Revised Abstract: """"""""Biostatistics and Bladder Cancer Database"""""""", provides statistical input for the design, conduct, and interpretation of all laboratory and clinical studies as well as maintenance of the Bladder Cancer Research Database, a resource essential to the translational aspects of the program project. In the previous review, concerns were expressed about suboptimal integration of core biostatistical efforts with those of Project 2 and a lack of innovative statistical approaches to integrate biomarker results. The revised approach shows better integration of Core B activities with the plan for the statistical analyses of data to be collected for this project. It is clarified that Drs. Yu and Groshen have a longstanding collegial relationship and will be meeting regularly for discussions regarding implementation of the research, quality control, and the analytic plan. Methods for exploring the co-expression of COX-2 and DNMTs are elaborated with stratification for smoking status, vitamin C intake, and other variables. Power calculations have been revised to include two-tailed tests of significance, and samples sizes appear sufficient to detect approximate 10 percent differences with sufficient statistical power. Sample size considerations are given in greater detail for not only the univariate comparisons, but also for examination of two-way and three-way interactions. There is also a plan to incorporate multivariate modeling of the data for tumor stage at detection to evaluation impact of the biomarkers on progression of bladder carcinogenesis. Dr. Groshen has proposed to use resampling methods (Westiall and Young, 1993) to address overfitting issues in model selection. Data base activities of the core have been enhanced to include the management of specimen tracking and management of intermediate data. In summary, all of the issues raised in the previous review have been adequately addressed. As a result, Core B is now well-integrated with other components of the program project and provides excellent statistical support crucial to the success of this program. This core was previously rated as satisfactory and is now rated as superior.
Showing the most recent 10 out of 33 publications
