Zaslat SMS: A Bayesian cost-benefit approach to sample size determination and evaluation in clinical trials