7062 Education & Debate BMJ Volume 313
Saturday 12 October 1996
Education and Debate
BMJ No. 7062 Volume 313 Saturday 12 October 1996
Passive smoking and health: should we believe Philip
Morris's "experts"?
George Davey Smith, Andrew N Phillips
A series of
adverts has recently appeared in newspapers across Europe comparing the risk of lung cancer from passive smoking
with a variety of other apparent risks from everyday activities (see
fig 1). The implication is that the increased risk of lung cancer among
those exposed to other people's tobacco smoke, of around 20%, is
minuscule in comparison with the apparent 500% increased risk of lung
cancer associated with a diet high in saturated fat, the 180% increase
with frequent cooking with rape seed oil, the 60% increase with
drinking 1-2 glasses of whole milk per day, or the 70% reduction in
risk associated with high fruit diet. The advert, entitled ``What
risks do you take?'' is cleverly tailored to the public's
scepticism about the apparent health risks of everyday
activities.(1) A few weeks after the adverts
appeared the main headline in a major British Sunday newspaper was
"Beefburgers linked to cancer."(2) The frequent
appearance of such news stories, which are then often contradicted or
reversed by subsequent reports, leads to distrust in pronouncements
from experts - what may be called the "now they're saying"
syndrome.
The central message of the advert is that passive smoking is not
"really a meaningful health risk to people who have chosen not to
smoke." Readers were asked to write for a copy of a report
"Environmental tobacco smoke and lung cancer: an evaluation of the
risk," from a team of authors referred to as "The European Working
Group on Environmental Tobacco Smoke and Lung
Cancer."(3) The report focused only on the risk
of lung cancer, while the adverts referred to an absence of risks of
health problems in general with passive smoking. The title of the
working group has the ring of authority to it, although unsurprisingly
it turns out to be an industry funded enterprise.
Passive smoking and lung cancer: the European working
group report
The review of the evidence linking passive smoking to lung cancer risk
produced by the tobacco industry sponsored working group is of limited
use as a scientific document. It is, however, interesting as a
demonstration of the rhetorical mode adopted to dismiss the possibility
of a link between passive smoking and lung cancer. The core of the
report relates to a summary and meta-analysis of observational
epidemiological studies regarding the risk of lung cancer in people who
have never smoked but who are exposed to environmental tobacco smoke.
These investigations have looked at never-smoking spouses (or
cohabitees) of smoking partners, never-smoking people whose parents
smoked, and never-smokers exposed to smoke at their workplaces. As with
other meta-analyses of this
issue,(4)(5) an increased risk of about
20% was seen in the spousal studies. The working group considers only
studies with female never-smokers with male smoking spouses, on
the grounds that there are only nine studies with male
never-smokers. This is odd as later the group goes on to carry out
subgroup meta-analyses of as few as four individual studies. The
suspicion is aroused that the male studies are ignored because they
show larger increased risks on average than the female studies.
Having found a significantly increased risk in its meta-analysis of
spousal studies, the working group goes on to attempt to discredit this
finding. This could be seen as a welcome antidote to the usual practice
of scientists attempting to shore up their embattled hypotheses against
challenges from newly accruing empirical data. More plausibly, however,
it could be seen as a way of getting the accumulated data to fit with
the already agreed conclusion. A series of possible biases are
considered, which will here be dealt with in turn.
Misclassification
The main criticism of studies showing an increased risk of lung cancer
among never-smokers exposed to environmental tobacco smoke is that some
current or former smokers may claim to be never-smokers yet be at
increased risk of lung cancer because of their previous smoking. If
this misclassification of never-smokers is associated with spousal
smoking, then it could generate an apparent, but spurious, increased
lung cancer risk among passive smokers. Since smokers are more likely
to marry other smokers it is plausible that such misclassification is
differential - that is, more declared never-smokers married to smokers
are actually ex-smokers or current smokers than is the case with
declared never smokers married to non-smokers. This issue has been much
discussed by Peter Lee.(6-9) Lee is the main
authority referred to by the working group, which accepts all his
propositions and applies his misclassification model to their data.
Nowhere do they point out that Lee is an enthusiastic recipient of
tobacco industry financial support and someone who, like the working
group, has presented models which are most favourable to the tobacco
industry case.
