Philip Morris
'health Effects of Passive Smoking: Assessment of Lung Cancer in Adults and Respiratory Disorders in Children' Notes on the Meeting of the US Environmental Protection Agency (Epa) Science Advisory Board (Sab) to Consider the External Review Draft Held at Crystal City, Washington Dc on 901204 and 901205
Fields
- Author
- Lee, P.N.
- Area
- LEGAL DEPT/EEMA ARCHIVE
- Document File
- 2501073421/2501073557/Ets General 900000 - 910000 Eema Legal Dpt.
- Type
- REPT, REPORT, OTHER
- Named Organization
- Americans for Nonsmokers Rights
- American Cancer Society
- Centers for Disease Control
- Ciar, Center for Indoor Air Research
- Coalition on Smoking or Health
- Doctors Ought to Care
- Epa Board
- Epa, Environmental Protection Agency
- Harvard
- Hri, Health Research Inst,Roswell Park
- Iarc
- Nrc
- RJR, R.J.Reynolds
- Science Advisory Board
- TI, Tobacco Inst
- Washington Post
- Women Vs Smoking Network
- 1990 World Smoking + Health Conference
- American Cancer Society
- Site
- E35
- Named Person
- Aviado
- Axelrad
- Bayard, S.
- Benowitz
- Blot
- Brown, K.
- Burns, D.
- Butler, W.J.
- Clayton
- Crawfordbrown, D.
- Cummings
- Curvall
- Davis, D.L.
- Debethizy
- Dockery
- Doll
- Farland, W.H.
- Flaak, R.
- Flamm
- Fleiss, J.L.
- Garfinkel
- Gibbons, J.D.
- Giovani
- Glantz, S.A.
- Gori, G.
- Gross, A.J.
- Haley, N.
- Hammond
- Hiller
- Hirayama
- Hoffmann
- Hood
- Houston
- Howard, G.
- Huber, G.L.
- Humble, C.
- Idle
- Janerich
- Kabat, G.
- Katzenstein, A.W.
- Kilpatrick, S.J.
- Knudson
- Koppikar, A.
- Laties
- Layard, M.
- Lee, P.N.
- Levois, M.
- Lewtas, J.
- Lippmann
- Lopez
- Lowrey
- Makosky
- Mantel, N.
- Moolgavkar
- Okeeffe
- Parrish
- Peto
- Pettiti
- Poland, T.M.
- Reasor
- Repace
- Ritchie
- Robertson
- Rockette
- Rosenbaum, W.L.
- Sanet
- Schneider, B.
- Shimizu
- Sobue
- Springall, A.
- Sterling, T.D.
- Stewart
- Surgeon General
- Switzer, P.
- Thorslund, T.
- Trichopoulos
- Tweedie, R.
- Uberla, K.
- Varela
- Wald
- Weinkam, J.J.
- Wells, A.J.
- Will
- Williams
- Witorsch, P.
- Witorsch, R.
- Wu
- Wynder, E.
- Axelrad
- Attachment
- 2501073503/2501073556
- Master ID
- 2501073503/3556
Related Documents:- 2501073503
- 2501073529 U.S. Environmental Protection Agency Science Advisory Board Indoor Air Quality and Total Human Exposure Committee (Iaqthec) Environmental Tobacco Smoke Review
- 2501073530-3532 U.S. Environmental Protection Agency Science Advisory Board Indoor Air Quality and Total Human Exposure Committee Environmental Tobacco Smoke Review Final Draft Agenda
- 2501073533-3535 Science Advisory Board Review of the Draft Report 'health Effects of Passive Smoking: Assessment of Lung Cancer in Adults and Respiratory Disorders in Children.' Epa/600/6-90/006a
- 2501073536-3555 Draft Talk to Epa Sab
- 2501073556 Passive Smoke A Cause of Cancer, Panel Concludes Epa Told Risk of Respiratory Illness in Children Also Increased by Involuntary Exposure
- Litigation
- Stmn/Produced
- Date Loaded
- 31 Jan 1999
- UCSF Legacy ID
- xhi33e00
Document Images
"Health Effects of Passive Smokinz:
Assessment of Lung Cancer in Adults
and Respiratory Disorders in Children"
Notes on the Meeting of the
US Environmental Protection A encv (EPA)
Science Advisory Board (SAB)
to consider the External Review Draft
held at Crystal City, Washington DC on
4 and 5 December 1990
Author : P N Lee
Date : 9.12.90
1. Backg.round
Earlier in the year the EPA had circulated for public comment a
draft document entitled "Health Effects of Passive Smoking:
Assessment of Lung Cancer in Adults and Respiratory Disorders in
Children" and also a further draft document entitled "Environmental
Tobacco Smoke (ETS): A Guide to Workplace Smoking Policies". The
principal points to note in the main document were that:
(i) It concluded based on the epidemiological evidence on spouse
smoking and lung cancer that ETS was a Group A Carcinogen,
i.e. an agent known to cause cancer in humans).
