Community Medicine, Zone 3 Primary Health Care Centre, Albacete, Spain.
PLoS One. 2013 Sep 5;8(9):e73529. doi: 10.1371/journal.pone.0073529. eCollection 2013.
The question about what risk function should be used in primary prevention remains unanswered. The Framingham Study proposed a new algorithm based on three key ideas: use of the four risk factors with the most weight (cholesterol, blood pressure, diabetes and smoking), prediction of overall cardiovascular diseases and incorporating the concept of vascular age. The objective of this study was to apply this new function in a cohort of the general non Anglo-Saxon population, with a 10-year follow-up to determine its validity.
The cohort was studied in 1992-94 and again in 2004-06. The sample comprised 959 randomly-selected persons, aged 30-74 years, who were representative of the population of Albacete, Spain. At the first examination cycle, needed data for the new function were collected and at the second examination, data on all events were recorded during the follow-up period. Discrimination was studied with ROC curves. Comparisons of prediction models and reality in tertiles (Hosmer-Lemeshow) were performed, and the individual survival functions were calculated.
The mean risks for women and men, respectively, were 11.3% and 19.7% and the areas under the ROC curve were 0.789 (95%CI, 0.716-0.863) and 0.780 (95%CI, 0.713-0.847) (P<0.001, both). Cardiovascular disease events occurred in the top risk tertiles. Of note were the negative predictive values in both sexes, and a good specificity in women (85.6%) and sensitivity in men (79.1%) when their risk for cardiovascular disease was high. This model overestimates the risk in older women and in middle-aged men. The cumulative probability of individual survival by tertiles was significant in both sexes (P<0.001).
The results support the proposal for "reclassification" of Framingham. This study, with a few exceptions, passed the test of discrimination and calibration in a random sample of the general population from southern Europe.
在初级预防中应该使用哪种风险函数的问题仍然没有答案。弗雷明汉研究提出了一种新的算法，基于三个关键理念：使用权重最大的四个风险因素（胆固醇、血压、糖尿病和吸烟）、预测整体心血管疾病和纳入血管年龄的概念。本研究的目的是在一个非盎格鲁-撒克逊人群的队列中应用这种新功能，并进行 10 年的随访，以确定其有效性。
该队列于 1992-94 年和 2004-06 年进行了研究。样本包括 959 名随机选择的年龄在 30-74 岁之间的代表性人群。在第一个检查周期中，收集了新功能所需的数据，在第二个检查周期中，记录了随访期间的所有事件数据。使用 ROC 曲线研究了区分度。进行了预测模型与现实的三分位数（Hosmer-Lemeshow）比较，并计算了个体生存函数。
女性和男性的平均风险分别为 11.3%和 19.7%，ROC 曲线下面积分别为 0.789（95%CI，0.716-0.863）和 0.780（95%CI，0.713-0.847）（P<0.001，均）。心血管疾病事件发生在风险最高的三分位数。值得注意的是，在两性中，阴性预测值均为阳性，并且在女性（85.6%）和男性（79.1%）心血管疾病风险较高时，特异性较好，敏感性较高。该模型高估了老年女性和中年男性的风险。两性的个体生存累积概率按三分位数显著（P<0.001）。
结果支持“重新分类”弗雷明汉的建议。除了一些例外，本研究通过了对来自南欧的一般人群的随机样本的区分和校准测试。
