Prediction Model Construction and Validation of Cardiovascular Event Risk in Patients with Type 2 Diabetes

WANG Rui; WANG Cuijuan; YIN Fuzai; ZHANG Mengmeng; HAN Gailing; LU Qiang; MA Chunming; LU Na; WANG Jiamei

doi:10.12102/j.issn.1672-1349.2025.19.020

Abstract：Objective: To analyze the influencing factors of cardiovascular events (CVE) in patients with type 2 diabetes (T2DM) and to construct and validate a risk prediction model. Methods: A total of 330 T2DM patients admitted to the First Hospital of Qinhuangdao, Hebei Province from December 2017 to December 2019 were selected. Among them, 280 patients were assigned to the modeling group, and those who experienced CVE during a 2-year follow-up were classified as the CVE group, while those without CVE were classified as the non-CVE group. Clinical data of the two groups were compared. After modeling, the remaining 50 patients were used as the validation group. Multivariate logistic regression analysis was used to identify the influencing factors of CVE in T2DM patients, and a nomogram prediction model was established using R software. The model was then validated. Results: Among the 280 T2DM patients included in the study, 34 patients were lost to follow-up, and 51 patients experienced CVE, resulting in a CVE incidence rate of 20.73%. The CVE group had higher age and waist circumference compared to the non-CVE group (P < 0.05). The proportion of smoking history, diastolic blood pressure, triglycerides (TG), low-density lipoprotein cholesterol (LDL-C), and glycated hemoglobin (HbA1c) were also higher in the CVE group (P < 0.05). Multivariate logistic regression analysis showed that age (OR = 1.121), smoking history (OR = 5.946), diastolic blood pressure (OR = 1.087), waist circumference (OR = 1.424), TG (OR = 8.690), LDL-C (OR = 7.186), and HbA1c (OR = 2.411) were independent risk factors for CVE in T2DM patients. These factors were further incorporated into a nomogram to predict the risk of CVE in T2DM patients, with a C-index of 0.921. The receiver operating characteristic (ROC) curve showed that LDL-C and waist circumference had higher diagnostic efficacy in predicting the risk of CVE in T2DM patients compared to age, diastolic blood pressure, TG, and HbA1c. The calibration curve of the nomogram had a slope close to 1, indicating good discrimination and consistency between the predicted and observed risks of CVE in T2DM patients. Conclusion: The nomogram prediction model incorporating variables such as age, smoking history, diastolic blood pressure, waist circumference, TG, LDL-C, and HbA1c can effectively predict the risk of CVE in T2DM patients.