-




    I-
    REVIEW 702 CONFIDENTIAL
    Subiect ref 8a
    "Environmental tobacco smoke
    and the risk of cancer in adults"
    J Trddaniel et al
    European Journal of Cancer (1993), 29A, 2058-2068
    Main conclusions
    This review paper is of importance in that three of its authors,
    Trddaniel, Boffetta and Saracct are from the International Agency for
    Research on Cancer (IARC) in Lyon. Although the paper is not long, it
    covers a considerable amount of the relevant literature, and considers
    many of the important arguments. Its main conclusions are as follows:
    Lung cancer. "All the available data seem to fulfill, at the present
    time, and at least to reasonable degree, the criteria needed to accept a
    causal link between ETS and lung cancer among lifelong non-smokers."
    However it is noted that "the strength of this association has still to
    be estimated".
    Sinonasal cancer. "The suggestion of an association is present" ... "If
    there were to be a carcinogenic effect of ETS exposure, the prime site of
    action might be the nasal airway rather than the bronchi."
    Bladder cancer. "Does not seem to be associated to ETS exposure."
    pShg_K sites. "Positive studies are available' for cancers from other
    sites. including the breast, the uterine cervix and the brain, but these
    are difficult to interpret."
    CD
    C~
    GO
    C:1
    ~_n
    rQ
    BATCo document for Legal Services : Health Canada 19 October 1999

    




    I would not disagree with the conclusions regarding sites other than
    the lung, but I do not believe their conclusions regarding lung cancer
    are well founded. Below I give a number of criticisms of the paper,
    mainly in fairly general terms. Many relate to points I have made before
    regarding other papers so I will not elaborate in detail here. If
    necessary, I will elaborate on my criticisms at a later date.
    Criticisms
    There is a suggestion in a number of places that the dose received
    by passive smokers may be the equivalenc of smoking up to three
    cigarettes a day so chat their cancer risk "may approach the risk found
    for very light smokers". This erroneous conclusion is based on the
    Japanese cotinine data of Macsukura (their ref 117), known to be based on
    poor chemistry. In fact, in terms of cocinine, passive smokers on average
    (and averages are what is important for interpreting the epidemiological
    data) receive more like the equivalent of one cigarette a month, and
    maximal exposure is very unlikely to exceed the equivalent of one
    cigarette a day.
    Poor understanding of dosimetric issues is also shown when arguing
    that "health consequences of ETS exposure may differ from those of direct
    smoking". The authors fail completely to consider the point chat smokers
    have greater ETS exposure than do nonsmokers, so that even if there were
    a disease caused by ETS and not smoking one would expect to see a
    difference in risk between smokers and nonsmokers.
    (-rl
    i--)
    BATCo document for Legal Services : Health Canada 19 October 1999

    




    -3-
    1.
    A totally wrong impression of the magnitude of the observed
    relationship between ETS and lung cancer is given. There are three
    reasons for this:
    (I) They refer to results of meta-analyses of data on spousal smoking
    giving relative risks of 1.35 without giving up-to-date figures,
    which are more like 1.15-1.20.
    (U) The data cited in Table 2 are extremelv selective. Thus, for
    example. they conceal the fact that the Brownson 1992 study found
    D2 excess risk if the husband smoked and they select out results
    for the Shimizu study (smoking mother 4.0, smoking husband's father
    3.2) without citing many other lower risks for the study (which
    cites eight ETS relative risks, six in the range of 0.8-1 2).
    (iii) They fail to make it clear that the overall evidence for childhood
    and workplace ETS exposure shows no relationship at all with lung
    cancer risk.
    In discussing the possibility of confounding by diet of the ETS/lung
    cancer relationship. they cite results of only one relevan: study (Le
    Marchand, ref 114). from which they conclude that the confounding effect
    would not be gr eater. In fact (see pp 139 and 140 of my Karger book),
    according to Le Marchand's data. an observed relative risk o--" 1.13 (not
    far short of the total association to be ex-olained) could be wholly
    explained by confounding by beca-carotene.
    The discussion on bias due to misclassification of active smoking
    status is totally inadequate and completely underestimates its likely
    Importance.
    CD
    cc
    ~Jr
    BATCo document for Legal Services : Health Canada 19 October 1999

    




    -4-
    The possibility of publication bias is not even mentioned by the
    authors.
    In discussing criteria for a causal association, it is not pointed
    out that there are no data at all on one criterion - evidence chat risk
    reduces on cessation of exposure.
    In citing various authors' estimates of numbers of deaths from lung
    cance: due to ETS exposure the authors fail to mention the very much
    41 lower. dosimetrically based, estimate of Arundel jl.:, Al (J Environ Sci
    Healf- C, 1986, i, 93-118). Nor do they consider cany of ::'.-a obvious
    weaknesses of these estimates (see for example my review in Env~-.ronmental
    Technology [1991, .12, 193-208]).
    P N Lee
    2.2.96.
    9
    Co
    CD
    BATCo document for Legal Services: Health Canada 19 October 1999