I Used % (per cent) instead of count for subjects. 5 % is equal to 1 subject. Let's do it with subject counts:
In the beginning there is 20 subject distributed as 13 in IGA 3 (moderate psoriasis) and 7 in IGA 2 (mild psoriasis) subjects.
At the end there is still 20 subjects but distributed now: 3 in IGA 0, 8 in IGA 1, 3 in IGA 2 and 6 in IGA 3.
Those three in IGA 0 must have come either IGA 2 or IGA 3. The worst case for Prurisol and the original IGA 3 group is if all subjects in ending IGA 0 group come from the starting IGA 2 group. 3 subjects out of original 7 subjects leaves 4. At the end these remaining 4 can be in IGA 1, 2 or 3. The worst case for is, again, if they all occupy the lowest possible ending group, which is this time IGA 1. But IGA 1 slot has room for 8 and we have only 4 subjects left in original IGA 2 group. HENCE in order to fill the ending IGA 1 group we need to take additional 4 subjects from the starting IGA 3 group.
The above means that at least 4 subjects out of 13 in the starting IGA 3 group must have improved to clear/almost clear groups (IGA 0/1) in the end. In percent: 100 times 4 divided by 13 is 30.7 per cent. And that number represent the lowest possible percentage of moderate psoriasis subjects that DID improve to either almost clear or clear.
I am pretty sure that the above is correct.