UNITED STATES ENVIRONMENTAL PROTECTION AGENCY WASHINGTON D C 20460 EPA—SAB- IAQTHEC -91-007 April 19, 1991 THE AOMI IST A T - Honorable William K. Reilly Administrator US EPA REGION I LIBRARy U.S. Environmental Protection Agency IJFK FEDERAL BLDG 401 M Street, S .W. BOSTON, NA 02203_2211 Washington, D.C. 20460 Subject: Science Advisory Board’s review of the Office of Research and Development document Health Effects of Passive Smoking: Assessment of Lung Cancer in Adults and Respiratory Disorders in Children , EPA/600/ 6-90/006A, June 1990, and the Office of Air and Radiation’s draft doc- ument Environmental Tobacco Smoke: A Guide to Workplace Smoking Poli- cies , (EPA/400/6—90/004), June 1990. Dear Mr. Reilly: On November 1, 1990, the Offices of Research and Development and Air and Radiation requested that the Science Advisory Board (SAB) review the above referenced draft reports. The first document (here- after referred to as the risk assessment report) incorporates a health risk assessment of the impact of passive smoking (i.e., exposure to environmental tobacco smoke, or ETS) on adult lung cancer incidence, and a discussion of the effects of exposure to ETS on the incidence and prevalence of respiratory disorders in children. The risk as- sessment report was prepared at the request of the Indoor Air Divis- ion, Office of Air and Radiation, to provide information and guidance on the potential hazards of indoor air pollutants. The second doc- ument (hereafter referred to as the policy guide) reflects and para- llels the risk assessment report, but was not developed as a sci- entific document The Agency sought the advice of the SAB’s Indoor Air Quality and Total Human Exposure Committee (IAQTHEC) concerning the risk assess- ment’s accuracy and completeness, and the Committee’s opinion on whether the weight of available evidence supported the conclusions drawn concerning ETS’s roles in causing lung cancer in adults and res- piratory disease in children. The SAB was also requested to review that portion of the policy guide which presented a scientific database on ETS. ------- The IAQTHEC met on December 4 and 5, 1990, in Arlington, Virgi’ to conduct its review of the ETS draft documents. In summary, Committee found the risk assessment document to be a good faith effort to address complex and difficult issues affecting public health. The authors attempted to select and interpret the most relevant infor- mation from an enormous and diverse scientific data base, most of which was not designed or intended to yield the information needed for this task. Since the task is extremely difficult, it should come as no surprise that the Committee also found the document to be incom- plete in many respects. The situation is analogous to that for the Criteria Air Pollutants, wherein it has been necessary to prepare and review two or more draft criteria documents prior to their endorsement by the Clean Air Scientific Advisory Committee. The IAQTHEC has sug- gested changes both in the organization and specific technical content of the draft, that if followed, can result in an improved ETS risk assessment document. The Committee also suggested changes that would strengthen the use of the incorporated scientific database to support the recommendations contained in the policy guide. The Charge to the Committee, and associated findings of the Committee are outlined below: A. Carcinoqenicity Issues 1. Carcinoaenicity of ETS Has EPA met the requirements stated in its carcinogen guidelines for characterizing ETS in Category A, i.e., is the evidence sufficient to conclude that ETS is causally associated with lung cancer? The Committee concurs with the judgment of EPA that environmen- tal tobacco smoke should be classified as a Class A carcinogen, but notes that it had some difficulty in applying the Guidelines for Car- cinogen Risk Assessment (51 FR 33992), as they are currently formu- lated, to this complex and variable mixture. We advise EPA to place greater weight on the biological considerations and the extensive experience with active human smoking to support the classification. 2. Spousa.1 Smoking Is spousal smoking a proper measure of ETS exposure to assess lung cancer risk? Despite its various limitations as an indicator of ETS exposure, spousal smoking status seems to be a feasible method for identifying people with greater, versus lesser, ETS exposure. There are potential 2 ------- important confounders related to spousal smoking status as a measure of exposure, but such confounding concerns are present in other surrogates of exposure as well. 3. United States and Foreign Studies Are the differences in relative risk observed between studies in the U.S. and those overseas of concern, and if so, to what degree? The Committee believes that data from studies conducted in other countries, as well as in the United States should be utilized in eval- uating whether exposure to ETS increases risk of lung cancer, and does not find the observed differences to be of concern. It is appropriate to examine the totality of evidence from all the case-control and cohort studies, regardless of where they were conducted. 4. Use ot Meta—Analysis Is Meta—Analysis an appropriate tool to use in the document and has it been applied correctly? Have the epidemiological studies been properly evaluated and combined using this technique? Meta-analysis is a general term applied to a wide range of techniques intended to synthesize findings across related studies. Although it is an appropriate tool to summarize the epidemiological studies investigating the risk of ETS, the emphasis given the meta-analysis of ETS/lung cancer association in this report is not justified. Biological considerations related to respiratory carcin- ogenesis and extrapolations from human exposure via active smoking provide compelling evidence that is consistent with the results of the meta—analysis. 5. Confounders/Misc lassification Have the most important con- founders been properly addressed? Has the issue of misclassification (classifying current and former smokers as “never smokers”) been ade- quately addressed and the proper adjustments made? Are there other confounders which could be addressed in greater detail? Important potential confounders of the ETS—lung cancer relation- ship were addressed in the report mainly, by carrying out a separate meta-analysis of those studies which included adjusted analyses. The potential main confounders included in these adjusted analyses were age and surrogates for confounding factors, including education, and social class. Comparison of relative risks in those studies which analyze both factors suggests that these effects are not important 3 ------- confounders. As for other potential confounders of the ETS-lui cancer relationship, including occupation, radon exposure, and diel there is no way to evaluate their importance as confounders or to adjust for them, since virtually none of the studies contains information on them. 6. Characterization of Uncertainties Does the document charac- terize the uncertainties, both in the weight-of-evidence and the number of attributable deaths, appropriately? Vis-a-vis weight of evidence, the draft document’s conclusion that exposure to ETS sometimes leads to the development of lung cancer in humans rests upon two main arguments: (1) the biolo ical plausi- bility of such a causal association is high, given the known effects of active smoking and the known composition of ETS (e.g. the carcino- genicity of ETS in some animal studies, and the presence of known human carcinogens in ETS); and (2) the accumulating epidemiolo ic evidence on the relationship between exposure to ETS and lung cancer. These together appear to argue for a positive effect. Because the epidemiologic evidentiary base for drawing conclusions regarding ETS’s carcinogenicity consists mainly of studies of exposure levels produced by spousal smoking, the biological plausibility argument assumes great importance. Each step in that argument should therefore be careful addressed, with the uncertainties encountered being spelled o explicitly. 7. Quantitative Risk Assessment Has the quantitative risk of lung cancer been properly assessed? Would it be more prc perly as- sessed by a dose response assessment using either cotinine or res- pirable suspended particulate matter as surrogate measures of exposure (Appendix C)? Would it be more properly assessed with alternative modeling approaches (Appendix D)? Should a dose-response model be developed for ETS-radon interaction effects? The Committee generally agreed that the quantitative assessment of the risk of lung cancer due to exposures to ETS should be based on the human epidemiology studies and that meta-analysis is a suitable approach to combining the data. This approach is direct and makes the fewest assumptions. It should be noted that this approach is fully consistent with the risk assessments that have been done for many other carcinogens. Given that the epidemiology studies should be the basis of the risk assessment, some suggestions for refinements of the risk assessment are detailed in our report. 4 ------- 8. Home vs. Workplace Exposure Should the Draft Report attempt to distinguish between the effects of home vs. iiorkplace exposure to ETS? The Committee recognizes that there is little epidemiologic literature on the health effects of ETS in the workplace, and therefore on the relative impacts of home and workplace exposure. However, the report should review and comment on the data that do exist, if only to bring out the need for future research in this area. B. Respiratory Disorders in Children 1. Weight of Evidence Has the weight of evidence for ETS re- lated respiratory disorders in children been properly characterized? A draft report with a detailed description and analysis of 26 recent studies has recently been prepared and is enclosed. It is in a form similar to that of Appendix A. Should it be included in a revised report as Appendix E? In reviewing the weight of the evidence, the present Chapter 5 does not establish an appropriate framework for considering the data. The alternative explanations for association of ETS exposure with adverse respiratory effects need to be clearly listed. The weight of the evidence could then be judged to determine the causality of associations. The additional literature available since 1986 provides a basis for increased concern about the effects of ETS exposure on respiratory disorders. Thus, the Committee urges a thorough review of the entire body of evidence, including earlier reports covered in the 1986 re- ports of the Surgeon General and National Research Council. This review could be included in the revised risk document as Appendix E. 2. Confoundera Have confounders in the epidemiologic studies been adequately addressed? A number of confounders were mentioned by the report, but ad- dressed improperly, including j utero exposure, parental reporting bias, and active smoking. One must stress both the biological pre- cursors important to the effects of ETS in childhood, and the socio- economic and behavioral factors. 