Dear Doug,

In the case of a 3x2 design the post hoc contrast for evaluating the interaction effect would be as follows:
D1Male
D1Female
D2Male
D2Female
D3Male
D3Female

ANOVA to test the effect of diagnosis (D1, D2, D3) only for males
 [ 1  0  -1  0  0  0]/2
 [ 1  0  0  0  -1  0]/2

ANOVA to test the effect of diagnosis (D1, D2, D3) only for females
 [ 0  1  0  -1  0  0]/2
 [ 0  1  0  0  0  -1]/2

Is this correct?

Many thanks again.
Gabriel