On each successive run, yes, the chance to find it would be still be 0.5% since the events are independent. However, to find the total chance of not finding it in n runs, you use the multiplication rule. Say you want to find the chance of not finding it after two runs, which we will call events A and B:

P(A and B) = P(A) * P(B|A)

Since, as the events are independent, as you pointed out, P(B|A) = P(B),

P(A and B) = P(A) * P(B)

Furthermore, P(A) = P(B), so the probability of not finding it twice in a row is:

P(A)^2

The same logic applies for n runs, where we get the probability of not finding the item after n runs:

P(A)^n

Since A is the complement of our drop chance, 0.5% in this case, we end up with:

(0.995)^n

So, if we did 400 runs at 0.5% drop chance, we would end up with:

(0.995)^400 = 0.1346

or, ~13.5% chance of still not having found it after 400 runs. The complement of this, ~86.5% is the chance that one or more of those 400 runs the item drops.