Mthods For Missing and auxiliary Data in clinical trials

点击翻译按钮获取中文摘要

Randomized clinical trials are and will continue to be the key vehicle for evaluation of new and existing cancer therapies. This revolutionary era of advances in the biological sciences is leading to the discovery of novel biomarkers and complex genetic and genomic information that may be highly associated with various clinical outcomes, offering the tantalizing opportunity to exploit this information to both improve the precision of the analyses of trials and to develop models of longitudinal disease progression that may reveal important insights. A recurrent challenge is that missing data and subject drop-out are commonplace, presenting complications for analyses of these trials. Through a series of aims addressing these issues, this project proposes research that will have a significant impact on the quality and strength of inferences possible from current cancer clinical trials. That it is possible to improve efficiency of primary analyses of clinical trials by exploiting prognostic baseline auxiliary information is well known; however, such analyses are controversial because of the temptation to choose the analysis that leads to the most dramatic treatment effect. In the first aim, new methods for such "covariate adjustment" will be studied that circumvent this issue and can improve over existing approaches. In the second aim, these methods will be extended so that they may be used in the common case where outcomes are missing due to drop-out. Efficient methods for longitudinal analysis of measures such as quality of life and biomarkers in the presence of drop-out will also be developed. Understanding the relationship between such longitudinal measures and clinical outcomes such as time to recurrence or survival time is of key importance. The third aim focuses on development of methods for assessing the correctness of so-called joint statistical models used for this purpose and for assessing the influence of particular observations on the fit ofthe model, where the data used to develop the model may be missing. Finally, taking appropriate account of missing data sometimes requires unverifiable assumptions about why the data are missing, which are incorporated in models that thus cannot be checked based on the data. The fourth aim is devoted to development of a new statistical framework for assessing how sensitive conclusions are to the modeling assumptions made.