The CONSORT Statement aims to improve the quality of reports of randomized trials. It is encouraging that over a third (38%) of high impact factor journals, assessed in our study, refer to the CONSORT Statement in their published 'Instruction to Authors'. This represents a relative increase of 73% since this study was last conducted in 2003 when only 22% of journals mentioned CONSORT in their 'Instruction to Authors' . Some journals, however, still refer to the superseded 1996 version of the CONSORT Statement and we would encourage journals to keep their 'Instructions to Authors' up-to-date.
There is, however, still ambiguity in the wording of some journals' published 'Instructions to Authors' as to whether they require, or recommend, that authors comply with the CONSORT Statement . We believe that journals should provide a clear message to their authors and recommend authors submit a completed CONSORT flow diagram and checklist as a requirement to submission in an endorsing journal. Only by endorsement of the CONSORT Statement by more journals, and greater editorials efforts to ensure that authors comply, can the quality of reporting of randomized trials published in leading journals be improved .
In September 2004, the members of the ICMJE published a joint editorial stating that they would only consider a trial for publication if it has been registered before the enrolment of the first patient . In our study, over a third (37%) of journals required recent clinical trials to be registered as a requirement of submission to that journal. Disappointingly, a number of these journals did not provide the same endorsement of the CONSORT Statement despite recognising the importance of trial registration which, by itself, is not the best marker of trial quality.
It is also disappointing that so few journals endorse the recent extensions to the CONSORT Statement [4–6, 9]. These extensions reflect a considerable amount of work and were developed to improve the reporting and transparency of trials using specific designs and types of data. Other CONSORT extensions have very recently been published [7, 8] and others are being developed. We would recommend that journals reference these extensions in their 'Instructions to Authors' thereby helping authors wanting to improve the reporting of their randomized trials.
One obvious limitation of our study is that, despite several electronic reminders, there was a poor response rate (39%) to the on-line survey of journal editors' endorsement of CONSORT. As might be expected journals which endorsed CONSORT were more likely to respond (61%) to the survey than those journals which did not (39%). However, despite this limitation the results of this survey are still of interest, in particular, to those journals who said they would require, or recommend, authors comply with the CONSORT Statement but do not mention this in their 'Instructions to Authors'. This is particularly apparent for CONSORT extensions. In response to the survey two journal editors said that they were in the process of updating their journals' 'Instructions to Authors'. It is possible that these instructions were updated during the study period; however, the numbers are very small and would not affect our results. We note however, that there was inconsistency between editors' responses and the information on the journal's website under 'Instructions to Authors', especially in relation to the support of extensions to CONSORT.
Similar to the CONSORT Statement for reporting the results of randomized trials, other reporting guidelines exist which provide advice on how to report research methods and findings for other types of study designs. In our study journals which endorsed the CONSORT Statement were more likely to endorse other reporting guidelines, although the numbers were very low. The EQUATOR Network  is a new initiative which aims to increase awareness of good reporting guidelines in health research and thus improve the quality of scientific publications. We hope the work of this initiative will lead to better endorsement by journals of these important reporting guidelines.