[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Exchanges with Lubin and with Field et al in HEALTH PHYSICS
Some Comments by Fritz Seiler on the Continuing Controversy
about the Meaning of Bernhard L. Cohen's Data on Lung Cancer
vs. Radon Concentration.
Joe Alvarez is incommunicado on vacation, so I will comment
on this ‘oldie but goodie' issue by myself.
I have just received the September issue of the journal Health
Physics and have looked at the two Letters to the Editor by Bill
Field et al. and by Jay Lubin (pp. 328-329; and 330-331, resp.).
Also today (Friday), Bernie sent out a message on the RADSAFE
mailing list. It is attached here for those readers of RISKANAL
who did not get to see it. The letter speaks for itself, and there is
no need for me to comment, particularly because others have
rather ably done so already.
As for the two letters, the first thing to note is that while Bernie
Cohen's data are discussed, none of his data are shown. There are
data plots and fits shown in the figures of Lubin's letter, but they
show all the other data with their errors, but not Bernie's data set,
just his fitted line. The second thing to note is that nobody has ever
successfully attempted to question the quality of Bernie's data, just
their applicability to lung cancer risk assessment has been challenged.
These peculiar aspects have led Joe Alvarez and me to do a plot
of ALL the data available on the same, somewhat busy, graph, and
then we could see why Bernie's data are never shown together with
the other measurements: The other data have experimental errors
which are huge compared to those of Bernie's data set!
Now, using the basic requirements of the Scientific Method......
No! I am not going to start with another foray about the Scientific
Method because we have already done so several times on this
list and in a number of published papers.
Also, some of you may remember our recent paper "Is the
‘Ecological Fallacy' a Fallacy?" that we sent to the mailing list on
the subject of radon and lung cancer, and the graph that we showed.
We have since presented these arguments during the Health Physics
Society Meeting in Philadelphia at the beginning of July, and at present
the paper is being reviewed for the journal "Human and Ecological
Risk Assessment".
For those who have not yet read the "Fallacy" paper but would
like to, I will be glad to send you a preprint of the paper via e-mail.
It is formatted in Wordperfect 7/ Word 97, and the figure is done in
Excel 97. If the figure leads to problems with your software, let me
know, send me your fax number, and I can send it to you that way.
If all else fails, snail mail can do the trick. This striking figure is most
important in the context of the basic scientific question posed in our
paper: "What do all these data really tell us?"
Both the paper and figure show that the arguments of the "Linear
Enthusiasts" are fallacious, simply because they are irrelevant for a
risk assessment. All their arguments do not concern the data needed
for a risk analysis; they do concern the data that would be needed if
you wanted to show that the linear model might be correct after all.
Our discussion in the paper shows that this attitude is basically flawed,
and that it is the raw, uncorrected measurements by Bernie Cohen
which are the data actually needed for a risk assessment. Thus we
show that Bernie's data are the most important tool in the context of
radon and lung cancer, regardless of the shape of the underlying dose-
effect relationship. We do agree with Bernie though, that the "true"
underlying dose-effect relationship for the entire population is probably
very close to his data. However, in order to make a risk prediction,
no corrections of the raw data are called for.
The form of our argument may be a bit unfamiliar to some, as it is
a 'thought experiment' (in German: "ein Gedanken Experiment"), so
beloved of physicists. It is a process in which every step and every
aspect of an experiment is carefully discussed in a mental ‘dry run'. This
logical device is an extremely useful tool for clarifying one's ideas. Many
theorists, among them Albert Einstein, Niels Bohr and their colleagues,
used it in their well known discussions of the difficult but basic issues in
Quantum Mechanics and in many other disciplines (e.g. Schroedinger's
Cat, Hidden Variables, Maxwell's Demon, etc.). It takes time to think
things carefully through in this manner, but the results are well worth it
and are quite often astounding.
In a nut shell, our argument is quite simple: Cohen's measurements
show how the U.S. population (the test population) with all their idio-
syncrasies and peculiarities responds to a graded set of radon
exposures. Any substantial subset of that population (the prediction
population, such as the population of Colorado, for instance) or even
a bona fide individual in that subset, will react the same way, once we
make the crucial assumption which is always needed, namely that the
responses of the two populations are identical in all important respects
(Equivalent Population Assumption). If this assumption is not made,
NO RISK PREDICTION IS POSSIBLE. Conversely, in order to
make a risk prediction, you MUST MAKE THIS ASSUMPTION so
that you are able to use the data available. Once you think of the
problem in this manner, the conclusion seems inescapable. If it is not,
and you want to find out, please contact me to send you our paper.
Fritz Seiler
****************************************************
Copy of Bernie Cohen's letter, sent to RADSAFE,
Friday, 8| 20| 1999, 8:24 am:
Early in 1998, I accidently found out that papers were to be
published in Health Physics by Lubin and by Field et al critiquing my
study of the radon vs lung cancer relationship among U.S. Counties as a
test of the linear-no threshold theory, originally published in the Feb
1995 issue of Health Physics. I asked the Editor if I could reply to these
critiques, and he assented under the condition that my replies be received
within 2 weeks and Lubin and Field et al would be given a Rejoinder in the
same issue. All of these were published in the July 1998 issue.
I felt that the Rejoinders raised new issues and submitted a
response to these. The Editor, after some urging, agreed to publish my
response with the understanding that Lubin and Field et al would be given
an opportunity to reply to that response, and that there would be no
further publications on the exchange. My response was published in the
April 1999 issue, and their replies recently appeared in the September
1999 issue. I feel that their replies were highly misleading and am
frustrated by not having an opportunity to explain why.
As an outlet for my frustration, I would be happy to explain these
matters in brief notes on Radsafe if people are interested, or if people
have questions they might transmit thru Radsafe, I would be delighted to
respond. Any other suggestions would be most welcome.
Bernard L. Cohen
***************************
Fritz A. Seiler, Ph.D.
Principal
Sigma Five Associates
P.O. Box 1709
Los Lunas, NM 87031
Tel. 505-866-5193
Fax. 505-866-5197
e-mail: faseiler@nmia.com
***************************
************************************************************************
The RADSAFE Frequently Asked Questions list, archives and subscription
information can be accessed at http://www.ehs.uiuc.edu/~rad/radsafe.html