One thing I mentioned before that I did not cover on my second attempt is that while the system of testing is focused on side effects of a drug (and closely related ones), when it gets out of the lab and into the real world the number of possible interacting drugs rapidly increases. It is not rare for people to be taking, for example, a drug for depression, one for a heart problem, a blood thinner, something for osteoarthritis and the occasional aspirin or paracetamol. Another person might be taking statins, something for low or high blood pressure and some herbal compound which was not prescribed but has an active components. People mainly take combinations of drugs, not a single drug.

With the best testing in the world, and every good intention, it is simply not feasible to address all possible combinations of all possible drugs. The pharma companies will certainly test the more obvious possible sources of issues, but it is not practical to cover everything.