Bliley PM
Discussion Points
Abstract
Delineates discussion points which provide arguments in opposition of the EPA's classification of Environmental Tobacco Smoke as a Class A Carcinogen. Provides arguments that the EPA did not apply its assessment guidelines even-handedly to the tobacco industry and if the EPA subjected chlorinated water (tap water) to the exact same guidelines "chlorinated water would also be a Group A carcinogen." Argues that the EPA is "adjusting science to fit policy" by ommitting studies that don't fit its objective, i.e. numerous studies which found no statistically significant increased risk for disease. Duplicates Bates #2023854880 and #2023329136.
Fields
- Company
- Philip Morris Cos., Inc.
- Type
- Memorandum
- Named Person
- Alavanja, M.C.R.
- Brownson
- Farland, W.
- Hock, E.T.
- Loy, T.S.
- Named Organization
- American Journal of Public Health
- *EPA ( use United States Environmental Protection Agency)
- National Cancer Institute NCIDivision of Cancer Prevention and Control, National Cancer Institute located in Rockville, MD
- Office of Health and Environmental Assessment
- Science Advisory Board's Indoor Air Quality Total Health Exposure Committee [IAQTHEC]
- Region
- United States
- Keyword
- "Passive Smoking and Lung Cancer in Nonsmoking Women"
- Bladder Cancer
- Childhood respiratory disease
- Chlorinated water
- Confounding
- Diet
- Ex-smokers
- Group A Carcinogen
- Nonsmoking women
- Rectal Cancer
- Spousal smoking
- Spousal smoking
- Thesaurus Term
- animal research
- dietdiets low in fruits and vegetables might be a risk factor for cancer
- *environmental tobacco smoke (use secondhand smoke)
- epidemiology
- inhalation study
- lung cancer
- non-smoker
- Nutrition/Weight (Health Effects)
- pollution
- public place
- respiratory disease
- risk assessment
- risk factorsome will say only that smoking is a risk factor for disease rather than a cause
- workplace
Document Images
Discussion points
Isn't it a fact that 24 of the 30 studies reviewed by the
EPA, and on which it based its risk assessment calculations,
showed no statistically significant increased risk?
Discussion:
The EPA's response to this inquiry should be yes.
The concept of "statistical significance" is
important because it permits a scientist to infer
either that the data in a study support or do not
support a given hypothesis. For the studies on
spousal smoking and lung cancer, the data in 24 of
30 studies are compatible with the hypothesis that
there is no overall association between spousal
smoking and lung cancer. Typically, when results
do not achieve statistical significance, further
analysis oft he data is not meaningful or productive.
Apparently, the EPA does not ascribe to this accepted
statistical principle. While data from the six
remaining statistically significant studies permit
the scientist to reject the hypothesis of no
association, the scientist must further investigate
whether the statistically significant association
is due to the exposure in q~estion or to some other
risk factor.
2504100360
2 o
Isn't it trn/e that of the ll U.S. studies reviewed by the
EPA, not one originally reported an overall statistically
significant risk?
Discussion:
The EPA should confirm that none of the eleven
studies reviewed by EPA reported an overall
statistically significant risk. Moreover, even
when the EPA recalculated the statistical confidence
limits for the ii U.S. studies, only one of those
studies reportedly achieved statistical
significance. However, as indicated above, the raw
data in the i I studies do not permit chance to be
effectively ruled out.
3. Isn't it true that of the studies the EPA considered, ll
included estimates of workplace exposures, of which nine
reported no statistically significant increased risk for
nonsmoking females?
Discussion:
The EPA should clearly respond in the affirmative.
If the data on workplace exposures are pooled in a
meta-analysis, the risk estimate is below 1.00
(unity), which indicates no positive association
between reported workplace exposures to ETS and
- 2 -
2504100361
lung cancer in nonsmokers. However, the Draft Risk
assessment did not consider those data.
Isn't it a fact that the EPA omitted from its ETS risk
calculation data from the NCI-funded Brownson, et al., study,
one of the largest and most recent studies on ETS and lung
cancer, which found no increase in risk from exposure to ETS?
Discussion:
The EPA should confirm that it has not included the
Brownson study in its calculations.
This case-control study is among the largest
conducted on reported ETS exposure and lung cancer
incidence. It includes 432 "lifetime" nonsmokers
and 186 exsmokers, and 1,402 controls.
A~ OR of 1.0 (95% CI 0.8-1.2) was reported for
spousal smoking in nonsmokers (218 cases and 598
controls). This odds ratio is not statistically
significant.
Brownson, R.C., Alavanja, M.C.R., Hock, E.T., and
Loy, T.S. "Passive Smoking and Lung Cancer in
Nonsmoking Women," American Journal of Public Health
82: 1525-1530, 1992.
- 3 -
Isn't it true that if We EPA had include~ the Brownson study,
its risk assesslent would not have found a statistically
increased risk of lung cancer due to exposure to ETS?
Discussion:
T~e EPA should unequivocally answer yes. If the
Brownson study is added to
U.S. ~S-l~g cancer studio, ~ If ~e EPA's method
of adJ~en~ for misclassifica~ion Is applied to
all U.S. studies d~s not exce~ 1.07, a risk
estate which i~ not statls~i~lly ~ignlfi~t
a 95% confidence
In llght of the a~ove, isn't it true that the EPA would not
have classified ETS as a Group A carcinogen had the EPA used
the meth~olo~ies and ~uidellnas it e~ployed in
previou~ risk assessments?
