Forest Plots

How do I present the results of a meta-analysis?

Information from a cohort study can be displayed in a 2 x 2 table, see Table a.

After performing the meta-analysis, the results could be presented as text in the review. However, more commonly the results are displayed using a figure, called a ‘forest plot’.

Here is an example of a forest plot from a systematic review of randomised controlled trials which assessed the effect of giving an inexpensive course of corticosteroids to women who were about to give birth prematurely. Each of the circles with horizontal lines represents an individual study. Therefore there are seven studies in this review. The diamond at the bottom of the plot represents the pooled quantitative result from the meta-analysis. The pooled result from this meta-analysis indicates strongly that giving pregnant women who are about to give birth early a simple inexpensive course of corticosteroids significantly reduces the risk of babies dying from complications of immaturity.

At the top of the plot is the title of the review, and a label to tell you what the treatment comparison and the outcome of interest are. For this review, the treatment comparison is corticosteroids versus placebo, and the outcome of interest is early neonatal deaths. At the bottom of the plot there is a line which tells you the scale for the treatment effect you are measuring. In this example the outcome is the proportion of babies that died, and the effect measure is the Odds Ratio. Towards the left of the scale, the scale is less than 1 which means that the treatment has made the risk of neonatal death less likely. Towards the right of the scale, the scale is more than 1 which means that the treatment has made the risk of neonatal death more likely. The vertical line in the middle of the plot is where the treatment and the placebo have had the same effect on the risk of neonatal death. In other words, there is no difference in risk between the two treatment groups, so that the risk of neonatal death in the corticosteroid group is the same as the risk in the placebo group.

Here are the results of the first randomised controlled trial conducted in 1972. This is the label for the trial; usually given as the first author’s name, or the acronym of the study. The quantitative results from this study are presented for each treatment group. The lower case n refers to the number of babies that died in the corticosteroid group, and the uppercase N refers to the total number of babies that were recruited into the likely. Towards the right of the scale, the scale is more than 1 which means that the treatment has made the risk of neonatal death more likely. The vertical line in the middle of the plot is where the treatment and the placebo have had the same effect on the risk of neonatal death. In other words, there is no difference in risk between the two treatment groups, so that the risk of neonatal death in the corticosteroid group is the same as the risk in the placebo group. Here are the results of the first randomised controlled trial conducted in 1972. This is the label for the trial; usually given as the first author’s name, or the acronym of the study. The quantitative results from this study are presented for each treatment group. The lower case n refers to the number of babies that died in the corticosteroid group, and the uppercase N refers to the total number of babies that were recruited into the corticosteroid group. For the Liggins and Howie study, there were 36 deaths in the corticosteroid group and a total of 352 babies in the corticosteroid group. The data for the placebo group is then entered using the same format. The weight is the percentage weight given to this study in the pooled meta-analysis. The general rule is that the bigger the study in terms of sample size, the larger the percentage of weight given to that study. The figure in the middle of the plot shows pictorially the results for the individual studies, here the treatment effect for the individual study is represented by a square. The size of the square is proportional to the percentage weight of the study in the meta-analysis. So the larger the square the more weight the study has in the meta-analysis. The horizontal line represents the certainty of the treatment effect, where the shorter the line the more certain we are about the treatment effect. The horizontal line is called the confidence interval. Usually 95% confidence intervals are presented on forest plots. If the horizontal line touches or crosses the vertical line, it means that the trial found no significant difference between the two treatments. The position the circle to the left of the vertical line shows that this trial found the corticosteroid treatment was beneficial as compared to placebo.

Looking at the plot as a whole, at the bottom of the forest plot, a diamond is shown. The middle of the diamond is the pooled meta-analysis result for the treatment effect, and the width of the diamond is the certainty of the result, again usually presented as a 95% confidence interval. If the confidence interval crosses the vertical line, then there is no significant difference in the effectiveness of the corticosteroid treatment group and the placebo group. However, in this example, we can see that the effect is significant and a course of corticosteroids is beneficial in reducing the risk of neonatal deaths, since the horizontal line does not cross the vertical line, and the diamond is to the left of the plot.