The goal of this research project is to carry out an extensive investigation of several areas of statistical methodology for the design and analysis of biomedical studies, planned or observational, applicable to various areas of health research including cancer, toxicology, environmental health and epidemiology. The objective is to provide more efficient statistical methods to achieve valid conclusions at less cost in terms of time and sample size. The research falls into six main categories: (a) Interim monitoring of clinical trials: (b) Design and analysis of long-term animal tumorigenicity bioassays; (c) Detection and surveillance of disease clusters; (d) Performance measures for diagnostic tests; (e) Quality control procedures for laboratory analyses; (f) Modeling and analysis of multitype recurrent events in longitudinal studies. Specific projects include the use of group sequential designs and repeated confidence intervals in clinical trials with particular emphasis on multiple endpoints and longitudinal data. An important advantage of this approach is that its flexibility allows inferences to be drawn independent from any stopping rule. It is planned to investigate the design of efficient and robust interim sacrificing schedules in long-term animal studies. Statistical methodology for the detection of clusters of disease will be extended to account for heterogeneous populations and the existence of multiple putative sources of hazard. Applications will be made to geocoded cancer and suicide data. Nonparametric estimators of performance measures of health or nutritional status will be studied using asymptotic theory and applied to diabetes and anemia data from the NHANES-II survey. Optimal use of controls, standards and blind replicate samples will be investigated as part of a quality control system for laboratory, analyses.
