Experience with a multi crossover model in dyspepsia.

A multi cross over model (MCOM) has been designed for single case studies. The model which is only partly randomized, implies regular interchanges between treatment periods with active drug and placebo. The individual evaluation is based on the number of times the active drug is associated with less symptoms than the preceding or following placebo period (X-score), while the effect in a group is evaluated according to the X-score distribution and a paired t-test. The advantages of the single case approach and the impact of the MCOM is illustrated by the results from a study of the effect of cimetidine in non ulcer dyspepsia. Although there are several statistical objections to the model, the results from the study are reasonable and demonstrate a small degree of violence of preassumptions.