The glaring omission in Lee's model(9) and in the
working group report is consideration of the underestimation of the
association between exposure to environmental tobacco smoke and lung
cancer risk which will be generated by the use of spousal smoking as an
indicator of exposure. Lee considers only whether the smoking status of
the never-smoking spouses enrolled in studies could be misclassified
and states that this has little effect and "will be ignored
hereafter."(9) What this fails to acknowledge is
that the factor which may cause lung cancer in never-smokers is the
inhalation of environmental tobacco smoke, not simply marriage to a
smoker. As an indicator of the amount of environmental tobacco smoke
that gets into people's lungs, the smoking status of spouses is a
highly approximate measure. The ways in which spouses smoke - for
example, do they smoke in the house or outside the house only; do they
smoke in the same room as other household members; do they have smoking
friends who are frequently around the house? - will influence the amount
of tobacco smoke the non-smoking spouse inhales. The non-smoking
spouses will also be exposed to environmental tobacco smoke from
sources outside their homes and unrelated to the smoking status of
their partners. The strength of the association between inhaled
environmental tobacco smoke and lung cancer will be underestimated when
it is indexed by the association between the proxy exposure measure of
spousal smoking status and lung cancer. Since spousal smoking history
is a fairly distant proxy measure of the amount of environmental
tobacco smoke a person inhales this attenuation will be
considerable.(10)(11)
Using Lee's method the working group claims that the association in
spousal studies it sees in its meta-analysis, of 1.16 (95% confidence
interval 1.08 to 1.25) is reduced to 1.08 (1.00 to 1.16), which
they suggest is still an overestimate since in some
countries greater spousal misclassification occurs, particularly in
"traditional countries of Europe such as Greece," where
"culture...frowns on female smoking." If, however, they took into
account the misclassification of exposure to smoke when the proxy
measure of spousal smoking is used, the association would be found to
be considerably larger than the estimate from the meta-analysis.
Many problems exist in adjusting for measurement error in
epidemiological studies.(10-12) Researchers have commonly
applied correction from misclassification to the associations they are
interested in - hence increasing them(13) - while
ignoring any biases which may have led to overestimation of the
exposure-disease associations.(13) In the present
case, however, the working group members reversed the usual procedure,
so that any association between passive smoking and lung cancer risk
becomes less apparent.
Diet as a confounding factor
The working group makes much of the possible confounding of the passive
smoking-lung cancer association by dietary factors. They quote the
inverse association between intake of fruits and vegetables and lung
cancer in many epidemiological studies, which they consider to be due
to these food items being rich in antioxidants "for which an
efficient cancer protective effect has been hypothesised." They
suggest that vitamin E and carotenoids, such as Beta carotene, are
particularly important agents here and quote "LeMarchand
et al as stating that 'complete control of the
confounding effect of Beta carotene intake is essential in assessing the
true magnitude of the passive-smoking/lung-cancer
association.'"(14) They then go on to quote
evidence that non-smokers living with smokers have lower intakes of
these antioxidants and that such confounding may account for the
apparent association between environmental tobacco smoke and lung
cancer.
Again, what is remarkable about this account is what it leaves out.
Firstly, it ignores the fact that the apparent associations between
dietary intake and lung cancer could themselves be caused by incomplete
control of confounding by smoking. Secondly, the working group fails to
point out that virtually all of the studies of diet and lung cancer
relate to lung cancers occurring among smokers, lung cancer being very
rare among true never-smokers. Thus the effect of diet, at most, is one
which potentiates the effect of cigarette smoke. Thirdly, several
randomised controlled trials of long term supplementation with
antioxidants have produced very inauspicious results, with increases or
no change in lung cancer mortality being seen in the groups given Beta
carotene, vitamin E, or vitamin A
supplementation.(15-17) The results of these trials
created enormous scientific interest and it seems implausible that the
working group did not know of them. By far the most robust evidence
relating to the proposed confounding by dietary factors seems to be
ignored simply because it does not fit in with attempts to dismiss the
association between passive smoking and lung cancer.
