The objectives of the Biostatistics Core are to provide quantitative expertise in the design of experiments, to provide flexibility for data management and quality control for clinical studies, and to assure expertise in the statistical analysis and reporting of study results. The Biostatistics Core Resource will be under the direction of Dr. Steven Piantadosi who has extensive experience in applying each of these relevant areas to lung cancer research. With regards to biostatistical design, much input has already been obtained from the core resource concerning study feasibility for stated objectives, power, and sample size. The necessary interactions with a statistician for accomplishing these goals are facilitated by the location and structure of the Biostatistics Core Resource. We propose electronic access to the data in addition to physical access to statistical consultants to facilitate queries, hypothesis generation, and exploratory analyses. With regard to data management, the diverse projects require specialized needs. For example, the early detection study requires a full fledged coordinating center with database entry, maintenance, and quality control. Because there are 11 participating centers, communications and project management become special issues. Special equipment and detailed methods are described for dealing with coordinating center issues. For projects local to Johns Hopkins, a centralized database using state of the art relational database technology is proposed. This approach is essential to integrate the clinical outcome and research laboratory data which forms the very foundation of the project. Approaches to statistical analysis and summary of results are among the least demanding goals on the project. Following a careful design in quality control data collection, statistical analyses are greatly facilitated. However, we propose a structure in function which brings together the clinician, laboratory scientist, and statistician in an attempt to explore, analyze, and report data in relevant and creative ways. This approach is facilitated by having a substantial effort on the part of biostatisticians in the project and maintaining skilled supervision by the principal investigator.
Showing the most recent 10 out of 256 publications
