Introduction

Recent evidence of selective reporting of results has eroded public and academic confidence in publications of clinical trials, leading to renewed calls for trial registration.1^–5 The dangers of non-disclosure of trial results, although described for years, sparked an international furore last spring after the publication of two systematic reviews on the effects of selective serotonin reuptake inhibitors for childhood depression.1 6 Subsequent legal proceedings7 and policy statements by journal editors,8 9 medical associations,10 and industry11 have recognised the importance of trial registration. The rationale for registering trials is well known (box 1).12 13 Most importantly, the contribution to social good that justifies research on human participants is not realised when resulting knowledge remains invisible.

As an interested and neutral party that has been registering the trials that it funds,14 the Canadian Institutes of Health Research hosted an open meeting on 4 October 2004 in Ottawa, Canada, to foster international consensus on trial registration. The resulting Ottawa statement aims to establish internationally recognised principles for registration (box 2). The full statement is on bmj.com, but here we highlight and discuss some of the key principles. A statement on how to implement these principles (part 2) is still in development.

Summary of principles

The mandatory registration of all trials has three components:

• Obtaining an internationally unique identification number (unique ID)

• Registering the original protocol along with subsequent amendments

• Registering the trial results.