Are smokers more likely to be obese? Is stroke more common in people with hypertension?  Are old people more likely to have diabetes than young people?

The prospect of trawling through the scientific literature to get the answers to these questions isn’t terribly appealing – reams of tables and risk ratios aren’t helpful if you just want the information at a glance.   Technology and health care company GE have developed a new way to present complex epidemiology data in graphic form.

Taking a New Look at Health allows you to compare various demographics, risk factors and diseases in a random sample of 100,000 patient records from GE’s proprietary database.  Once you’ve picked the two variables you want to look at, legions of tiny men shoot across the graphic and align to show what proportion of people with variable  also have variable y.

In this example, I have looked at smoking and BMI to see whether smokers are more likely to be obese than are non-smokers:

Given that the figures are presented as an image rather than a table of numbers, it’s much easier to get a handle on the proportions and what they really mean.  I can now see at a glance that actually more non-smokers and ex-smokers than smokers are obese (28%, 29% and 27%, respectively); that is, smokers are less likely to be obese.

Here are a couple more examples:

Is stroke more common in people with hypertension?

As well as showing that people with hypertension are considerably more likely to experience a stroke than those without hypertension (5% vs 1%), this particular graphic also allows us to see roughly what the incidence of hypertension is in the GE sample – not huge judging by the slim column on the right hand side.

Are old people more likely to have diabetes than young people?

Yup, look how many little orange people there are in the 65-74 column and the 75+ column!

Why not have a play with the visualisation yourself? Is the link between heart disease and hypertension what you expect? And what was the male:female ratio in this sample anyway? If you like epidemiology you’ll probably have a lot of fun!