A Comment on 'health Effects of Passive Smoking: Assessment of Lung Cancer in Adults and Respiratory Disorders in Children'. Epa / 600/6-90/006a, 900500

Lorillard

A Comment on 'health Effects of Passive Smoking: Assessment of Lung Cancer in Adults and Respiratory Disorders in Children'. Epa / 600/6-90/006a, 900500

Date: May 1990 (est.)
Length: 31 pages
87655164-87655194
Jump To Images

http://tobaccodocuments.org/lor/87655164-5194.html

snapshot_lor 87655164-87655194

Fields

Author

Gori, G.B.

Type

REPT, OTHER REPORT

BIBL, BIBLIOGRAPHY

FOOT, FOOTNOTE

SCRT, SCIENTIFIC REPORT

Alias

87655164/87655194

Area

SPEARS,ALEXANDER/EXEC CONF ROOM STORAGE

Named Organization

Cdc

Division of Cancer Causes + Prevention

Epa, Environmental Protection Agency

Franklin Inst

Ftc, Federal Trade Commission

Hew, Dept of Health Education and Welfare

Instituto Superiore Di Sanita

Intl Society of Regulatory Toxicology +

NCI, Natl Cancer Inst

Nutrition + Cancer

Regulatory Affairs Comm

Society of Toxicology

Technical Affairs Comm

TI, Tobacco Inst

Academy of Toxicological Sciences

Named Person

Adami, H.O.

Albanes, D.

Beaglehole, R.

Benner

Brown, C.L.

Cheng, Y.

Correa

Coultas

Doll, R.

Dube

Edlin, C.

Ekwo, E.E.

Fisher

Gao, Y.T.

Garfinkel

Goodman, M.T.

Green

Grimm

Hatziandreu, E.J.

Hiller

Hinds

Jackson, R.

Jain, M.

Kerigan, A.T.

Kirk

Koo

Kulessa, Che

Kvale, G.

Lange, P.

Lavecchia, C.

Lebowitz, M.D.

Leeder, S.R.

Lees, R.E.

Lemarchand, L.

Li, W.X.

Lowrey

Lynch

Mettlin, C.J.

Miesner

Mumford, J.L.

Ooi, W.L.

Peto, R.

Pisani

Pollack, E.S.

Repace

Rothman, K.

Sakurai, R.

Samet, J.M.

Stanton, M.F.

Sterling

Surgeon General

Tager, I.B.

Tenkanen, L.

Turner

Varela

Weber

Whichelow

Willett

Witorsch, P.

Witorsch, R.J.

Wu, A.H.

Wynder, E.

Ziegler

Date Loaded

05 Jun 1998

Request

R1-004

R1-059

R1-132

Master ID

87653565/6821
Related Documents:

Litigation

Stmn/Produced

Author (Organization)

Health Policy Center

Characteristic

ATCH, ATTACHMENTS MISSING

Site

G65

UCSF Legacy ID

swr21e00

Document Images

Page 1: swr21e00 Log in for more options!

A COMMENT ON "HEALTH EFFECTS OF PASSIVE SMOKING: ASSESSMENT OF LUNG CANCER IN ADULTS AND RESPIRATORY DISORDERS IN CHILDREN". EPA/600/6-90/006a, MAY 1990 Dr. G.B.Gori. The Health Policy Center Bethesda, Maryland

Page 2: swr21e00 Log in for more options!

- 2 - I have been asked by The Tobacco Institute to review the draft EPA document entitled "Health Effects of Passive Smoking; Assessment of Lung Cancer in Adults and Respiratory Disorders in Children," which was released for public comment on June 25, 1990. My general conclusion is that this draft cannot be regarded as a credible scientific document. CREDENTIALS I am currently director of the Health Policy Center, Bethesda, Maryland, a consulting practice with a program of research in toxicology, epidemiology, risk assessment, and the formulation of public policy alternatives compatible with health, safety and the common welfare. For eight years I was director of the Franklin Institute's Policy Analysis Center, leading a similar program of basic and applied research for an independent examination of health hazards and risks, and the costs and benefits of regulation. Before joining the Franklin Institute I held executive positions from 1968 to 1980 at the National Cancer Institute. From 1972 to 1980 I was Deputy Director of the Division of Cancer Causes and Prevention, with simultaneous responsibilities as Acting Associate Director of the Carcinogenesis Program (1976-78), Director of the Smoking and Health Program (1968-1980), and Director of the Diet, Nutrition and Cancer Program (1972-1980). Before 1972 I served as the Associate Scientific Director of the Division, which is entrusted with developing an understanding of cancer causation and of the methods for its prevention. I have been

Page 3: swr21e00 Log in for more options!

3 directly active in environmental carcinogensis and health, nutrition and health, non-ionizing radiation and health, and the relationship between smoking and health. In recognition, I received the DHEW Superior Service Award in 1976. Prior to 1968 I held positions of increasing responsibility in industry, academia, and Istituto Superiore di Sanita, the national health research center of Italy. I am a diplomate of the Academy of Toxicological Sciences, Vice President designate of the International Society of Regulatory Toxicology and Pharmacology, past member of the Technical and Regulatory Affairs Committees of the Society of Toxicology, a member of several scientific societies and editorial boards, editor of Nutrition and Cancer, an international Journal, and the author of over 100 scientific publications. A copy of my curriculum vitae is attached. INTRODUCTION Reductionist assumptions have frequently been employed in the epidemiology of multifactorial diseases. Although it is recognized that such diseases are influenced by a multitude of factors, cost considerations often have dictated the assumption that only one or a small number of factors are involved. Public health policies based on such grossly simplistic epidemiologic foundations can easily lead to unfair treatment for the particular chemical or other substance that is deemed to be the "culprit" for regulatory purposes. Yet no impartial jury would convict a defendant who

Page 4: swr21e00 Log in for more options!

- 4 - happened to be present along with many other people at the scene of the crime, without ascertaining his actual involvement. Nevertheless, policy makers often do proceed in this fashion, offering the justification, as Kenneth Rothman, a leading epidemiologist, has pointed out, that "the exigencies of public health problems demand action and that despite imperfect knowledge causal inferences must be made." (Rothman 1986, p. 17). Surely there is a need for some bounds to the uncertainty permitted in the formulation of public health policy. At the very least it would seem reasonable to require that all or at least the most obvious competing hypotheses should be adequately explained away or factored in, before a suspicion of causality can be linked to a specific factor and before public policies can be imposed on grounds of reasonable prudence. On its face this would seem a minimum standard for fair and sustainable policies. As a foundation for public policy, the draft EPA ETS risk assessment does not meet this minimum standard. This document admits the presence of, but fails to account for, the multitude of factors that we know must play a part in this complex epidemiologic scenario. The ETS-attributed relative risks are very small, and indeed they usually are much lower than the reported RRs of suspected and likely confounders. Indeed, it appears that the EPA draft began with a conclusion and proceeded only with evidence that sustains that

Page 5: swr21e00 Log in for more options!

- 5 - conclusion: the use of one-tailed probabilities, the dogmatic exclusion of thresholds, the selective exclusion and undermining of contrary evidence, the fanciful mathematical pretensions as surrogates for a complex and difficult reality -- all justified in the name_of policy making prudence. By any measure of fairness, the EPA draft is not a scientific analysis. Rather, it reflects a manipulation of the evidence. Whatever the motivations of the authors, their methods and conclusions cannot be accepted by anyone with a modicum of appreciation for the uncertainties involved or for 0 the impartiality that the scientific method demands. ASSESSMENT OF LUNG CANCER IN ADULTS While it is cloaked as an open-minded scientific exercise, the EPA's draft ETS lung cancer risk assessment reveals itself after careful scrutiny as an apology for a preconceived notion of what the evidence means. It fails the test of scientific objectivity for the following reasons. * The stated null hypothesis is that ETS does not increase the lung cancer risk of exposed subjects. Yet studies supporting the null hypothesis are selectively excluded from the analysis. * Confidence intervals include values below 1.0 for RRs in most of the relevant epidemiologic studies, which implies the probability of either increased or reduced RR values. Nevertheless, one-tailed probability estimates are used in the meta-analysis exercise in the draft.

Page 6: swr21e00 Log in for more options!

- 6 - * The document claims consistency of responses and dose/effects gradients, but this only by selective evidence. * The document explanations when in fact claim can be sustained elimination of studies with contrary affirms that no alternative exist for the alleged excess RRs observed, no study has controlled for numerous suspected factors that have been associated with lung cancer risk. * In meta-analysis exercises, tests based on statistical symmetry apply only to hypothetical non-biased data and would not hold in the presence of uncontrolled non-random confounders. * The document recognizes that social and environmental differences tend to increase RR values in Japanese studies, and that the trend of association may be valid for such studies only. Yet it makes improper use of Japanese data to estimate RRs and lung cancer mortality projections for the United States. SELECTIVE EXCLUSION OF STUDIES The EPA document justifies the exclusion of the Varela study from table 3.5 (p. 3-15) and from the meta-analysis because the proportion of ETS-exposed controls is not reported. This exclusion is unwarranted, however, because the raw data "missing" from the written report were

Page 7: swr21e00 Log in for more options!

