Fewer than half of all clinical trials recently published in journals with high impact factors are adequately registered, and about a third of those that are registered properly show discrepancies between the registered primary outcome and the one that actually gets reported.