(ii) It estimated that annually 3800 deaths a year from lung cancer
in the USA among never and ex-smokers were the result of
exposure to ETS, with 95% confidence limits cited as
1800-6100. 2501073504
-2-
(iii) It concluded that exposure of young children to ETS from
parental smoking, particularly during infancy, is associated
with increased prevalence of acute lower respiratory tract
infections, respiratory symptoms of irritation, middle ear
effusions and reduced lung function, growth and pulmonary
development.
(iv) It did not include deaths from other cancers, heart disease,
non-neoplastic respiratory disease in adults, or from other
causes in its risk assessment, considering that evidence of a
causal relationship had not been demonstrated.
Although EPA has no regulatory power to limit exposure to ETS,
the document, when finalised, would carry great weight when
attempting to bring in new laws in US states.
The main purpose of the meeting in Crystal City was for the
EPA to receive public comments and comments from the SAB on the
draft so that they could either finalize the documents or perhaps
resubmit a further draft to go through the comment procedure.
Apparently EPA were actually under no obligation to ask for the
SAB's advice or to act on it, but in practice they usually did take
account of it. The documents could in fact not be finalized until
they had been approved by a higher Committee (which met 3-monthly)
after which they would become official EPA publications.
The EPA had already received over 3000 pages of comments from
over 100 individuals on the draft. At the meeting they promised
they would provide a considered reply to the
comments, but the level
of detail this would go into was unclear. As I understood it the
-3-
written comments
had not gone in full to the board, although each
board member was sent by the Tobacco Institute a set of
presentations to the SAB. This consisted of a volume giving 13
primary submissions on the ETS Lung Cancer Risk Assessment, a volume
giving 3 submissions on the ETS Workplace Policy Guide and 7
Appendix Volumes giving the full texts of comments mainly by those
who would be speaking or who had provided primary submissions. With
the exception of Richard Tweedie's material from Australia, I was
the only non US based consultant used by the Institute. I had
provided a 12 page summary of
my main comments (which related to
lung cancer) in the primary submissions, and a commentary of some 38
pages, with Annexes which included my draft book on ETS and my draft
paper on risk assessments in the Appendix Volumes.
The Tobacco Institute had also provided a press release (for
the day before the meeting) which outlined some of the flaws in the
report. It also included a summary paragraph for each of the
comments of all the people who had filed documents critical of the
draft EPA ETS risk assessment.
I have in my possession, for those who are interested, copies
of:
1) The TI primary submissions volume 1, which includes comments by
de Bethizy (Overall), Reasor/Will and Aviado
(Toxicology/Chemistry/Dosimetry), LeVois/Layard, Tweedie and
Lee (ETS Epidemiology General), Butler (Confounding),
Fleiss/Gross, Tweedie, Howard and Switzer (Meta-analysis),
Flamm (Regulatory Aspects), and Hood/R. Witorsch/P. Witorsch
(Childhood Respiratory Diseases). 2501073506
-4-
2) The TI press release
3) Detailed submissions of:
W J Butler
J L Fleiss and A J Gross
J D Gibbons
S A Glantz
G Gori
A J Gross
N Haley
G L Huber
A W Katzenstein
S J Kilpatrick
M Layard and M LeVois
N Mantel
R J Reynolds Tobacco Company (including Appendix by Dr George
Howard)
B Schneider
A Springall
T D Sterling, J J Weinkam, W L Rosenbaum and T M Poland
P Switzer
R Tweedie
K Uberla
A J Wells
E Wynder
I also of course have copies of the original draft reports, and
my own submissions, including a later commentary on Wells'
submission. The only paper presented at the meeting which I have
copies of is that by A J Wells, although I am on a mailing list for
copies of other materials. Ultimately a transcript of the whole 2
days will become available.