Evaluating the role of confounding in epidemiological studies is highly
problematic.(10)(12) Associations which
remain after adjustment for apparent confounders can still be spurious,
caused by misclassification of confounders or to the presence of
unknown - and thus uncontrollable - confounding
factors.(18) The usual situation is that
investigators do not adequately consider the degree to which
confounding can have produced their findings.(19)
but the working group bends over backwards to claim that the positive
association they find is due to confounding, ignoring important
considerations and evidence in the effort. Evaluation of associations
in epidemiological studies requires improved study design, with repeat
exposure measurements,(10)(13) rather
than a partisan attachment to one explanation or another.
Environmental tobacco smoke: can it really be bad for you?
The working group devotes much space to details of the components of
environmental tobacco smoke. They suggest that the level of carcinogens
is too low to be of concern and that even a heavily exposed passive
smoker inhales considerably less than one cigarette per day. The
composition of such smoke is, however, different from that of
mainstream smoke, with some toxins being at lower concentrations and
some at higher ones.(4) The importance of this has
been shown with respect to atherosclerosis, another illness associated
with environmental tobacco smoke.
Penn et al performed studies in cockerels, supported by
the tobacco industry funded Centre for Indoor Air Research. They found
that exposure for six hours a day for 16 weeks led to a worsening of
the state of atherosclerotic plaques.(20) The
tobacco industry claimed the exposure levels were "unrealistically
high"(21) and the experiment was repeated at
lower exposure levels, with identical results.(22)
The investigators monitored environmental tobacco smoke in bars and
restaurants and showed that levels which were associated with
accelerated atherosclerosis in cockerels were seen in such places.
Following the second study the tobacco industry funded body refused to
continue supporting this group.(21)
Publication bias
The working group echoes the oft made claim that the apparent
association between passive smoking and lung cancer is due to failure
to publish negative findings. Such biases can influence
meta-analyses,(23) but an exhaustive investigation
of whether this occurs with regard to the health effects of
environmental tobacco smoke shows that it is not the
case.(24)
Are the increased risks plausible?
The implication in the working group's conclusions is that the 20%
increased risk of lung cancer in passive smokers is not plausible. It
is, however, consistent with findings from epidemiological studies of
active smoking. Relative risks of death from lung cancer according to
number of cigarettes smoked are presented in figure
2.(25) (James Neaton, personal communication.) In
this study of a third of a million men a 40% increased risk is seen in
those smoking only 1-4 cigarettes per day, compared with those smoking
no cigarettes. This increase in risk is underestimated because the no
smoking group contains ex-smokers, whose lung cancer risk is raised by
their previous smoking. A 20% increased risk of lung cancer in passive
smokers compared to never-smokers is fully compatible with the findings
of such studies. A necropsy study showed an increased prevalence of
epithelial, possibly precancerous, lesions in women married to smokers
compared to women married to non-smokers,(26) and
tobacco related carcinogens have been identified in the urine of
non-smokers exposed to tobacco smoke.(27) In this
latter experimental study confounding and misclassification are not
issues.
Thresholds and hormesis: is a little bit of exposure good for
you?
The working group engages with the literature on thresholds - that is,
the question whether there is a level of exposure below which
substances that are carcinogenic at high concentrations may not be
dangerous. This is a complex and much debated
issue.(28) As shown above, there is no suggestion
of such a threshold in relation to the association between cigarette
smoking and lung cancer. The working group also makes much of a
phenomenon referred to as hormesis: the suggestion that low level
exposures are beneficial because they stimulate bodily defences and
thus reduce the risk of cancer.(28) This argument
has been advanced by the nuclear industry with respect to low dose
radiation exposure.
