Having worked closely with pharmacovigilence in a veterinary company, I learned that post-marketing surveillance is all about detecting signals. It usually requires more investigating to ascertain cause for the signal. I’ve seen both true and false signals, which were only determined by digging into the data deeper, along with smaller prospective studies. It’s very complex, and I’ve seen regulatory agencies jump to incorrect conclusions based on just the signal. I’ve also seen companies pull a product based on the signal because it just wasn’t worth the potential bad reputation, accurate or not.