Might be considered "not answering the question" but:
Option 1:
Depending on the type of cells you are using, you can use milder methods to detach cells. Example include Versene which is essentially PBS + EDTA. Milder still is to just use PBS that has no Mg or Ca (no chelation).
Option 2:
Duplicate your setup - instead of two dishes - one with drug, one without, you make 4 dishes (dispensed from a common source - i.e. if your dish is a 6cm dish with 3ml media, you make 4.5x3mls of cells at the final dilution you want) each plate gets 3mls from the same source so you know they have the same number of cells. Two dishes you add drugs, two you don't. On the day you want to count, you take the replicate plates and count them. This way, you get the count without touching the final dishes at all.
As a side experiment, if you put those trypsinized cells back on two dish (and replace the drug media in the drug treated replicate), and you count all plates on the last day of your experiment, provided cells are not at 100% confluency, any difference you see from your trypsinized vs untrypsinized can be viewed as potential error that you would have introduced if you went through your original methodology.