Xueting Wang wrote this blog post. Xueting Wang developed this add-in as part of the Biostatistics Undergraduate Summer Internship (BUSI) program at the University of North Carolina at Chapel Hill.
Introduction
What data visualization do you use when you want to view the evidence from different studies and interpret the overall findings of a systematic review? The forest plot is a good choice, as it displays several confidence intervals or credible intervals vertically in a compact space. The JMP Forest Plot Add-In (valid for JMP 17 and later releases) provides a straightforward way to produce forest plots and has numerous applications, helping you present a variety of results, including:
- A single endpoint for multiple studies, as well as the overall estimate of that endpoint across studies.
- Multiple endpoints of one study.
- A single endpoint within subgroups.
- Observed results or any changes from baseline for a single endpoint across time.
Dialog and roles
The roles of the dialog include:
- Subgroup is the value of a character column that designates a column to be a subgroup factor. When it is applied, the individual rows with the same subgroup will be grouped together. It will be presented on the Y axis with values of Label nested within values of Subgroup.
- Label (required) is the value of a character column that is displayed on the Y axis. The values of the Label column are assumed to be unique across rows of the data table if Subgroup is not used, or unique within the rows for each value of Subgroup if used.
- Lower Limit (required) is the value of a numeric column that presents the lower limit value of the credible or confidence interval for the estimate.
- Center Value (required) is the value of a numeric column that contains the estimate for which the credible or confidence interval is being summarized.
- Upper Limit (required) is the value of a numeric column that presents the upper limit value of the credible or confidence interval for the estimate.
- Confidence Interval Text is the value of a character column that shows the text on the right axis, usually including the estimate and the confidence interval.
- Size is the value of a numeric column that contains the sample size of each row that will be used to modify the size of the marker for the Center Value.
There are several other options in the dialog:
- Add Null Reference
Checking this box will add a dashed vertical reference line based on the null value and enable:
- Color Preferences for Intervals Above/Below Null Value: There are three options for the color of the intervals. It can help accentuate the intervals that exclude the null value. If the interval includes the null value, it will be black.
- Null Value: This box sets the null value to your preference.
- Data Table Order
When this option is checked, the forest plot will present the rows in the order that they are in the data table. Otherwise, the rows will be in the reverse order of the data table.
- Add Confidence Diamond for Overall or Total Label
When the value of the Label for the last row is equal to Overall or Total, a diamond instead of a regular marker will represent the confidence or credible interval.
How to use
- Download the Forest Plot Add-In from the JMP Marketplace and double-click to install. Once installed, it can be found in the Add-Ins menu.
- Open a data table. Then select Add-Ins > Forest Plot to open the dialog.
- Drag columns into Roles in the dialog and select other appropriate options.
- Click Run to produce the plot.
Now we will show how to use the Forest Plot Add-In with several data tables from the medical literature.
ForestPlotSample_HBV.jmp contains the primary and six secondary binary endpoints from a clinical trial in patients with chronic hepatitis B viral (HBV) infection. It includes the columns Endpoint Type (subgroup), Label, the Difference in proportions between entecavir and lamivudine, the Lower and Upper limits of the 95% confidence intervals, and a formatted field containing text of the estimate and 95% confidence interval. The basic forest plot with columns assigned to roles is presented in Figure 2 with Null Value = 0.
The forest plot produced using the HBV data is presented in Figure 3. Confidence intervals fully to the right of the reference line are colored blue as a visual cue that these intervals to not contain the Null Value.
We can apply the Subgroup (Endpoint Type) and Confidence Interval text roles to this plot (Figure 4). Now, the Label is nested within Subgroup on the left axis, and the Confidence Interval text is located along the right axis.
We use the data table ForestPlotSample_RosaMeta.jmp, which contains a frequency column and an “overall” row to demonstrate the Size role and Add Confidence Diamond for Overall or Total Label option for a meta-analysis of patients with type II diabetes mellitus (T2DM).
Figure 6 presents a forest plot with the Size role and Add Confidence Diamond for Overall or Total Labelapplied. The size of the marker of the estimate of each interval varies according to the count column, which represents the sample size. A confidence diamond is drawn for the meta-analysis result in the Overall row at the bottom of the plot.
This new add-in addresses several issues identified in early versions of the Forest Plot Add-In (details are here). Further, except for the Confidence diamond, all elements are produced with native Graph Builder functionality, making it easier for users to modify forest plots for color or symbol size beyond what is initially produced by the add-in. Hate the blue and red color scheme for emphasizing intervals? These colors can now be changed according to user preferences, and the script of the modified presentation can be saved to the data table to reflect any post-add-in modifications.
If you like the add-in, please give us a five-star review! New features can be requested in the Q&A.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.