A list of the SAB members is attached as Annex A, with the
agenda attached as Annex B. Annex C is a copy of the official
letter requesting advice from the SAB. It includes 8 questions
relating to lung cancer in adults and 3 questions relating to
-5-
1)
respiratory disorders in children on which the SAB's advice was
particularly sought. I also have a copy of a document summarizing
what is known about the board members and their views on smoking and
ETS.
A red herring
Before the meeting there had been allegations in the press that
members of the Board had been "bought" by the industry. Some of
these allegations arose out of the fct that some SAB members had
conducted peer reviews for the CIAR (The Centre for Indoor Air
Research set up a couple of years ago by inter alia parts of the
tobacco industry) and had received money for this. The net result
of all this was that the 100 or so of us who were present had to
waste an hour while every board member related all his work on
tobacco and ETS and who had provided funding and to whom. There was
in fact no indication that the SAB were biassed towards industry,
rather the reverse as one of the members, Dr David Burns, was well
known for his militant anti-smoking beliefs. It was a pity no-one
made the simple point that when considering an industry's problems
it helped to have scientists who actually knew something about the
subject. As most research on an industry's problems was generally
paid for by that industry, one really had to ask whether restricting
attention to squeaky-clean ignorami, however intelligent, was really
desirable. To his credit the chairman of the SAB did point out that N
[A
there were a lot of knowledge gaps and that the CIAR work was very 0
O
useful, given EPA only have a budget of 3 mn. dollars a year for ca
Cn
0
indoor air research, only 30-40% of which is on ETS. OD
I
-6-
3. Introductory talks by the EPA
Atielrad pointed out that the EPA's policy guide was not
dependent on the ETS risk assessment. Their view was that, even in
the absence of complete scientific data, "it is prudent to minimize
exposure to indoor pollutants". They also could have relied on the
"credible" reports of the Surgeon-General and the NRC. However they
felt their policy guide would carry more weight if they evaluated
the ETS studies since 1986, quantified the extent of risk and made a
classification of ETS as a carcinogen.
He also noted the EPA had received 115 submissions on time and
23 late and that they would be preparing a detailed response
document.
Steve Bayard who was the EPA project officer in charge of the
ETS work noted the people who had prepared the various sections of
the draft risk assessment. Kenneth Brown had been responsible for
the epidemiological assessment, the meta-analysis, the risk
assessment, and Appendix B which gave relevant mathematical formulae
relating to the adjustment for smoking habit misclassification.
Charles Humble had prepared Appendix A, which provided a detailed
summary and analysis of the post-1986 case-control studies of ETS
and lung cancer. Douglas Crawford-Brown had prepared Appendix C,
which discussed the dosimetry of ETS, while Todd Thorslund had
prepared Appendix D, which described a potential framework for
dose-response modelling for ETS and lung cancer. Advice had also
been received by Joellen Lewtas and Agatha Koppikar. Bayard noted
-7-
their estimate of 3800 deaths per year was incorrect due to a
typographical error! The correct figure should have been 3700!! His
discussion on the difficulties of interpreting the ETS epidemiology
and the meta-analysis was rambling and often incoherent (at one
stage Lippmann told him to
speed up and get to the point). Amazingly
he cited evidence from a 1990 paper by Wu-Williams and Sanet (an SAB
member) of a significant dose-related trend from the overall
proving a cause and effect relationship!
data as
Thorslund briefly presented his dose-response model fitting
work. Based on data from Doll and Peto and from Hammond he fitted
the Moolgavkar-Knudson model to obtain a mathematical relationship
between lifetime cancer risk and number of cigarettes per day given
age of starting and stopping. If one plugged in ;i to 1 cigaretts
per day which he (totally unreasonably) felt was appropriate for the
cigarette equivalent for ETS exposure, then one ended up with risk
predictions consistent with the epidemiology. He then gave
estimates of risk in extreme situations, including an absurd figure
of a lifetime risk of over 50% of lung cancer for a heavily ETS
exposed individual who was an extensive metaboliser of debrisoquine.
