[tsimionescu]

The decision is not as centralized as you make it out - various governments and other super-national bodies can and have taken different decisions, based on various assessments and, at least to some extent, the will of their populace (though I'll be the first to point out that representative democracy and technocracy make the effect very indirect).

The argument for not making this decision fully decentralized is that assessing the risk-to-benefit ratio requires a significant level of expertise, both in the subject matter itself (were the studies conducted well? what possible blind spots may they have methodologically? what plausible effects could these blind spots have on public health?), and in the more political/social side (are the doctors running these studies trustworthy? are the results being accurately reported between observation and collation? will this medicine be over-sold to patients in disregard of the studied benefits/risks? what is the cost/benefit estimate for using this in public hospitals?).

Now, it would be possible in principle for the state to provide the current centralized verification services, but then only give consultation-level opinions to the populace, doctors (and/or the public health system). But given the strength of PR and direct-to-consumer marketing, this seems to be extremely risky in terms of people ending up swindled to spend huge amounts for terrible medication sold by unscrupulous corporations (a free-for-all version of the opioid crisis).