That seems like selection bias.

You do a lot of studies and experiments, and filter out most proposed medicine because it causes harm quickly, or doesn't cause benefits quickly enough or at all. Then you market whatever survived testing. Obviously, if it's still harmful, the harms will show up only slowly, while the benefits will show up quickly - otherwise you would have filtered it out before it reached the consumer.

This is like saying engineering disproportionately channels optimism, because almost all the appliances you buy in the store work now and only fail later. If they had failed immediately, they would have been flagged in QC and never got to the shop.