Evidence-Based Dermatology: A Brief Introduction
David A Barzilai MD PhD
Providence, RI, USA
on January 21, 2007
Resident in Dermatology, Rhode Island Hospital, Brown University Department of Dermatology, Providence, RI, USA
Since “evidence-based medicine” (EBM) was coined in 1992, much information and misinformation has been circulated about what evidence-based medicine is and isn’t (1). This is perhaps particularly true in dermatology where we have only recently begun to appreciably introduce its language and tools into our residency programs and practices (2). David Sackett defines EBM as “integration of the best research evidence with our clinical expertise and our patient’s unique values and circumstances (3).” This description emphasizes that EBM is not “cookbook medicine.” Rather, EBM combines the “state of the science” (which is often lacking in dermatology), with sound clinical judgment and knowledge of what makes our patient unique.
Clinicians have incorporated evidence into practice since long before Hippocrates . What makes EBM unique as a paradigm is the formalization of the process by which we assimilate, evaluate, and employ data (1). EBM can be formalized into a series of logical steps (4):
1. Formalizing your question into a well-built answerable question
Step 1 helps formalize your question, making your inquiry more explicit, and fosters the next step to search for the evidence. Most clinical questions are in PICO format, involving a Patient, an Intervention, a Comparison, and a clinical Outcome. For example, “In a 22 year old female with mild chronic non-comedonal acne (the Patient) is benzoyl peroxide monotherapy (Intervention) superior to salicylic acid monotherapy (Comparison) in preventing inflammatory papules from developing?” The more specific the PICO question, the more useful the answer, thus specifying the strength of benzoyl peroxide vs. salicylic acid may also be helpful in corresponding data. Online resources helpful in designing well-built clinical questions include Anatomy of a well-built clinical question (University of Sheffield) and Constructing a well-built clinical question using PICO (University of Washington).
Step 2 involves systematically searching for relevant data to answer our question (5). The objective is to have a comprehensive search that won’t miss the highest quality sources. When better sources are available, this step shuns textbooks and “experts” which tend to be outdated and vulnerable to bias. This search thus is geared preferentially towards the pinnacle of the “hierarchy of evidence.” The best source of information, when available, is the systematic review (particularly comprised of randomized controlled trials when the clinical query pertains to therapy). Systematic reviews answer focused study questions through explicit a priori methods, and are exhaustive searches incorporating study quality when appropriate. In dermatology, Cochrane Systematic Reviews compiled by the Cochrane Skin Group (Figure 1) generally provide the best answers (6). When these are not available and the question of interest pertains to a common skin condition, secondary journals such as Evidence-Based Medicine and ACP Journal Club provide structured abstracts to high quality studies and commentary helpful in their critical appraisal. Four times a year Archives of Dermatology features an evidence-based dermatology section with similar content. PubMed Medline is the most popular primary source, but most searches have a high noise to signal ratio. Pubmed’s clinical query database is designed for searches with fewer false positive (undesired or irrelevant) references (Figure 2). In recent years there has been increased attention to evidence-based references, which in contrast to traditional textooks formally integrate quality of evidence and are frequently updated. These include UpToDate and Clinical Evidence. A high-quality text, Evidence-Based Dermatology (7) is also available, and can provide an excellent starting point for an up-to-date search. General and dermatology-specific guides on how to perform evidence-based searches are available (1, 5, 7-9).
Figure 1: Cochrane Skin Group (CSG) CSG is part of the Cochrane Collaboration, dedicated to preparing, maintaining and promoting the accessibility of systematic reviews of the effects of health care interventions. This site features abstracts of Cochrane systematic reviews which by design meet the highest review standards.
Figure 2: PubMed Clinical Query Tool. PubMed’s Clinical Query tool was designed to clinical searches relevant to practice. Clinical searches can focus on etiology, diagnosis, therapy, prognosis, or clinical prediction rules. A narrow (specific) search will display the most relevant results only whereas, a broad (sensitive) search prioritizes comprehensiveness. For most dermatology searches, a narrow search is most appropriate to avoid being overwhelmed with the number of results.
