Gravar-mail: Strategies for handling missing data in randomised trials