What Do The Results Mean?

It is quite reasonable for authors just to report trial results in a systematic review. For instance ‘We found that all the trials reported a beneficial effect’.

Mathematically combining the results (Meta-analysis) is useful when the trials are all pretty similar. Meta-analyses can be usefully summarised in a Forest plot or ‘Blobbogram’, like the one below. There are a couple of key points to identify on a Forest plot. The individual box and whisker icons represent individual trials. The more patients there are in a trial, the bigger the box is. This is referred to as the weight of the trial. The location of the box indicates the size of the effect and the edges of the whiskers represent the confidence intervals.

The diamond shape at the bottom of the diagram below represents the pooled estimate. The lateral edges of the diamond represent the confidence intervals. The middle of the diamond corresponds to the estimate of the effect. The maths of generating a pooled estimate of effect are very complicated and, though there is freely available software, it is best left to someone with statistical expertise. Very basically there are two methods of pooling results, ‘fixed effects’ and ‘random effects.’ Random effects is best used when there is a lot of heterogeneity and fixed effects when there is little heterogeneity. Most meta-analyses will report both random effects and fixed effects results.

Meta-analysis is not simply adding and splitting trial results.