Expert opinion: supplementing the gaps in evidence-based medicine.

Readers will note the appearance of a new type of article in this issue of Annals, the Expert Clinical Management series, and may wonder, what’s this about? Since the days of Hippocrates, physicians have relied on experienced and wise colleagues to advise them on how to manage difficult cases. In the past 50 to 60 years, however, some of the limitations of such advice have become better understood. As a consequence, much effort has gone into trying to make sources of expert information more objective, scientific, unbiased, and “evidence based.” Detailed and expanding guidelines and standards have been established about how this should be done. That’s all to the good, and we plan to continue to implement higher standards; Annals has worked hard for years to make our content as evidence based, and as free of bias, as possible. The proportion of randomized controlled trials (RCTs), systematic reviews, and meta-analyses in our journal has grown steadily, and we will continue that trend. However, the current model of evidence-based medicine fails to address the fact that we don’t have solid evidence for the majority of the care we provide, and no concrete plan for remedying that problem exists. Such a solution would take decades to achieve, and I personally think it will not ever be realized with the current approach. A good example of why is provided by reviewing any sound meta-analysis and noting how authors thoroughly search all conceivable databases for relevant articles, usually finding hundreds. However, many of these then do not meet the authors’ thresholds for number of subjects, study design, adequate controls, accurate analysis, etc. This leads to most being eliminated from further assessment so that most meta-analyses are based on only a few dozen studies at best. All the rest are discarded; no doubt many of them were junk, but probably not all. We’re just not smart enough to know how to extract the nuggets of truth and innovative ideas from less-than-perfect research. By the time the final careful analysis is done, we usually are left with a dozen or so articles deemed good enough to include. In many of these cases, the final conclusion is either that the results are contradictory or the studies have too much heterogeneity and no conclusions can be drawn. This is so common as to be a cliché, and yet it is the best medicine can do at present.