Systematic Reviews: How to Analyze a Systematic Review

Forest plots are visual summaries of essential information found in a meta-analysis. Each study included in the MA is plotted with its own line and includes the size of the intervention group and control group, the confidence interval, and the weight. These are all plotted against the Line of No Effect (either 0 or 1). 

Higgins JPT, Thomas J, Chandler J, et al., eds. Cochrane Handbook for Systematic Reviews of Interventions. 2nd ed. Hoboken, NJ: Wiley-Blackwell|; 2019.

Above is an example of a forest plot from the Cochrane Handbook for Systematic Reviews of Interventions. In order to read this, you'll need to understand how forest plots are built.

• Each study in the meta-analysis is represented as its own line on the plot, usually denoted by the author and year of the study.
• The tick mark represents the confidence interval of the study, so the longer the horizontal line, the larger the CI.
• The box or dot on that line represents the sample size; big box is a higher sample size, small box is a smaller sample size.
• The diamond at the bottom is the representation of the combined CIs and sample size (this is calculated during the meta-analysis process).
• The vertical line is called the Line of No Effect. To the right of the line is the intervention side; the left is the no intervention/placebo side. Studies that fall on the intervention side of the plot show that the intervention was effective. Studies on the placebo side did not have findings supporting the efficacy of the intervention. The Line of No Effect can either start at 0 or 1. This is determined during the statistical calculations of the meta-analysis. 

An important part of understanding a forest plot is understanding what the combined results show. If the diamond representing the combined results falls on the right of the Line of No Effect, the meta-analysis finds the interventions to be effective. If it touches the Line of No Effect or is completely on the left side (no intervention), the meta-analysis finds that the intervention was not effective, or is not statistically significant enough to make that determination. 

Another thing that a forest plot can show is whether or not the studies are homogenous. In order to do a quality meta-analysis, the studies must be homogenous. If they are not, it will appear on a forest plot. In order to check this, see if you can draw a line through all the tick marks (CIs) of all the studies. If you can, the meta-analysis is using homogenous studies. 

PRISMA flow charts are often included in systematic reviews and meta-analysis. The completed PRISMA chart below from Yang & Deng (2020) shows and example of what you're likely to see.

Yang C, Deng S. Laparoscopic versus open mesh repair for the treatment of recurrent inguinal hernia: a systematic review and meta-analysis. Annals of Palliative Medicine 2020;9(3):1164-1173. doi:10.21037/apm-20-968

Funnel plots are the graphical representation of the size of trials plotted against the effect size they report (Lee & Hatopf, 2012). They are scatter plots of the intervention effect estimates from a study against the study's size or precision (Cochrane Handbook, 2019). Chapter 13 of the Cochrane Handbook for Systematic Reviews of Interventions explains funnel plots in more detail. 

Higgins JPT, Thomas J, Chandler J, et al., eds. Cochrane Handbook for Systematic Reviews of Interventions. 2nd ed. Hoboken, NJ: Wiley-Blackwell|; 2019.

Risk of Bias tables summarize the judgment of bias for the studies used in the systematic review. Each study's judgments for types of bias are represented with colored dots and symbols. Green + dots denote low risk bias in an area, red - dots denote high risk of bias, and yellow ? dots denote unclear of unknown risk of bias. The types of bias included in these tables are determined by the researchers and what areas they believe to be most important within the context of the review. 

The RoB 2 tool from Cochrane is a quick way to make a Risk of Bias chart like the one below.