Re: I'm get tired of it too, but...

[steve.rima@DOEGJPO.COM (Steven Rima)]

     Vince has expressed, much better than I, what I have been wondering 
     for a long time with regards to this topic. I too am not an 
     epidemiologist, so can someone who is explain something to me? I think 
     that I understand, at least in principle, the "ecological fallacy." 
     However, if one were to study a very large group of people, as Cohen 
     has done, except comparing smoking to lung cancer rates, I am quite 
     certain that one would observe a very clear cause-effect relationship.
     
     Question: If this is true, why would a large ecological study support 
     the number one cause of lung cancer, but completely contradict the 
     (purported) number two cause?
     
     Steven D. Rima, CHP, CSP
     Manager, Health Physics and Industrial Hygiene
     MACTEC-ERS, LLC
     steve.rima@doegjpo.com
     
     Disclaimer: The above is a personal opinion from an operational HP who 
     refuses to believe that 10 millirem each to 100,000 people causes the 
     same number of cancers as 1000 millirem each to 1000 people. :-)


______________________________ Reply Separator _________________________________
Subject: I'm get tired of it too, but...
Author:  Vincent.King@doegjpo.com (Vincent King) at Internet
Date:    3/15/00 11:39 AM


     Dear group,
     
     Can someone make the radon/lung cancer issues here a little simpler 
     so the average HP can understand them?
     
     A linear, no threshold theory is a straight line going through zero. 
      Only the slope is a variable.  And, at the exposure levels we're 
     discussing, it doesn't matter how you spread the exposure out in a 
     group (if risk is truly proportional to dose in a linear manner). 
     Anything else is not LNT.
     
     Since it seems to be a given that the number one cause of lung 
     cancer is smoking, if you can do a reasonable job of adjusting for 
     smoking, you MIGHT be able to see other lung cancer cause/effect 
     relationships if they are 'big' enough. (If you can't properly 
     adjust for smoking, it will overwhelm any other effects you're 
     looking for).
     
     The SECOND leading cause of lung cancer is (according to assertions 
     by the EPA) radon. IF this is true (and if you can properly adjust 
     for smoking as stated above), then looking at average radon 
     exposures and lung cancer rates in similar populations should either 
     show (1) a positive correlation or (2) nothing but random scatter if 
     the effect is too small to see.  It should NOT consistently show you 
     a negative correlation, regardless of 'confounding' influences 
     (which, by the definition at the top of this paragraph, must be 
     smaller than the number two cause of lung cancer, and which would 
     have no reason to 'join forces' and confound in the same direction.)
     
     Turning the last sentence around, if you DO see a negative 
     correlation between radon exposures and lung cancer rates, you can 
     say either (1) the negative correlation is due to the radon 
     exposure (which would be suggestive of hormesis) or (2) the 
     negative correlation is due to something other than radon (which 
     implies that radon exposure should be much further down the list 
     than the 'Number two cause of lung cancer'. Notice that this ALSO 
     implies that the currently accepted LNT regarding radon is wrong, 
     if it can be overwhelmed so easily by 'something else'). 
     
     If you misinterpret confounding factors, you shouldn't get a 
     negative correlation - you should get 'noise'.  For example, I 
     suppose you can take issue with whether average radon 
     concentrations in a county are truly representative of average lung 
     exposures, but how likely is it that average radon concentrations 
     in a county are REVERSED from average individual exposures?
     
     What Dr. Cohen seems to be saying (and please correct me if I am 
     wrong, Dr. Cohen) is simply that the relationship between county 
     radon concentrations and lung cancer rates is the opposite of what 
     you would expect if the current LNT regarding radon is true. He 
     does not say that the cancer rates are 'caused' by anything. The 
     arguements against this conclusion, while they may also be valid, 
     have been far too obscure for me to follow. 
     
     I do not fancy myself to be an epidemiologist, but I have been 
     exposed to some of the basic concepts, so, IN RELATIVELY SIMPLE 
     TERMS PLEASE, is my understanding right or wrong, and if wrong, why 
     is it wrong?  (And if it is wrong, we'd better look at HP education 
     in this country, because I base my thinking on what I thought the HP 
     education process was trying to impart.)
     
     Thanks,
     
     Vincent King
     vincent.king@doegjpo.com
************************************************************************ 
The RADSAFE Frequently Asked Questions list, archives and subscription 
information can be accessed at http://www.ehs.uiuc.edu/~rad/radsafe.html
************************************************************************
The RADSAFE Frequently Asked Questions list, archives and subscription
information can be accessed at http://www.ehs.uiuc.edu/~rad/radsafe.html