Koppikar discussed briefly the evidence on respiratory disease
in children. When mentioning sources of bias it was notable that
she totally failed to refer to transmission of infection or
confounding by social class.
2501073510
-8-
Brown, who had written the key part of the document, had
clearly read my detailed submission as he referred to it a number of
times. He presented some slides in which he had included data from
studies that had been omitted from the report - most notably the
large US case-control study of Varela and Janerich, but also the
studies of Shimizu and Sobue and the second study of Kabat.
He
presented estimates based partly on data in my report and partly
using misclassification corrections provided by Wells showing that,
whereas the Asian studies showed a statistically significant
increased risk in relation to spouse smoking (relative risk 1.37,
95% limits 1.17-1.61), the US studies did not (relative risk 1.02,
95% limits 0.80-1.28). He did not comment on the fact that
essentially all the data on misclassification came from Western
populations so that the Asian relative risk estimates were highly
dubious.
Presentations by industry
The intention had been for there to be 12 presentations. By
the time Parish, de Bethizy, LeVois, Tweedie and Butler had
talked, we were then told we only had half an hour left. After
discussion it was agreed that we would drop Clayton, Witorsch and
Robertson, concluding with Switzer, Fleiss, myself and Flamm. In
practice this made little difference as there was very small direct
response of the SAB to points made and there were written
submissions anyway. The basis of my own presentation, including
slides, is attached as Appendix D, though in practice I modified it
2501073511
-9-
somewhat to comment on points already raised. I will not describe
the points made by the other speakers here, although I will refer to
some issues in section 8, "Questions of interpretation". Generally
the great majority of what was said came over as reasonable,
although the extent to which the points will be taken up is open to
question.
5. Presentations by coalition on smoking or health
Trichopoulos, now Professor of Epidemiology at Harvard, went
through some of the issues of interpretation. He contrasted the ETS
situation with that of alar, use of which was banned without any
real evidence, and with that of fat in diet, reduction of which
was national policy, based on evidence no stronger than that for
ETS. Although his talk purported to be an academic discussion of
issues of interpretation, there were major omissions (any mention
of smoking habit misclassification bias) and major distortions (only
advantages of using Asian data in meta-analysis were cited, with no
mention of the very relevant disadvantages - see section 8).
Garfinkel, who has now retired from the American Cancer
Society, was very disappointing. I had half expected him to present
a whole lot of new results from the two huge Cancer Prevention
Studies. Instead he only described results from his 1981 and
particularly his 1985 lung cancer studies. I had heard him give
this talk about 3 times before!
2501073512
I
-10-
Dockery gave an extremely superficial view of the ETS evidence.
Essentially he was saying that because there was a preponderance of
positive lung cancer relative risks
in relation to spouse smoking,
because the confidence limits didn't indicate any obvious
inconsistencies between the studies, and because according to his
funnel plot analysis there was no suggestion of publication bias, it
was clear that ETS caused lung cancer. There was no discussion of
various relevant sources of potential bias.
6. Public comments
Devra Lee Davis, who was involved in the NRC report, quoted
estimates of Lopez and Peto that active smoking caused 3 mn. deaths
a year worldwide now, predicted to rise to 10 mn. in 30 years time.
She offered the SAB copies of the NRC report (it tells one a lot
that some didn't have it already!) and emphasized the children's
data, drawing attention to evidence of effects of paternal smoking.
Cummings of Roswell Park Cancer Institute in Buffalo, New York,
presented evidence from his own studies of cotinine levels in
smokers, non-smokers married to smokers, and non-smokers married to
non-smokers. His data showed low levels of smoking habit
misclassification, low levels of between spouse smoking habit
concordance and a relatively high level of cotinine in non-smokers
married to smokers. I will refer to these data, which considerably
increase uncorrected relative risk estimates, in section 8.
Hiller's talk was aimed at clarifying confusion on estimates of
relative exposure in ETS exposed non-smokers and active smokers, but
was in fact rather unclear, partly because he never discussed
average exposure. He did, however, refer to his well known work
demonstrating that pulmonary retention of particles from ETS is
substantially less than that from mainstream smoke.
