As mentioned in the August 1, 2001, “Inside AFP, ” we are strengthening the process for incorporating an evidence-based approach in our clinical review articles. We now ask all authors to rate the level of evidence for key clinical recommendations on diagnosis and treatment. As far as I know, American Family Physician is the first journal to request, where possible, ratings of evidence for all of its articles.
We realize that this is an ambitious undertaking, with several potential pitfalls. First, not all articles are equally suited to evidence ratings. For example, an article on interpreting electrocardiograms or performing certain procedures may not be supported by enough evidence-based resources to warrant the inclusion of ratings.
Second, there are many different rating systems in the medical literature, and these continue to evolve over time. Several are complex, with some including as many as 10 different levels. There are pros and cons to adopting any one system, and challenges in developing a new one. In the interest of simplicity, we have decided to use a three-tier, ABC system, with explicit modifiers. At each rating of evidence, we will tell our readers what the rating means. Consider this a trial period, subject to modification as we and our authors gain experience with this new process. I am hopeful that readers will appreciate a good-better-best approach and get a feel for the strength of the evidence behind important recommendations on therapy.
The third potential pitfall is the subjective nature of applying any rating system to the complex universe of science. Is a fair-quality randomized controlled trial (RCT) “better” than a high-quality retrospective cohort study? What if seven RCTs conclude that a treatment works, but three disagree? And when it comes to expert opinion, whose expertise counts? The literature contains many examples of flawed RCTs, meta-analyses that ultimately prove to be false, and studies with similar objectives but contradictory findings.
We undertake this new process with some humility, knowing that science is inexact; that authors and editors are imperfect; that the process itself is, at best, an attempt to get closer to a standard of best practice, and that this goal may be difficult to achieve. Ultimately, we believe that this process will encourage authors to review the medical literature more critically and provide more informative articles for readers.
For a more detailed description of this process, please see the “Information for Authors” on AFP's Web site,www.aafp.org/afp/authors, which refers to a sample article showing how the levels of evidence are to be incorporated in the text and offers additional background information. Please also see the article on page 251 of this issue,1 which provides guidelines on writing an evidence-based review article.