what would a genuinely convincing study of one of these compounds look like?

Question

Constructive for once. If someone wanted to settle whether one of the popular preclinical compounds does anything, what would it take?

My sketch: randomized, placebo controlled, double blind (non-negotiable for any subjective endpoint). Pre-registered with a single stated primary endpoint. An objective primary endpoint where possible, imaging, function, a measured quantity, not a rating scale. Adequately powered for a PLAUSIBLE effect size, not the inflated preclinical one. A defined population with stated entry criteria. Results reported regardless of outcome.

That's an unremarkable controlled human trial design. The notable thing is that essentially nothing in the preclinical tier of this space has had one. What am I missing?

n_of_what · Answer

"Powered for a plausible effect size, not the inflated preclinical one" would sink most of these before they started. Power on the animal effect size and you get an underpowered trial.

deload_week · Answer

so the answer is nobody will ever pay for it?

unconvinced_still · Answer

Which is arguably worse than no trial, because an ambiguous result gets cited by both sides forever.

same_tbh · Answer

objective endpoint. yeah thats the hard bit

SteelmanIt · Answer

The funding counterpoint applies here: this design costs a lot and nobody owns the compound. Real explanation, and also the unfalsifiable argument from the other thread. Both true, which is what makes it hard.

what would a genuinely convincing study of one of these compounds look like?

6 replies