Transformation of evidence to knowledge: a neglected task within the medical profession.

A common and far-reaching fallacy among physicians is to believe that the terms evidence and knowledge mean the same thing. Evidence means the result(s) of research, whereas knowledge stands for 'true justifiable belief.' This difference becomes obvious when we take a look at the results from different studies on the same issue. Results of studies on the same issue sometimes vary considerably and make it difficult for the physician to know which results are relevant for the patient sitting or lying in front of him. Two examples for illustration; within the last ten years about 600 studies evaluating the diagnostic accuracy of D-Dimer in the diagnosis of venous thrombosis have been published and the values for the sensitivity of the test varies between 70% and 98%. Trial results reporting the effect of steroids on survival in patients with septic shock differ between no effect at all and a significant reduction of mortality. Which of these results should physicians trust? As a consequence the transformation of evidence to knowledge, including the handling of varying and partially contradictory results, could and should be one of the main tasks of academic medicine. The transformation of evidence to knowledge makes the results from research applicable to daily practice. An attempt to accomplish this demand is to synthesize the results of various primary studies in systematic reviews with meta-analyses. The efforts of the Cochrane Collaboration can be mentioned favorably, but some reservations against the assembly line like production of systematic reviews are justified. It is not understandable why primary studies of low methodological quality, rated as such by the reviewers, are still included in systematic reviews. Further, the concern that results from studies with very different patient populations are pooled in systematic reviews pertains. Two approaches to generate medical knowledge that is helpful for physicians will be presented in the lecture for further discussion; the broader utilization of consensus methods, and the derivation of probability functions. Medical knowledge can be defined as 'shared experts belief' and consensus methods (Delphi technique and others) are appropriate to specify these beliefs. Beside consensus a valuable result of a carefully performed consensus process is the precise specification where and why disagreement remains. The ideal form of medical knowledge is a comprehensive set of probability functions, for the diagnostic, the prognostic and the etiognostic domain. These probability functions, stored in cyberspace can be used to calculate the probabilities about the presence of an illness, the future course of an illness and the etiology of an illness. Such probability functions can be derived from harvesting experts' knowledge - the quasi scientific method -, or from patient data - the scientific method. The emerging opportunities in computational science and information technology will make the dream - a comprehensive set of probability functions - to become reality. [ABSTRACT FROM AUTHOR]