Standardized Reporting of Clinical Trials into Electronic Trial Banks: In Support of Computer-assisted Evidence-based Medicine

Note: This technical report is intended for readers who are generally familiar with critical appraisal of the clinical literature, but who do not have any background in database or artificial intelligence technologies. ABSTRACT Though clinical trials are our best source of clinical evidence, it is logistically difficult to practice medicine based on the results of the most up-to-date clinical trials. Frontline practitioners simply do not have the time nor the expertise to formally synthesize evidence from the clinical literature, a synthesis which is increasingly done with a statistical technique called meta-analysis. With the advent of widespread digital publication, now is an opportune time to build an infrastructure for intelligent computer systems to assist us in meta-analysis and the practice of evidence-based medicine. This paper proposes an architecture for the standardized reporting of randomized clinical trials directly into electronic knowledge bases that can be used by many different poeple and computer systems. For these knowledge bases, or trial banks, to be well-integrated and practical, they must share a common conceptual model of clinical trials. This conceptual model must be able to record information about the design, execution, and results of trials in any stage of completion, from planned to fully completed. We report on a conceptual model of clinical trials that incorporates trial reporting recommendations from expert trial commentators. Our model supports three of the major tasks involved in using the clinical literature: (1) accurate retrieval of trial reports, (2) judging the validity and generalizability of the trials, and (3) quantitative synthesis of the trial results. The model is a blueprint for building knowledge bases that will have the form and content necessary for supporting computer-assisted meta-analysis and evidence-based medicine.