Discussion:
The EPA's answer shoul~ ~ yes. The EPA has
establishe~ a precedent-sAatterlng framework for
the ETS and other future risk assessments. As
proclaime~ by Dr. William Farland, fr~mEPA'e Office
of Health andEnvironmen~alAssessment to the Science
Advisory B~ard'~ IA~F~EC meeting on July 21, 199~:
This (the ETS risk assessment) is a high
visibility assessment . . . because of
2504100363
- 4 -
its implications for the future of the
way we do business. (Meeting transcript,
at 1-31.)
First of all, we have a focus on human
data which is fairly unique in terms of
dealing with environmental pollutants.
(Meeting transcript, at 1-33.)
Another fairly unique situation with regard
to environmental risk assessments is we've
taken the opportunity to use some new
techniques, meta-analysis for lung cancer,
that we think will be important to us in
terms of combining information from various
studies as we do risk assessments in the
future. (Meeting transcript, at 1-34 .)
So, we think that there are some
interesting and important features and an
OpPOrtunitY to do some innovative risk
assessment work in this particular
~uq~K~l~. (Emphasis added.) (Meeting
transcript, at 1-35.)
So these are all features of this
particular assessment that we think are
going to have a great impact on the way
we do future assessments in the Agency.
. . . (Meeting transcript, at 1-36.)
If the EPA had followed its own 1986 draft
guidelines, it would have included: (1) a hazard
evaluation which would have examined data regarding
the physical and chemical characterization of ETS,
as well as the results from published animal
- 5 -
inhalation studies and in vitro studies; (2) an
exposure evaluation which would have included the
data from well over i00 studies in the published
literature which monitored ETS constituents in the
air of public places and work places; (3) a dose-
response evaluation which would have included an
examination of the actual data reported in the
epidemiologic studies on spousal smoking; and (4) a
risk characterization which would have included the
range of uncertainty in numbers of lung cancer deaths
reportedly attributed to ETS exposures. The
guidelines also require, for the evaluation of
epidemiologic studies, that chance must be ruled
out statistically (i.e., the results should be
statistically significant), and that all possible
biases and possible confounding factors in such
studies are to be considered. The EPA's current
strategy to classify ETS as a Group A carcinogen
based solely upon epidemiolo~ic studies would have
failed had they carefully adhered to their own
guidelines.
Isn't it true that if the EPA subjects chlorinated water, the
ordinary tap water consumed by most Americans, to the exact
2504100365 _ _
same methodology applied to ETS, chlorinated water would also
be a Group A carcinogen?
Discussion:
EPA must clearly answer yes. According to the
results of a meta-analysis on the chlorination of
water and chlorination by-products, published in
the American Journal of public ~@alth (July 1992),
the authors reported that "a sample meta-analysis
of all cancer sites yielded a relative risk estimate
for exposure to chlorination by-products of 1.15."
These results were statistically significant, as
were results reported for "organ-specific neoplasms"
such as bladder cancer and rectal cancer. The meta-
analysis was based upon the adjusted relative risk
estimates from epidemiologic studies, precisely the
same basis used in the EPA risk assessment on ETS.
The estimated results were reportedly statistically
significant, even after the apparent adjustment for
confounding factors.
Isn't it true that the NCI has completed a major study that
finds poor diet among non-smoking women is a significant risk
factor for lung cancer? If that's true, doesn't this mean
that ali ETS studies that didn't take diet into consideration
must be re-evaluated?
- 7 -
2504100366
Discussion:
A finding by the NCI that poor diet among nonsmoking
women is a significant risk factor for lung cancer
should compel EPA to answer yes. Several of the
published studies on spousal smoking considered by
the EPA also have adjusted for the importance of
diet. While the results are mixed, several suggest
that a healthy diet, or, conversely, the avoidance
of a poor diet containing fat and spicy foods, will
affect risk estimates for nonsmokers married to
smokers. Other studies suggest that diet is an
independent risk for lung cancer.
Isn't it true that this is the first time th~ EPA has
classified any substance a Group A carcinogen based on such
weak epidemiological data and without corroborating animal
data?
Discussion:
EPA must respond in the affirmative. This is the
first EPA risk assessment based solely upon
epidemiologic data in which a substance has been
designated a Group A carcinogen. Public comments
on the risk assessment by scientists pointed out
that the epidemiologic data on spousal smoking,
when taken as a whole, do not convincingly support
2504100367 - 8 -
a Group A carcinogen classification. Most of the
studies are not statistically significant, and study
biases and confounders (e.g., occupation, diet,
heredity, etc.) were not effectively zn/led out as
contributing factors to the reported associations
between spousal smoking and lung cancer. The snmmary
risk estimate achieved by combining those studies
is very low (less than 1.20), which is deemed "weak"
by epidemiologists and statisticians. Moreover, the
Group A carcinogen classification has not
corroborated by any published animal inhalation
studies.
i0.
Isn't it true that the EPA is adjusting science to fit policy
as it was criticized of doing in its audit report
"Safeguarding the Future" by excluding studies that don't
fit its objective, changing the confidence level from 95% to
90% and not disclosing its methodologies for external analyses
and verification?
Discussion:
The EPA's SAB review draft risk assessment on ETS
(May, 1992) excluded from consideration over 25
studies on childhood respiratory disease and parental
smoking which did not report an affect from parental
smoking. It also excluded from its analysis
2504100368 - 9 -
published criticisms of epidemiologic studies on
spousal smoking and lung cancer. Perhaps even more
important, it presented a meta-analysis of the
available epidemiologic studies on spousal smoking
which is not readily reproducible, and one in which
the confidence interval was narrowed from the
scientifically accepted 95% level for statistical
significance to a 90% level.
- lO -
