[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Cohen's Observation
On Sat, 14 Jun 2003 epirad@mchsi.com wrote:
> ----------------------------
> We addressed the validity of Cohen's mortality data before. For the SEER
> states, his mortality data had only mediocre agreement with good incidence
> data. See our papers in Health Physics. Cohen used mortality data while
> incidence data is more appropriate. We showed in Iowa (a SEER state) that when
> more valid lung cancer incidence data was used, the inverse association
> disappears. An analyses limited to SEER states
> http://seer.cancer.gov/registries/index.html would be an vast improvement
> over the mortality data that Dr. Cohen uses.
--I used the mortality data for 1970-1979 and for 1979-1994. These
are the only sets of data for which lung cancer mortality rates are
available for anywhere nearly all of the counties. I did not use the SEER
data because it is not available for most counties, because I don't
understand why it is better than mortality data, and because entering it
into my files would require a lot of time and effort and I doubt if the
results would be publishable.
>
> Radon Measurements
>
> Look at a county map of the United States.
> http://www.epa.gov/iaq/radon/zonemap.html Look at the variation in county
> sizes per state. Imagine how poorly a few short-term (2-7 days tests) radon
> tests represent the radon progeny exposure for individuals in any county, let
> alone huge counties as can be seen on the map. In fact, many of the
> measurements were made in basements where people spend limited time.
--My radon data are derived from 3 sources, our own 350,000
measurements, EPA measurements, and measurements sponsored by individual
states. If either of these 3 were used alone, the same results would be
obtained. The correlations between the 3 sets is very good. These things
are discussed in more detail in my 1995 paper in Health Physics, and
elsewhere.
> We showed that if we would have used year long basement measurements in the
> Iowa Study, we would have found a protective effect from radon for lower
> exposures. http://www.aarst.org/news_pdf/2002_IowaU_Follow-up_Study.pdf
> I think most people would agree that short-term screening or long term
> basement measurements do a poor job of predicting radon exposure for an
> individual let alone a county.
--The validity of our measurements is discussed extensively in
published papers that have not been disputed -- references available on
request.
> In a paper we published in American Journal of Public Health. 80(8): 926-30,
> Aug 1990 showed Dr. Cohen's LSC radon detector had very poor precision.
> However, I would hope that this spot check was not reflective of the overall
> performance of this detector.
--We always passed the EPA quality assurance tests without
difficulty. To the best of my knowledge, they are highly respected in the
radon measurement community.
> A better question then the one you posed may be why is there an inverse
> association between Cohen's county radon concentrations and his county
> estimated smoking rates. Smoking is correlated with so many other factors
> that also affect lung cancer incidence so adjustment is needed for the total
> combined effects of smoking related factors.
--Uncertainties in our smoking data are discussed extensively in
our papers, and reviewed in Sec $ of Item #7 on my web site
> Remember, we already showed that Cohen's summary smoking information alone
> does a poor job of explaining the lung cancers within the counties.
--On the contrary, it does an excellent job of inndicating that
the great majority of lung cancers are caused by smoking.
There is
> little surprise that residual confounding exists in his data. The proof of
> that confounding has been pointed out by Drs. Gilbert and Puskin who clearly
> showed that the inverse association is found for other smoking related cancers.
--They claim that the Puskin observation would be resolved by a
stronger negative correlation between smoking and radon. My response to
Puskin (on my web site) shows that this is not true. The stronger negative
correlation do a poorer job of explaining the smoking vs lung cancer
correlation.
************************************************************************
You are currently subscribed to the Radsafe mailing list. To unsubscribe,
send an e-mail to Majordomo@list.vanderbilt.edu Put the text "unsubscribe
radsafe" (no quote marks) in the body of the e-mail, with no subject line.
You can view the Radsafe archives at http://www.vanderbilt.edu/radsafe/