新危险因素在预测中年男性缺血性心血管病20年发病风险中的作用

Role of novel risk factors in predicting risk of ischemic cardiovascular diseases in middle aged men in twenty years in Shanghai

目的 评价Framingham冠心病预测模型（Framingham模型）与国人缺血性心血管病发病风险预测模型（国人模型）在预测上海市中年男性缺血性心血管病20年发病中的准确性,及预测模型引入新危险因素后的潜在附加值。方法 在1992年参加＂中国11省市队列人群心血管病发病前瞻性研究（CMCS） ＂基线时无冠心病与脑卒中的840例上海市男性作为随访队列（CMCS上海队列）,每年随访一次,记录本年度发生的心血管病事件和死亡。2007年复查基线危险因素,并增加测量新的危险因素。采用C-statistic与Hosmer-Lemeshow χ^2检验比较Framingham模型及国人模型的预测能力,并以国人模型为基本预测模型,将新危险因素依次独立添加至基本模型。通过计算受试者工作特征曲线下面积（AUC）、重新分类净改善（NRI）指数和综合判别改善（IDI）指数评价新危险因素的预测能力。结果 截止2014年12月,CMCS上海队列平均随访22.3年,在可供分析的17 617人年中,共发生冠心病事件（心肌梗死或/和冠心病死亡）24例和缺血性脑卒中45例。国人模型预测上海市中年男性20年缺血性心血管病发病风险表现良好, 其中Framingham模型：AUC=0.657 6（95%CI：0.594 2～0.724 0）,国人模型：AUC=0.726 5（95%CI：0.664 3～0.788 7）,国人模型优于Framingham模型,AUC增量为0.068 9（95%CI：0.019 6～0.117 1）,P=0.006。以国人模型为基本模型,未发现新危险因素可显著改善模型的再分类能力。＂高敏C反应蛋白＂是唯一导致NRI显著增加的危险因素。2007年风险评分可显著改善IDI,但净变化很小。结论 国人模型具有良好的预测上海市中年男性20年缺血性心血管病发病风险能力,增加＂高敏C反应蛋白＂因素可小幅度改进风险预测,但对改变临床风险的重新分类或鉴别策略收效甚微。

Objective To examine the existing Framingham Risk Score （FRS） and Chinese Risk Score （CRS） in predicting the development of ischemic cardiovascular diseases （ICVD）, and determine potential added value of novel risk factors. Methods The China Multi-Provincial Cohort Study （CMCS） was a population-based prospective cohort study in 11 provinces of China. An annual follow up was conducted in 840 men aged 35 to 64 years in Shanghai cohort, who were without coronary heart disease and stroke at baseline examination in 1992, to collect the incidence data of ICVD events （coronary death, myocardial infarction, and ischemic stroke）. The detection of novel risk factors were conducted for the cohort in 2007. The basic Framingham and Chinese prediction scores power were assessed by using C-statistic of ICVD events associated with risk scores, then the novel risk factors were evaluated by adding them independently to the basic Chinese models. The area under the curve （AUC）, net reclassification improvement （NRI）, and integrated discrimination improvement （IDI） were calculated to determine if each of the novel risk factors improved risk prediction. Results By the end of December 2014, 24 cases of coronary heart disease （myocardial infarction or/and coronary death）, 45 cases of ischemic stroke had occurred in 840 subjects in Shanghai cohort with a follow-up of 22.3 years averagely. Both the FRS and CRS had predicting power for ICVD, the AUCs were 0.657 6 （95%CI： 0.594 2-0.724 0） and 0.726 5 （95%CI： 0.664 3-0.788 7）, respectively. The incremental AUC was 0.068 9 （95%CI： 0.019 6-0.117 1） （P=0.006）. None of the novel risk factors significantly improved the AUC. High-sensitive-CRP （hs-CRP） was the only novel risk factor resulting in a significant increase of NRI. CRS in 2007 significantly improved the IDI, but net changes were small. Conclusions CRS had high power in the 20-year risk prediction for ICVD in middle-aged men in Shanghai. The inclusion of hs-CRP could make some improvement in risk prediction, but is unlikely to be meaningful when reclassification or new discrimination strategy are made which can change the clinical risk.