Radiology Editor Urges Authors to Use Checklists to Improve Reporting Accuracy
Tuesday, Dec. 01, 2015
Radiologists can publish more effective research and will more readily get published if they follow standard checklists of essential information to include in studies, said presenters at a Monday session.
The presenters reviewed a toolbox of checklists that researchers should follow to ensure they include key information in their studies to show that their conclusions are reproducible and generalizable.
"The goal of the course is not to be alarmist but rather to help attendees understand the importance of these checklists and how they can be used to improve the quality of the reporting of studies submitted to our journal," said Herbert Y. Kressel, M.D., editor of Radiology and professor of radiology at Harvard Medical School. "The major focus has been on reproducibility of results."
Dr. Kressel provided examples. In working with original authors, the pharmaceutical company Amgen, could only reproduce results on five of 53 key studies, he said. And in another review of studies, Bayer Healthcare could only reproduce results in 25 percent of 63 studies, he said.
To address similar problems in radiology research such as bias and applying an inappropriate analysis method, presenters offered an overview of various checklists, including Standards for Reporting Diagnostic Accuracy Studies, or STARD, and Preferred Reporting Items for Systematic Reviews and Meta-Analyses, or PRISMA. The session focused on STARD and PRISMA because they were developed for the most common types of studies that RSNA publishes.
STARD is a checklist for reporting studies on the accuracy of a diagnostic procedure. It outlines the standard sections for a study and includes a checklist of more than 30 required items such as an index test, participants' eligibility criteria and where the full study protocol can be accessed.
It's better to consult, PRISMA, an evidence-based minimum set of items for reporting in systematic reviews and meta-analyses, if a researcher is conducting a systematic review of research or a meta-analysis, said presenter Matt McInnes, M.D., of the University of Ottawa.
The PRISMA checklist suggests including such items as presenting a full electronic search strategy for at least one database; discussing methods used for assessing risk of bias of individual studies; and discussing limitations at study and outcome level and at review level.
Other guides include CONSORT, a checklist for reporting randomized controlled trials, and STROBE, a checklist for reporting observational studies.
Although the checklists might seem burdensome, researchers should view them as tools to ensure the research is done correctly, said presenter Patrick Bossuyt, M.D., a clinical epidemiologist from the University of Amsterdam who helped draft the first STARD checklist 15 years ago. "Publishing quality research is about increasing value, reducing waste and disseminating more information," he said.
Senior Radiology Editor Deborah Levine, M.D., agreed, saying that raising the standard of publishing is good for the researchers. "Our goal is to help you build and optimize your research study and have the most clinical impact," she said
Dr. Kressel noted that the NIH is increasingly requiring funding candidates to follow these standards and that Radiology will begin to require that all diagnostic accuracy studies follow STARD guidelines starting in January.
The details of all of these guides are in the EQUATOR network website, www.equator-network.org, which includes a tool to help researchers determine which reporting checklist is best for their research.