A look at the statistical overview (or meta-analysis)

Abstract

In essence, the statistical overview (or meta-analysis) does little more than many medical scientists and practising clinicians have been attempting for years: estimating the average size of an effect from an apparently disparate medical literature. The statistical overview provides a method of pooling the results of comparable trials (or observational studies) which, on their own, may be too small to demonstrate a 'statistically significant' effect. The role for the overview is obvious, the method is relatively simple, but the findings have to be interpreted cautiously, since joining the hypothesis post hoc contravenes one of the essential principles of scientific enquiry. The use of the statistical overview and some of its pros and cons are illustrated in four clinical examples.