An analysis in which the effects observed in the subjects of a trial are evaluated according to the intervention to which they were randomised, whether or not they received it, or whether or not they complied with the study protocol.