[Why evidence-based medicine? 20 years of meta-analysis].

BACKGROUND: Meta-analysis, described within evidence-based medicine, has become a frequent issue in recent medical literature. An exhaustive search of reported meta-analysis from any medical specialty is described. MATERIAL AND METHODS: Search of papers included in Medline or Embase between 1973-1998. A study of intra and inter-reviewers liability about selection and classification have been performed. A descriptive analysis of the reported papers (frequency tables and graphics) is described, including differences of mean of reported meta-analysis papers by medical specialty and year. RESULTS: 1,518 papers were selected and classified. Most frequently found (45.91%) were: methodology (15.7%), psychiatry (11.79%), cardiology (10.01%) and oncology (8.36%). Inter personal agreement was 0.93 in selecting papers and 0.72 in classifying them. Between 1977-1987 overall mean of reported studies of meta-analysis (1.67 + 4.10) was significatively inferior to the 1988-1998 (49.54 + 56.55) (p < 0.001). Global number of meta-analysis was positively correlated (p < 0.05) with the number of studies about fundamentals and methodology during the study period. CONCLUSIONS: The method used to identify meta-analysis reports can be considered to be adequate; however, the agreement in classifying them in medical specialties was inferior. A progressive increase in the number of reported meta-analysis since 1977 can be demonstrated. Specialties with a greater number of meta-analysis published in the literature were: psychiatry, oncology and cardiology. Diffusion of knowledge about fundamentals and methodology of meta-analysis seems to have drawn and increase in performing and reporting this kind of analysis.