This study characterizes discrepancies between primary outcomes specified in trial protocols, registrations, and publications, and the association between prospective registration and consistency of published outcomes.