Thursday, 24 March 2011

The best places to hide trees is in forests.



  1. Fredrik,
    Hi. I got your question about this from my blog. Can you clarify what your hypothesis is? Basic epidemiology in a US context pretty much always involves the observation that there is that area of higher cancer rates, which is plausibly attributed to higher rates of smoking and unhealthy dietary choices.

    As I think you are noting, it lines up pretty well with what might be described as the green bits of the country that were most hospitable to settle and grow stuff in the 18th and 19th centuries. There is probably something to this more than communication, since that area is still disproportionately populated by particular social groups (rural blacks, Appalachians) who have less healthy than average behaviors. As for the lights, these are the densest non-urban areas in the country (with a few urban bits -- Houston, New Orleans, Atlanta, etc.) and the lights reflect that -- lots of light, but not huge pools like New York, Chicago, or Minneapolis/Saint Paul (I have always marveled at how bright the latter, my former home, centered left-to-right and near the top, shows up on that light map -- it is not nearly as big a city as the other ones that are that bright

  2. Hi Carl, thanks for your reply. I don't really know that I have an hypothosis as such but I am trying to see if there is a relationship between preciptitaion an lung cancer over time and space. The above map is an elevetion map
    but there also seems to be a corrolation between precipitation and lung cancer. Of course, rain will cause green bits on maps! But thinking about your analogy of drunk drivers and iatrogenic death in the context of causation, I think it unlikely that unhealthy people seek out green pastures more than healthy people but that does not mean to say that that they are not more likely
    to get ill when they get there! My long term plan is to write a visualisation tool so that I can see what happens over time (I am a software engineer) using data from the CDC (Wonder database) and environmental data. Sadly, it's not the writing of the code that is the problem, it's getting the numbers right that takes up the time! I was really interested in asking you because if you could have told me straght away what is going on then I could save myself some time! I am sure such tools already exist but I doubt if I could get my hands on them!
    Many Thanks and thanks for your blog!