Stopping early for benefit

Trials stopped early (before protocol, in particular in the absence of adequate stopping rules) overestimate treatment effects.