The dilemma of crossover versus long term results in randomised controlled trials