Gravar-mail: Risk of selection bias in randomised trials