The working group approvingly quotes the suggestion that cancer
mortality is decreased by exposure to low dose radiation and that
"this scientifically-proven benefit invalidates imagined risks from
erroneous interpolations of low-dose radiation." This
implication - that low level radiation and perhaps a small dose of
environmental tobacco smoke are beneficial (with no evidence at all
given to support the latter proposition) - is an addition to the
attempts, often made by interested parties, to dismiss concerns about
low level
exposures.(28)(29)
Passive smoking and other risks to health
The tobacco industry clearly wants the opinion of the working group
that environmental tobacco smoke does not increase lung cancer risk to
be generalised into a general clean bill of health for environmental
tobacco smoke. There are, however, many other potential health effects
of inhaling smoke. The reduced birth weight and higher risk of
perinatal and sudden infant death among the offspring of smokers can be
considered the earliest effects of passive
smoking.(30)(31) In later childhood
there is evidence of poorer respiratory health and increased risk of
asthma, middle ear infections, and sore throats resulting from exposure
to environmental tobacco smoke.(30)(32) In adults the risk of chronic obstructive airways disease is higher
among those exposed to environmental tobacco smoke and lung function is
worse.(33) A considerable body of epidemiological
and experimental data suggests that ischaemic heart disease risk be
increased by exposure to environmental tobacco
smoke.(21) The tobacco industry and its friends
clearly hope that doubts cast on the association between such smoke and
the risk of lung cancer will influence considerations regarding the
effects of environmental tobacco smoke on other forms of ill health.
Passive smoking: why is the tobacco industry so concerned?
The tobacco industry is clearly going to considerable trouble to
discredit evidence that environmental tobacco smoke is detrimental to
health. Two main reasons for this can be discerned. Firstly, the threat
of legal cases resulting in compensation for those whose health has
been damaged by passive smoking represents a considerable economic
threat. For smokers themselves the tobacco companies can, and do, argue
that the risks to health have been made widely known and that the ill
effects are at least partially self inflicted. For passive smokers,
however, claims that the risks are self inflicted are less sustainable
and the legal sanctions potentially more damaging. Several cases in
Australia and the UK have resulted in damages being awarded against
employers to individuals who consider that their health or employment
prospects were damaged by environmental tobacco
smoke.(34) A major class action by non-smoking
airline cabin crew who suffer from illnesses attributable to passive
smoking is currently under way in the United
States,(34) with damages of $5 billion being
claimed.
Secondly, threats of legal action against employers who allow their
employees to be exposed to others' smoke will lead to the imposition
of stringent smoking restrictions at workplaces, which will in turn
reduce cigarette sales.(35 )
How the tobacco industry defends your right to smoke
Considerable resources are devoted to funding apparent "experts" to
cast doubt on the health damaging effects of smoking. The working group
is a typical example. A Medline search on the working group's members
revealed no evidence of expertise in epidemiology; indeed the
industry rarely funds epidemiologists.(28) Thus a
review which focused largely on epidemiological findings was carried
out by a group whose relevant expertise is in
doubt.(36)
Professor J R Idle, chairman of the working group, thanks his team
"for their tireless efforts in otherwise extremely busy agendas. The
integration of such varied expertise into a cohesive and unified
statement is a testimony to the dedication and both personal and
professional qualities of the working group."(3 )
These "tireless efforts" were presumably encouraged by the funding
which was received for this undertaking. As we saw above, when
researchers funded by the industry come to conclusions counter to their
sponsors' needs, such funding can be removed. The expenses of the
working group are tiny compared with the potential loss of revenue
through reduced cigarette sales and potential legal fees in litigation,
making these experts a very cheap investment indeed.
The way such investments can pay off is shown by news coverage of a
recent study showing a link between passive smoking and coronary heart
disease,(37) a study which added to the extensive
evidence in this area.(21) In an Independent news story a report of the positive link between
environmental tobacco smoke and coronary disease risk was qualified
with the statement that "last May the European Working Group on
Environmental Tobacco Smoke analysed 48 studies and concluded that
passive smoking did not cause cancer."(38 ) A
spokesman from the tobacco industry sponsored organisation FOREST was
quoted as saying that the claims regarding "exposure to environmental
tobacco smoke have been demolished over the years, shown to be bogus
and based on rotten science." The fact that, by the nature of its
funding, the authoritative sounding European working group was unlikely
to come up with conclusions other than it did is lost in such
reporting.
| "Doubt is our
product"
The main product of the tobacco industry is to promote the impression that
there is considerable uncertainty and scientific controversy about the damaging
affects on health of smoking. In 1969 an internal industry document read:
Doubt is our product since it is the best means of competing with the 'body of
fact' that exists in the mind of the general public. It is also the means of
establishing a controversy. If we are successful at establishing a controversy
at the public level then there is an opportunity to put forward the real facts
about smoking and health"(28)
In 1978 a report to the US Tobacco Institute stated that public worries about
smoking were "the most dangerous development to the long term viability of the
tobacco industry that has yet occured" and that the strategic and long run
antidote to the passive smoking issue is.... developing and widely publicising
clear-cut, credible medical evidence that passive smoking is not harmful to the
non-smoker's health" (40) |
Conclusion
Investigating the links between exposure to environmental tobacco smoke
and disease is beset with methodological problems, while the public
response to reports of new health risks is understandably a sceptical
one. The tobacco industry capitalises on this situation to protect its
commercial interests, through the promotion and magnification of
confusion. The industry is guarded about its real knowledge on the
health damaging effects of tobacco smoke(39) and
tries to influence opinion through the selective funding of research,
support of the publication of pro-industry opinions, and intimidation
of its opponents. The partial and biased nature of the adverts and
"expert" report at the heart of the latest industry campaign
represents a continuation of its characteristic behaviour.
REFERENCES
1 Frankel S J, Davison C, Davey Smith G. Lay epidemiology and
the rationality of responses to health education. Br J General
Practice 1991;41:428-30.
2 Richmond C, McKie R. Beefburgers linked to cancer.
Observer 1996;18 Aug:1.
3 European Working Group. Environmental tobacco smoke and
lung cancer: an evaluation of the risk. Trondheim: European Working
Group, 1996.
4 Office of Health and Environmental Assessment and Office of
Research and Development. Respiratory health effects of passive
smoking: lung cancer and other disorders. Washington DC: US
Environmental Protection Agency, 1992.
5 Law M R, Hackshaw AK. Environmental tobacco smoke. Br Med
Bull 1996;52:22-34.
6 Lee P N. Passive smoking and lung cancer association. A result
of bias? Human Toxicology 1987;6:517-24.
7 Lee P N. Marriage to a smoker may not be a valid marker of
exposure in studies relating environmental tobacco smoke to risk of
lung cancer in Japanese non-smoking women. Int Arch Occup Environ
Health 1995;67:287-94.
8 Lee P N. Environmental tobacco smoke and mortality .
Basle: Karger, 1992.
9 Lee P N, Forey V A. Misclassification of smoking habits as a
source of bias in the study of environmental tobacco smoke and lung
cancer. Statistics in Medicine 1996;15:581-605.
10 Phillips A N, Davey Smith G. How independent are independent
effects? Relative risk estimation when correlated exposures are
measured imprecisely. J Clin Epidemiol
1991;44:1223-31.
11 Phillips A N, Davey Smith G. Bias in relative odds estimation
due to imprecise measurement of correlated exposures. Statistics
in Medicine 1992;11:953-61.
12 Davey Smith G, Phillips A N. Confounding in epidemiological
studies: why "independent" effects may not be all they seem.
BMJ 1992;305:757-9.
13 Davey Smith G, Phillips A N. Inflation in epidemiology: "The
proof and measurement of association between two things" revisited.
BMJ 1996;312:1659-61.
14 LeMarchand L, Wilkens L R, Hankin J H, Haley N J. Dietary patterns
of female non-smokers with and without exposure to environmental
tobacco smoke. Cancer Causes and Control
1991;2:11-6.
15 The Alpha-Tocopherol, Beta Carotene Cancer Prevention Study
Group. The effect of vitamin E and beta carotene on the incidence of
lung cancer and other cancers in male smokers. N Engl J Med 1994;330 :1029-35.
16 Hennekens C H, Buring J E, Manson J E, Stampfer M, Rosner B, Cook
NR, et al. Lack of effect of long-term supplementation
with beta carotene on the incidence of malignant neoplasms and
cardiovascular disease. N Engl J Med
1996;334:1145-9.
17 Omenn G S, Goodman G E, Thornquist M D, Balmes J, Cullen M R, Glass A, et al. Effects of a combination of beta carotene and
vitamin A on lung cancer and cardiovascular disease. N Engl J
Med 1996;334 :1150-5.
18 Davey Smith G, Phillips A N, Neaton J D. Smoking as
"independent" risk factor for suicide: illustration of an artefact
from observational epidemiology? Lancet
1992;340:709-12.