O'Keeffe, who represented Americans for Non-Smokers Rights,
added nothing to the science. Apart from encouraging the EPA to act
rapidly, she noted that grapes from Chile had been banned because
they contained 3 pg of hydrogen cyanide, which contrasted with 100
~Lg for ETS. (There seemed to be a problem with units here!)
Pettiti, an epidemiologist from California, discussed some
issues related to metaanalysis. She noted that exclusion of the
Varela study did not affect the conclusion regarding the statistical
significance of the overall meta-analysis relative risk. She also
noted that the study relative risks did not show statistically
significant evidence of heterogeneity. She did not, however,
mention the fact that global tests of heterogeneity are very weak in
power, and that in fact there was clear evidence that US studies
gave much lower relative risk estimates than Asian studies.
Houston, of "Doctors ought to Care", pointed to industry
N
UI
O
r+
O
V
W
fJi
r-
A
-12-
bringing in epidemiologists who claim epidemiology does not mean
anything. He cited the (totally bogus) consistency of the various
estimates of risk of lung cancer from ETS claimed by Repace and
Lowrey in their recent risk assessment.
Giovani, talking on behalf of the CDC (Centers for Disease
Control), said the draft report was excellent, comprehensive and
objective. This view, though patently absurd, is quite important,
as CDC have a reputation as being a strong base for good
epidemiology.
Makosky, of "Women vs Smoking Network",
the debate.
added nothing new to
Lowrey, Repace's well-known colleague, discussed briefly some
of the well-known claims of the two of them, since Repace, who was
at the meeting, being a member of EPA, could not talk. He
emphasised the point that they felt the evidence showed that
workplace exposure of women in the US, but not of women in Japan,
was higher than that at home. This increased background exposure in
the US studies explained why Hirayama found a larger relative risk
for spouse smoking than in Japan.
2501073515
The final speaker was Judson Wells (neither Stewart nor
Hoffmann on the list attended). He discussed misclassification of
active smokers as never smokers. He was strongly critical of much
of my work, as he had been at Perth in the 1990 World Smoking and
-13-
Health Conference, despite the fact that I had essentially answered
all his relevant criticisms in the work submitted to EPA. It was
notable that he presented estimates of the effect of
misclassification, study by study, that were substantially lower
than those he provided in his written submission (which I have
already commented on in detail). I will comment on his new material
at length separately, but for the moment it was clear that
(i) even he was coming up with estimates of bias in US studies that
were similar in magnitude to the relative risks observed, and
(ii) that he had no good data on misclassification in Asian
populations.
i. Comments of the SAB
The procedure was to go through the various sections of the
draft report, starting with detailed (clearly prepared) comments by
the members of the SAB best qualified to discuss particular areas,
and then to have comments from anyone else on the SAB that wanted to
comment. At intervals, the conclusions would be summarised for the
benefit of the EPA staff members, and on occasion Dr.Bayard asked
questions to clarify what the committee ment and he also at times
explained why EPA had taken the course they did. I do not intend to
go through all the points made in detail. Rather, in the rest of
this section I summarize general impressions and conclusions,
leaving discussion of some of the scientific points to the net
section. 2501073516
A number of general points came over very clearly. On the one
hand, it was evident that the SAB found
the document highly
unsatisfactory for numerous reasons and that it wanted to see it
~
-14-
considerably rewritten - there seemed no likelihood of an approved
version appearing inside the next 6 months. On the other hand, it
was also evident that a document revised along the lines they
envisaged would be perhaps more damaging to the industry than the
current draft. Although it seemed quite likely that the 3800 deaths
per year from lung cancer may change substantially - in particular,
it may be given a much larger aura of uncertainty - the majority
opinion of the SAB was that ETS caused not only lung cancer but a
range of other effects. I suspect that there will be another
similar meeting in perhaps 9 months time to go through the next
draft, but, given the views of the SAB, I think that the broad
conclusions of the report will be along the lines predicted from the
conclusions in the Surgeon-General and NRC reports. It is
interesting to note that there appears to be a situation where the
public perceives the EPA as being more authoritative than the
Surgeon-General or NRC, despite the fact that the EPA staff and
consultants who prepare the report are clearly very inexperienced
and are of inadequate scientific stature to go against conclusions
of the Surgeon-General or NRC!
