The objective is to develop statistical methods for specific application to clinical trials in arthritis, diabetes, digestive and kidney diseases. In most of these clinical trials, the principal outcomes are measures of morbidity rather than simple mortality; multiple, possibly recurrent, responses are evaluated; data are partially missing; and restricted randomization procedures are employed with small sample sizes. Seven specific research projects are identified: (1) early termination of clinical trials comparing long-term time trends for dependent continuous responses; (2) multivariate one-sided tests for censored data; (3) optimal combinations of two-sample tests based on incomplete repeated measurements of categorical data: (4) estimation and testing of recurrence rates based on interval counts for censored data; (5) generalized EM (GEM) approach to proportional hazards models when covariate data are partially missing; (6) restricted randomization designs with continuous covariates; (7) testing and estimation for censored paired observations. In each case, the methods developed will be illustrated by application to data from clinical trials in arthritis, diabetes, digestive and kidney diseases in which the G.W.U. Biostatistics Center has been involved. It is hoped that this program of statistical research will further enhance the scientific validity of the results of such clinical trials.
