A meta-analysis is a statistical analysis that synthesizes and summarizes the outcomes of similar individual studies. Not all systematic reviews contain a meta-analysis, because not all topics have sufficient research evidence to allow a meta-analysis to be conducted. A meta-analysis should also not be conducted when there are significant differences in the populations or interventions of the individual studies.
Why Do a Meta-analysis?
Combining information from all relevant studies provides a more precise estimate of the findings than information from individual studies.
Answer questions not posed by individual studies.
Look at the comparative effectiveness of multiple interventions for the same condition.
Settle controversies arising from conflicting studies.
Generate new hypotheses.
Assess the homogeneity or heterogeneity of the results.
Quantify between study variations.
Reduce problems of interpretation due to sampling variations.
Decide what would be meaningful to analyze.
Think about both the qualitative (summary and discussion of the studies' characteristics) and quantitative elements (statistical analysis).
Include the specification of the comparison.
Is the data continuous, dichotomous or categorical?
A forest plot is a graphic representation of a meta-analysis displayed in a single image. It summarizes the results from all the relevant studies that ask the same question, and shows their confidence intervals, and the estimated common effect statistic. A forest plot also gives a visual representation of the amount of homogeneity or heterogeneity among the studies.
The left-hand column of the forest plot lists studies by author names and date in chronological order from the top down.
The right side displays the effect size of the individual studies (indicated by a black square), and the the 95% confidence interval (indicated by a line through the black square). The size of each square is proportional to the study's weight in the meta-analysis. Studies with greater sample sizes and smaller confidence intervals are indicated by larger squares, and they contribute to the pooled result to a greater degree. Smaller studies are represented by smaller squares with long lines through them, because the 95% confidence interval is broad.
The meta-analytical result is the diamond on the bottom of the forest plot, which quantifies the effect size as well as its uncertainty. The lateral points of the diamond indicate confidence interval for this estimate.
The following software can be useful for creating forest plots and is available to purchase: