Des Principes De La Méta-Analyse Non Statistique Pour UNE Revue Critique De La Recherche Biomédicale

Abstract

In most circumstances, considerable time lags before results from biomedical research are integrated in general medical practice are inevitable. Several economic, technical factors and educational orientation are responsible for this gap. Effective information management at the institutional level could provide clinicians the means to adopt new perspectives. At present, the link between basic research and clinical application relies upon the motivation of individuals to develop new skills in literature retrieval and analysis and yet their occupational responsabilities inhibit this endeavour. Academic librarians, educated and trained in information and biomedical sciences, have a role in providing physicians with current and readily applicable knowledge from the critical analysis of the literature.

The Faculty of Medicine Library is developing a research programme to design an updated and comprehensive picture of the diagnostic and prognostic biological markers in the field of human prostate cancer. The final product is envisioned as a database accessible on the web. As this product is designed to have a direct impact on clinical decision making and patient care, it should guarantee thoroughness, relevancy and objectivity and not become yet another narrative review of the literature. Over the last decade, there is a move towards using statistical meta-analyses to synthesize literature. This has proven to be useful when similar strategies are used to produce similar quantitative outcomes that can be merged for statistical analysis.¹ Even though studies in cancer research assess the same model, they are often too fragmented or heterogeneous in their design, methods and instrumentation. Approaches have been developed to conduct non-statistical meta-analysis of such publications.² Quantitative and qualitative meta-analyses are based on common principles largely discussed in the literature, and are related to model development, literature searching, rating references for quality and annotating and synthesizing high quality references. In practice, these basic principles require translation into actual instructions adapted to every domain of biomedical research. Model definition is a very critical step that governs the quality of subsequent analyses. Specific examples illustrate the guidelines acquired from our experience.