Brennan Kahan NIHR Doctoral Research Fellow Using re-randomisation in clinical trials to increase patient recruitment Brennan Kahan NIHR Doctoral Research Fellow
Challenges in recruitment 45% of publically funded UK trials fail to recruit to target1 Implication: Best method of patient care unknown 1. Sully BG, Julious SA, Nicholl J. A reinvestigation of recruitment to randomised, controlled, multicentre trials: a review of trials funded by two UK funding agencies. Trials. 2013;14:166.
The SWIM trial RCT comparing ibuprofen vs. placebo for sickle cell patients presenting to A&E with a pain crisis Poor recruitment -> trial terminated early
The SWIM trial Small patient group, large number of episodes Parallel group designs miss out on >60% of episodes
Re-randomisation trials Parallel group Re-randomisation
Estimated recruitment (sickle cell trial) Re-randomisation vs Estimated recruitment (sickle cell trial) Re-randomisation vs. parallel group
Observed recruitment benefit in 11 re-randomisation trials % increase in recruited episodes Median: 31% IQR: 25% - 47% Range: 16% - 119%
Does re-randomisation work? Does it: Provide unbiased estimates of treatment effect Correct type I error rate Adequate power Yes (if used appropriately) Setting Design Analysis
Setting requirements Some patients may require treatment on multiple occasions The intervention(s) would be used for each new treatment episode
Design requirements Patients are only re-enrolled and re-randomised when they have completed the follow-up period from their previous randomisation Randomisations for the same patient are performed independently
Analysis issues Research still ongoing to determine best analysis approach Analyse episodes independently (i.e. ignore ‘patient’) Counter-intuitive, but it works More complicated analyses (GEEs, mixed-effects models) can introduce bias
Treatment effect estimates in re-randomisation vs Treatment effect estimates in re-randomisation vs. parallel group trials (febrile neutropenia)
Settings where re-randomisation has been used Episodes of febrile neutropenia Asthma exacerbations Sickle cell pain crises Influenza vaccine Smoking cessation Complications from cirrhosis Pregnancy Release from prison
Summary Re-randomisation is applicable to a wide variety of clinical areas It can increase the recruitment rate Can provide valid results
Acknowledgements and references Collaborators Sandra Eldridge Richard Hooper Rupert Pearse Caroline Doré Tim Morris Andrew Forbes Shaun Seaman Erica Harris Funding Brennan Kahan is funded by a National Institute for Health Research Doctoral Research Fellowship References Kahan BC. Using re-randomization to increase the recruitment rate in clinical trials – an assessment of three clinical areas. Trials 2016 17:595. Kahan BC, Forbes AB, Doré CJ, Morris TP. A re-randomisation design for clinical trials. BMC Medical Research Methodology 2015 15:96.