Pacific D
Sample Size Calculations in n-of-1 Trials (306788)
Christopher Henry Schmid, Brown UniversityJon Arni Steingrimsson, Brown University
*Jiabei Yang, Brown University
Keywords: personalized medicine, n-of-1 trials, sample size, multilevel models, shrinkage estimates
Randomized controlled trials are the gold standard for estimating average treatment effects in a population, but are less effective in describing treatment effect heterogeneity. N-of-1 trials, trials carried out in individuals, are able to estimate efficacy for individuals directly; combining n-of-1 trials increases precision for individual as well as population average treatment effects. We derive sample size formulae for estimating population average treatment effect and precision for individual-specific treatment effect under both fixed- and random-effects analysis, assuming various randomization schemes and correlation structure among repeated measurements taken on individual patients. We will present figures to inform designs balancing the relative effects of: 1) number of crossover periods and number of measurements taken per period in individual trials and 2) number of individuals and number of measurements per individual in a series of N-of-1 trials.