[ RadSafe ] Radon: POWERFULLY associated with LESS lung cancer by B.Cohen

Chris Hofmeyr chris.hofmeyr at webmail.co.za
Sun Jun 19 05:10:06 CDT 2011

Dear Howard, Radsafers,
I devised a sensitive way to assess Cohen's data which indicated very
convincingly that domestic radon concentration, particularly at the high end,
had no effect - either positive or negative - on lung cancer mortality. This
fits neatly with Raabe's findings about protracted low-intensity exposure - we
simply do not survive long enough to see the effect, even at the high end
average levels.

I must emphasise that the above findings pertain to the upper ranges of average
radon concentration (Cohen's data) and I therefore cannot make any pronouncement
on the lower ranges.
However, Howard, you yourself mentioned that the 'apparent' negative
correlation is similarly strong with or without smoking correction. Exactly
because there happens to be a negative correlation between smoking prevalence
and radon concentration (for his purposes Cohen would certainly have preferred
no correlation), the smoking corrected LC data should, in my estimation, show a
less strong negative correlation?  To me that is a strong indication that the
smoking correction is problematic, but it definitely does not preclude the
possibility of a negative correlation per se.

Chris Hofmeyr
chris.hofmeyr at webmail.co.za

On Thu, 16 Jun 2011 08:29:55 -0700 Howard <howard.long at comcast.net> wrote

> Chris, your "APPARENT inverse relationship between lung cancer mortality and
> average radon concentrations" (emphasis added), disqualifies you as a serious
> critic of this most powerful epidemiologic study.
> I and many others have tried to torture out of Prof. Cohen (at meetings of
> Doctors for Disaster Preparedness and here, on line) a spurious "garbage in"
> explanation. Analyses assuming a hundred potential social skews results in
> the same: POWERFUL association in all stable USA
> county populations, with and without smoking and 100 other potential skewers.
> I just took out of my wallet a cc of Cohen's graphs with bars of 95%
> significance used as points!
> The "corrected for smoking" graph seems identical to the total graph, bpth
> showing
> mortality just 2/3 at the 4 Bq/meter-cubed level touted as dangerous, as at
> 1/4 that exposure.
> This is evidence for hormesis, not "apparent", but as definite as
> epidemiology can give.
> LNT seems disproven by this (all Cohen will claim).
> Politics should follow science, not vice versa (I like this use of the word
> "vice").
> Howard Long MD MPH Family Doctor and Epidemiologist
> On Jun 16, 2011, at 6:22 AM, "Chris Hofmeyr" <chris.hofmeyr at webmail.co.za>
> wrote:
> > Dear Proff Raabe and Cohen, Radsafers,
> >
> > Prof Cohen's US-wide county-based radon studies caused quite a stir since
> > 1995,
> > mainly because of the apparent inverse relationship between lung cancer
> > mortality and average radon concentrations. However, Cohen's main
> > conclusion
> > was that the result was at variance with the linear-no threshold (LNT)
> > model.
> > My own analysis of the data sets for two periods kindly provided by Prof
> > Cohen
> > showed adamantly that there was in fact NO indication of a dependence on
> > average radon concentration up to the maximum recorded. Such a finding
> > would
> > concur with Raabe's model of protracted low-intensity exposure (HPJ, July
> > 2011).  Human lifespan is too short to show discernible lung cancer
> > induction
> > from domestic radon concentrations in the USA.
> >
> > However, uncertainty remains with respect to the question of the apparent
> > inverse relationship in Cohen's data between lung cancer mortality rates
> > and
> > average county radon concentration, which some people wanted to interpret
> > as
> > proof of hormesis. Significantly, in a 2006 paper on cancer risk from low
> > level
> > radiation, Cohen did not cite his own radon data in support of hormesis or
> > even
> > rejection of LNT.  In Cohen's data average county radon concentration was
> > to
> > some degree anti-correlated with smoking prevalence, thus explaining the
> > inverse relationship in the uncorrected (for smoking) data. The smoking
> > correction to the data was consequently important for the 'real' dependence
> > curve. Unfortunately it was not possible to analyse the correctness of the
> > smoking correction, nor associated uncertainties, but I became convinced
> > that
> > Cohen was not able to correct the data adequately for smoking, possibly
> > resulting in the persistent apparent inverse dependence of the 'corrected'
> > data.
> >
> > A stark illustration of the overwhelming importance of extraneous factors -
> > most probably smoking - is to compare the female lung cancer data sets
> > 1970-79
> > with 1979-94. There was a mortality increase of almost a factor of 2,
> > whereas
> > male lung cancer declined very slightly between the two data sets (during
> > 1970-79 it was about a factor of 5 higher than the female figure).
> > Regards.
> > Chris Hofmeyr
> > chris.hofmeyr at webmail.co.za
> >
> > ____________________________________________________________
> > South Africas premier free email service - www.webmail.co.za

South Africas premier free email service - www.webmail.co.za 

For super low premiums, click here http://www.dialdirect.co.za/?vdn=15828

More information about the RadSafe mailing list