Journal A received a manuscript—a randomised, controlled, double anonymous, parallel clinical trial. The manuscript was reviewed by two specialist reviewers who suggested acceptance after revision. One of the important points that was asked was to provide the “registration number of the RCT”. As our journal is a member of the ICMJE, we reviewed the instructions of the ICMJE and found that there may be rare exceptions of failure to registration. Accordingly, we asked the authors to provide the registration number if they had already registered it, or to register it (after implementation) if not yet and to declare the reasons for delayed registration. We also provided some links to registries that accept this type of delayed registration.
We received a reply from the authors stating that for reasons of business confidentiality, safety and possible conflicts of business interest they find it very complex to register the study as a clinical trial.
Questions for COPE Council
- If the authors insist on “No registration”, should the paper be rejected?
- Does publishing a randomized controlled trial without inclusion of the registration number greatly affect the credibility of the journal?
- If the authors accept delayed registration (post-implementation), can the journal proceed with the next steps in the peer review process while steps are taken to register the study?
Advice on this case is from a small number of COPE Council Members. Most cases on the COPE website are presented to the COPE Forum where advice is offered by a wider group of COPE Members and COPE Council Members. Advice on individual cases is not formal COPE guidance.
All journals that publish trials should have a policy on trial registration and COPE would recommend that this be clearly stated in their instructions for authors. Hence the journal should have specific author instructions that state that all primary studies, regardless of design, should be registered before participant enrolment and that no studies will be admitted to peer review if no registration has been made. They could also state that retrospective registration is not suitable, as it defeats the purpose of registering clinical trials—that is, to prevent post hoc changes to the protocol, especially to the statistical analysis regarding primary and secondary outcomes as well as subgroup analysis.
The excuse given by the authors seems rather weak, and the absence of a proper explanation is concerning. The journal should probe for more details—and apply some pressure for a more detailed explanation. The researchers may not know what kind of explanation is warranted, or why. It may therefore be necessary for the journal to explain to them why registration is important (without registration it can be very hard to know, for example, what trials should be in systematic reviews and that papers in the systemic reviews are not counted twice); what it's trying to achieve; how common it actually is (ie, pretty normal, pretty low risk); why retrospective registration is not really good enough but why it can be an acceptable approach in some instances as a way to model better practice and thus encourage wider adoption of prospective registration.
COPE would suggest if the authors still refuse to retrospectively register, the paper should be rejected. If the study is registered late, the journal should consider a declaration on the paper about why, and provide a guarantee to the journal that any related papers will be registered