The draft report on DOE worker risks

["Dukelow, James S Jr" <jim.dukelow@pnl.gov>]

Thanks to Michael McNaughton's kind posting of the location of a bootleg copy of
the National Economic Council's (?!) prelimnary draft report  _Occupational
Illness Compensation for Department of Energy Contractor Personnel: Report of
Task Group I_, I have been able to begin looking at the draft report, rather
than just reading relatively uninformative media reports about it.  Here are
some preliminary reactions to the preliminary draft, dated 23 December 1999.

Although the most common phrase in the report is something on the order of
"There were no causes of death with rates greater than those seen in the U.S.
general population", media reports would have you believe that former DOE
contractor workers are dropping like flies.

The second sentence of the Executive Summary says, "The DOE is a
government-owned, contractor-operated Federal government agency."  This sentence
is a good predictor of the logical precision of the rest of the report.

There is a certain intellectual incestuousness evident. Of the 27 studies
forming the basis of the draft report, at least 13 are written by researchers
associated with ORISE or the University of North Carolina School of Public
Health or both.  Of the remaining 14 studies, 4 were performed by some
combination of Mancuso, Kneale, and Stewart.  On the other hand, the draft
report notes, "Although data from other radiation exposed cohorts, most notably
shipyard workers, were presented to the panel, the results of these studies are
not included in the report due primarily to the limited focus and charge of this
panel."

The specific cancers and other causes of death for which "positive trends" are
reported dance around from study to study.  There is no consistent association
of a type of cancer or cause of death with radiation exposure over the entire
set of studies.

One analysis that should be done on each of the data sets, but which, as far as
I can tell, has not been done on any of them, is a Chi-square goodness of fit
test of the entire data set to the null hypothesis that there is no biological
effect operating on the data.  This test could be done using either general
population, local population, or worker population data to establish the
expected frequencies for each analysis category.  If the overall data set is
well-fit to the null hypothesis, then we can be properly cautious about our
interpretation of any "statistically-significant positive results" within the
data set.

It is impossible to tell from the draft report, but the description of some of
the categories for which "positive results" or "positive trends" are reported,
strongly suggests the industrial strength data dredging that is obviously
present in some of the 27 studies the draft report is based upon.  An example is
the report of two "sensitive age-at-exposure groups" in an Oak Ridge population.
They were those exposed under the age of 25 and those exposed over the age of
45.  Three subsequent analyses by the same research team also reported
"sensitive age-at exposure groups", but variously defined as over 58, over 62,
and age 55-65.  Several positive results are for worker subgroups excluding
those workers exposed to plutonium, where there is no positive result listed for
the same group including those exposed to plutonium.  Are we to conclude that
exposure to plutonium is protective?  Another example reports a higher rate of
leukemia among Savannah River hourly workers who were employed before 1955 who
worked from 5 to 15 years.  Again, there was no report of a higher rate for all
hourly workers or hourly workers employed after 1955 or hourly workers employed
less than 5 or more than 15 years.  What is it about all of these conditions
that is protective against leukemia?

Appendix 1 of the draft report provides a laundry list of all of the reported
"statistically significant positive results or trends" in the 27 studies.  The
laundry list does not provide citations to identify in which of the 27 studies
each positive result is found.  No information is provided on some of the
interesting statistically significant negative results and trends in the data or
on the number of overall tests of hypothesis or calculations of SMRs, etc. done
in each of these studies.

Several times, the draft report notes that workers hired in 1944 and 1945 show
statistically significantly higher death rates than the general public for a
number of disease categories.  Finally, after several of these "positive
results", the report notes that "Because these workers were not engaged in
military service during wartime, it is not possible to know whether or not these
conditions reflect occupational exposures or underlying health conditions that
would have excluded them from military service."

The draft report states that, using the BEIR V dose reponse assumptions,  DOE
contractor employees (with an estimated average exposure of  1mSv per year)
would experience an excess of 288 cancers per 100,000 workers (95% CI = 215 to
546) over the expected number 20,560 of cancer deaths per 100,000 individuals.

The members of the panel (all federal government employees) are listed in an
appendix, but no indication is given of their backgrounds.

The figures and tables are either missing or cobbled in the PDF file of the
report on the Government Accountability Project web site.

Best regards.

Jim Dukelow
Pacific Northwest National Laboratory
Richland, WA
jim.dukelow@pnl.gov

These comments are mine and have not been reviewed and/or approved by my
management or by the U.S. Department of Energy.
************************************************************************
The RADSAFE Frequently Asked Questions list, archives and subscription
information can be accessed at http://www.ehs.uiuc.edu/~rad/radsafe.html