Scientists often fail to publicly register plans for clinical trials or to publish the results, and the outcomes they do share may mask instances when new treatments are unsafe or ineffective, a small study suggests.