12:00 - 12:05 pmSunday, September 18
Upper Lobby
Not just "tilting at windmills": the role of communication and collaboration in the statistical review process
Upper Lobby
Not just "tilting at windmills": the role of communication and collaboration in the statistical review process
Journal of the American Veterinary Medical Association
Multiple studies have demonstrated the ongoing problem of methodological errors in the biomedical medical literature, with the availability of statistical reviewers a related challenge. Errors in published... Read more

Description

Multiple studies have demonstrated the ongoing problem of methodological errors in the biomedical medical literature, with the availability of statistical reviewers a related challenge. Errors in published articles may go unnoticed because of author and reader lack of knowledge of statistical methods. Furthermore, journal editors frequently may not recognize statistical errors during the peer review and editorial processes. In 2005 Ioannidis published his much cited and discussed article “Why Most Published Research Findings Are False,” in PLOS Medicine. However, since that time concerns regarding the lack of reproducibility in science, and related discussions of the need for statistical rigor remain; affecting “large” and “small” journals in science and medicine. Lack of sufficient emphasis on rigorous statistical review can lead to highly publicized retractions and the attendant fallout. Inclusion of statistical checklists in the Instructions for Authors is generally ineffective and whereas many journals endorse various reporting guidelines (eg, STROBE, SQUIRE, CONSORT etc.) editors may not enforce a requirement for adherence, and submission of a completed checklist is not an indicator of study quality.

This presentation will highlight efforts to implement a rigorous statistical review process at the Journal of the American Veterinary Medical Association (JAVMA), the most widely circulated clinical veterinary medical journal in the world; an effort that formally commenced in 2013. Statistical review was available prior to 2013 but was generally performed on an “ad-hoc” basis, in part because of a lack of resources (ie, a very limited pool of statistical reviewers); in part because, as for many medical journals, it may have been unclear that there was a greater need for statistical rigor.

From the outset particular emphasis was placed on engaging in a highly collaborative process, valuing the the contribution of all involved, acknowledging each individual for their willingness to submit to the process, and with the editor (this author) being available for feedback and discussion. It was emphasized to authors that this was not about “finding mistakes” and statistical reviewers (who volunteered their time and all of whom were highly experienced) also received frequent communication versus form emails only. Whereas this has required considerable commitment, we have found notable success and enthusiasm from authors, biostatisticians, and members of the profession. It is our opinion that supportive and respectful communication may be valuable in promoting engagement and commitment to a rigorous statistical review process. A small number of comments can require an extensive amount of work as well as continued input from volunteer biostatistician reviewers. Nonetheless, after approximately 3.5 years a variety of examples of progress are evident because of the engagement and hard work of individuals with a shared passion and respect for the process, and appreciation for the effort of all involved. This has not been easy, or without obstacles. However, we believe that pursuing this is the right thing to do. Our patients deserve this. I hope that by sharing some examples of this process from our journal it may be helpful to many others working to improve statistical rigor and the quality of the peer-reviewed biomedical literature.

Contact Us

We're not around right now. But you can send us an email and we'll get back to you, asap.

Not readable? Change text. captcha txt

Start typing and press Enter to search