Design and analysis of randomised controlled trials not seeking marketing authorisation: when and how to adjust for multiplicity