My lab is funded at around $300,000 a year and runs ~300 toddlers a year, although with non invasive cognitive testing and single visit. They are recruited pretty randomly from county birth record phone numbers, though of course not everybody agrees (they tend to be too RICH actually, not too poor in my case).
I'm not sure how you're managing to spend 1,000's of times more money per person, but okay, sure. Never done injections or whatever, maybe it is that much more. Even still, if it's say $1,000,000 for 100 subjects, 50,000 = $500,000,000. Which is like, less than 1/6th of what we spend on developing and distributing every typical vaccine. Seems worth doing by comparison at least once.
But alternatively, sure, more extensive post marketing studies are better than nothing if that's all I can get. Even those, we could be doing far better. The biggest I've seen are like ~1-2 million people, and usually 1/3 to 1/2 of those studies included in meta analysis are ones exclusively looking at autism (eyeroll), so more like 0.5 to 1 million for what we want. Where the hell is everybody else? If nothing else, the government could at least mandate simple records of this crap so that we could have hundreds of millions instead of 1... It doesn't even have to be identifiable data at any level of the database. Private doctors could keep their own records of actual people like they do already and then submit summaries yearly or whatever and the very first thing the government gets at all is already aggregate anonymous data with privacy intact.
Edit: Locking topic because I'm tired of it and apparently so are other people since nobody has participated in 2 days. Also more and more frequent circles. Thanks everybody for an awesome discussion, though. Only one flamer post really in 30 odd pages! Lots learned all round I hope, myself included.