Multiplicity in randomised trials I: endpoints and treatments