Re: misc

[Bernard L Cohen <blc+@pitt.edu>]

On Sun, 11 Feb 2001, Ted Rockwell wrote:

>  Cohen, as I understand it, is disagreeing
> with Seiler only in saying that he has not proved a pure relationship
> between radon levels and lung cancer, only a practical relationship between
> people living in homes with radon and the resulting risk of cancer.  

	--My study shows that average radiation dose to groups of people
(populations of counties) has a certain relationship to the lung cancer
rate in those groups. In general, that cannot be interpreted as meaning
that this relationship appllies when considering the relationship between
the radiation exposure to an individual person and his
lung cancer risk, even leaving aside issues of confounding factors. To
assume that they are the same is an example of the "ecological fallacy".
	For example, if there were a sharp threshold at 50 pCi/L, the
cancer rate in a county would depend on the fraction of the population
exposed to levels above 50 pCi/L, which is not necessarily related to the
average radon level in the county, which might be 2 pCi/L.
	However, if the Linear-no threshold theory is valid (as opposed to
the threshold assumed above), the two relationships should be identical.
For example, we are all familiar with the idea that person-rem determines
the number of deaths - after dividing by the population, person-rem is the
average dose and number of deaths gives the cancer mortality rate. This is
a direct consequence of LNT.
	Thus I treat my study as a test of LNT. If LNT is correct, my data
should show the same linear increase with average dose as is predicted for
risk to individuals. Since it most definitely does not show this linear
increase, I conclude that LNT fails very badly, and this conclusion is not
affected by the "ecological fallacy". Avoiding the "ecological fallacy" is
of paramount importance in convincing epidemiologists, so I make this
point paramount. 

************************************************************************
The RADSAFE Frequently Asked Questions list, archives and subscription
information can be accessed at http://www.ehs.uiuc.edu/~rad/radsafe.html