They call it the “Proteus effect.”
Researchers tracking clinical trials have long been aware that the first studies to be published sometimes show benefits that aren’t borne out in later trials. Like Proteus, the Greek god of the sea who could change his shape, evidence can shift over time to show more modest improvements.
This isn’t shocking. Lots of early phase studies, half show higher than true effect, half show lower (and more than half if they study multiple doses due to multiplicities). Then only the high-effect studies move forward. Then upon subsequent, larger, studies we get regression to the mean as expected. Precisely why so many late-stage studies fail — trial designers fail to account for regression to the mean that’s bound to occur in later stages.
Comments are closed.