5 ------- 3. Use of Meta—Analysis Should a ineta-analytic approach be attempted as in the lung cancer analysis? The Agency should give serious consideration to meta-analysis of those studies of sufficiently similar design to warrant it. However, it was not clear that there is a body of studies suitable for such an analysis. C. Review of the Policy Guide The Committee found, with some exceptions detailed in our report, that the scientific database incorporated in the policy guide is cor- rect and appropriate. The policy guide should be revised to reflect changes made to the risk assessment report. We appreciate the opportunity to review these issues, and stand ready to provide review comments on any significant revisions to the subject documents. We look forward to your response on the major points we have raised. Dr. R ymond Loehr, Chairman Science Advisory Board Dr. 4orton Lip ann, Chairman Indoor Air Quality and Total Human Exposure Committee ENCLOSURE 6 ------- U.S E vIro, m.ntII W..hington, DC PrGt.ctlon Aqsncy EPA-SAB -IAQTHC-9 1-007 REPORT OF THE INDOOR AIR QUALITY AND TOTAL HUMAN EXPOSURE COMMITTEE REVIEW OF THE OFFICE OF RESEARCH AND DEVELOPMENT’S DRAFT REPORT UHEALTH EFFECTS OF PASSIVE SMOKING: ASSESSMENT OF LUNG CANCER IN ADULTS AND RESPIRATORY DISORDERS IN CHILDRENU (EPA/600/6-90/006A) A 3CIENC ADVIBORY BOARD REPORT APRIL, 1 9 1 ------- ABSTRACT The Indoor Air Quality and Total Human Exposure Committee (IAQTHEC) met on December 4/5, 1990, to conduct its review of the environmental tobacco smoke (ETS) draft documents. In summary, the Committee found the risk assessment document to be a good faith effort to address complex and difficult issues affecting public health. Since the task is extremely difficult, it should come as no surprise that the Committee also found the document to be incom- plete in many respects. The IAQTHEC has suggested changes both in the organization and specific technical content of the draft, that if followed, can result in an improved ETS risk assessment docu- ment. The Committee also suggested changes that would strengthen the use of the incorporated scientific database to support the recommendations contained in the policy guide. The Charge to the Committee, and associated findings of the Committee are outlined below: A. Lung Cancer in Adults 1. Carcinogenicity of ETS The Committee concurs with the judgment of EPA that environmental tobacco smoke should be classified as a Class A Carcinogen. 2. Spousal Smoking Despite its various limitations as an indicator of ETS exposure, spousal smoking status seems to be a feasible method for identifying people with greater, versus lesser, ETS exposure. 3. United States and Foreign Studies The Committee believes that data from studies conducted in other countries, as well as in the United States should be utilized in evaluating whether exposure to ETS increases risk of lung cancer. 4. Use of Meta—Ana],ysis It is an appropriate tool to summarize the epidemiologica]. studies investigating the risk of ETS, but the emphasis given the meta-analysis of ETS/lung cancer association in this report is not justified. 5. Confounders/MisclassifiCatiOfl Important potential con- founders of the ETS-lung cancer relationship were addressed in the report. Comparison of relative risks (RR5) in those studies which analyze relevant factors suggests that these effects are not important. 6. Characterizatic of Uncertainties The draft document’s conclusion that exposure ETS sometimes leads to the development of lung cancer in humans rests upon two main arguments: (1) the biological plausibility of such a causal association is high; and (2) the accumulating epidemiologic evidence on the relationship i. ------- between exposure to ETS and lung cancer. Because the epidemiologic evidentiary base for drawing conclusions regarding ETS’s carcino- genicity consists mainly of studies of exposure levels produced by spousal smoking, the biological plausibility argument assumes great importance. Each step in that argument should therefore be carefully addressed, with the uncertainties encountered being spelled out explicitly. 7. Quantitative Risk Assessment The Committee generally agreed that the quantitative assessment of the risk of lung cancer due to exposures to ETS should be based on the human epidemiology studies and that meta-analysis is a suitable approach to combining the data. 8. Home vs. Workplace Exposure The Committee recognizes that there is little epidemiologic literature on the health effects of ETS in the workplace. However, the report should review and comment on the data that do exist, if only to bring out the need for future research in this area. B. Respiratory Disorders in Children 1. Weight of Evidence In reviewing the weight of the evi- dence, the present Chapter 5 does not establish art appropriate framework for considering the data. The alternative explanations for association of ETS exposure with adverse respiratory effects need to be clearly listed. The weight of the evidence could then be judged to determine the causality of associations. 2. Confounders A number of confounders were mentioned by the report, but addressed improperly, including utero exposure, parental reporting bias, and active smoking. 3. Use of Meta-Artalysis The Agency should give serious consideration to meta-analysis of those studies of sufficiently similar design to warrant it. However, it was not clear that there is a body of studies suitable for such an analysis. C. Policy Cuide The Committee found, with some exceptions detailed in the report, that the scientific database incorporated in the policy guide is correct and appropriate. KEYWORDS: Environmental Tobacco Smoke (ETS); Carcinogenicity; Passive Smoking; Sidestream Smoke; Meta-analysis; Confounders; Lung Cancer; Respiratory Disease 11 ------- U. 8. ENVIRONMENTAL PROTECTION AGENCY NOTICE This report has been written as a part of the activities of the Science Advisory Board, a public advisory group providing extramural scientific information and advice to the Administrator and other officials of the Environmental Protection Agency. The Board is structured to provide balanced, expert assessment of scientific matters related to problems facing the Agency. This report has not been reviewed for approval by the Agency and, hence, the contents of this report do not necessarily represent the views and policies of the Environmental Protection Agency, nor of other agencies in the Executive Branch of the Federal government, nor does mention of trade names or commercial products constitute a recommendation for use. iii ------- U.S. Environmental Protection Agency Science Advisory Board Indoor Air Quality and Total Human Exposure Committee Environmental Tobacco Smoke Review Chairman Dr. Morton Lippmann, Professor, Institute of Environmental Medicine, New York University Medical Center, Tuxedo, New York 10987 Vice chairman Dr. Jan A.J. Stolwijk, Professor, School of Medicine, Department of Epidemiology and Public Health, Yale University, P.O. Box 3333, 60 College Street, New Haven, Connecticut 06510 Members of the IAQTHEC Dr. Joan Daisey, Senior Scientist, Indoor Environment Program, 90—3058, Lawrence Berkeley Laboratory, One Cyclotron Road, Berkeley, California 94720 Dr. Victor G. Laties, Professor of Toxicology, Environmental Health Science Center-Box EHSC, University of Rochester School of Medicine, Rochester, New York 14642 Dr. Jonathan M. Samet, Professor of Medicine, Department of Medicine, The University of New Mexico School of Medicine, and The New Mexico Tumor Registry, 900 Camino De Salud, NE, Albuquerque, New Mexico 87131 Dr. Jerome J. Wesolowski, Chief, Air and Industrial Hygiene Laboratory, California Department of Health, Berkeley, California 94704 Dr. James E. Woods, Jr., Professor of Building Construction, College of Architecture and Urban Studies, 117 Burress Hall, Virginia Polytechnic Institute and State University, Blacksburg, Virginia 24061—0156 Consultants to the IAOTMEC Dr. Neal L. Benowitz, Professor of Medicine, Chief, Division of Clinical Pharmacology and Experimental Therapeutics, University of California, San Francisco, Building 30, Fifth Floor, San Francisco General Hospital, 1001 Potrero Avenue, San Francisco, California 94110 iv ------- Dr. William J. Blot, National Cancer Institute, 9000 Rockville Pike, Bethesda, Maryland 20892 (Federal Liasion to the Committee) Dr. David Burns, Associate Professor of Medicine, Department of Medicine, H772-C, University of California, San Diego Medical Center, 225 Dickenson Street, San Diego, California 92103—1990 Dr. Delbert Eatough, Professor of Chemistry, 276 FB Brigham Young University, Provo, Utah 84602 Dr. S. Katharine Hammond, Associate Professor, Environmental Health Sciences Program, Department of Family and Community Medicine, University of Massachusetts Medical School, 55 Lake Avenue, North, Worcester, Massachusetts 06 155 Dr. Geoffrey Xabat, Senior Epidemiologist, American Health Foundation, 320 East 43rd Street, New York, New York 10017 Dr. Michael D. Lebowitz, Professor of Internal Medicine, University of Arizona College of Medicine, Division of Respiratory Sciences, Tucson, Arizona 85724 Dr. Howard Rockette, Professor of Biostatistics, School of Public Health, 318 Parran Hall, University of Pittsburgh, Pittsburgh, Pennsylvania 15261 Dr. Scott T. Weiss, Channing Laboratory, Harvard University School of Medicine, Boston, Massachusetts 02115 Science AdvisorY Board Staff Mr. Samuel R. Rondberg, Designated Federal Official, Science Advisory Board (A—1O1F), U.S. Environmental Protection Agency, 401 M Street, SW, Washington, DC 20460 (202) 382—2552 FAX: (202) 475—9693 Mr. A. Robert Flaak, Designated Federal Official, Science Advisory Board (A-1O1F), U.S. Environmental Protection Agency, 401 M Street, SW, Washington, DC 20460 (202) 382—2552 FAX: (202) 475—9693 Ms. Carolyn Osborne, Staff Secretary, Science Advisory Board (A-1O1F), U.S. Environmental Protection Agency, 401 M Street, SW, Washington, DC 20460 (202) 382—2552 FAX: (202) 475—9693 V ------- 1.0 EXECUTIVE SUMMARY TABLE OF CONTENTS 1 2.0 INTRODUCTION 2.1 Background 2.2 Scope of Issues/Charge to the Committee 2.3 Conduct of The Review 3.0 REVIEW OF THE RISK ASSESSMENT DOCUMENT 3.1 Chapter 3--Epidemiologic Evidence of Lung Cancer from ETS 3.2 Chapter 4--Assessment of Lung Cancer Risk from ETS 3.3 Chapter 5--Environmental Tobacco Smoke and Respiratory Disorders in Children 3.4 Appendix A--Summary Descriptions of Eleven Case- ControlStudies 3.5 Appendix B--Mathematical Formulas and Relationships 3.6 Appendix C--Dosimetry of ETS 3.7 Appendix D--Alternative Approaches for Estimating the Yearly Number of Lung Cancer Deaths in Non- Smokers Due to ETS Based on Dose—Response Modeling 3.8 Appendix E--Summary Descriptions of Twenty-Six Studies on Environmental Tobacco Smoke and Respiratory Disorders in Children 4.0 REVIEW OF THE POLICY GUIDE 5.0 SPECIFIC REVIEW ISSUES . 