Comparison between publicly accessible publications, registries, and protocols of phase III trials indicated persistence of selective outcome reporting.