Turning to more specific issues, a number of conclusions were
evident from listening
to what the SAB had to say:
(i) The report should follow its own published recommended
procedures for hazard identification.
(ii) The report should extend the list of diseases considered, to
consider all those to which an association with ETS has been
claimed. In that context, Lippmann suggested that because
ETS may be deposited proportionately less in the lung than is
2501073517
-15-
active smoke, other organs may be more affected by ETS.
(iii) When reviewing evidence, the report should discuss the merits
of each study and not only rely on the Surgeon-General or NRC
for pre-1986 studies.
(iv) The report should carefully discuss sources of potential bias
relevant to each association considered.
(v) The report should present an argument that the associations
seen with effects in children are cause and effect, the SAB's
view being that they were.
(vi) the report should clearly state, discuss and justify all
assumptions made.
(vii) The report should quantify the effect of all sources of
uncertainty - confidence limits resulting merely from
sampling variation are an inadequate expression of this.
(viii) For all diseases where a causal relationship is demonstrated,
the EPA should use risk assessment to quantify the effect on
the population.
(ix) Appendix C (Dosimetry of ETS), which was heavily criticised
by Lippmann (Constants used are "off the wall", "references
don't support statements", "A lot of nonsense"), should be
withdrawn. Instead, new sections on the chemistry and
physics of ETS should be included.
(x) Appendix D (Alternative approaches to risk assessment based
on dose-response modelling) was also heavily criticised. The
notion that polycyclic aromatic hydrocarbons (PAHs) could be
used as an adequate marker for ETS was strongly criticised,
it being noted that the epidemiology was not consistent with
-16-
PAHs being a major source of the carcinogenicity of direct
smoking. Most of this Appendix was to be assigned to the
bin, though certain parts could be used elsewhere in the
revised report.
(xi) Appendix B (Mathematics of misclassification) would need
extensive editing, partly to make it
comprehensible, partly
to make the formulae correct.
(xii) Appendix A (Summary descriptions of eleven case-control
studies) should, as noted above, be extended to cover all
relevant studies, not just recent ones.
(xiii) There was considerable discussion, but no conclusion, about
the possibility of limiting attention (as IARC do) only to
publications in peer reviewed journals. Personally, I think
this a bad idea (especially with regard to publication bias).
EPA said it was not their general policy to limit data in
this way.
(xiv) The question of studies to be included or excluded in the
meta-analyses was given some thought. There seemed to be
fairly general agreement that EPA were wrong to exclude the
Varela thesis. Amazingly, Laties (an SAB member who was not
an epidemiologist) said he had read Kilpatrick's submission
criticisinz the Hirayama study and could not see how EPA
justified including it in their meta-analysis!! No-one on
the SAB agreed or disagreed with him, though Blot had said
earlier EPA were wrong to argue Hirayama was a better study
than Garfinkel. 2501073519
(xv) In discussing the draft policy guide it was noted more than
-17-
once that the comparison of sidestream and mainstream smoke
was misleading. As Hammond noted there were not higher
concentrations in ETS of any compound. lore attention needed
to be given to the massive dilation of sidestream smoke by
room air. Concentrations in sidestream smoke were of little
or no direct relevance.
Two very important points should be noted.
First, Bayard of the EPA made it clear a number of times that
the SAB were asking for so many changes and envisaged such a
dramatically different and more comprehensive document (really an
updated S-G report with multiple meta-analyses and risk assessments
thrown in for good measure) that their existing staff and available
funds would be unable to cope.
Second, there was likely to be quite a strong switch of tack
about how the case was put forward to try to demonstrate that ETS
was a Group A carcinogen. The actual panel discussion on ETS
started with a nervously delivered but very sound commentary on the
limitations of the epidemiological evidence on lung cancer by
Geoffrey Kabat. Among points he noted were:
(i) the draft did not acknowledge that one is dealing with a weak
association,
(ii) that confounders may matter,
(iii) that epidemiology as a science was being stretched to its
l imi ts , 2501073520
(iv) that his own new data showed results that were "unbelievably
flat",
-18-
(v) that Garfinkel's case-control study found no association with
ETS unless someone other than the subject or the spouse
reported the data,
(vi) that one cannot overcome problems by meta-analysis,
(vii) that one cannot take the S-G and NRC reports as gospel!,
(viii) that even active smoke was a weak carcinogen compared with
such as BCME,
(ix) that ETS, even if it was a carcinogen, was not in the same
league as others on the group A list, e.g. asbestos and vinyl
chloride, and that
(1) he was happiest with the IARC conclusion which labelled the
epidemiological findings as consistent with both an effect
and with no effect. He noted the nature of the evidence had
not changed since 1986.
