The basis of my little clip there that “large doesn't make good” is that it really depends on the source of the data. That kind of study is very typical of retrospective epidemiologic studies. The universal weakness of those studies is that one does not have information about all the things you really want to have.
You want to know what the subjective symptoms and behaviours were that were exhibited. What you often end up with is little things in a chart; you have a diagnosis, but you don't know when it started, so you end up basically with an inference that that particular study is at odds with what we know from other studies that have been properly controlled.
Professor Quinn referred to one such study, but there are others. In the literature as a composite, going back as early as the 1970s, it is clear that drugs that have this chemical structure are associated with this again and again and again.
When you see somebody leaving a bank, and the money isn't there, you say “woah”. It happens the next day. There's a robber. Something is going on. This is a repetitive pattern with this drug.