Menotti A, Puddu P E, Lanti M, Kromhout D, Blackburn H, Nissinen A
Division of Epidemiology, University of Minnesota, Minneapolis, Minnesota, USA.
Eur J Epidemiol. 2003;18(2):113-22. doi: 10.1023/a:1023092415353.
The 'accelerated failure time model' (AFT) was tested in the 25-year experience of the Seven Countries Study, to describe the shape of hazard for coronary heart disease (CHD) mortality during long-term follow-up. Sixteen cohorts of men aged 40-59 at entry were enrolled in eight countries (USA, Finland, The Netherlands, Italy, Croatia, Serbia, Greece and Japan) for a total of 12,763 subjects. Risk factors were measured and CHD mortality data collected during 25 years. The AFT incorporating the Weibull survival distribution was applied to the pooled cohorts of each country, with CHD mortality as endpoint. Cumulative hazards and the annual hazard for CHD mortality were estimated during 25 years and compared among countries. The annual hazard for CHD in 25 years tended to reduce relatively or flatten out in the northern American and northern European countries, exponentially increased in Serbia and Japan, and increased moderately in the other countries of southern Europe. Positive correlations were found between the shape of the hazard curve and population mean risk factor changes (serum cholesterol and systolic blood pressure) observed during the first 10 years follow-up. The Japanese group was an outlier. The countries with the largest relative decline in the annual hazard function were the same where, during the same historical period, the largest decreases in official death rates from CHD were recorded and vice versa. The curve shape in the annual hazard for CHD mortality in different countries seems related to changes in mean levels of serum cholesterol and systolic blood pressure.
在七国研究的25年经验中对“加速失效时间模型”(AFT)进行了测试,以描述长期随访期间冠心病(CHD)死亡风险的形状。八个国家(美国、芬兰、荷兰、意大利、克罗地亚、塞尔维亚、希腊和日本)招募了16组年龄在40 - 59岁之间的男性队列,共计12763名受试者。测量了风险因素,并收集了25年期间的冠心病死亡率数据。将包含威布尔生存分布的AFT应用于每个国家的合并队列,以冠心病死亡率作为终点。估计了25年期间冠心病死亡的累积风险和年度风险,并在各国之间进行了比较。25年期间冠心病的年度风险在北美和北欧国家往往相对降低或趋于平稳,在塞尔维亚和日本呈指数增长,在南欧的其他国家适度增加。在随访的前10年中观察到的风险曲线形状与人群平均风险因素变化(血清胆固醇和收缩压)之间存在正相关。日本组是一个异常值。年度风险函数相对下降最大的国家与在同一历史时期冠心病官方死亡率下降最大的国家相同,反之亦然。不同国家冠心病死亡年度风险的曲线形状似乎与血清胆固醇和收缩压平均水平的变化有关。