The following speaker, Rockette, who discussed meta-analysis,
also made it clear the epidemiology was not convincing and that
there was need to investigate why there
was an association with lung
cancer in other countries and not in the US.
After that, the next speakers made it clear they felt that a
major limitation of the EPA report was that they had not made it
clear that:
(i) active smoking caused lung cancer,
(ii) no threshold had been demonstrated for active smoking, and
(iii) ETS and active smoking contained the same carcinnogens and
were "more similar than different",
2501073521
-19-
and that this argument alone justified the claim that ETS was
carcinogenic. Despite the fact that the EPA had highlighted the
differences between ETS and active smoking, and had pooh-poohed the
cigarette equivalent approach, there were some who seemed to be
implicitly arguing that they might be happier with a dosimetric
based risk assessment. It must be noted, however, that a number of
the SAB, probably the majority, felt the epidemiology did indicate a
risk. It is interesting to note whether, since the EPA Group 1
classification demands epidemiological evidence, evidence of risk
from active smoking is actually sufficient to satisfy the criteria
as defined.
8. Questions of interpretation
8.1 Misclassification
There was no technical discussion or criticism of the
presentations on Wells and myself on the extent of bias due to
misclassification. It was generally agreed that it would be
appropriate to adjust study by study. I gave Brown, who did the
misclassification adjustment work in the draft, a copy of my
document criticising Wells' submission. I think the point went home
to the board that there was a lack of good Asian data on
misclassification rates with which to conduct accurate corrections.
2501073522
8.2 "Backg,round correction"
Taking the lung cancer risk of a totally non ETS-exposed never
smoker as 1 unit, let us define 1 + x as the risk of an average
never smoker not married to a smoker and 1+ y as the risk of an
-20-
average never smoker married to a smoker. Since we observe the
relative risk (RR) according to marriage to a smoker and since we
can estimate, via ETS marker uptake, relative exposure for the two
groups (RE), we can estimate x
and y via the formulae
RR ~ 1 ±~
1 + x
RE - y/x
to give x= RR - 1
RE - RR
and
y = RE (RR-1)
RE - RR
In the report this "background correction" applied to an RR
value of 1.28 (based on meta-analysis adjusted for smoker
misclassification) and an RE value of 3 (based on Wald and Ritchie
cotinine data) gave a value of 1+ x of 1.16
and of 1 + y of 1.48.
In other words 16/116 - 14% of lung cancer risk among never
smokers not married to a smoker was attributed to ETS as was 48/148
= 32% of risk among never smokers married to a smoker.
While the Wald and Ritchie RE figure of 3 was close to what I
got from the TAC Cotinine Study, Cummings cited much lower RE values
based on urinary cotinine from his study in New York, 1.55 for women
and 1.27 for men. Had one used 1.55, and the same RR value, one
would have got
l+x
- 2.04 N
fl
~
1 + 3x
= 4.11 O
W
tn
N
w
-21-
with attributable fractions of about 51% and 76% for the two groups,
almost tripling the total number of deaths attributable to ETS.
Though I cannot imagine EPA will use only Cummings' data to
provide an RE estimate, they may well include it for an upper
limit.
The question of background correction was in fact more
discussed as a means of trying to explain why the Asian data gave
higher relative risks than the US data. It was speculated that, in
Japan, women met hardly any other men than their husbands (do no
Japanese women work?) and that most of the women they met did not
smoke. Ergo, their background exposure would be low, so that RR's
for spouse smoking would be less easily obscured. Also the extent
of exposure from the husband in Japan might be higher than in the
US. Most (all?) of this seemed to be speculation rather than fact,
but it seemed likely that arguments along this line might appear in
the revised version (unless hard evidence refuting it comes out
first).
