Within-Trial Decisions

3. Premature Termination

There are many reasons why a study may be terminated early, but in most cases these are pre-determined scientific or medical reasons. For example:
  • Emergence of a serious adverse reaction (ADR) where the risk to the patient is considered to be too great to continue, or
  • A pre-planned interim analysis that clearly identifies superiority of one treatment or lack of effect of a treatment.
In most cases stopping a clinical trial requires the approval of senior medical management (in most pharmaceutical companies this is the chief medical officer). There may be cases where recruitment is proving so challenging that it is considered unethical to continue the trial. In these situations, it is unlikely that the scientific question will ever be answered. The regulatory authorities that approved the study need to agree to the termination.

In most organisations conducting clinical trials there are detailed instructions within the protocol on how interim analyses should be approached and what the statistical and decision-making processes will be to evaluate the data. Under no circumstances can an interim analysis be planned to have a quick check of the data during a trial. All interim analyses have to be clearly defined; including the time point when the data will be analysed and why.

In the majority of cases an interim analysis is usually conducted to:
  • Identify whether there is an imbalance between the safety profile of the groups and therefore potentially placing some patients at risk, or
  • To assess whether there is an imbalance in efficacy against a comparator treatment.
In all cases where the possibility exists to terminate a trial following an interim analysis, there has to be clearly defined stopping rules detailed in the protocol. Stopping rules need to be described in the informed consent documentation as well. To decrease any bias, the decision on whether the data produced from an interim analysis meets the stopping criteria should be made by an independent committee.