Stefan Sauerland1, Dirk Stengel. 1. Institut für Forschung in der Operativen Medizin, Universität Witten/Herdecke. S.Sauerland@uni-koeln.de
Abstract
INTRODUCTION: Critical appraisal of medical literature is a key component in the practice of evidence-based medicine (EBM) and in guideline development. Various schemes have been proposed to grade scientific evidence. We examined whether such grading can be taught. METHODS: Thirteen participants in an EBM course were asked to classify 10 short study abstracts into Sackett's evidence levels (I to V). The studies related to therapeutic, diagnostic, and aetiological questions (n = 4, 3, and 3, respectively) in the field of trauma surgery. After a four-hour course, another 10 abstracts were classified in order to enable a randomised pre-post comparison. RESULTS: The course increased the proportion of correctly classified articles from 47% to 56%. This effect of +9% was nonsignificant (p = 0.14; 95% confidence interval [95% CI]: -2% to + 19%). The rating of diagnostic studies was more difficult (46% correct; 95% CI: 37% to 55%) than the rating of therapeutic studies (58% correct; 95% CI: 52% to 64%). Only two studies reporting either level-I evidence (randomised trial) or level-IV evidence (case series) were correctly classified by all participants. CONCLUSIONS: A half-day EBM training for clinicians is not sufficient to enhance the ability to categorise study abstracts to evidence levels.
INTRODUCTION: Critical appraisal of medical literature is a key component in the practice of evidence-based medicine (EBM) and in guideline development. Various schemes have been proposed to grade scientific evidence. We examined whether such grading can be taught. METHODS: Thirteen participants in an EBM course were asked to classify 10 short study abstracts into Sackett's evidence levels (I to V). The studies related to therapeutic, diagnostic, and aetiological questions (n = 4, 3, and 3, respectively) in the field of trauma surgery. After a four-hour course, another 10 abstracts were classified in order to enable a randomised pre-post comparison. RESULTS: The course increased the proportion of correctly classified articles from 47% to 56%. This effect of +9% was nonsignificant (p = 0.14; 95% confidence interval [95% CI]: -2% to + 19%). The rating of diagnostic studies was more difficult (46% correct; 95% CI: 37% to 55%) than the rating of therapeutic studies (58% correct; 95% CI: 52% to 64%). Only two studies reporting either level-I evidence (randomised trial) or level-IV evidence (case series) were correctly classified by all participants. CONCLUSIONS: A half-day EBM training for clinicians is not sufficient to enhance the ability to categorise study abstracts to evidence levels.