5.1 Lung Cancer in Adults 5.1.1 Carcinogenicity of ETS 5.1.2 Spousal Smoking 5.1.3 United States and Foreign Studies 5.1.4 Use of Meta-Analysis 5.1.5 Confounders/Misclassification 5.1.6 Characterization of Uncertainties 5.1.7 Home vs. Workplace Exposure . 5.2 Respiratory Disorders in Children . . 5.2.1 Weight of Evidence 5.2.2 Confounders 5.2.3 Use of Meta—Analysis 6.0 SW1MARYANDCONCLUSIONS 26 • . . . 27 27 • . . 27 • . • 28 31 • . . . 32 • . • . 33 • . . 35 • • . • 40 • • . . 40 41 • • . . 41 • . . . 43 • • . . 44 7.0 REFERENCES CITED 51 8 8 9 10 11 11 14 15 20 20 22 23 25 vi ------- 1,0 EXECUTIVE SUMMARY The Committee’s review of the environmental tobacco smoke (ETS) Risk Assessment document and Policy Guide found them to be good faith efforts to address complex and difficult issues affecting public health. The authors attempted to select and interpret the most relevant information from an enormous and diverse scientific data base, much of which was not designed or intended to yield the information needed for this task. Since the task is extremely difficult, it should come as no surprise that the Committee also found the documents to be incomplete in many respects. The situation is analogous to that for the Criteria Air Pollutants, where it has been necessary to prepare and review two or more draft documents prior to their endorsement by the Clean Air Scientific Advisory Committee (CASAC). This Committee has suggested both organizational and specific technical changes and additional analyses that, if followed, should result in improved ETS Risk Assessment and Policy Guide documents. The Committee stands ready to provide further review comments on the revised drafts. The SAB was asked to address the following issues in reviewing the documents. A. Luna Cancer in Aaults The Committee noted that Chapters 3 and 4 draft risk document addressed only the issue of lung cancer risk for non-smoking women due to spousal smoking. It is suggested that the revised document be expanded to include the full range of cancer impacts of ETS. The Committee also noted a number of areas where considerable improvements could be made organizationally, and in terms of substantive content--particularly regarding material that was not adequately covered or not covered at all. We urge the EPA staff to redraft those chapters as well. Furthermore, we recommend the addition of a new chapter addressing addressing the physical, chemical, and dose considerations of ETS in relation to the same considerations for active smoking. Findings on specific issues within the broader context of lung cancer in adults follow: 1. Carcinoaenicitv of ETS Has EPA met the requirements stated in its carcinogen guidelines for characterizing ETS in 1 ------- Category A, i.e., is the evidence sufficient to conclude that ETS is causally associated with lung cancer? The Committee concurs with the judgment of EPA that environ- mental tobacco smoke should be classified as a Class A Carcinogen, but notes that it had some difficulty in applying the Guidelines for Carcinogen Risk Assessment (51 FR 33992), as they are currently formulated, to this complex and variable mixture. The draft risk assessment document’s conclusion that exposure to ETS sometimes leads to the development of lung cancer in humans rests upon two main arguments: (1) the biological plausibility of such a causal association is high, given the known effects of active smoking and the known composition of ETS (e.g. the carcinogenicity of ETS in some animal studies, and the presence of known human carcinogens in ETS); and (2) the accumulating epidemiologic evidence on the relationship between exposure to ETS and lung cancer. These together appear to argue for a positive effect. We advise EPA to place greater weight on the biological considerations and the ex- tensive experience with active human smoking to support the classi- fication. 2. Spousal smoking Is spousal smoking a proper measure of ETS exposure to assess lung cancer risk? Despite its various limitations as an indicator of ETS ex- posure, spousal smoking status seems to be a feasible method for identifying people with greater, versus lesser, ETS exposure. The problems in not accounting for background exposure from other sources would, if anything, bias against finding increased risk of lung cancer. Bias related to misclassification associated with smoking status has been addressed and corrected for in this draft report. There are possible confounders related to spousal smoking status, but such confounding concerns are present in other surro- gates of exposure as well. The potential importance of these confounders has been determined not to be sufficient to alter the conclusion that ETS increases the risk of lung cancer. 3. United States and Foreign Studies Are the differences in relative risk observed between studies in the U.S. and those overseas of concern, and if so, to what degree? The Committee believes that data from studies conducted in other countries, as well as in the United States, should be 2 ------- utilized in evaluating whether exposure to ETS increases risk of lung cancer. It is appropriate to examine the totality of evi- dence from all the case-control and cohort studies, regardless of where they were conducted. 4. Use of Meta-Analvsis Is Meta-Analysis an appropriate tool to use in the document and has it been applied correctly? Have the epideiniological studies been properly evaluated and combined using this technique? Meta—analysis is a general term applied to a wide range of techniques whose objective is to synthesize findings across related studies. Although, it is an appropriate tool to summarize the epidetniological studies investigating the risk of ETS, the emphasis given the meta-analysis in this report in attempting to demonstrate that ETS is causally associated with lung cancer is not justified. Biological considerations related to respiratory carcinogenesis (e.g., biologic plausibility) are equally compelling. Given the similarities in composition between mainstream smoke and ETS, biological considerations related to respiratory carcinogenesis and the extensive evidence on active smoking should receive greater weight. 5. Confoundars/MIsclasSifiCatiOfl Have the most important confounders been properly addressed? Has the issue of misclas- sification (classifying current and former smokers as “never smo- kers”) been adequately addressed and the proper adjustments made? Are there other confounders which could be addressed in greater detail? Important potential confounders of the ETS-lung cancer rela- tionship were addressed in the report mainly by carrying out a separate meta-analysis of those studies which included adjusted analyses. The potential main confounders included in these ad- justed analyses were age and surrogates for confounding factors, including education, and social class. Comparison of unadjusted and adjusted relative risks (RRs) in those studies which present both factors suggests that these effects are not important con founders. As for other potential confounders of the ETS-lung cancer relationship, including occupation, radon exposure, and diet, there is no way to evaluate their importance as confounders or to adjust 3 ------- for them, since virtually none of the studies contains information on them. The issue of misclassification should not be restricted to misclassification of current smokers and ex-smokers as “never smokers.” It should also be mentioned that misclassification of diagnosis (diagnoses other than lung cancer being incorrectly classified as lung cancer; or vice versa) will cause a biasing of the RR toward the null. Not enough attention was given to possible non-differential misclassification of ETS exposure. This is an important issue, since marriage to a smoking spouse is an imperfect proxy for total ETS exposure. In the case of dichotomous exposure, such misclassi- fication would have the effect of biasing the RR estimate toward the null. 6. Characterization of Uncertainties Does the document characterize the uncertainties, both in the weight-of-evidence and the number of attributable deaths, appropriately? Vis-a-vis weight of evidence, the draft document’s conclusion that exposure to ETS sometimes leads to the development of lung cancer in humans rests upon the two main arguments noted earlier: (1) the biological plausibility of such a causal association; and (2) the accumulating epidemiologic evidence on the relationship between exposure to ETS and lung cancer. With exposure levels that are usually quite low, it is not surprising that the association is likely to be weak although, given the size of the exposed popula- tion, societally important. Because the epidemiologic evidentiary base for drawing conclusions regarding ETS’s carcinogenicity consists mainly of studies of exposure levels produced by spousal smoking, the biological plausibility argument assumes great impor- tance. Each step in that argument should therefore be carefully addressed, with the uncertainties encountered being spelled out explicitly. Epidemiologic evidence on the relationship between exposure to ETS and lung cancer should be described more completely, with the deficiencies of individual studies used to weight their contribu- tions to any conclusions that are drawn. The assumptions and un- certainties associated with each step of the risk assessment pro- cess ought to be explicitly stated. 