Abstract: The Childhood & Adolescent Migraine Prevention (CHAMP) trial was designed to test common prevention strategies and treatments for children and adolescents with migraine, with an ultimate goal of determining the optimal treatment for the prevention of migraines in this population. At the time the study was initiated, the existing evidence for the treatment of childhood and adolescent migraine was based on results from adult headache studies. There were many differences of opinions among clinical experts as to whether results from these adult studies would apply to children. Many argued that there was no reason to believe that this was not the case, and it would be improper to expose children to risk needlessly. Others argued that children are not “little adults”, and that what might be effective and safe in adults may not have the same properties in children. The study randomized kids to either amitriptyline or topiramate (both commonly used to treat migraines, but unproven to be effective in kids) or placebo in a 2:2:1 manner. Thus, the study essentially involved a “3 trials in 1” approach. At the first interim analysis, the Data and Safety Monitoring Board elected to stop the study early for futility. Surprisingly, it turned out that placebo did just as well as either active treatment in preventing migraines. Regardless of treatment, the majority of children and adolescents enrolled into the study responded to treatment. Stopping the study early saved taxpayer dollars, but also stopped children from taking powerful drugs that did nothing to prevent their migraines. The results of CHAMP were published in the New England Journal of Medicine, and named as one of the top 10 advances in pediatric medicine in 2016 by the journal editors. The story was significant enough to be picked up by the New York Times and the Washington Post.
This type of trial is one that a pharmaceutical company would likely never run; doctors were already prescribing these adult migraine drugs to children and, if they were proven ineffective, the company would lose money. Given the high stakes of such a trial, it is critical for the study to be designed and analyzed appropriately. In this presentation, I review many of the statistical aspects that had to be addressed during the design (choice of primary outcome, approach for handling missing data, sample size calculations), implementation (interim monitoring, early stopping), and dissemination of the trial results. In particular, since the primary objective of the trial was to empirically determine the best choice of prevention medication for current practice, the trial required an ‘a priori’ decision making process to incorporate data from all three primary comparisons of interest. This involved a detailed simulation study to thoroughly examine the properties of the ‘3 trials in 1’ approach under a variety of different scenarios. Through the simulation study, we show that the “power” of the study for addressing the primary objective is greater than the “sum of its parts” – i.e., the power for any single hypothesis test.