Third TRY for Radon Post!!!

[Fritz Seiler <faseiler@nmia.com>]

Hi all,

We finally found the culprit, responsible for the cut-off!  The line
where the message was cut off started with the not so innocent
word: 'From'!  Some servers such as one on the link to
RADSAFE will interpret that as the start of a new message,
leading to a lot of nonsense.  Others such as the one linking up
with RISKANAL do better, there was no trouble there.  I am
sorry for the two failed tries clogging up your mailbox.  So we
will try a third time, and this time we start that line with the
hopefully more innocent word 'Using'! Here it goes:

Comment # 2 by Fritz Seiler and Joe Alvarez on the
"De Minimis, Radon and Societal Risk" Thread.

The third comment which was announced in our last post, concerns
the risk of exposures to radon and its progeny.  Here, three main
points need to be made:

One, as long as the data base consists of nearly all of the U.S.
population, as in Bernie Cohen's data set, any individual in a sub-
group of N persons, exposed to a particular radon daughter
concentration, shares the same risk m / N, where m is the number
of expected lung cancer fatalities estimated as an ‘a priori' risk in
that sub-population.  Again the average number of days of lost
life expectancy can be calculated and is equivalent to the estimated
number of m fatalities.  A division by the number N cannot change
that equivalence.  Evaluating the ‘a posteriori' risk is like 'Monday
Morning Quarterbacking', it shows quite clearly what the best
estimate should have been, by inspecting what the actual loss of
life and life expectancy was.  But then, we talked about the 'a
priori' and 'a posteriori' problem in the first post.  Here, we have
to make another important point: Although the interpretation of
Bernie's data has been savagely attacked, nobody has seriously
questioned the validity of the experimental data, whatever they
are allowed to mean.

Two, with regard to these much maligned data, we have made
a rather interesting discovery (Seiler, F.A., and J.L. Alvarez,
"Is the ‘Ecological Fallacy' A Fallacy?"; to be submitted shortly).
It will influence risk calculations and thus the cost-risk-benefit
calculations discussed on this thread in a rather decisive way.
Using some ‘ab ovo' considerations of the data needs for a risk
assessment of radon daughters, we were able to derive three
conclusions: 1) The data which are really needed for a radon
risk assessment are the raw Cohen data, uncorrected for
smoking or any other confounding effect!  2) The ‘true' shape
of the dose-effect relationship is almost irrelevant for risk
calculations!  For a goal-oriented activity like risk assessment,
the ‘true' shape is mainly of academic interest.  The more
confounding factors there are, the less importance can be
attached to the ‘true' dose-effect relationship.  3) The term
‘Ecological Fallacy' applies only to the ‘true' dose-effect
relationship, not to the dose-effect relationship which is
needed for a risk assessment.

These counter-intuitive results become understandable only
if we carefully consider what we really need to know in
order to estimate the number of health effects in a prediction
population.  Also, this result holds true quite generally for
ecological studies of any health effect due to an exposure
to any toxic agent in a sufficiently large test population,
results which are then applied to a sufficiently large
prediction population.

As the Cohen data are thus the data to be used in a radon
risk assessment, we are dealing with a U-shaped dose-effect
relationship with some interesting consequences, particularly
in a cost-risk-benefit evaluation.  First it leads to decreasing
relative risks as a function of increasing exposure, followed
- after crossing the point where the relative risk is 1 - by
beneficial effects, which then level off and diminish.  Then the
relative risk begins to rise again and after passing through a
relative risk of 1 (Zero Effects Point!), it begins to yield
actually harmful effects!  In this context, we also discuss in
our paper the derivation of average risks and of the risk to
an individual in this population, all derived from ecological data.
The latter is a big No-No according to conventional wisdom
but - as so often - the conventional wisdom is wrong, at least
for some well defined cases.

Three, in the lone figure in our ‘Ecological Fallacy' paper, we
have assembled all measurements done at radon concentrations
below 500 Bq m-3.  If the Cohen data are ignored for the time
being, we have the BEIR VI linear fit to the uranium miner data
and the data actually measured at these exposures.  There can
be no question that the fitted risks in the lower half of the
concentration interval depicted are dramatically smaller than
their real errors or the errors of the real data points measured
at these concentrations. This is a classical example of risks way
below significance, and thus of ‘de  minimis' risks.  In this
situation, the typical comment of many risk assessors is: "But
we know that the risk is there!"  Such comment are mostly
made by persons who cannot or - for various reasons - will
not distinguish between risk assessment and risk management.
Our usual answer is solely based on science: "You do not
know anything about what is there!  It could be almost anything
within the errors, and you would not know the difference!"
A corollary to that statement is: "..and it is your duty as a
responsible ethical scientist to point out that fact!"

Now, finally, we have an experimental example to back up our
contention: We have shown that Cohen's curvilinear dose-effect
relationship, which has comparatively tiny errors, is the only
data that should be used in a risk assessment. And they flatly
contradict the models such as the BEIR VI extrapolation which
leads to relative risks which are always larger than one.  This is
mainly due to the rationale that "we know that the risk is down
there!"  This attitude is a prime example of an unscientific
procedure leading to a totally wrong conclusion.  As the Cohen
data demonstrate, there exists not a linear but a clearly nonlinear
dose-effect relationship, and there is a statistically highly
significant beneficial or hormetic effect, and all of that is buried
in the huge errors of the "acceptable" data.  So much then for
the value of statements of the type "We can't see it, but we
know it is there!"

Our paper mentioned here is complete and all we need to do
now is to decide on a journal to which we will submit it.
Given the ‘politically incorrect' results in the paper, it is not
an easy choice!  We are also trying to put it on a webpage
which is under re-construction.  Meanwhile we can send it
out as a preprint by e-mail, but only in Wordperfect 6 & 7
(because of the figure!).


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

Fritz A. Seiler, Ph.D.
Principal
Sigma Five Associates
P.O. Box 14006
Albuquerque, NM 87191-4006
Tel.    505-323-7848
Fax.    505-293-3911
e-mail: faseiler@nmia.com

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

Joe Alvarez
Auxier & Associates, Inc
10317 Technology Dr, Suite 1
Knoxville, TN 37932
Email: jalvarez@auxier.com
Tel: 423-675-3669
FAX: 423-675-3677

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


************************************************************************
The RADSAFE Frequently Asked Questions list, archives and subscription
information can be accessed at http://www.ehs.uiuc.edu/~rad/radsafe.html