- 7 - obviously used in the construction of the results published. This massive study supports the null hypothesis. Although the 19 studies in table 3.5 are used in the meta-analytic construction of RRs, statistical support of a positive association is also contrived from a different set of selected studies from which all but one of the < 1.0 RR studies are excluded (table 3-6). The report itself concedes that the inclusion of studies in the meta-analysis exercise "... may be subjective." (p. 3-21 bottom). Indeed it is subjective. ONE-TAILED PROBABILITIES The null hypothesis is that ETS exposure does not affect lung cancer risk. Epidemiologic studies generally show RR confidence interval bands on both sides of 1.0, which does not justify rejecting the null hypothesis. The authors of the EPA document show an understanding of the meaning of such confidence intervals when they state (p. 3-35 bottom): "confidence intervals are consistent with a wide range of possibilities," but they resist the obvious possibilities of RR values less than 1.0. The situation mandates the use of two-tailed probabilities in the statistical analysis of the data. Instead the EPA document contains statements that must be astonishing even to an elementary student of statistics, such as (p. 2-2 top): "Statistical significance is evidence that an effect is at a sufficiently high level to be detected with the data

Page 8: swr21e00 Log in for more options!

8 available;-lack of significance only supports the conclusion that it is below a level that the data have adequately high power to detect with assurance." And again (section 4.4, p. 4-22): "The data from epidemiologic studies currently available are evaluated for evidence of an elevated occurrence of lung cancer associated with ETS exposure." Accordingly, the document improperly uses one-tailed probability criteria in several occasions. For instance (p. 3-21, mid-page), the report states that over half of the 19 case-control studies have P5 values less than 0.1, when the use of two-tailed probability shows that to be true for only 4 studies. STATISTICAL SYMMETRY TESTS The tests of trend significance based on statistical symmetry of meta-analysis data (Figures 3-2 and 3-3 of the report), are valid under the hypothetical assumption of unbiased data, namely that non-random confounders ounders are not at play. Since this assumption cannot be validated and is actually likely to be false, such tests would also give a false impression. Publication bias alone would tend to invalidate the assumption of a symmetrical distribution. The significance of the Wilcoxon signed-rank test (p. 3-22 bottom) might be relevant, but the report does not indicate whether in fact appropriate weight was applied to each study under two-tailed probability criteria.

Page 9: swr21e00 Log in for more options!

- 9 - DOSE-EFFECT GRADIENTS The EPA document states (p. 3-38, mid-page): "[w]ith a consistent upward trend across exposure categories and age groups that cannot be ascribed to chance alone, one has some assurance that the sources of variability are not obscuring a dose-response relationship." This could be true, unless of course the confounders are non-random ones. As will be seen later, no study has provided reasonable controls for probable non-random confounders of lung cancer risk that are well-described in the literature. Moreover, this assurance fades when considering that the dose-response gradients of the epidemiologic studies reviewed are anything but consistent. In fact, the erratic behavior of such gradients (figure 3.4, p. 3-28) strongly suggests interferences from non-random confounders. The EPA document goes to great lengths to explain away the absence of a dose-response gradient in Garfinkel (1981). This is especially evident in the statement at p. 3-35 (bottom): "The statistical evidence supporting an association between lung cancer incidence and ETS exposure in GARF(coh) is inconclusive -- it is consistent with either the presence or absence of a true dose-response relationship." In fact the evidence tends to support the absence, unless one accepts the unbelievable rewriting of the meaning of confidence intervals that follows in the EPA document. The evidence shows that, with the exception of the weak Correa study (only 8 cases reported), none of the U.S. epidemiologic studies provides

Page 10: swr21e00 Log in for more options!

evidence of an upward dose-response gradient (figure 3-4, p. 3-38). Thus, one of the fundamental points of this report, namely that of an upward trend in dose-response (see p. 1-4), is in fact without foundation. POTENTIAL CONFOUNDING RISK FACTORS The EPA draft maintains that "[t]he statistical results solidly support the conclusion that the observed association between lung cancer and ETS exposure in case-control studies is not attributable to chance occurrence " (p. 3-43, top). And again (4-38 mid-page): "Consideration of plausible confounding factors and covariables has not produced an alternative explanation ...." And (4-I mid-page): Review and analysis . . . of the epidemiologic studies ... have not indicated a correlate of ETS that may explain the observed association . The truth, however, is that available ETS epidemiologic studies controlled for none or very few of the many common factors that increase or decrease the risk of lung cancer. Separate studies indicate that such independent factors may actually cluster and selectively segregate in families with smokers (Pisani et al., 1986; Whichelow, 1988; Koo et al., 1988). A partial list of such factors is given in table 1 below.

Tobacco Documents Online

Lorillard