LAUSR

In this edition of Anaesthesia, we have published an articulate editorial by Smith and Dworkin [1]. They argue strongly in favour of mandatory prospective trial registration, enforced by journals, and they present a robust case for why this is a good thing. Since Anaesthesia has resolved to continue to accept papers without prospective (e.g. retrospective) trial registration [2], it is important to explain to readers the basis for the honest disparities in opinion. The main implication of Smith and Dworkin’s argument is that journals should automatically reject trials not prospectively registered, as itself is a sign of poor scientific practice. We suggest that this would be counter-productive. The journal is not taking a principled stance against the concept of prospective trial registration; it is something to be strongly encouraged. Yet, we must recognise that although proportionate ethical review is supported by an overwhelming intellectual argument, prospective trial registration is not. The strongest counter-argument is that, if registration were deemed so essential by regulators, as is ethical review, then there is an extremely simple solution which regulators have long had within their power to apply. All human clinical trials need ethical review. It would be a very simple matter for ethics committees, at the time of granting ethical permissions, automatically to upload the study protocol onto a trial registry. This would be transparent, and, also, by removing any further action on the part of authors, would ensure 100% compliance. Indeed, the entirety of the ethical application could be automatically uploaded and made publicly accessible – including the blow-byblow correspondence and all the amendments made during the course of ethical review. There is a potential problem created by the existing system of separate ethical review and trial registration. It is theoretically possible for an author to submit, perhaps in subtle but important ways, one set of details to an ethics committee, but different details in the registration. Because ethical applications are currently not publicly accessible, we simply do not know how common this problem is (except that deviation from ethical permissions features in celebrated cases of author misconduct, even if originally suspected on statistical or other grounds) [3–7]. Expressed another way, the fact that regulators have chosen not to adopt a simple solution within their grasp could be taken to suggest that they themselves are lukewarm about registration. Automatically rejecting trials not prospectively registered itself has ethical implications. The patients have already been subjected to the intervention and in an ethically approved manner. Journals cannot undo what has been done. Risks have been taken, evidence has been collated, often to a very high standard, as El-Boghdadly et al. confirm [2]. Automatic rejection prevents sound data from being assessed that could otherwise help future patients. What is to be done with the data of a trial that was not prospectively registered? It is unconvincing to view all such results as contaminated. Changing a primary endpoint is not any sort of ‘sin’, so long as there is a valid explanation for it. Robotic adherence to a preselected end-point does not make good science. Good science is the recognition of the story revealed by the experiment [8]. Authors are to be encouraged to write: ‘We planned a primary end point X but This editorial accompanies an editorial by Smith and Dworkin, Anaesthesia 2018; 73: 538–41, and an article by El-Boghdadly et al., Anaesthesia 2018; 73: 556–63.