19 Davey Smith G, Ben-Shlomo Y. Flavonoid intake and coronary
mortality (letter). BMJ 1996;312:1479-80.
20 Penn A, Snyder CA. Inhalation of sidestream cigarette smoke
accelerates development of arteriosclerotic plaques.
Circulation 1993;88:1820-5.
21 Glantz S A, Parmley W W. Passive smoking and heart disease:
mechanisms and risk. JAMA 1995;273:1047-53.
22 Penn A, Snyder C A. Inhalation of steady-state sidestream smoke
from one cigarette promotes arteriosclerotic plaque development.
Circulation 1994;90:1363-7.
23 Egger M, Davey Smith G. Magnesium, myocardial infarction and
misleading meta-analysis. BMJ 1995;310 :752-4.
24 Bero L A, Glantz S A, Rennie D. Publication bias and public
health policy on environmental tobacco smoke. JAMA
1994;b:133-6.
25 Kuller L H, Ockene J K, Meilahn E, Wentworth D N, Svendsen K H,
Neaton JD. Cigarette smoking and mortality. Preventive
Medicine 1991;20:638-54.
26 Trichopoulos D, Mollo F, Tomatis L, Agapitos E, Delsedime L,
Zavitsonas X. Active and passive smoking and pathological indicators of
lung cancer risk in an autopsy study. JAMA
1992;268:1697-1701.
27 Hecht S S, Carmella S G, Murphy S E, Akerkar S, Brunnemann K D,
Hoffmann D. A tobacco-specific lung carcinogen in the urine of men
exposed to cigarette smoke. N Engl J Med
1993;329:1543-6.
28 Proctor R N. Cancer wars: how politics shapes what we know
and don't know about cancer. New York: Basic Books, 1995.
29 Wolff S. Are radiation induced effects hormetic?
Science 1989;245:575,621.
30 Charlton A. Children and smoking: the family circle. Br
Med Bull 1996;52:90-107.
31 Blair P S, Fleming P J, Bensley D, Smith H, Bacon C, Taylor E,
et al. Smoking and the sudden infant death syndrome:
results from 1993-5 case-control study for confidential enquiry into
stillbirths and deaths in infancy. BMJ
1996;313:195-98.
32 Strachan D P, Jarvis M J, Feyerabend C. Passive
smoking, salivary cotinine concentrations, and middle ear effusion in 7 year old
children. BMJ 1989;298:1549-52.
33 Hole D J, Gillis C R, Shopra C, Hawthorne V M. Passive smoking and
cardiorespiratory health in a general population in the west of
Scotland. BMJ 1989; 299:423-7.
34 Howard G. Tobacco and the law: the state of the art. Br
Med Bull 1996;52:143-56.
35 Borland R, Chapman S, Owen N, Hill D. Effects of workplace
smoking bans on cigarette consumption. Am J Public Health
1990;80:178-80.
36 Boffetta P, Vainio H, Saracci R. Epidemiology versus a smoke
screen. Lancet 1996;348 :410.
37 Ciruzzi M, Esteban O, Rozlosnik J, Montagna H, Caccavo A, De La
Cruz J, et al . Passive smoking and the risk of acute
myocardial infarction. Eur Heart J 1996;17
(suppl):309.
38 Cooper G. Health risks of a smoker in the home.
Independent 1996; 1996;28 Aug:3.
39 Glantz S A, Slade J, Bero L A, Hanauer P, Barnes D E. The
cigarette papers. Berkeley: University of California Press, 1996.
40 Chapman S, Borland R, Hill D, Owen N, Woodward S. Why the
tobacco industry fears the passive smoking issue. Int J Health
Serv 1990:20:117-27.
(Accepted 25 September 1996)
We thank Dr James Neaton for providing the updated analyses of
lung cancer mortality according to smoking status from the MRFIT
screening study and Anne Rennie for help in preparing the manuscript.
University of Bristol,
Department
of Social Medicine,
Bristol BS8 2PR
George Davey Smith,
professor of clinical epidemiology
University
Department of Primary Care and Population Sciences,
Royal Free Hospital
School of Medicine,
London NW3 2PF
Andrew N
Phillips, reader in epidemiology and biostatistics