OK, I promise this is the last article on Forest Plots (at least for a while).
In the previous article on Subgrouped Forest Plot with Font Attributes, I discussed how to use bold text for subgroup headings. I mentioned that increasing the font size would not work as it would misalign the subgroup values from the headings.
Clearly, the result shown in the link above is less than ideal, and I have not yet been able to come up with a better way to do this using SAS 9.2. If you have a better solution, I am sure we all would be very interested to hear of it.
But with SAS 9.3, we can do better because we can use the new HighLowPlot statement. This statement supports a LowLabel and/or HighLabel. In this case, the label is drawn starting from the appropriate end of the plot line. So, all the contortions needed for the MarkerCharacter solution with SAS 9.2 above can be avoided.
Forest Plot using SAS 9.3 HighLowPlot. Here is the graph. Click on it for a bigger view:
Note the following improvements in this graph:
- The subgroup heading and values use the same font family as the rest of the columns. It is not necessary to use a non-proportional font.
- The subgroup headings use a bigger font size and bold weight.
- The subgroup values are indented.
- The error bars do not have serifs.
In this graph, we have used the HighLowPlot to draw the subgroup headings and values. The actual drawing of the high low line itself is hidden by using line thickness of zero. The other text strings are still drawn using ScatterPlot (with MarkerCharacter), but that too can be changed to HighLowPlot if we need left or right aligned strings. The error bars for the Hazard Ratio are also drawn using the HighLowPlot.
With SAS9.3, the ScatterPlot itself also has a new option to draw DataLabels with DataLabelPosition of LEFT | RIGHT, etc. So, we could have used that for these labels. But there are some interesting interactions with axis offsets that need to be considered. So, at this time, HighLowPlot is preferred.
GTL code snippet for the subgroup labels:
highlowplot y=obsid low=zero high=zero / highlabel=heading
lineattrs=(thickness=0) labelattrs=(size=7 weight=bold);
highlowplot y=obsid low=zero high=one / highlabel=subgroup
lineattrs=(thickness=0);Forest Plot without horizontal bands:
Full SAS 9.3 Code: ForestPlot_93
9 Comments
Hi Sanjay,
At long last a SAS 9 way of making a forest plot with sub-group headers and indentation using a proportional font.
Excellent!
I've been banging my head against a wall trying to get a SG version of this for over a year. I had used PROC REPORT with included graphical elements each created individually, before settling on a much more straightforward PROC GPLOT/ANNOTATE solution.
Now all I've got to do is to wait for SAS 9.3 to be available at my company.
Kind regards
-- Jeff
Hi Sanjay,
Thank you very much for providing the code for the forest plot, its been incredibly useful! I was just wondering if there is any way to make the xaxis label of the plot dynamic so that it is centered around the Hazard ratio of 1 when the Low and High values change. Thanks.
Kind regards,
Anish
At this time you could do this by turning off the display of the axis label and using two DRAWTEXT statement with DRAWSPACE of DATAVALUE and the appropriate anchor point. You could use the unicode symbol for "<-" and "->" for better results.
DRAWTEXT textattrs=(size=7 weight=bold) "<---PCI Better" / drawspace=wallpercent widthunit=percent
x=0 y=-4 width=100 anchor=topleft;
DRAWTEXT textattrs=(size=7 weight=bold) "Medical Therapy Better--->" / drawspace=wallpercent widthunit=percent
x=100 y=-4 width=100 anchor=topright;
I will write up an example in the next blog post. With SAS 9.4, you will be able to use SGAnnotation.
Thanks a lot Sanjay!! Wish my brain worked half as well as yours
. For one of my plots, I have a lower HR of 0.91 and an upper HR of 20 odd which makes the plot "squish" to the left. Would you be able to suggest a way to "unsquish" the plot? Thanks.
Kind regards,
Anish
Hi Sanjay,
I want to create the following plots in SAS 9.3 - Funnel plot, Forest plot, ROC curve. I am doing Metaanalysis on diagnostic tests. THis is just what I need to graduate and I am running out of options as no one in my class knows how to do it. I did SAS in my Biostat class 2 years ago. My former prof has retired. A programmer offered his help but we both couldn't figure out the code. Could you please give me the correct code for the Forest Plot, Funnel Plot and ROC. My data includes author's name, laboratory test, sensitivity and specificity. Thank you so much for your help. -Khay
I have included code for many versions of the FOREST plot in previous posts in this blog. I am sure you can find one that is suitable to your need. Send me examples of the Funnel plot and ROC plot, with data, and I can help you.
Re. Anish's question:
I chose a similar but different approach: I created a second row in the lattice in which I added the two text labels in an entry statement, left and right aligned:
entry textattrs=(family='Arial' size=7pt) halign=left "<---Control better" halign=right "Test better--->";
But what I am wondering is if there is any way to get rid of the vertical white lines that separate the horizontal bands at the column borders? walldisplay=none does not seem to help.
Kind regards,
Juergen
Hi Sanjay,
Thank you for your reply. Below is the sample data which I would like to create the Forest Plot and Funnel Plot in SAS 9.3. I found similar Forest plots from your previous posting but none has the same variables. I hope you can tell me the full code for this example. Thanks, Khay
Author Assay Sensitivity Specificity
Convert, M. et al RT-PCR Roche 1.00 0.855
Chan, K.L. et al RT-PCR Roche 0.45 0.99
Atkins, K.L. e al RT-PCR IDI GBS 0.868 0.952
Davies, D.H. et al RT-PCR IDI GBS 0.939 0.959
Uhl, J.R. et al RT-PCR -LC 1.00 0.969
Uhl, J.R. et al RT-PCR -Compact 0.925 0.991
Goodrich, J.S. et al RT-PCR- LC Roche 1.00 0.959
Goodrich, J.S. et al RT-PCR -BD Cepheid 0.925 0.925
Have implemented the above and got many laudits for an up-market graphic!
However just read Breadhseet et al in Pharmaceutical Statistics (A course in constructing effective displays of data for pharmaceutical research personnel), and they call this a "grable", coz it is more table than graph!