The paper by Willis et al [1] contains, amongst other things, maps at a county-by-county level showing the distribution of the incidence and prevalence of Parkinson's among Medicare beneficiaries.

Incidence: http://www.ncbi.nlm.nih.gov/pmc/arti...395/figure/F2/

Prevalence: http://www.ncbi.nlm.nih.gov/pmc/arti...395/figure/F1/

There is a way to use the difference between these two maps to extract more information. Let me explain how this can be done.

With a disease such as Parkinson's, where there are few, if any, people cured, we would expect the prevalence to be directly related to the incidence. For instance, if the average PwP lives 10 years after diagnosis, the prevalence will be approximately 10 times the incidence.

As far as the maps are concerned, we would expect high incidence counties to be high prevalence counties and low incidence counties to be low prevalence counties. There will always be statistical "noise", so a certain variation is to be expected. There are five levels on the maps (dark green to red). I suggest that any difference, up or down, of at three levels is worth looking at.

The annual incidence figures are for 2002-2005. People don't usually know when they "caught" PD, but most feel that the onset was several years before the first presentation and even longer before the first diagnosis. So, to be a statistic in 2002-2005 we need to look at events in the period from, say, 1995-2005.

What I'd like all of you with knowlege of the US to do is this:
- open both maps, look at areas you are familiar with;
- identify any counties with at least a three step difference (up or down);
- find the population of the county, prioritize the counties with populations above 100000;
- see if you can find any events that may explain the change.

The sort of thing I have in mind are:
- a mine opens or closes;
- fluoride is added or taken away from tap water;
- a pollution event occurs, possibly up-wind;
- a new road is built;
- major lay-offs occur.

Why am I asking you in the US to do this? Simply because you have the best statistics. I don't know of any other paper with the detail of the Willis paper. Also, your county system seems to have just the right granularity for this problem.

Happy hunting. We may just get lucky.

[1] "Geographic and Ethnic Variation in Parkinson Disease: A Population-Based Study of US Medicare Beneficiaries"
Allison Wright Willis, Bradley A. Evanoff, Min Lian, Susan R. Criswell, and Brad A. Racette
Neuroepidemiology. 2010 April; 34(3); 143-151.
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2865395/

John