The other day, we did a piece entitled, Some Advice For CDC And FDA. In the article we basically raised two points. First, that FDA and CDC need to examine their use of language. The constant announcement of “new” or “additional” cases is deceptive because consumers interpret that to mean somebody got sick today when, in fact, the “new” cases can be a month old and they are “new” only in the sense that the CDC didn’t know about them before.
We actually think FDA and CDC did much better on this issue in the last two press conferences. They clearly tried to emphasize that these were mostly old cases. So kudos to them for trying to do a better job.
We also critiqued the CDC map:
The problem with this map is that, like an Election Day map on TV, it is one of two colors — in this case, white or green.
For Election Day, this makes sense since the state is either Democratic or Republican.
But for a map of illness, this model — on or off, white or green — makes no sense. If you look at this map, it looks like a broadly dispersed problem. However, if you study the numbers, you see how deceptive that is.
New York may be green — but only with two people ill. Vermont, just one; Connecticut, only two; Florida, only one, etc. By way of contrast, Texas and New Mexico have 68 each!
We are not there yet, but we thought we could try a few pictorial alternatives. Here is the CDC Map and caption:
Since April, 383 persons infected with Salmonella Saintpaul with the same genetic fingerprint have been identified in 30 states and the District of Columbia. These were identified because clinical laboratories in all states send Salmonella strains from ill persons to their State public health laboratory for characterization. The marked increase in reported ill persons is not primarily due to a large number of new infections. The number of reported ill persons increased markedly mainly because some states improved surveillance for Salmonella in response to this outbreak and because laboratory identification of many previously submitted strains was completed. The number of ill persons identified in each state is as follows: Arkansas (2 persons), Arizona (26), California (8), Colorado (2), Connecticut (2), Florida (1), Georgia (8), Idaho (3), Illinois (34), Indiana (8), Kansas (9), Kentucky (1), Maryland (10), Michigan (3), Missouri (9), New Hampshire (1), New Mexico (70), New York (9), North Carolina (1), Ohio (3), Oklahoma (5), Oregon (4), Pennsylvania (2), Tennessee (4), Texas (131), Utah (2), Virginia (17), Vermont (1), Washington (1), Wisconsin (5), and the District of Columbia (1).
Now here are some alternative map concepts we have been working with:
Our first thought was to use pins to represent each illness. CDC hasn’t told us the counties, so we just stuck the pins in the map in the state where the illness occurred. With more information, the map could show more detailed information. In other words, if all the illnesses in Illinois are limited to Chicago, that is a very different outbreak than one with illnesses spread throughout the state.
The map definitely has some advantages over the CDC one. A state such as Florida shows up bright green, just like Texas on the CDC map. Here you instantly see that there was one isolated Floridian who got sick from eating tomatoes — implying he or she was probably a tourist who visited the outbreak area.
On the other hand, we still aren’t thrilled because the map is still deceptive. A state like Texas looks like it is covered in victims from wall to wall — but, in actuality, the 131 pins could fit easily in one neighborhood or apartment building.
There is also no sense of proportionality between state population and the outbreak. New Mexico, with a small population, is hit much harder than Texas on a proportional basis, even though Texas has more victims.
This is much more accurate but isn’t very graphically appealing.
This gives some good information graphically, but you really don’t want to have to look at a legend to understand the map.
This is basically the CDC’s map with the states color-coded to match cohorts of the outbreak. It is not perfect. Like the CDC, you have to look at the legend to know what it means, but it does give a lot more information at a glance than the CDC map.
We’ll keep working on developing options… if anyone has some ideas for a map that is accurate, information-rich and won’t cause panic when none is required, we welcome ideas here.
Many thanks to Matt Winans for help with the graphics.
