James
I have a couple of suggestions for your calculator, which I think is a great communications tool.
But, before I make the suggestions, I would like to compare the visual metaphors used by your calculator (http://bestsciencemedicine.com/chd/calc2.html) and the example given by Richard Hockey (http://robslink.com/SAS/democd5/cardio.htm)
Your graphic shows relative and absolute probabilities of good outcomes, bad outcomes, bad outcomes that are potentially avoidable, and bad outcomes that a chosen intervention would prevent.
Now, one of these images cannot show how risk changes with risk factor or intervention. To see changes you have to compare a new image with a remembered previous image. And, you have to scroll up to make changes, and then scroll down to see the new image.
With the visual metophor in Richard's example, changes in shading show how risk changes with systolic blood pressure and total cholesterol. Changes in risk with sex, age, and smoking are shown by juxtaposing the fundamental BP v TC images. This style gives a good impression of changes in relative risk, but poor impression of benefits of modifying risk factors, and no impression of absolute risks and benefits (although there are numbers in each of the cells).
So, my first suggestion for James is to have two images next to each other (side-by-side if there is room).
And on the right-hand side (apologies for any discrimination perceived by Lefties), have a neat vertical list of risk factors and interventions.
With this layout it would be hard not to see the sensitivity of risks to changes in risk factors and interventions.
My second suggestion is to show confidence intervals for the estimates. I realize that any method of calculating them would itself have wide confidence intervals. But, at least it would help get the message across that the point estimates are smudgy points.
You could visualize this by having two graphics side-by-side, one showing the lower bound and the other the upper.
Michael
|