[The size of the effect of the difference between two means: statistically significant or clinically relevant?].