8.3 Confounding
Butler's presentation for the industry argued that there were
numerous other risk factors for lung cancer and that, if modestly
correlated with spousal smoking, might cause important bias. In
discussion, Sanet asked how many of the risk factors were proven
risk factors according to the criteria Butler used to prove risk of
ETS. Butler should have, but did not, point out that one does not
need to demonstrate the risk factor is causal in order to illustrate
important confounding. A number of the industry submissions, e.g.
25pi073524
-22-
Gori, Katzenstein, Sterling, Tweedie, gave lists of risk factors for
lung cancer. What is needed, of course, and research is relatively
easy to do, is good evidence that such risk factors are positively
correlated with indices of ETS exposure.
8.4 US vs non-US
Over and over again came the point that the association of ETS
with lung cancer was evident in countries other than the US, but
not in the US itself. The general impression I gained was that the
SAB did not think the meta-analysis should be restricted to the US -
however the US/non-US difference should be brought out in the
report, and reasons for it discussed.
8.5 Dose-response
A number of the industry submissions argued that evidence of
dose-response was needed to prove causation, that such evidence
could only validly be obtained from trend analysis omitting the
non-exposed group. Since such analysis did not show a significant
trend, causation cannot be considered proved. The SAB were clearly
unimpressed by this argument, preferring analyses includinZ the
non-exposed group which did show a significant trend in many
studies. I must say I agree with the SAB here. It is well known
that, as a statistical test against an alternative hypothesis that
risk increases linearly with dose, a trend test including the non-
exposed group is most powerful. Why reject part of the relevant
data? I and most epidemiologists (and statisticians analyzing
toxicological data) use the latter test routinely. In the ETS
2501073525
-23-
context it is easy to see reasons why, if a true effect exists, it
is quite likely not to be picked up
.
as significant by a test for
trend omitting the non- exposed group. One is that one is trying to
detect a smaller increase in risk, another that one can more
reliably categorise subjects according to spouse smoking than one
can categorise subjects whose spouse smokes by degree of°exposure.
8.6 Histology
This was hardly mentioned. I was told by someone at the
meeting that Trichopoulos had data on adenocarcinoma and ETS but had
never published it.
8.7 Relative ETS exposure at home to at work
It was clear estimates were highly variable by study, with
information lacking at home and at work. This of course relates to
8.2.
8.8 Dosimetry
Benowitz tried to argue cotinine underestimated particulate
exposure from ETS relative to active smoking. This was based on
differences in decay rates in air between nicotine and particles.
A counter argument was expressed that cotinine may overestimate
since there is nicotine present in a room even when no-one has
smoked for some time, due to emission from furnishings, etc.
A number of industry written submissions and de Bethizy's
spoken presentation referred to nicotine in tea, potato skins,
tomatoes, aubergines, peppers, etc. as confusing the cotinine data.
2501073526
-24-
The fact that Idle's paper was known to be completely in error (due
to a confusion of dry and wet weight of tomatoes) was not generally
understood. I strongly suspect nicotine from food is at most only
equivalent to that from minimal
ETS exposure. Benowitz said an
experimental study should be conducted. (My wife, who loves
tomatoes, would gladly volunteer!)
Benowitz also referred to new data from Curvall which very
clearly, in his view, showed that the half-life of cotinine was
virtually identical in smokers and non-smokers.
He also pointed out many of the problems in cotinine analysis
(e.g. overestimation by RIA) but felt that, properly done by GC, it
was very valuable as a marker of ETS exposure.
The complexities behind using any marker came over well in the
discussion. One relevant point made was the difficulty of getting
good data on background exposure in the past.
8.9 "Own data bias"
Kabat noted that people tend to believe their own data more
than others. It was interesting, in that context, to note that all
the members of the SAB who had carried out epidemiological studies
of ETS and lung cancer (Kabat, Blot, Sanet) had found no significant
association!
9. Press coverage
All I saw while in the US was an article in the Washington Post
(Annex E). 2501073527
Annexes
A Science Advisory Board Members
B Final Draft Agenda
C Letter from Dr W H Farland to R Flaak seeking the advice of the SAB
D Basis of my presentation with slides used
E Article in the Washington Post
N
cn
0
~
0
~
cn
N
oO
f