Step 3 involves critical analysis of the data collected in step 2 to determine overall quality. For this purpose it is important to familiarize yourself with the basic terminology and concepts in clinical epidemiology as pertains to study quality.(10-12). A good study is reasonably free from bias and confounding (systematic errors), includes an adequate number of patients (i.e. is adequately powered to detect a clinically important difference), and is relevant to your patient.. Although a full review of critical appraisal is beyond the scope of this overview (entire books are published on the subject) the hierarchy of evidence offers a general rule to start with: systematic reviews of randomized trials are superior randomized trials, which are superior to cohort and other study designs, which in turn are superior to case reports and expert opinion. It is important to emphasize here that this rule by itself is inadequate since well-designed observational study may be more meaningful than a poorly designed or executed trial. For further reading on the critical appraisal of study design, the Centre for Health Evidence (CHE) offers an online User Guide (based on a JAMA series by the same title).
Step 4 relates to individualizing care and critically evaluating the context of the clinical problem (13-15). In step 3, we examined internal validity (how free it is from bias), but just as important is external validity, or how generalizable it is to your patient group and setting, which may be very different. Your 25 year old patient with mild 10% BSA psoriasis probably won’t respond the same as a 50 year old female patient in a tertiary care setting. Incorporating patient preferences and social settings into clinical decisions is imperative for compliance, patient rapport, and better outcomes. The Centre for Health Evidence offers a brief checklist to assist with these considerations.
Step 5 has us ask ourselves what we learned from Steps 1-4 in order to improve our next search. Storing our results for future reference is also valuable. Citation managers like EndNote store this information electronically and permit efficient sorting and searching of references.
Online, the most comprehensive listing of evidence-based dermatology resources can be found at the United Kingdom’s National Library of Health Skin Disorders Specialist Library . This massive initiative, funded by the United Kingdom’s National Health Service (NHS) indexes high quality, evidence-based information on all of aspects of skin disorders for patients and providers (Figure 3).
Figure 3: United Kingdom’s National Library of Health Skin Disorders Specialist Library. This comprehensive high-quality resource, funded by the United Kingdom’s National Health Service (NHS) was designed to meet the information needs of dermatologists, and allied health professionals in the search for high quality, evidence-based information. This unique database has search capabilities and is browsable by disorders, epidemiology and quality of life, diagnosis and investigation, treatment and management service delivery.
ebDerm.org (Figure 4) is another online evidence-based dermatology resource with a slightly different focus - a mission to teach and disseminate evidence-based dermatology. While the current version of this website includes an annotated guide to selected web-based evidence-based dermatology resources and PowerPoint guide, it is undergoing a major expansion with grant support from the Sulzberger Institute. This update will permit multimedia resources and collaborative evidence-based learning projects on August 1st 2007. This will include ebDerm Learning, a comprehensive guide to web-based resources, ebDerm Library, a digital library of evidence-based dermatology materials, and the ebDerm Community. A key highlight of this expansion will be the teaching of EBM via participation in a Critically Appraised Topic (CAT) clinical query based learning tool. Residents, dermatologists, and dermatology training programs interested in participating may contact the author of this publication at email@example.com.
Figure 4: ebDerm.org. The current version of ebDerm.org offers annotated hyperlinks to Internet resources and tools pertinent to learning and practicing evidence-based dermatology. The image above shows ebDerm.org as it will appear on its formal release date on August 1st 2007. This release will feature multimedia and interactive tools such as searchable Cochrane Skin Group abstracts and a Critically Appraised Topic (CAT) bank of community generated evidence-based dermatology summarizes for practice-based learning of evidence-based dermatology.
Like any other skill, proficiency in EBM searches (and efficiency obtaining your answer!) increases with consistent practice. This brief informal narrative was not intended to be comprehensive, but rather to whet your appetite for further reading on how EBM can be operationalized into practice. The author sincerely hopes that this most important learning objective was encouraged by the overview presented. Of the references below, 8 and 9 provide EBM guides focused on dermatology.
Disclosure Statement: Dr. Barzilai is Founder of Ebderm.org, a non-profit website devoted to promoting EBM in dermatology and is author of Barzilai DA, Freiman A, Dellavalle RP, Weinstock MA, Mostow EN. Dermatoepidemiology. J Am Acad Dermatol 2005;52(4):559-73; quiz 574-8.