OBSERVATIONAL ASPIRIN USE AND CVD IN THE PHS

这个应用程序建议分析来自医生的现有数据 健康研究(PHS)，一项低剂量随机一级预防试验 阿司匹林和β胡萝卜素在22071名美国男性医生中进行了调查， 解答有关阿司匹林和心血管疾病的问题 在随机服用阿司匹林期间，这一问题不能得到充分解决 句号。试验中的阿司匹林成分于1月终止 1988年25日，由于阿司匹林对心肌有明显的益处 脑梗塞。然而，在那个时候，中风和心血管死亡的数量 受试者所经历的不足以确定 评估这些终端。 试验中的β胡萝卜素成分持续不间断，直到 它原定于1996年12月终止。在此期间 继续收集有关试验后使用阿司匹林的详细信息 通过年度问卷调查。截至1995年10月， 包括心血管死亡在内的死亡人数增加了四倍 在随机化期间，中风的数量增加了 3.5倍。因此我们建议使用来自随机阿司匹林的数据 试验后的观察期和观察期以评估 阿司匹林的使用对心血管疾病和总死亡率的影响 使用阿司匹林对随后的中风和心肌梗死的长期影响。 建议的方法将包括对两种随机阿司匹林的分析 分配和时变阿司匹林使用情况，按年度评估 问卷调查。由于潜在的偏见，倾向于 必须服用阿司匹林，尤其是在观察期。 考虑到了。分析将包括使用比例风险模型。 允许使用阿司匹林的时变效应和控制 对于时变混杂因素，以及使用 因果模型。

This application proposes to analyze existing data from the Physicians Health Study (PHS), a randomized primary prevention trial of low-dose aspirin and beta carotene conducted among 22,071 U.S. male physicians, to address questions concerning aspirin and cardiovascular (CV) disease that could not adequately be addressed during the randomized aspirin period. The aspirin component of the trial was terminated on January 25, 1988, due to a demonstrated benefit of aspirin on myocardial infarction. At that time, however, the number of strokes and CV deaths experienced by trial participants was inadequate to definitively evaluate these endpoints. The beta carotene component of the trial continued uninterrupted until its scheduled termination in December, 1996. During this period detailed information continued to be collected on post-trial aspirin use through annual questionnaires. As of October, 1995, the number of deaths, including cardiovascular deaths, had increased fourfold from that in the randomized period, and the number of strokes had increased 3.5 times. We thus propose to use data from both the randomized aspirin period and the observational period following the trial to assess the impact of aspirin use on cardiovascular and total mortality, and the long-term impact of aspirin use on subsequent stroke and MI. The proposed methods will include analyses of both randomized aspirin assignment and of time-varying aspirin use, as assessed on the annual questionnaires. Because of the potential for bias, the propensity for aspirin use, particularly during the observational period must be taken into account. Analyses will include use of proportional hazards models allowing for both time-varying effects of aspirin use and controlling for time-varying confounders, as well as more complex procedures using causal modeling.