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

[Ted de Castro <tdc@xrayted.com>]

YES - I'd like this same question answered!

LINEAR is indeed a special case - it is the ONE case where distribution
does NOT matter and averaging or accumulating over populations IS
mathematically justified.

It doesn't seem that Cohen is trying to say anything about various
slopes in the causal relationship - just that IF linear an effect should
be seen in the population - but it isn't.

I THINK I understand the Cohen argument clearly - I don't understand the
others.

I also understand that if the repsonse to confounders is NOT linear -
then averaged data will not properly address those other agents.  But I
haven't seen that explaination offered either.

Vincent King wrote:
> 
>      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