Meta-analysis

Abstract

Meta-analysis corresponds to all systematic methods which use statistical techniques for combining results from several independent studies. The aim is to get a consistent estimation of the global effect of a procedure on a specified outcome. The technique allows us to increase the power of statistical testing, and to get information which cannot be drawn from one individual study. Two approaches are possible, and often combined: the qualitative approach consists of weighing various studies according to their methodological quality; the quantitative approach consists of pooling the results of different studies, in order to generate results with a higher statistical power. A meta-analysis is a long and rigorous process, which follows several steps: statement of objectives; definition of articles inclusion and exclusion criteria; literature search; collection of data and evaluation of the quality of each study; tests for homogeneity; pooling; sensitivity analyses; presentation of results; and conclusions. Meta-analysis has several advantages: it estimates the size of an effect; it improves the generalizability; it compels to rigor; it lessens the part of subjectivity. When meta-analysis is cautiously and properly done, it brings new useful information, and helps physicians and health policy makers in answering to a specific question.