It is not uncommon for investigators to analyze clinical trial data several times as they accumulate during the planned course of the study. Thus, e.g., in a trial planned to last for a period of five years, groups may be compared annually or, even, every six months. This is done first for safety, to ensure that the participants in the trial are not being harmed by the intervention, but also to enable the early stopping of trials when efficacy is clearly established; or in case it is unlikely that efficacy will be established when all the data are in. These ethical safeguards require that adjustments to analysis strategies be made to account for the fact that multiple testing affects the overall significance level of the tests. For example, if five tests are done, each at the nominal 5% level of significance, the probability that one or more will be significant even when the groups do not differ is over 20%.

In this Workshop we elaborate on the philosophical and ethical dimensions of study monitoring, and present some of the statistical methods that have been developed to control inflation of the type I error rate associated with multiple testing. Particular attention is paid to clinical equipoise as the ethical anchor of clinical trials and the use of Data and Safety Monitoring Boards as exclusive, apartheid keepers of the accumulating data.

Anyone interested in monitoring clinical trials.

Familiarity with basic statistical concepts and procedures, say at the level of CSCAR’s Workshop, “Statistics: A Review.” Knowledge of the basic structure of RCTs, as covered in CSCAR’s Workshop, “Randomized Controlled Trials” would be helpful, but not necessary.