The Use of Systematic Review and Meta-Analysis in Modern Epidemiology Nuno Lunet

The extent to which systematic reviews/meta-analyses are used properly, their results interpreted correctly, and recognized as useful resources in biomedical research depends on the understanding of their methodological bases and on the pursuing of objectives compatible with the level of inference allowed both by these methods and by the characteristics of the primary data sources. This comes ultimately from the comprehension of “the nature of meta-analysis” [4], which has been summarized beautifully as the “epidemiology of results of independent studies” [5] or “observational study of the evidence” [6], where the subjects are independent investigations, just as in ecological designs the group replaces the individual as the unit of analysis. In Modern Epidemiology, Greenland and O’Rourke [4] state that “meta-analysis can be viewed as the transference of good analytic practice from the single-study to the multiple-study context”, and recognize that the identification, abstraction and analysis of data from different studies “parallels the need of single studies to identify eligible subjects, abstract their information, and analyze the resulting data by summarizing information across subjects”.