Evidence-based medicine.

The quality and credibility of medical publications vary substantially. The greatest problem associated with publishing information that is not entirely valid is that it may mislead physicians in their clinical management of patients. This is true particularly because some published literature looks quite believable at first reading, even though it does not stand up to closer scrutiny. Recently, emphasis on evidence-based medicine (EBM) has sought to address this problem.

Movement toward EBM does not necessarily mean that every article published must be the result of a double-blind, placebo-controlled trial. This is fortunate for publishers, because if that were the case, we would have far fewer journals. Nevertheless, the growing awareness of the importance of EBM emphasizes the need for critical analysis and scrutiny of every article we read. It also provides a structure to help evaluate the credibility of a publication. Hence, it is valuable for all otolaryngologists to acquire a basic understanding of EBM.

In order to approach a problem using the EBM paradigm, the information sought on a typical topic is structured into questions that can be answered. These questions should be addressed efficiently, using the best possible evidence (see information below on levels of evidence) to elucidate the answers. The evidence obtained is evaluated critically to determine its validity and to determine whether it answers the questions in a useful way. The results are then combined with information based upon clinical expertise and previous experience gleaned from the literature. Then, outcomes are evaluated. This systematic approach to expanding scientific knowledge is the essence of the EBM "movement," although a great deal of attention is paid to one portion of this paradigm, specifically: the level of evidence (LOE).

The LOE helps a reader understand the credibility of the results presented. Are they valid? That is, are they unbiased and true? Validity does not necessarily predict clinical relevance, but it is certainly important to understand whether reported information is believable before deciding whether to rely upon it for changes in clinical management.

Evidence may be classified into five categories. Level 1 evidence (the highest category) includes randomized, controlled trials; meta-analyses; systematic reviews; and reports of diagnostic sensitivity and specificity. In randomized, controlled trials, two identical groups are divided randomly into a control group and an experimental group; both groups are followed prospectively for specific endpoints. Meta-analyses involve quantitative synthesis (or pooling) of data from numerous independent clinical trials. Systematic reviews utilize specific methods to identify primary studies, assess their quality, and review their data. Diagnostic sensitivity and specificity studies report the true positive and negative rates for diagnostic tests.

Cohort studies are classified as level 2 evidence. They involve groups of people who are identified because of their exposure to a particular agent, for example, and who are followed for selected outcomes. Case control studies (level 3 evidence) compare subjects who have a specific condition with matched "controls," retrospectively analyzing differences between the groups. Cross-sectional surveys (surveys or interviews of a subject population) and case reports for short case series are examples of level 4 evidence. Expert opinion, in the form of opinion from a preeminent authority or consensus from a group of authorities, constitutes level 5 evidence. Anecdote and conversation are not classified with a defined evidence level.

It should be obvious immediately that lower rank does not necessarily mean invalid or "bad" information. In fact, well-synthesized expert opinion from a panel of world leaders may provide invaluable information, as may well-written case reports. Nevertheless, it is important for readers of medical literature to recognize that the validity of conclusions drawn from level 3 or level 2 studies (or lower level reports) may not be as certain as the validity of conclusions derived from level 1 studies. Skepticism is wise when interpreting any medical literature, and physicians are always cautious about what they believe.

The old adage "Save me from the doctor who has just read a paper" calls to mind the pitfalls associated with less-than-critical absorption of information from the medical literature. Evaluating each paper for LOE helps physicians sort out articles on which they should rely from those with insufficient power to support a clinical decision.

In addition to using LOE to evaluate published literature, otolaryngologists should also remember classification of evidence when designing a study or preparing to write a paper. Often, it takes surprisingly little extra effort to increase the credibility of a study by modifying the design to increase its evidence level. This can sometimes make the difference between valid and invalid results, and between acceptance and rejection of a manuscript.

Medical writers and publishers are still struggling with sensible adoption of the principles of EBM. Overall, the trend toward a more rigorous study design appears healthy, and heightened awareness of the strengths and weaknesses of the study design of each published paper is certainly advantageous to physicians and our patients.

ROBERT T. SATALOFF, MD, DMA 

Editor-in-Chief

EAR, NOSE & THROAT JOURNAL