4 ------- 7. quantitative Risk Assessment Has the quantitative risk of lung cancer been properly assessed? Would it be more properly assessed by a dose response assessment using either cotinine or respirable suspended particulate matter as surrogate measures of exposure (Appendix C)? Would it be more properly assessed with alternative modeling approaches (Appendix D)? Should a dose- response model be developed for ETS-radon interaction effects? The Committee generally agreed that the quantitative asses- sment of the risk of lung cancer due to exposures to ETS should be based on the human epidemiology studies and that meta-analysis is a suitable approach to combining the data. This approach is direct and makes the fewest assumptions. It should be noted that this approach is fully consistent with the risk assessments that have been done for many other carcinogens and that those assessments are generally based on fewer studies. Given that the epidemiology studies should be the basis of the risk assessment, some refinements of the risk assessment are recommended with respect to: 1. Criteria for Including Individual Studies in the Meta Analysis 2. Adjustment for Smoker Misclassification 3. Misclassification of Exposure 4. Uncertainties in the Estimate of Annual Lung Cancer Deaths Due to Passive Smoking 5. Dose-Response Estimation of Risk 8. Home vs. Workplace Exposure Should the Draft Report attempt to distinguish between the effects of home vs. workplace exposure to ETS? The Committee recognizes that there is little epidemiologic literature on the health effects of ETS in the workplace, and therefore on the relative impacts of home and workplace exposure. However, the report should review and comment on the data that do exist, if only to bring out the need for future research in this area. The report should also review and comment on the data that exist on exposure to ETS in public places. 5 ------- B. Respiratory Disorders in Children Chapter 5 on respiratory disorders in children was a com- mendable first effort for a very difficult task. Nevertheless, we found that it could be substantially improved and that the conclu- sions drawn in it could be made much stronger if the chapter were revised in the manner suggested in this report. The Committee found the evidence for respiratory health ef- fects in children to be stronger and more persuasive than that stated in Chapter 5 of the draft ETS Risk Assessment document, and recommends that the new draft contain a more comprehensive discus- sion on quantitative risk assessment for these effects. Specific issues are addressed below: 1. weight of Evidence Has the weight of evidence for ETS related respiratory disorders in children been properly charac- terized? A draft report with a detailed description and analysis of 26 recent studies has recently been prepared and is enclosed. It is in a form similar to that of Appendix A. Should it be included in a revised report as Appendix E? The additional literature available since 1986 provides a basis for increased concern about the effects of ETS exposure on respiratory disorders. Thus, the Committee urges a thorough review of the entire body of evidence, including earlier reports covered in the 1986 reports of the Surgeon General and National Research Council, and its incoporation as Appendix E. In reviewing the weight of the evidence, the present Chapter 5 does not establish an appropriate framework for considering the data. The alternative explanations for association of ETS exposure with adverse respiratory effects need to be clearly listed. The weight of the evidence could then be judged to determine the caus- ality of associations. 2. Confounders Have confounders in the epidemiologic studies been adequately addressed? A number of confounders were mentioned by the report, but addressed improperly. These include utero exposure, parental 6 ------- reporting bias, and active smoking. One must also stress both the biological precursors important to the effects of ETS in childhood, and socio—economic and behavioral factors. 3. Usa of Meta-Analysis Should a meta-analytic approach be attempted as in the lung cancer analysis? The Agency should give serious consideration to meta-analysis of those studies of sufficiently similar design to warrant it. However, it was not clear that there is a body of studies suitable for such an analysis. If one is warranted, it should be guided, to the extent possible, by the same considerations outlined for meta- analysis for lung cancer. The Committee was also asked to examine whether the draft Policy Guide’s statements on health contained within the first 20 pages were scientifically defensible. With some exceptions, de- tailed in the this report (section 4.0), the scientific data and interpretations contained in the draft Policy Guide were appro- priate. The Policy Guide draft will need to be revised to reflect the changes being made in the Risk Assessment. 7 ------- 2.0 INTRODUCTION 2.1 Background On November 1, 1990, the Offices of Research and Development and Air and Radiation requested that the Science Advisory Board (SAB) review the draft report “Health Effects of Passive Smoking: Assessment of Lung Cancer in Adults and Respiratory Disorders in Children,” which incorporated a health risk assessment of the impact of passive smoking on lung cancer incidence. The document was prepared by the Human Health Assessment Group, Office of Research and Development, at the request of the Indoor Air Division, Office of Air and Radiation, under the authority of Title IV of Superfund (The Radon Gas and Indoor Air Quality Research Act of 1986) to provide information and guidance on the potential hazards of indoor air pollutants. The draft risk report reviews and analyzes the data on the respiratory effects of environmental tobacco smoke (ETS) with heavy emphasis on the epidemiologic data and statistical (meta) analysis. One major portion of the Report (Chapters 3 and 4) examines the weight of evidence for lung cancer in adults. It concludes that under EPA’S carcinogen assessment guidelines, ETS should be classified as a Category A or known human carcinogen. It also estimates from epidemiology (not modeling) data that, on average, 3,800 lung cancer deaths per year in U.S. nonsmokers are attributable to ETS. The final chapter of the report examines the epidemiological evidence for non-cancer respiratory disorders in children and concludes that the detrimental respiratory effects described are associated with exposure to ETS, but that a causal association has not been established. The draft report also contains four appendices. Appendix A provides a detailed summary and analysis of eleven recent case- control studies of ETS and lung cancer. Appendix B presents pertinent mathematical formulae and relationships. Appendix C describes the dosimetry of ETS, and Appendix D presents a potential framework for dose-response modeling for ETS and lung cancer. The draft risk report was made available for public review and comment on June 25, 1990, with a 90-day comment period which closed Oct. 1, 1990. Over 3,500 copies were distributed and 107 public comments were received as of Oct 10, 1990. A summary of those comments were prepared and provided to the SAB Committee. 8 ------- 2.2 Scope of Issues/Charge to the Committee The Agency sought the advice of the SAB on the draft risk document’s accuracy and com- pleteness. The Agency also wanted an opinion on whether the weight of available evidence supported the conclusions drawn concerning ETS’s role in causing lung cancer and respiratory disease. In ad- dition, the SAB was asked to address the following specific issues: A. Lung Cancer in Adults 1. Has EPA met the requirements stated in its carcino- gen guidelines for characterizing ETS in Category A, i.e. is the evidence sufficient to conclude that ETS is causally associated with lung cancer? 2. Is spousal smoking a proper measure of ETS exposure to assess lung cancer risk? 3. Are the differences in relative risk observed between studies in the U.S. and those overseas of concern, and if so, to what degree? 4. Is meta—analysis an appropriate tool to use in the doc- uinent and has it been applied correctly? Have the epi- demiological studies been properly evaluated and com- bined using this technique? 5. Have the most important confounders been properly ad- dressed? Has the issue of misclassification (classify- ing current and former smokers as never smokers) been adequately addressed and the proper adjustments made? Are there other confounders which could be addressed in greater detail? 6. Does the document characterize the uncertainties, both in the weight-of-evidence and the number of attributa- able deaths, appropriately? 7. Has the quantitative risk of lung cancer been properly assessed? Would it be more properly assessed by a dose response assessment using either cotinine or respirable suspended particulates as surrogate measures of expos- ure (Appendix C)? Would it be more properly assessed with alternative modeling approaches (Appendix D)? Should a dose-response model be developed for ETS-radon interaction effects? 8. Should the Draft Report attempt to distinguish between the effects of home vs. workplace exposure to ETS? 9 ------- B. Respiratory Disorders in Children 1. Has the weight of evidence for ETS related respiratory disorders in children been properly characterized? A draft report with a detailed description and analysis of 26 recent studies has recently been prepared and is enclosed. It is in a form similar to that of Appendix A. Should it be included in a revised report as Appendix E? 2. Have confounders in the epidemiologic studies been ade- quately addressed? 3. Should a meta-analysis approach be attempted as in the lung cancer analysis? The SAB was also asked to comment on the scientific foun- dations of a second draft document, “Environmental Tobacco Smoke: A Guide to Workplace Smoking Policies,” (hereafter refered to as the “Policy Guide”) produced by the Indoor Air Division of the Office of Air and Radiation. 2.3 Conduct of The Review The review was assigned to the SAB’s Indoor Air Quality and Total Human Exposure Committee. The Committee met on December 4 and 5, 1990 in Arlington, Virginia to receive briefings from EPA staff, hear extensive comments from members of the public, and discuss the several issues embodied in the charge. Following the discussions, the Chair requested that designated Members of the Committee provide written materials organized to respond to the charge, and reflecting the preceding interactions. Those materials, after editing and review by all Members of the Committee constitute this report. 