a training group (706 cases) and a validation group (303 cases). General clinical data were compared, and key variables were screened using logistic regression (AIC). A CHD risk nomogram for Tacheng was constructed. Model performance was assessed using ROC AUC, calibration curves, and DCA.
In the training group, non-Han Chinese (OR = 2.93, 95% CI: 2.0-4.3), male (OR = 1.65, 95% CI: 1.0-2.7), alcohol consumption (OR = 1.82, 95% CI: 1.2-2.9), hyperlipidemia (OR = 2.41, 95% CI: 1.7-3.5), smoking (OR = 1.61, 95% CI: 1.0-2.6), diabetes mellitus (OR = 1.62, 95% CI: 1.1-2.4), stroke (OR = 2.39, 95% CI: 1.6-3.7), older age (OR = 1.08, 95% CI: 1.1-1.2), and larger waist circumference (OR = 1.04, 95% CI: 1.0-1.1) were the risk factors for coronary heart disease (all < 0.05). The area under the curve (AUC) of the work characteristics of the subjects in the training group and the validation group were 0.80 (95% CI: 0.8-0.8) and 0.82 (95% CI: 0.8-0.9), respectively. The Hosmer-Lemeshow test indicated = 0.325 for the training group and = 0.130 for the validation group, with calibration curves closely fitting the ideal curve. The predicted values aligned well with actual values, and decision curve analysis results suggest that the model offers a net clinical benefit.
The CHD risk prediction model developed in this study for general hospitals in Tacheng Prefecture, Xinjiang, demonstrates strong predictive performance and serves as a simple, user-friendly, cost-effective tool for medical personnel to identify high-risk groups for CHD.
分析新疆塔城地区综合医院住院患者冠心病（CHD）的危险因素，并构建和验证CHD风险的列线图预测模型。
回顾性选取2022年6月至2023年6月新疆塔城地区489例CHD患者（CHD组）和520例非CHD个体（对照组）。按照7:3的比例将患者分为训练组（706例）和验证组（303例）。比较一般临床资料，采用逻辑回归（AIC）筛选关键变量。构建塔城地区CHD风险列线图。使用ROC AUC、校准曲线和DCA评估模型性能。
在训练组中，非汉族（OR = 2.93，95%CI：2.0 - 4.3）、男性（OR = 1.65，95%CI：1.0 - 2.7）、饮酒（OR = 1.82，95%CI：1.2 - 2.9）、高脂血症（OR = 2.41，95%CI：1.7 - 3.5）、吸烟（OR = 1.61，95%CI：1.0 - 2.6）、糖尿病（OR = 1.62，95%CI：1.1 - 2.4）、中风（OR = 2.39，95%CI：1.6 - 3.7）、年龄较大（OR = 1.08，95%CI：1.1 - 1.2）和腰围较大（OR = 1.04，95%CI：1.0 - 1.1）是冠心病的危险因素（均P < 0.05）。训练组和验证组受试者工作特征曲线下面积（AUC）分别为0.80（95%CI：0.8 - 0.8）和0.82（95%CI：0.8 - 0.9）。Hosmer - Lemeshow检验显示训练组χ² = 0.325，验证组χ² = 0.130，校准曲线与理想曲线拟合良好。预测值与实际值吻合良好，决策曲线分析结果表明该模型具有净临床效益。
本研究为新疆塔城地区综合医院开发的CHD风险预测模型具有较强的预测性能，是医务人员识别CHD高危人群的一种简单、易用、经济有效的工具。
