When randomisation goes horribly wrong: examples of major failures of randomisation and strategies to avoid them

Randomisation forms the foundation of clinical trials, but its implementation can be prone to error. Often randomisation errors affect few participants and do not seriously compromise the integrity of the trial. However, in some cases randomisation errors can have widespread consequences and call into question the validity of trial conclusions. Published articles may be retracted as a result. Valuable insight can be gained from studying past errors to minimise the risk of similar errors and their disastrous consequences impacting future trials. The aims of this article are to (i) describe examples of major failures of randomisation, and (ii) provide guidance on how to avoid them in practice.

Major failures of randomisation were defined as inadvertent errors that affected many trial participants and occurred during the process of designing the randomisation scheme, generating the randomisation schedule, allocating participants to treatment groups, or providing the assigned treatment. Examples of major failures of randomisation were drawn from author experience and through a review of the published literature, which included a systematic search of the Retraction Watch Database for serious randomisation problems that led to the retraction of a published article. Practice points to avoid such errors were developed by consensus among the authors.

Examples are provided of seven broad types of major failures of randomisation: randomisation schedule followed incorrectly, randomisation schedule sorted incorrectly, randomisation schedule too short, clusters handled incorrectly, incorrect or unknown treatment provided at randomisation, poorly designed randomisation scheme, and programming errors in adaptive randomisation schemes. Practice points for avoiding such errors are presented, including suggestions for written documentation, staff training, and thorough testing of the randomisation process prior to trial commencement.

Randomisation is of fundamental importance in clinical trials. Greater consideration should be given to the potential for major failures of randomisation and strategies to avoid them. When major failures of randomisation do occur, greater transparency in reporting is needed.