The evidence-based medicine (EBM) movement—perhaps officially launched in 1991 with the first article in the medical literature that used the term,1,2 but with antecedents that go back considerably further—advocates that systematic summaries of the highest quality relevant evidence should inform our clinical care.3-5 With this premise, clinicians should have some understanding of what constitutes more vs less trustworthy evidence. The widespread acceptance of the principle, and the corollary regarding physician familiarity with evidentiary standards, has resulted in a major change in medical education. Requirements exist throughout North America, and in many other countries, that both undergraduate and postgraduate medical training include attention to skills of accurately interpreting the medical literature.
As EBM was being launched, JAMA became a champion for the development and dissemination of the new approach to clinical practice.4-6 JAMA's Users’ Guides to the Medical Literature provided the core educational materials for EBM, and became established as required reading for frontline clinicians, and medical educators, in understanding the literature and using the literature to inform clinical practice. The series has provided the core material for all subsequent EBM educational resources.
Although the Users’ Guides have played a key role in EBM education they have, as any resource will, limitations that require complementary material. The Users’ Guides all have a broad scope, addressing how clinicians should read primary articles and systematic reviews informing choice of therapy, evaluating claims of harm, establishing prognosis, evaluating diagnostic tests, and so on. JAMA, again wisely and with the initiative to innovation, recognized the limitation: the articles do not focus on specific methodologic and statistical issues that clinicians would ideally understand when using the medical literature to guide their care.
This is particularly relevant because the statistical and methodologic world has not stood still. Clinicians may not be familiar with research methods introduced after they completed training. Recognizing the unmet clinician needs, JAMA introduced its Guide to Statistics and Methods series to provide a more granular and specific discussion regarding statistics and research methodology. The series has proved very successful, with over 55 publications to date.
The articles address a number of crucial themes in understanding the research literature. Within the broad category of interventional studies, issues addressed have included trial strategy and design; enrollment, allocation of treatment, and ethics; measurement of outcomes, analyses, and interpretation of results; and application of results. Issues relevant to observational studies have included study strategy and design; assessment of risk factors and exposure; measurement of outcomes, analyses, and interpretation of results; and application of results. Another group of 13 articles within the series offer practical guides to surgical data sets.
These guides on statistical and methodologic issues are highly relevant to clinicians. First, they address statistical and analytic approaches and methods used in research reported in JAMA and JAMA Network journal articles. These research articles use the particular statistical tests or methodologic approaches under discussion, and thus provide—a key to any EBM educational material—relevant examples. In addition, the articles address concepts that clinicians will repeatedly see in the medical literature.
Another great merit of the JAMA Guide to Statistics and Methods series is its accessibility. The articles are written in plain English, using language practicing clinicians can easily understand, avoiding complex mathematics or the use of arcane technical language, and presenting material graphically whenever possible. Linking the methods articles to published research reports further enhances accessibility and usefulness of the statistical and methodologic concepts discussed.
Although I am a methodologist, as with any scientist, the scope of my expertise is limited, and I am vulnerable to ignorance in areas that have developed and expanded after I received my training. Thus, I have a number of articles in the series among my favorites: those in which my understanding was limited, and in which I needed lucid explanations to gain a grasp of new concepts. For example, neural networks have been around since the 1980s, but have only gained popularity since the increasing sophistication that has accompanied advances in machine learning, and in particular in application to medical imaging. An article addressing convolutional neural networks applied to medical imaging opened the black box, providing a clear and compelling explanation of the fundamental principles underlying the method. The article also links to a very helpful video.
Although I have been involved in primary interventional studies, they have almost exclusively been randomized trials. As a clinical epidemiologist, I am also familiar with standard observational cohort and case-control studies, but much less so with less common and more complex observational study designs such as stepped-wedge clinical trials. The design made feasible a study of a quality improvement intervention to reduce complications after myocardial infarction that could not have succeeded within a conventional randomized trial. The article in this series on stepped-wedge clinical trial design clearly explains the advantages and limitations of this approach. As a methodologist, I was gratified to learn that although a naïve analysis suggested a benefit, the sophisticated analysis that was possible because of the stepped-wedge design showed a different result. While the intervention was ongoing, clinical care was improving independently, and the optimal adjusted analysis demonstrated that it was this secular improvement, rather than the intervention, that improved outcomes.
Now, JAMA has wisely and helpfully collected the articles in the series into a single book, which has also been incorporated into JAMAevidence online.7 Clinicians interested in deepening their understanding of the articles that guide their clinical practice will find the collection a highly accessible and extremely informative source of information.
Gordon Guyatt, MD, MSc, McMaster University
Department of Clinical Epidemiology & Biostatistics, McMaster University. How to read clinical journals, I: why to read them and how to start reading them critically. Can Med Assoc J
G. Evidence-based medicine. ACP J Club (Ann Intern Med). 1991;114(suppl 2):A-16.