10 ------- 3.0 REVIEW OF THE RISK ASSESSMENT DOCUMENT The preparation of a risk assessment document for ETS rep- resents a formidable challenge, and the Conunittee recognizes the quite considerable efforts put forth by EPA staff in preparing the draft. The document contains some excellent summary materials on a large and diverse set of relevant literature, as well as some skillful and pertinent analyses that serve to address the critical issues related to the public health impact of exposures to ETS. Although we commend EPA for its efforts, we find that the document could be substantially improved, and recommend a series of specific changes in organization and content that, if followed, would make the revised document a much stronger basis for policy guidance on ETS exposure and its health effects. Since the impact of ETS on public health is comparable to that of some of the criteria air pollutants, we recommend that the revised document follow more closely the format of the Air Quality Criteria Documents. It should include additional chapters addressing the physics and chemistry of ETS, its relation to mainstream smoke, the exposures of various populations of interest to ETS, and as appropriate to the discussions of biological plausibility and weight of evidence, those aspects of dosimetry which will be needed to support other parts of the document. The contents of these additional chapters should strengthen the basis for any actions or recommendations. The Committee also reviewed the utility, format and adequacy of the five appendices to the ETS Risk Assessment Document. We found them to be of varying utility and quality, and made specific suggestions for revisions and deletions. 3.1 Chapter 3--EpidemiologiC Evidence of Lung Cancer from ETS The focus of this chapter is on hazard identification: that is, de- termining if the available evidence on ETS warrants the conclusion that exposure to ETS increases the incidence of lung cancer. As described in the Agency’s “Guidelines for Carcinogen Risk Assess- inent” hazard identification is a qualitative process that involves review “...of the relevant biological and chemical information bearing on whether or not an agent may pose a carcinogenic hazard.” The scope of hazard identification is broad, involving review of information on 1) physical-chemical properties and routes and pat- terns of exposure; 2) structure-activitY relationships; 3) 11 ------- metabolic and pharmacokinetic properties; 4) toxicologic effects; 5) short-term tests; 6) long—term animal studies; and 7) human studies. The present chapter, comprising the hazard identification step of the risk assessment, seems limited in scope when measured against the encompassing process set out by the Agency. Although the appendixes do address exposure and toxicology to an extent, this material needs more direct discussion in the chapter. The components of ETS possibly relevant to this risk assessment should be reviewed along with the characteristics of mainstream and side- stream smoke, and similarities in composition and from .j vitro bioassay should be discussed. The complexity of ETS merits emphasis; it is not a single chemical agent, but a mixture with varying characteristics by place and time in relation to its for- mation in the burning cigarette. The toxicologic effects of in- dividual components merit further discussion. Exhaustive review is not needed, since the Surgeon General’s Reports provide comprehen- sive documentation. In reviewing the human evidence, the chapter fails to draw on the voluminous evidence on active smoking and lung cancer. The 1986 Surgeon General’s Report, for example, concluded that there was enough toxicologic similarity between mainstream smoke and ETS to justify using the evidence from active smoking in reaching conclusions concerning ETS and lung cancer. The evidence on active smoking and lung cancer documents the consequences of a higher level of exposure to a mixture, mainstream smoke, that resembles ETS in composition. The epidemiologic evidence on ETS and lung cancer in nonsmokers should be considered as addressing the risks of lower levels of exposure. Thus, the evidence on active smoking and lung cancer needs to be reviewed in this chapter. The causal nature of the association between active smoking and lung cancer should be described, as should exposure-response relations for active smoking and lung cancer. The Surgeon General’s Reports could serve as the basis for developing this material. The existing chapter reviews case-control and cohort studies providing information on the association of ETS with lung cancer. Characteristics of these studies are considered. Several statis- tical approaches are used to assess the aggregate significance of the evidence and a pooled relative risk estimate with associated confidence limits is calculated. The chapter concludes with a 12 ------- review of potential biases affecting interpretation of these studies, with an emphasis on misclassification. The roster of studies selected represents those available through the time at which this draft report was released; in re- vising the draft, consideration should be given to substituting the data from the recent report by Janerich et al. (1990) for the earlier analysis of these same data (Varela, 1987). The features of the individual studies are adequately reviewed. The discussion of bias (Section 3.5) needs expansion and some consideration of types of bias other than misclassification. The types of bias potentially affecting any epidemiological study include selection bias, information bias (which includes both differential and non-differential misclassification), and con- founding bias. Selection bias, particularly likely to affect case- control studies based on cases and controls derived from specific institutions, should be addressed. The possibility of confounding bias merits review because of evidence that smokers are increas- ingly distinct from nonsmokers in socioeconomic characteristics that may have implications for health. Thus, those more highly exposed to ETS may differ from those less exposed in other relevant characteristics. Confounding, however, is an unsatis- factory explanation for the general pattern of the reported stud- ies, with the majority showing increased risk. These studies have been conducted in a wide variety of locales with consistent find- ings of positive association; this consistency weighs against con- founding as an explanation for the increased risk associated with marriage to a smoker. The discussion of misclassification should be expanded to include studies that have addressed the quality of information on passive smoking derived from questionnaires as well as the relation between questionnaire—based measures of exposure and biological markers of exposure. The 1990 Report of the Surgeon General includes reviews of the quality of information on smoking from surrogate respondents as well as of the validity of self- report of smoking history; this recent report should be considered and cited in the discussion of misclassification. Two major cohort studies providing evidence on passive smoking and lung cancer have been published; a lengthy discussion is provided concerning the comparative findings of the two studies. Unfortunately, we lack information on the comparative exposures to ETS of subjects in the two populations. For both studies, follow- 13 ------- up periods began well before present methods for atmospheric moni- toring or for assessing biological markers were available. Thus, arguments concerning the possibility that passive smoking is more “direct” in Japan are speculative and the lack of data should be cited and uncertainty added in drawing any conclusion concerning the comparative levels of exposure in the U.S. and Japan. The discussion of the two studies should be markedly shortened. In summary, Chapter 3 of the May 1990 External Review Draft provides a generally adequate review and assessment of the epi- demiologic evidence on ETS and lung cancer in never smokers. A complete hazard identification is not conducted, however. The chapter needs to be expanded to address more fully the toxicology of ETS and the evidence on active smoking and lung cancer. This expanded review, coupling more closely the evidence on biological plausibility that ETS is a carcinogen with the supporting epidemi- ological evidence would adequately support the conclusion that ETS is a Group A carcinogen, a determination that should be moved from the quantitative risk assessment (current Chapter 4) to Chapter 3. The Committee accepts this overall conclusion, in spite of the limitations of the current chapter; a more comprehensive review as suggested by the Committee should strengthen the determination that ETS is a Group A carcinogen. 3.2 Chapter 4--Assessment of Lung Cancer Risk from ETS In reviewing published quantitative risk assessments, Chapter 4 of the review draft properly dichotomizes the approaches that have been taken -— the cigarette-equivalent approach, and analyses of epi- demiologic studies in which the excess lung cancer risk in non- smokers is observed as a function of exposure to ETS. As indicated in Chapter 4, there are serious difficulties in both of these ap- proaches. The cigarette-equivalent approach has the great ad- vantage that it is based on relatively abundant and consistent relative risk (RR) determinations in active smokers, which can be used to project the risk in non-smokers exposed to ETS in the form of a percentage of the risk in active smokers. The assessment of the cigarette-equivalent in non-smokers due to exposure to ETS has a considerable level of uncertainty embedded in it. The ratio of sidestream (SS) to mainstream (MS) emissions is highly variable among the components of cigarette smoke, so that the number of cigarette-equivalents to which a passive smoker is exposed varies greatly with the compound used as a marker or expo— 14 ------- sure surrogate. Neither cotinine concentration in body fluids nor the measurement of tobacco smoke particulate matter can be used with great confidence for quantitative assessment of the carcino- genic potential of ETS. There is a suggestion that the uncertainty in exposure assessment for either approach is about an order of magnitude (It should be noted here that the Guideline for Carcino- genic Risk Assessment anticipates that numerical risk estimates will have no more than one significant digit). Neither cotinine nor smoke particulate matter levels are direct indicators of car- cinogenic components. The other type of assessment is based on inferences from the epidemiologic studies of the association of exposure to ETS and carcinogenic risk in non-smokers. Since spousal smoking is a very important exposure proxy used in many studies, the utility of a categorical classification (married to a smoker/not married to a smoker) for quantitative exposure assessments needs to considered. Physical proximity of smokers to non-smokers), daily length of exposure, and exposure outside the home to ETS may be quite dif- ferent in different cultures and over decades of time. Misre- porting of smoking status for cases and controls in these studies may also introduce a bias. Various attempts have been made to apply corrections for these sources of bias. These attempts re- quire further assumptions and are based on limited data available on misreporting rates and cotinine concentrations in various groups. The assessment presented in the last section in Chapter 4 of the Review Draft does not appear to be in conflict with procedures established in other reviews, and states all assumptions made in the quantitative assessment with considerable care. The results are given in too many significant figures however. 