How much weight should we give safety sections when the evidence is this limited?

Your instinct is correct and it's basically a statistical-power point: a study of 40 people over 8 weeks simply cannot detect a rare or slow-developing harm. "Well tolerated" in that context means "we didn't see anything common and fast," which is genuinely useful but a narrow claim. Absence of evidence, etc.

I try to read the safety section together with the sample size and follow-up length every time. A glowing tolerability line attached to n=30 and 6 weeks tells you almost nothing about long-term or uncommon risk.

Agreed with the above. I'd also flag publication and reporting bias: adverse events are inconsistently collected in small studies, and the ones that get written up skew positive. So the safety picture in the literature is often rosier than reality even setting aside the power problem.

The honest summary for most of these compounds is "limited data, mostly short-term, interpret cautiously." That's not a satisfying answer, but it's the accurate one, and it's why anyone making a real decision should be doing it with a clinician who can weigh their individual risks.

The reporting-bias angle is the piece I under-weight. So even the modest reassurance from a small study is probably a slight overestimate. Good — that's the direction of error I'd rather know about. Thanks both.