[Comparation of the performance of different cardiovascular risk models in predicting cardiovascular diseases risk among Chinese patients with rheumatoid arthritis]

Abstract

Objective: To compare the performance of four different cardiovascular risk models in predicting cardiovascular diseases (CVD) risk among Chinese patients with rheumatoid arthritis (RA). Methods: This retrospective cohort study included RA patients treated at Sun Yat-sen Memorial Hospital between January 2001 and July 2024. Clinical data were collected, and the main outcome was the events of CVD. Receiver operating characteristic (ROC) and calibration curves were used to evaluate the predictive performance of four CVD risk models in predicting CVD risk among Chinese patients with RA, including Framingham Risk Score (FRS), the Systematic Coronary Risk Evaluation 2 (SCORE2), the Prediction for Arteriosclerotic Cardiovascular Disease Risk in China (China-PAR) and the Expanded cardiovascular Risk Score for RA (ERS-RA). Results: A total of 1 334 RA patients were included in this study, 1 099 females (82.4%), with a mean age of (52.0±12.6) years. During a median follow-up of 6.16 (2.81, 9.07) years, 143(10.7%) patients developed CVD (incidence rate: 16.6 per 1 000 person-years). The ROC analysis revealed good discrimination across four models, with area under the curve (AUC) values ranging from 0.756 to 0.806. Among them, the China-PAR model showed the highest AUC (0.806, 95%CI: 0.760-0.852), which was significantly higher than SCORE2 (0.756, 95%CI: 0.704-0.808) and ERS-RA (0.763, 95%CI: 0.711-0.814), but was comparable to the FRS model (0.790, 95%CI: 0.743-0.838). The calibration curves revealed significant deviations between predicted risk and actual risk for all the four models (goodness-of-fit test, all P<0.05), all the standardized incidence ratios (SIR) were less than 1, indicating that all models underestimated the 10-year CVD risk in the RA patients. Based on the low-risk thresholds of the four models, 30.1% (China-PAR, 43/143) to 44.8% (ERS-RA, 64/143) of Chinese RA patients who experienced CVD events were misclassified as low-risk. Conclusion: The China-PAR model may be currently more suitable for CVD risk assessment in Chinese RA patients with caution of underestimation.