[The practice of systematic reviews. IV. Pooling results from separate studies]

Meta-analysis comprises the analysis of the data gathered in a systematic review. Meta-analysis can be defined as the combination of pooling results of similar studies into one overall estimate of the effect on the one hand and the exploration of sources of heterogeneity between studies on the other. A general statistical model is available for pooling the estimates of various types of effect parameters. Pooling enables the estimation of the effect of an intervention or exposure as precisely as possible (with narrow confidence intervals) which allows detection of minor effects. The Mantel-Haenszel method is also an excellent method for pooling effect parameters of dichotomous outcomes. Pooling provides an efficient means of summarizing the available knowledge.