At the SAS Global Forum 2013 in San Francisco, I presented a paper on making Patient Profile Graphs using SAS. You can follow the link to see the graphs presented in this paper.
In that paper, I presented ways to create AE, CM, VS and Lab graphs. I discussed aligning the x-axes of the two independent AE and CM graphs, and also carrying reference lines for each AE start event into the CM graph. Here are the AE and CM graphs from the paper. Click on the graph for a high resolution view.
Note these features:
- These are two separate graphs.
- The zero values on the x-axis of both graphs are correctly aligned.
- Reference lines for each AE start event are carried into the CM graphs.
After the presentation, and also recently at the Minnesota SAS users' group meeting, users stopped by to discuss these graphs, and ways to further improve the graph to help the PI clearly view each AE event and associate it with the CM prescriptions exactly.
This discussion led to the idea of plotting both AE and CM events together in one graph. Here is the first installment of this graph. Click on the graph for high resolution view.
Note the following features of this graph:
- Both the AE and the CM events are plotted in the same graph.
- The events are sorted by start date, with earliest event on the left.
- Each subsequent event (by date) is plotted in ascending fashion.
- AE's are classified by severity, using a distinct color for the each.
- CMs are colored in light blue to distinguish these from the AE.
- No need for any reference lines.
In my opinion, this graph is clearly superior to the previous version. It follows the principles of effective graphics, were related information is presented together. I would like to reduce clutter, and highlight the appropriate items.
Do you work with such graphs in your work? Would you find this useful? Your suggestions would be very valuable.
Similar combined graphs for VS and LAB would also be useful.
I have not included the code yet, but this graph is created using SAS 9.3. I would like to include the suggestions from readers into this graph.