3.3 Chapter 5--Environmental Tobacco Smoke and Respirator? Disorders in Children The Committee recommends that this chapter be re-organized to reflect directly the biological effects of Passive/Involuntary smoking (This terminology reverts to the orig- inal discussion of exposure-response and its impact, as per the Surgeon General reports of 1982-1986). The chapter should treat Jfl utero exposure as a precursor to extra uterine/post birth/childhood effects, and not as a confounder. It should also be extended to address effects in adults, since the sequelae of effects in chil- 15 ------- dren, as well as direct effects in adults, naturally proceed from the discussion of effects in children. The Committee’s recommendations as to the structure for a revised chapter follow. In outline, the proposed sections would be: an Introduction; 1) Biological Mechanisms; 2) Exposure and its Assessment; 3) Annoyance and Irritation; 4) Acute Illnesses (middle ear, upper respiratory, lower respiratory); 5) Chronic Respiratory Systems and exacerbations of chronic obstructive pulmonary disease (COPD); 6) Asthma; 7) Pulmonary Function; and 8) Health Hazard Assessment. The Introduction of the Chapter should include some reference to the overall problem of acute and chronic respiratory diseases, and the potential attributable risk of ETS/passive smoking (Chronic respiratory disease is the fifth leading cause of death, with an age—adjusted death rate of 15.7/1O in 1985; it is increasing still. Acute lower respiratory disease is the sixth leading cause of death, with an age—adjusted death rate of 13.4/10 in 1985. The prevalence rates of related conditions are signifi- cant——the rate for asthma, for example, is 4.1/102 (NCHS, 1986). Acute respiratory disease is the leading cause of morbidity and disability in the U. S. (as per NCHS)). The first section of the Chapter could be called Biological Mechanisms (5.1). It should discuss the biological plausibility of the respiratory responses (akin to the discussion of carcinogen- icity), and a brief discussion of the comparable response to active smoking. Such topics as irritant responses to pollutants (as are found in ETS) have been discussed at length in EPA Office of Research and Development and Office of Air Quality Planning and Standards (OAQPS) documents (e.g., National Ambient Air Quality Standards (NAAQS) Criteria Documents and Staff Papers, and National Emissions Standards for Hazardous Air Pollutants (NESHAP) Criteria Documents. It should start with the effects of in utero exposure effects as precursors to childhood effects (5.1.2.): reduced fetal oxy- genation, poorer lung (and brain) development, low birth weight, immunological and biochemical effects (e.g., changes in T cells and immunoglobulin levels, changes in prostaglandin regulation, pro- 16 ------- tease inhibitors) (Amer. Acad. Peds. J. Peds., 1976; ALK report, 1983; Wall et al., 1985; Rantakallio, 1978; McIntosh, 1984.; Tager, 1988). The next discussion (5.1.3) could be of the effects of low lung function at birth (due to genetics and j utero exposure) as a precursor of lower respiratory infections; poorer lung develop- ment (disposing to greater effects of ETS on lung growths) (Mar- tinez et al., 1989; Sherrill et al., 1990; Lebowitz, 1991), as discussed at the Committee’s meeting). There should be a discussion (5.1.4) of the potential re- duction in host defense mechanisms due to ETS (going further than that induced by in utero passive smoking), which is of a similar nature (though a different dose) to that induced by active smoking. This topic relates to the increased predisposition to and prev- alence rates of acute illnesses (middle ear, upper respiratory, lower respiratory, other exacerbations of chronic respiratory disease/chronic obstructive pulmonary disorder/airway obstructive disease (COPD/AOD)). The next logical discussion (Section 5.1.5) could concern how these experiences could lead to chronic respiratory disease (e.g., chronic cough, persistent wheeze) in childhood, and how the se- quelae of such would be chronic respiratory disease in adult life. The pathophysiological and anatomical mechanisms would be featured, and some discussion of biochemical mediators would occur. There should be a discussion (5.1.6) of the biological reasons why ETS would produce or exacerbate bronchial lability and respon- siveness (BR) (coupled to the lower airway caliber, and possibly genetics, discussed above), and how this BR, especially in conjunction with increased Iminunoglobulin E and lower respiratory infections (both discussed above) could lead to childhood wheezy bronchitis and asthma (Tager, ARRD, ‘88; 138:507; Burrows and Martinez, ARRD, ‘89 140:1515). The role of atopy (also genetically regulated) in this process should be discussed (ibid.). The role of these above-mentioned conditions (BR, bronchitis, asthma/persistent wheeze) on decreased lung growth (Sherrill, op cit.) should be discussed. This could lead into a general discussion (5.1.7.) of why lower lung function is related pathophysiologically to ETS, starting with deceased lung growth in 17 ------- utero (see above), and how this would lead to low lung function/COPD in adults (Tager et al., 1987; Lebowitz et al., 1987). Appropriate biochemical and anatomic mechanisms would be discussed. Unless the relevant material is covered generically in an earlier chapter, Exposure and its Assessment could be covered in a seperate second section (5.2). Such a section would cover reported exposures, measurements of indoor nicotine, other related pollu- tants (PM, CO , etc.), and biological markers such as cotinine (Jarvis et al., 1987). Factors associated with ETS (cf., Sandier et al., 1989) should be presented as relevant. Other methods of exposure estimation and assessment should also be discussed, including utilization of models based on source characterization from chamber studies. The relationships between reported exposures and monitoring results, and biases in reporting (cf., Friedman et al., 1983) could be discussed also. Confounding needs to be discussed (Quackenboss et al., 1989; Lebowitz, 1990) as well (other key references include National Research Council (NRC) 1981; World Health Organization (WHO) 1982; NRC 1986, and WHO/EURO Proceedings on Indoor Air 1984, 1987, and 1990.) Further, exposure-dose estimation would be presented (Hiller et al., 1982). The third section (5.3) should address Annoyance and Irri- tation. Annoyance is important per Se , and annoyance also affects subjective reports (Lebowitz, 1989; Department of Health, Educa- tion, and Welfare 1971; National Institute of Occupational Safety and Health 1971; National Clearinghouse on Smoking and Health 1976; NRC 1981, Surgeon General’s Report 1986) (Odor topics should be included in this section, introducing the concept of sensitive individuals). Irritation effects are well-documented (Weber, and Hugod, 1984), and occur more quickly at lower doses in those more “sensitive” (ibid., op cit). Acute irritant symptoms should be a major topic. This section should document such effects in children and adults, and differentiate irritant from infectious and allergenic effects. The fourth section (5.4) should cover Acute Illnesses . This includes middle ear effusions (5.4.1), upper and lower respiratory illnesses (including such exacerbations of COPD) (5.4.2—.4), and sequelae of lower respiratory illnesses. The effects of LRI’s on lung function (e.g., Yarnell and St. Leger 1979), and the possi- bility of LRI’s leading to asthma in children (Gregg, 1973) war- 18 ------- rants discussion. The effects of LRIs leading to asthma in children (Gregg, 1973) should also be discussed. The effects of LRIs also includes AOD in adults (the overall effects of childhood respiratory troubles (CRT)), especially as documented in longi- tudinal studies (Lebowitz et al. 1987, 1988: and Sherrill et al. 1990); this could be a separate section (5.4.6). Airway obstructive disease should be addressed in a major section because of the well-documented effects of ETS on lower respiratory tract illnesses, and the above-mentioned sequelae. The EPA report did a reasonable job in this area, and can be expanded (by incorporating some of the comments included in the reviews by Samet and Lebowitz). Further discussion of biases in reporting are available (Colley 1974; Cederlof and Colley 1974), as well as for confounding by other exposures (Hammer et al. 1976; Anderson, 1979; Speizer et al. 1980; Comstock et al., 1981; Melia et al. 1982; Koo et al. 1988), and interactions with other exposures (Lebowitz et al. 1989 and 1990). There should also be discussion of the concurrent effects of breast feeding and socio-economic status (Martinez et al., ‘89, ‘90). The fifth section should cover Chronic Restiratorv SvlnDtoms (and increases of symptomatology in COPD as exacerbations). Some of this topic was covered in the EPA report, but it could be improved by clarification and expansion (see reviews provided). Again, biases in reporting, confounding by other exposures, and interactions of exposures producing responses (ibid., op cit.) warrant discussion. Effects of active/self-smoking interacting with passive smoking should be discussed (Bland et al., 1978; Lebowitz et al. 1987 and 1988). Effects of family history (ibid., Schilling et al. 1977; Weiss et al. 1980) should also be covered. Sequelae (op cit.) could be discussed as well, and direct effects in adults also (Coastock et al. and Schilling et al., ibid.; Lebowitz and Burrows, 1976: Schwartz and Zeger, 1990). The sixth section (5.6) should cover all the aspects of Asthma . This section is one of the most important, and was one that was insufficiently discussed (in all, aspects) in the EPA report. It needs to discuss genetic and in utero aspects, the evidence for bronchial responsiveness, high IgE and atopy related to ETS (Weiss et al. 1983 and 1988; Tager 1988; Burrows & Martinez 1989; and Lebowitz et a].. 