Shi Rong, Wu Birong, Niu Zheyun, Sun Hui, Hu Fan
School of Public Health, Shanghai University of Traditional Chinese Medicine, Shanghai, People's Republic of China.
Diabetes Metab Syndr Obes. 2020 Dec 18;13:5025-5036. doi: 10.2147/DMSO.S273880. eCollection 2020.
This study aimed to study risk factors for coronary heart disease (CHD) in type 2 diabetes mellitus (T2DM) patients and establish a clinical prediction model.
A total of 3402 T2DM patients were diagnosed by clinical doctors and recorded in the electronic medical record system (EMRS) of six Community Health Center Hospitals from 2015 to 2017, including the communities of Huamu, Jinyang, Yinhang, Siping, Sanlin and Daqiao. From September 2018 to September 2019, 3361 patients (41 patients were missing) were investigated using a questionnaire, physical examination, and biochemical index test. After excluding the uncompleted data, 3214 participants were included in the study and randomly divided into a training set (n = 2252) and a validation set (n = 962) at a ratio of 3:1. Through lead absolute shrinkage and selection operator (LASSO) regression analysis and logistic regression analysis of the training set, risk factors were determined and included in a nomogram. The C-index, receiver operating characteristic (ROC) curve, calibration plot and decision curve analysis (DCA) were used to validate the distinction, calibration and clinical practicality of the model.
Age, T2DM duration, hypertension (HTN), hyperuricaemia (HUA), body mass index (BMI), glycosylated haemoglobin A1c (HbA1c), high-density lipoprotein (HDL-C) and low-density lipoprotein (LDL-C) were significant factors in this study. The C-index was 0.750 (0.724-0.776) based on the training set and 0.767 (0.726-0.808) based on the validation set. Through ROC analysis, the set area was 0.750 for the training set and 0.755 for the validation set. The calibration test indicated that the S:P of the prediction model was 0.982 in the training set and 0.499 in the validation set. The decision curve analysis showed that the threshold probability of the model was 16-69% in the training set and 16-73% in the validation set.
Based on community surveys and data analysis, a prediction model of CHD in T2DM patients was established.
本研究旨在探讨2型糖尿病（T2DM）患者冠心病（CHD）的危险因素，并建立临床预测模型。
2015年至2017年，共有3402例T2DM患者由临床医生诊断，并记录在花木、晋阳、银航、四平、三林和大桥六个社区卫生中心医院的电子病历系统（EMRS）中。2018年9月至2019年9月，采用问卷调查、体格检查和生化指标检测对3361例患者（41例缺失）进行调查。排除不完整数据后，3214名参与者纳入研究，并按3:1的比例随机分为训练集（n = 2252）和验证集（n = 962）。通过对训练集进行套索（LASSO）回归分析和逻辑回归分析，确定危险因素并纳入列线图。采用C指数、受试者工作特征（ROC）曲线、校准图和决策曲线分析（DCA）对模型的区分度、校准度和临床实用性进行验证。
年龄、T2DM病程、高血压（HTN）、高尿酸血症（HUA）、体重指数（BMI）、糖化血红蛋白A1c（HbA1c）、高密度脂蛋白（HDL-C）和低密度脂蛋白（LDL-C）是本研究中的显著因素。基于训练集的C指数为0.750（0.724 - 0.776），基于验证集的C指数为0.767（0.726 - 0.808）。通过ROC分析，训练集的曲线下面积为0.750，验证集的曲线下面积为0.755。校准测试表明，预测模型在训练集的S:P为0.982，在验证集的S:P为0.499。决策曲线分析表明，模型在训练集的阈值概率为16% - 69%，在验证集的阈值概率为16% - 73%。
基于社区调查和数据分析，建立了T2DM患者CHD的预测模型。
