中国多省市心血管病危险因素队列研究与美国弗莱明翰心脏研究结果的比较

Comparison between the results from the Chinese Multi-provincial Cohort Study and those from the Framingham Heart Study

目的 对比分析我国多省市心血管病危险因素队列研究 (ChineseMulti ProvincialCohortStudy ,CMCS)与美国弗莱明翰心脏研究的结果 ,探讨弗莱明翰预测模型在我国人群中的适用性。方法 分析CMCS队列 35～ 6 4岁 30 12 1人 10年的随访结果和弗莱明翰队列 30～ 74岁 5 2 5 1人的 12年随访结果 ,比较两人群 10年冠心病发病危险、基线危险因素水平、各危险因素的相对危险度以及CMCS模型和弗莱明翰模型的预测效果。结果 (1)CMCS队列男女两性的 10年发病危险分别为1 5 %和 0 6 % ,显著低于弗莱明翰队列男女两性的 8 0 %和 2 8%。 (2 )CMCS队列男女两性高胆固醇血症、高血压、男性低高密度脂蛋白胆固醇 (HDL C)血症和女性的吸烟率低于弗莱明翰队列 ,男性吸烟率高于弗莱明翰队列。 (3)与弗莱明翰队列中各危险因素的相对危险度 (RR)相比 ,CMCS队列中男性年龄的RR较高且差异有显著性 ;CMCS队列男性总胆固醇 (TC)为 5 2 0～ 6 2 3mmol L时、HDL C低于 0 91mmol L时以及女性吸烟时的RR较低 ,差异有显著性。其他多数危险因素的RR在两队列中的差异未达统计学显著性。 (4)CMCS模型和弗莱明翰模型都有较好的判别病例和非病例的能力。CMCS模型和弗莱明翰模型男性ROC曲线下面积分别为 0 736和 0 70 5 。

Objective To evaluate the validation of incorporating the Framingham predication model in Chinese population, by comparing the results from Chinese Multi-provincial Cohort Study (CMCS) with the results from the Framingham Study. Methods In CMCS, 30 121 participants of Han ethnic, aged 35 to 64 years, were recruited from 11 provinces in China. The subjects of the Framingham Heart Study were 5251 white Americans in Framingham, Massachusetts, US, who were 30-74 years old at baseline. The 10-year “hard” CHD risk, baseline risk factors, the association of risk factors with CHD in the two cohorts, as well as the performance of CMCS model and the Framingham model in this Chinese population were compared. Results (1) The 10-year hard CHD risks were much lower in CMCS cohort (1.5% for men and 0.6% for women, respectively), compared with the rates in the Framingham cohort (8.0% for men and 2.8% for women, respectively). (2) The risk factor profile in CMCS baseline were more favorable than those of Framingham baseline, except for smoking in men. (3) For most risk factors, differences in the magnitude of the Cox regression coefficients were not statistically significant, with a few exceptions. (ie. age, TC in range of 200 to 239 mg/dl and HDL-C below 35 mg/dl in men, and smoking in women). (4) The discrimination of the Framingham model in CMCS cohort was nearly as good as the CMCS model. Areas under the receiver operating characteristic curves (AUC) were 0.705 for men and 0.742 for women if predicted by Framingham model, whereas the AUCs were 0.736 and 0.759 for men and women, respectively if predicted by CMCS model. (5) The CMCS model predicted the CHD risk in CMCS cohort accurately, while the unadjusted Framingham model systematically overestimated the risk. Recalibration of the Framingham model (adjusted), by using the mean values for risk factors and average CHD event rates of CMCS cohort, substantially improved the performance of the Framingham model in CMCS cohort. Conclusions The CHD rates and risk factor levels in the CMCS cohort were lower than those in the Framingham cohort. The CMCS model accurately estimates the CHD risk of CMCS cohort. The unadjusted Framingham model systematically overestimated the risk of CMCS subjects, so it should not be incorporated into Chinese guidelines directly, unless a recalibration procedure is taken.