Abstract
OBJECTIVE: To illustrate the current meta-analytic approach to extracting clinically useful information from the glut of randomized controlled trials (RCTs) of the treatment of acute attacks of Crohn's disease.PROCEDURE: Acceptable RCTs were classified in a matrical format so that the feasibility of a meta-analysis of each treatment could be quickly determined. Each potential inclusion was then photocopied after blinding the source and outcome to minimize bias by the investigators.RESULTS AND CONCLUSIONS: The search was started in 1990, when a previous meta-analysis was completed. Over 100 citations in MEDLINE contained 21 RCTs of the treatment of acute attacks of Crohn's disease, and 11 more were culled from the references of recent papers. The meta-analysis most indicated by the matrix was a comparison of several different immunosuppressive regimens with a placebo. The previously demonstrated statistical significance was confirmed and the point made that it is no longer justified to compare any new treatments with a placebo. Cumulative meta-analysis, the process of performing a new analysis every time a new trial is published, revealed the importance of employing both random and fixed effect models when heterogeneity of treatment results is appreciable, as in this case.