1989 and 1990, op cit.). The effects of social status, breast feeding, other exposures, and sequelae should 19 ------- be discussed as well (op cit.; Rantakallio 1979; Sherman et a l. , 1990). Other studies, in press, could also be used (e.g., Martinez et al. , Peds.). Exacerbations can be studied as well, including those seen in field studies (Quackenboss et al. and Lebowitz et al. 1989—1991, op cit.) and chamber studies (Shephard et al. 1979; Dahms et al. 1981; Stankus et al. 1988; Danuzer et al. 1991). The seventh section (5.7) should cover effects on Pulmonary Function alluded to above, in the EPA report, including all the previous effects (and confounders, etc.) discussed above. It should be more precise, and include amount of change found. There could be a section at this point covering other, miscellaneous topics, as in the previous report. Alternatively, these topics could be put into other sections. The eighth section (5.8) would be a Health Hazard Assessment , which would include attributable risk and population impact. Further discussion is needed concerning the initial aspects of the section. 3.4 A endix A--Summary DescriDtions of Eleven Case-Control Studies The Committee agreed that Appendix A made a valuable contribution to the document, and that it should be included in the final draft. Much of the information contained in the appendix might be more useful however, if it was organized as a series of tables rather than a running text description. For example, a table that described the important characteristics of the study, e.g., population size, number of lung cancers, measure of ETS exposure used, characteristics of the control population and criteria for selecting the cases. Other tables might include potential biases addressed or not addressed in each of the studies and smoking characteristics in the background and control populations. 3.5 Appendix B—-Mathematical Poraulas and Relationships The Committee agreed that Appendix B is important to the overall report. However, in its present form, it contains several errors, both typographical and substantive, which should be corrected. Moreover, its format is difficult to read, and it is incomplete. The Committee recommends that Appendix B be restructured and rewritten in a more “reader friendly” style to include, as a minimum, the following three sections: 20 ------- 1. An Introduction to describe the purpose of the Appendix and its objectives. 2. An overview of the Mantel-Haenszel procedure that was used in this report for the rneta-analysis. This section should also include a rationale for the selection of this pro- ceedure, rather than the method used in the previous NRC report. Appendix B in the NRC report is suggested as a guide for the presentation of this material. 3. Description and derivations of the risk assessment equa- tions used . A rationale for the specific equations, a discussion of the validity of these equations for case- control as well as cohort studies, and explicit assump- tions pertaining to the equations should be included in this section. The Committee also suggests that the deri- vations be presented in a systematic format for ease of reader understanding: a. A new section on the derivation of the unadjusted relative risk equation. (This could be incorporated with item 2, above). b. A revised section on the derivation of the relative risk equation adjusted for misclassification (Bi). c. A revised section on the derivation of the relative risk equation adjusted for background exposure (32). d. A revised section on the derivation of the equation to estimate the population-attributable risk (83). Specific written comments pertaining to Appendix B were submitted by several Committee members. These comments identified several errors, typographical and substantial, in the equations. The Committee therefore recommends that these errors be corrected and that the results based on these equations be carefully reviewed before final publication. For example, it was acknowledged at the Committee meeting in December that the correction of one such error resulted in a slight downward shift of the predicted annual lung cancer deaths due to passive smoking. 21 ------- 3.6 Appendix C--Dosimetry of ETS The Committee agreed with EPA staff that Appendix C should be deleted in its present form. Some of the issues that are addressed in Appendix C should be addressed in at least one or more of the new chapters, but in a format appro- priate to a chapter, rather than a format considered appropriate to an appendix. When incorporating the discussion now located in Appendix C into appropriate text chapters, it should be noted that the Ap- pendix, as written, has serious technical errors and limitations. It is seriously deficient in that it focusses entirely on carcino- gens and their dosimetry in healthy adults. This is inadequate even when the endpoint of concern is lung cancer, as evidenced in the recent report of Janerich et al. (op. cit.) on the association between lung cancer in adults and their childhood exposure to ETS. It is even more inadequate in that it ignores the respiratory disorders in children that are reviewed in great detail in Chapter Five. The whole section C-5, “INTERNAL ORGAN BURDENS FOR THE LUNG,” is based upon a simplistic set of models and assumptions that produce regional lung retention times and dose estimates that are truly fanciful. It correctly states “that removal from the tracheo-bronchial region generally may be characterized by two phases. The first is a rapidly cleared phase, dominated by par- ticles deposited on the mucus of the upper passageways. The second is dominated by particles deposited on the slowly moving mucus of distal passageways.” However, the calculated half-times (C 1 and C 2 ) of 450 and 710 minutes, respectively, for the fast and slow phases of ETS particle clearance, differ considerably from the actual radio-aerosol study data on which the model is supposedly based, and which show much faster rates. The literature on the effect of active smoking on particle deposition and clearance rates is misinterpreted. One study of Albert et al. (1975) concerning the short-term effects of smoking on overall tracheobronchial clearance in humans, and one study by Wanner et al. (1973) on the effects of chronic smoke exposure on mucociliary transport velocity in the trachea of the dog are cited as a basis for doubling the retention times of particles on the tracheobronchial tree as a whole. The conclusion was drawn that smoking has no effect on the regional deposition of particles. The reality is quite different. Smokers have much greater tracheo- 22 ------- bronchial deposition than nonsmokers, and the short-term effect of smoking is to greatly accelerate the clearance of particles deposited on the tracheobronchial tree. It is curious to note that the C 2 value calculated for the second “fast-phase” component in smokers is 1400 mm (23.3 hours), while a 17 hour half-time was used for the “slower” alveolar region half—time in the dose calculations. 3.7 A andix D--Alternative Approaches for Estimating the Yearly Number of LUng Cancer Deaths in Non-Smokers Due to ETS Based on Dose-Response Modeling The major purpose of Appendix D is to bolster the risk assessment document. Much of the data referenced in this appendix provide further evidence of the carcinogenicity of environmental tobacco smoke, and should be clearly presented in this light. For instance, the Grimmer study clearly demonstrates the lung carcinogenicity of ETS in animals. Other sections are currently incomplete and point to future directions for research. While interesting, these sections are not as supportive of the main document and may be distracting. The first two approaches for deriving ETS dose-response models, the relative potency approach and the cigarette equivalent approach, share an implicit assumption that particle phase com- pounds, and polynuc].ear aromatic hydrocarbons in particular, are the carcinogens of interest. Other carcinogens have been identi- fied in ETS, and many of these are in the vapor phase, e.g., ben- zene, vinyl chloride, formaldehyde, and several N-nitrosamines. To the degree that vapor phase carcinogens have been ignored, or incompletely collected or extracted for administration in animal experiments, the potency of ETS has been underestimated. Further- more, the use of benzo- [ a)-pyrene as a reference standard of human lung cancer is highly problematic and should be reconsidered. The uncertainties in the relative potency approach are too great to support the derivation of an ETS dose-response model that would be an improvement over any that can be calculated from epi- demiologic data. The relative potency in animals is not rieces- sarily the same as the relative potency in humans, especially to the degree that metabolism may be involved. Furthermore, the data do not exist to support calculating the relative potency by a straightforward comparison, e.g., ETS and compound X in animal system A, compared with a known dose-response relationship for compound X in humans, so other intermediate comparisons are re- 23 ------- quired, such as compound X and mixture Y with different routes of exposure, on different animals, and with different tumors. The uncertainties of each step quickly overwhelm the uncertainties in the epidemiologic studies. The complexities of tobacco emissions complicate the cigarette equivalent approach. The referent mainstream emissions should be those of unfiltered cigarettes, upon which most of the active smo- king epidemiologic data is based. The variable ratios in side- stream to mainstream emissions of toxins lead to differences in the calculated cigarette equivalents to which a passive smoker is exposed, and these may range over two orders of magnitude (see Hammond, 1990). These different emission ratios are a source of variability in the ratio of biomarkers in smokers and nonsmokers. (Metabolism rates are another