Re: Cohen's Ecologic Studies

[epirad@mchsi.com]

Dr. Cohen,



I am not sure who you have been speaking with, perhaps 

they could take part in this discussion?  Nonetheless, 

epidemiology is a bit more involved than determining 

dose response relationships for individuals (for 

examples, see a good basic text on epidemiology).  



Epidemiology also involves the use of grouped, aggregate 

or ecologic data similar to your studies.  An example of 

this would be an ecologic study.  As an epidemiologist, 

I also perform these (for an example) – see:



http://www.junkscience.com/news/radon-breast-cancer2.html



Health Phys 1994 Mar;66(3):263-9, 	

Residential radon exposure and lung cancer: evidence of 

an urban factor in Iowa. Neuberger JS, Lynch CF, Kross 

BC, Field RW, Woolson RF.



Ecologic studies do not and can not examine the dose 

response association for individuals.  To assume the 

dose response relationship for grouped data is equal to 

the dose response relationship for individual data, 

would be opening yourself up to the ecologic fallacy.  

Epidemiologists use ecologic studies all the time to 

formulate hypotheses, not to test them.  To state that I 

do not understand your studies because they are ecologic 

in nature is ludicrous.  The reason I question your 

studies is because I DO understand the strengths and 

limitations of ecologic studies.  I understand that an 

ecologic study all of a sudden does not loose is 

limitations, because you say you are testing the LNTT 

with the study.  Testing the LNTT does not eliminate the 

potential for ecologic fallacy.  Especially, when in 

your derivation of your LNTT formula, which is far less 

rigorous then either the BEIR IV or VI LNTT formula, you 

are not able to determine the average dose. You would 

need to know the average dose to determine if it equals 

the average risk.  



The BEIR IV equation  implies that there is a linear no-

threshold effect of WLM (W) on the risk of lung cancer.  

On the other hand, your formula implies that the risk is 

linear in the living area radon level (r).  These two 

implications are not the same.  They coincide if and 

only if you adequately control for the determinants of 

W.  r would be equivalent to WLM only if it was the true 

time-weighted average of radon levels in all areas of 

the home, outdoors, and at other locations where persons 

spent their time over their lifetime.  Only then could W 

and r be used interchangeably in your equation.  In 

reality, r is probably a crude estimator of W.  You use 

short-term radon measurements to estimate r only in 

either the basement or living area of the home.  



In addition, you can not account for either smoking 

intensity or duration in your LNTT equation, but BEIR 

can.  Smoking has many co-correlated factors such as 

socioeconomic status; poorer health care, increased 

occupational exposure to lung carcinogens, increased 

residential exposure to asbestos, poorer diet, etc.  You 

may try to use various aggregate factors to account for 

this confounding, but you try to treat each one at a 

time and fail to consider cross level bias or additive 

effects of each factor.



It is not surprising that you find an inverse 

association for other smoking related cancer unrelated 

to residential radon exposure.  



Regards, Bill Field

> 	In communicating with other epidemiologists, I believe I may

> better understand your problems in understanding my papers. The goal of

> epidemiology is to determine the dose-response relationship for

> individuals. For this one must have a variety of information on those

> individuals, and cannot use ecological data, that is data for groups of

> people, such as averages for counties.

> 	My study is NOT an epidemiological study in that definition, but

> rather a test of the BEIR-IV LNT formula for risk vs dose from radon. I

> use that formula to derive mathematically a relation between lung cancer

> mortality rates, m, average radon exposure, r, and smoking prevalence, S.

> This is a relation among ecological variables. If my test is invalid,

> there must be a confounding factor, CF, that confounds the relationship

> between these 3 quantities. In most cases, these CF can be represented by

> ecological data, but I have identified and treated a few cases where

> this is not sufficient.

> 	You and other epidemiologists are disturbed by the fact that I do

> not treat CF on the level of individuals. But there is no logical need for

> considering them unless it can be shown that they lead to confounding of

> the relationship betweer m, r, and S. When you express such objections, I

> ask for an example of how they can affect that relationship.

> 	I will be out of town for the rest of this week so will not be

> reading my e-mail for a few days.

> 

> Bernard L. Cohen

> Physics Dept.

> University of Pittsburgh

> Pittsburgh, PA 15260

> Tel: (412)624-9245

> Fax: (412)624-9163

> e-mail: blc@pitt.edu

> 

> 

> 

> ************************************************************************

> 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/

> 

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

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/