Correlation in pooled groups

#1
Hello,

I evaluted 2 variables in 3 groups of subjects (~ 8 subjects/group) : (1) ill subjects, (2) a different group of subjects recovered from the same disease and (3) controls.
ANOVA and post-hoc showed
for Variable 1: (1)=(2)<>(3)
for Variable 2: (1)<>(2)<>(3)
Do we have the right to calculate correlation between these two varibles in pooled groups : (1)+(2), (2)+(3), (1)+(3) or (1)+(2)+(3)?
If not, what statistics should be use to evaluate the relationship between these two variables in these subjects?

Thank you