Thanks for the clarification. Yes, an RCT does attempt to allow for confounders, employ longitudinal studies for context, track changes in behaviour, etc to reduce exposure to accidental correlations, but that's really limited to known knowns. As we've seen recently (e.g. in the WHO's 180° reversal on the supposed benefits of red wine), it's really easy to miss a little detail that would completely change the conclusion.

(Irrelevant nerd note: The gold standard was a major contributing factor to the Great Depression and several economic depressions before that, which is why all rich countries abandoned it after WW2 in favour of fiat currencies, but that doesn't seem to have harmed its use as a metaphor. 😉)