How do we know that innovations in healthcare delivery would work? In this paper, we discuss the idea of applying the methodology of group-randomized intervention studies to evaluation of surgical care policies using data from simulation experiments. We argue that a new interdisciplinary framework, which links health services research, operations research, and computer sciences, is required. Specifically, the methodological rigor of evaluative studies should be applied to the analysis of simulation experiments. In turn, the evaluation of policy initiatives should include the simulation of health-system operations. We introduce the framework and study design to evaluate methods for improving the peri-operative process with the use of patient flow simulations.