STATISTICAL METHODS FOR CASE CONTROL SURVIVAL DATA

申请人的描述这个为期两年的项目的目的是开发和测试新的统计方法来分析患病率病例对照生存数据。 患病率病例对照生存研究可能是研究疾病发作时间以及评估危险因素与患病概率之间关系的有效方法。 在许多情况下，患病率病例对照生存研究比发病率病例对照或前瞻性队列研究更有效。 不幸的是，目前没有统计方法可用于对患病率病例对照生存研究的数据进行有效分析。该项目的具体目标是： 1) 将在单样本问题（分析中不包含协变量）的情况下开发的患病率病例对照研究数据的分析方法扩展到双样本问题的情况下，即开发估计和比较两个不同风险组的生存分布的方法； 2) 使用目标 (1) 中开发的方法分析腺瘤性息肉（腺瘤）患者患病率病例对照样本的生存数据； 3) 开发回归方法来分析患病率病例对照生存数据； 4) 开发用户友好的软件，用于分析患病率病例对照研究的生存数据。新方法的开发基于将病例对照可能性嵌入到预期可能性中的病例对照数据分析的常用方法的扩展，以及右截断生存数据的分析方法。该提案的方法学和流行病学相关性在于，它提供了对患病率病例对照生存数据进行分析的适当且有效的方法。

APPLICANT'S DESCRIPTION The aim of this two year project is to develop and test new statistical methods for the analysis of prevalence case-control survival data. Prevalence case-control survival studies may be a powerful approach to study time to onset of disease, and to evaluate the association between risk factors and probability of developing the disease. In many cases, prevalence case-control survival studies are more efficient than incidence case-control or prospective cohort studies. Unfortunately, no statistical methods are currently available for an efficient analysis of data from prevalence case-control survival studies. The specific aims of this project are: 1) To extend methods for the analysis of data from prevalence case-control studies developed in the setting of the one-sample problem (no covariates included in the analysis) to the setting of the two-sample problem, i.e., to develop methods for estimating and comparing survival distributions for two different risk groups; 2) To use the methods developed in Aim (1) to analyze survival data from a prevalence case-control sample of patients with adenomatous polyps (adenomas); 3) To develop regression methods for the analysis of prevalence case-control survival data; 4) To develop user-friendly software for the analysis of survival data from prevalence case-control studies. Development of the new methods is based on an extension of the usual approach to the analysis of case-control data of embedding the case-control likelihood into a prospective likelihood, and on methods of analysis of right truncated survival data. The methodological as well as epidemiological relevance of this proposal is that it provides appropriate and efficient methods of analysis of data from prevalence case-control survival data.