Probable deficits of randomisation and of proper application of the test cannot be repaired by performing a follow-up study. The results of such a study remain biased.