Table 1_Underestimation of cardiovascular risk by QRISK3 and RA-adapted SCORE2 in a Chinese rheumatoid arthritis population.docx

BackgroundPatients with rheumatoid arthritis (RA) face a substantially increased risk of cardiovascular disease (CVD), yet existing risk prediction models often perform poorly in this population. QRISK3 and RA-adapted SCORE2 incorporate RA in their frameworks, but their validity in Asian cohorts remains uncertain.
MethodsWe conducted a retrospective observational study using electronic hospital records from The First People's Hospital of Zhangjiagang City in China (2020–2025). Adults with confirmed RA who subsequently experienced a first major CVD event (coronary heart disease, ischemic stroke, or transient ischemic attack) were included. QRISK3 and RA-adapted SCORE2 were applied to the conventional thresholds of ≥10% and ≥5% respectively, to define high risk. Agreement between tools was assessed with Cohen's kappa and McNemar's test. Adjusted logistic regression examined demographic, RA-related, and traditional risk factors associated with risk underestimation.
ResultsA total of 249 patients with RA and CVD were included. Both tools substantially underestimated risk, with underestimation more frequent for QRISK3 than RA-adapted SCORE2. Agreement between the two was moderate (κ = 0.44), with discordance most marked across age, glucocorticoid exposure, and disease activity subgroups. Patients with high baseline DAS28 scores were particularly likely to be misclassified as low risk. In adjusted models, diabetes, chronic kidney disease, and systemic steroid use were associated with greater underestimation.
ConclusionsQRISK3 and RA-adapted SCORE2substantially underestimated cardiovascular risk in Chinese patients with RA, especially those with active disease. European-derived tools may not be reliable in this setting, underscoring the need for recalibrated or RA-specific models.

研究背景 类风湿关节炎（rheumatoid arthritis, RA）患者罹患心血管疾病（cardiovascular disease, CVD）的风险显著升高，但现有心血管风险预测模型在该人群中往往表现不佳。QRISK3与适配RA的SCORE2模型已将RA纳入其预测框架，然而二者在亚洲队列中的有效性仍未明确。
研究方法 本研究采用中国张家港市第一人民医院2020—2025年的电子住院病历开展回顾性观察研究。纳入确诊RA且随后发生首次主要CVD事件（冠心病、缺血性脑卒中或短暂性脑缺血发作）的成年患者。分别以≥10%和≥5%的常规阈值作为QRISK3与适配RA的SCORE2的高危判定标准。采用Cohen's Kappa检验与McNemar检验评估两种工具间的一致性。通过校正后的logistic回归分析与风险低估相关的人口学、RA相关及传统危险因素。
研究结果 本研究共纳入249例合并CVD的RA患者。两种模型均显著低估了患者的心血管风险，且QRISK3的风险低估发生率高于适配RA的SCORE2。二者一致性中等（κ=0.44），在年龄、糖皮质激素暴露及疾病活动度亚组中差异最为显著。基线DAS28评分较高的患者尤其容易被误判为低危人群。在校正模型中，糖尿病、慢性肾脏病及全身糖皮质激素使用与更严重的风险低估相关。
研究结论 QRISK3与适配RA的SCORE2均显著低估了中国RA患者的心血管风险，尤其是疾病活动期患者。源自欧洲的风险预测工具在该人群中可能并不适用，这凸显了对经过重新校准或针对RA特异性的心血管风险模型的迫切需求。