Re: Confounders and Coincidences

[BERNARD L COHEN <blc+@PITT.EDU>]

On Mon, 5 May 2003, Otto G. Raabe wrote:

>

> I agree with you for the most part, but unknown confounders are not the

> problem in epidemiological studies that they are in ecological studies. I

> cannot agree with your blanket statement that "...all epidemiologic studies

> up to and including those relating smoking to

> lung cancer could be discounted on the basis of possible unknown

> confounding factors...." The key difference is that the exposure of each

> individual in an epidemiological study is evaluated and the statistical

> probability of erroneous results is also evaluated. In an ecological study

> we have no way of knowing the dose to any particular person, nor do we know

> the chance that the results are spurious even if they are not.



	--Your argument is correct if the purpose is to determine the risk

to an individual. As I have emphasized many times, that is not the goal of

my studies. They are designed to test the linear-no threshold theory, LNT.

The procedure is to use LNT to derive a relationship between ecological

variables --cancer rates, smoking prevalences, and average radon exposure

in counties. I then use data on these to test that relationship. This

method --testing a derived relationship-- is widely used in other areas of

science, but not in epidemiology.

	It is a "religious conviction" among many epidemiologists, and a

justifiable one, to not trust an ecological study to determine risks to an

individual. The problem has been that they extend this to all ecological

studies, including those not designed to determine risk to individuals.

	These matters are discussed at some length in Item #7 on my web

site.



************************************************************************

You are currently subscribed to the Radsafe mailing list. To unsubscribe,

send an e-mail to Majordomo@list.vanderbilt.edu  Put the text "unsubscribe

radsafe" (no quote marks) in the body of the e-mail, with no subject line.

You can view the Radsafe archives at http://www.vanderbilt.edu/radsafe/