Current research in statistical methodology for clinical trials

Statistical Methodology for Clinical Trials has advanced primarily in those domains opened up by the advent of high speed computation. Data acquisition, data management, and data monitoring have been revolutionized, and statistical analysis in clinical trials has advanced from simple cross tabulation and lists of averages to the use of special regression and other methods for estimation and for covariance adjustment. Research on sequential designs has had little effect on practice, largely because the usual sequential methods fail to address the key problems posed by ethical constraints, but a related development of research on "stopping rules" seems promising. Although considerable attention has been given to the problems of multiplicity--analysis of subgroups, choice of endpoints, post-randomization exclusions--little progress in this area has been made as yet. The paper explores these issues in some detail with the aid of illustrative examples.