Reporting guidelines: How can they be implemented by veterinary journals?

Abstract

Poor reporting matters because it reduces the transparency, repeatability and reliability of research. It creates particular problems in assessing the risk of bias, and when attempting to compare and combine results in systematic reviews and meta-analyses. Over the last 6 years there have been several studies that have shown poor reporting in many research papers published in refereed veterinary journals. Perhaps the most relevant of these for equine veterinary medicine is a study that looked at the quality of published literature on reproduction in cattle, horses and dogs 1. Typical issues in reporting of clinical trials include definition of primary and secondary outcomes, sample size determination, allocation of study units to interventions, randomisation method and blinding. As a result, a number of reporting guidelines have been developed that provide guidance and checklists for authors, peer reviewers and editors in reporting research methods and findings in a range of study types. Details of these can be found via the website of the EQUATOR Network, http://www.equator-network.org. Although the original impetus to develop reporting guidelines came from human medicine, the principles also apply to veterinary research. In addition, some reporting guidelines specifically geared to animal studies have recently become available. The main reporting guidelines include CONSORT for randomised controlled trials, ARRIVE for preclinical and experimental animal studies, REFLECT for livestock and food safety trials, PRISMA for systematic reviews and meta-analyses, and STROBE for observational studies, but there are many more. Given the evidence of some poor reporting in veterinary research, we at the Centre for Evidence-based Veterinary Medicine wondered how aware veterinary journal editors are of reporting guidelines. In 2012, we carried out a survey of editors-in-chief through the International Association of Veterinary Editors (IAVE) and published the results in January 2014 2. The response rate was 36.8% (68/185) and we found that 47% of respondents had not heard of reporting guidelines before the survey, and only 35% said their journal referred to reporting guidelines in its instructions to authors. However, 68% believed that reporting guidelines should be adopted by all refereed veterinary journals where appropriate. Factors said to be barriers to using reporting guidelines included lack of knowledge among authors, reviewers and editors, resistance to change, perceived workload issues, and the fear of losing submissions. The need for promotion and communication to increase awareness of reporting guidelines among the veterinary community was frequently mentioned. Some editors suggested that there should be a consensus on implementation across all veterinary journals to ensure a level playing field, as authors might otherwise favour submitting to journals that did not require adherence to reporting guidelines. So how can the more widespread use of reporting guidelines be achieved in veterinary research? The first step is to increase awareness. We hope our survey of editors-in-chief has played a part in this, and the IAVE is also promoting reporting guidelines through its meetings and website (http://www.veteditors.org). The second step is for journals to endorse reporting guidelines and to include them in their instructions to authors and peer reviewers – in this respect the EVJ is in the vanguard among veterinary journals. In its Author Guidelines 3, the EVJ states that prior to submission of a paper, authors are recommended to consult the appropriate reporting guideline, with links given to guidelines for a wide range of study types. However, is endorsement of reporting guidelines by journals enough to ensure a high quality of reporting in published research? There has recently been much interest in this question in the medical world. A 2012 Cochrane systematic review 4 found evidence that journal endorsement of CONSORT did help contribute to the completeness of reporting of randomised controlled trials, but there was still suboptimal reporting, and it was suggested that journals are not giving a clear message to authors. Another study found that, although over 300 journals had endorsed the ARRIVE guidelines for preclinical animal studies, there had been little improvement in reporting after 2 years 5. The authors of that report concluded: ‘This suggests that authors, referees, and editors generally are ignoring guidelines, and the editorial endorsement is yet to be effectively implemented’. It would therefore appear that a necessary third step is for journals not just to endorse, but also to enforce reporting guidelines. This can happen at several stages in the publication process: authors can be required (rather than advised) to follow the relevant reporting guideline; editorial offices can use checklists at manuscript submission; peer reviewers can be asked to assess papers against the reporting guideline; and editors can take reporting guidelines into account in making their final decision. The EQUATOR Network has published some interesting case studies on its website of how medical journal editors have implemented reporting guidelines 6. Many medical journals put the onus on the authors, with some requiring a completed reporting guidelines checklist, and others requiring a formal statement of compliance. The EQUATOR Network case studies suggest that this approach has not created much extra work for the journals involved. EQUATOR also recommends the use of reporting guidelines by reviewers: peer reviewers have a key role in assessing research methodology and reporting quality, and reporting guidelines provide a useful framework and guide for this 7. Without a checklist, reviewers may miss things especially omissions in the text. Arguably, the use of reporting guidelines increases the consistency and fairness of the peer review process. Some might believe that enforcing reporting guidelines is unnecessarily prescriptive and restricts the freedom of authors and editors. However, I suggest that the items included in reporting guidelines are almost always necessary for completeness and clarity of reporting. Reporting guidelines should not be a hindrance, but a valuable guide. Furthermore, reporting guidelines do not necessarily need to be followed slavishly in every instance, provided that authors can give a valid justification for omitted items. As also proposed in a 2010 commentary by Simon More in The Veterinary Journal 8, I suggest there is no good reason why all veterinary journals should not require that the research they publish complies with the relevant reporting guideline. This is the only consistent way to ensure that research reports are complete and transparent, so avoiding the unnecessary waste and ethical implications of unreliable and unusable research. Already, there are some veterinary journals that require adherence to reporting guidelines. For example, Preventive Veterinary Medicine and Veterinary Record Open ask authors to submit reporting guideline checklists. Journals that enforce reporting guidelines should benefit through improved quality and attracting papers from the best researchers. Finally, as suggested by several editors in our survey, is the idea of a consensus on reporting guidelines across veterinary journals a feasible proposition? A recent example from human medicine suggests that it is. In early 2014, 28 rehabilitation journals all published the same editorial on reporting guidelines and agreed to mandatory use of guidelines and checklists by January 2015 9. So it can be done… The author would like to thank Dr Rachel Dean and Dr Marnie Brennan for their valuable comments on the first draft of this paper. The Centre for Evidence-based Veterinary Medicine is supported by an unrestricted grant from Novartis Animal Health and The University of Nottingham.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 7 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.