In contrast to new medicinal products, surgical interventions have many features that complicate their formal assessment through Randomised Clinical Trials. For example, surgery is delivered by multidisciplinary teams; hence, differential effects on the outcome are not solely caused by differences in the leading operator's skill but are also induced by surgical team differences and patient characteristics. This study focuses on how statistical methods can be used to accommodate the multicomponent nature of the delivery of surgical interventions. Hierarchical models with cross-classifications between components of surgery, applied to historic datasets, can be used during the trial planning phase to establish the effects and interactions between different components. Methods are illustrated using two influential components of the intervention, the surgeon and the anaesthetist, in a cohort of cardiac surgery cases. The statistical implications for trial design and analysis are presented. Copyright © 2016 John Wiley & Sons, Ltd.