potential difference.) Thus, co- tinine in nonsmokers is typically less than 1% the level found in smokers, while the median level of 4-aminobiphenyl hemoglobin ad- ducts in nonsmokers was 14% the median in smokers (Hammond et al, 1990). This corresponds to the emission ratios of these compounds, which differ by a factor of 15. Russell and coworkers (1986) (page 4—19) based their estimates of the risk of premature death from passive smoking on the ratio of cotinine in passive smokers to that in active smokers, 0.007, and assumed the same ratio held between premature deaths in passive and active smokers. The use of 4- aminobiphenyl hemoglobin adducts instead of cotinine to estimate relative exposures would have led to a higher predicted premature death rate due to passive smoking. Several studies have been conducted on the deposition of MS and SS particles in the human lung. These should be discussed rather than relying solely on biomarkers, where the exposures are not known. A few caveats are required regarding the use of DNA adducts to estimate dose. DNA adducts are subject to repair mechanisms, and the rate of repair may differ in smokers and nonsmokers. Since nonsmokers have very different exposures to ETS, one expects a wide range in the ratio of adducts in smokers and nonsmokers. A disadvantage in the exclusive use of DNA and protein adducts as biomarkers of dose is that such markers are available for only a few suspected agents. The use of benzo [ a]pyrene (BaP) DNA adducts is further complicated by the many other sources of BaP, including diet and various combustion products. 24 ------- Determination of the dose-response effect of ETS based on the epidemiologic studies of Hirayama would be most valuable. The data gathered by Hirayama and colleagues could have been greatly enhan- ced and more generalizable if measurements of ETS levels (especial- ly respirable particle and nicotine concentrations) had been taken in Japanese homes with varying amounts of smoking. Some of the methods used in Appendix D might be useful in estimating the importance of ETS in respiratory diseases in children. Appendix D has information which is supportive of the main document. Some of this information can be improved some is suggestive of future research directions. The release of the final document should not be delayed for these data. Finally, the data may be best incorporated into the relevant sections of the main document, rather than exist as an independent appendix. 3.8 Appendix E--Summary Descriptions of Twenty-Six Studies on Environmental Tobacco Smoke and Respiratory Disorders in Children The Committee concluded that an Appendix E, similar to the Appendix A, should be included in the revised document. As before, the Committee recommends that the information by organized as a series of tables rather than as a running text description with a similar format. 25 ------- 4.0 REVIEW OF THE POLICY GUIDE The Policy Guide was not prepared as a scientific document, but its recommendations are based upon summary statements of scientific knowledge. On this basis the Committee was asked to examine whether the Guide’s statements on health contained within the first 20 pages were scientifically defensible. The Committee did note that there is much technical content in other sections of the Policy Guide, including technical statements on ventilation, room and building ventilation. For the most part, the scientific data and interpretations contained in the draft Policy Guide were appropriate, but there were some notable exceptions -— an incorrect definition as to what constitutes a small particle, an erroneous statement as to the depth of penetration of mainstream smoke vs. sidestream smoke, and a misstatement of the current particulate matter NAAQS, to cite a few. Furthermore, there were statements about cardiovascular mortality, cancers at other sites, and aggravation of cardiovas- cular and respiratory disease that were not addressed in the ETS Risk Assessment. Thus, without having any supporting documenta- tion, the Committee could not endorse these statements. The Policy Guide draft will need to be revised to reflect the changes being made in the Risk Assessment. If the Committee is to review the Policy Guide again, it should be sent to the Committee with a supporting document that explicitly states the technical basis for each of its summary statements on the state of scientific knowledge. 26 ------- 5.0 SPECIFIC REVIEW ISSUES 5.1. Luna Cancer in Adults The Committee noted that Chapters 3 and 4 addressed the issue of lung cancer risk due to spousal smoking only for non-smoking women. It is suggested that the revised doc- ument be expanded to include the full range of cancer impacts of ETS. The Committee also noted a number of areas where substantial improvements could be made in the organization of the document, as well as in its content -- some material was not adequately covered or not addressed at all. The Committee urges the EPA to redraft those chapters as well. Some specific suggestions follow. 5.1.1 Carcinogenicitv of ETS The Indoor Air Quality and Total Human Exposure Committee concurs with the finding of the draft report that Environmental Tobacco Smoke (ETS) should be classified as a Class A Carcinogen. The Committee believes, however, that the case could be made more persuasively than does the current draft document. Part of the difficulty may be found in the language and the rationale of the Guidelines for Carcinogen Risk Assessment as they are currently formulated (51 FR 33992, August 22, 1986). The Guidelines address the case of a single chemical compound which may contain contaminants or impurities. The process envisioned in the Guidelines consists of Hazard Identification and “. .should include a review....to the extent that it is available” of: 1. Physical-Chemical Properties and Routes and Patterns of Exposure 2. Structure-Activity Relationships 3. Metabolic and Pharmacokinetic Properties 4. Toxicologic Effects 5. Short-Term Tests 6. Long-Term Animal Studies 7. Human Studies In the Guidelines, the Long-Term Animal Studies section is covered in 25 column-inches, and the Human Studies section takes up about seven column—inches, an indication of the emphasis on long- term animal toxicology studies. The evidence for the carcinogenicity of tobacco smoke is not based on long-term animal studies, which are negative. In this case, the strongest evidence is that obtained in a large number of epidemiologic studies of active smoking and lung cancer. The causality of the connection between inhalation of tobacco smoke and excess risk of lung cancer cannot be in doubt. It has been 27 ------- demonstrated that cessation of inhalation of tobacco smoke leads to a reduction of the excess lung cancer risk. The risk has been shown to be proportional to the amount of smoke inhaled. In ter ns of overall impact it has been shown that a very high proportion of the observed lung cancer incidence is due to inhalation of tobacco smoke. If the Guidelines for Carcinogenic Risk Assessment can be used to cast doubt on a finding that inhalation of tobacco smoke by humans causes an increased risk of lung cancer, the situation suggests a need to revise the Guidelines. The inhalation of ETS by children, by non-smokers or former smokers represents a risk that is much smaller than that experi- enced by active smokers, but it is an involuntary exposure. It is not uncommon to derive quantitative risk assessments of exposures to carcinogens from data obtained in more heavily exposed occupa- tional populations, and in that sense smokers represent a more heavily exposed population, providing data for extrapolation to the lower exposures imposed on children and adult non-smokers. There are both differences and similarities in the charac- teristics and the composition of mainstream smoke, sidestream smoke, exhaled tobacco smoke and environmental tobacco smoke. It is important to deal both with the differences and similarities as they might affect the quantitative risk which is most accurately known for mainstream smoke. The difference in carcinogenic po- tential is not such that any one of these other categories could be considered as non-carcinogenic in humans. The very clear carcino- genicity of mainstream tobacco smoke directly implies carcinogen- icity of ETS, particularly in view of the similarities in chemical composition and sizes of particulates between mainstream and sidestream smoke. Meta—analyses of epidemiologic studies in non—smokers and for- mer smokers are sensitive to decisions about exclusions and inclu- sions, and are primarily oriented towards increasing the overall statistical power and more precisely describing risk. Such analy- ses cannot effectively take into account any differences in quality of the study, differences in the way exposures were determined or classified, etc. Biases will be reduced only in that they will be averaged. 5.1.2 Spousal Bmokin All of the studies cited in the report on ETS and risk of lung cancer have made observations on married women 28 ------- who have been classified as “never-smoking”. Those married to a smoker are assumed to be exposed to greater levels of ETS than those married to a nonsmoker (p. 3 —l2). As noted in the report, this relative risk comparison implicitly compares women exposed to both spousal and other ETS to those exposed to other ETS only. The ideal measure of ETS exposure for lung cancer studies would include all sources of ETS with data on both dose of ETS and exposure over time for a lifetime, or at a minimum over the past 20 to 30 years. Spousal smoking is believed to be a useful and valid marker for ETS exposure because (1) it often indicates many years of exposure (this contrasts with biological markers such as urinary cotinine, which indicate exposure at only one point in time); (2) the level of ETS exposure in the home when the spouse smokes ap- pears to be greater in magnitude than the exposure from other, non- domestic, sources. Several studies exploring urinary cotinine as a measure of ETS exposure have found higher levels in non-smokers married to smoking spouses than to those married to nonsmoker spouses. The statement that ETS in the home is greater than that of other ETS exposures may be more or less true according to a variety of factors as noted below. The use of spousal smoking data is highly attractive because such data are easy and inexpensive to collect. For most studies spousal smoking is the only available measure of ETS exposure. There are potential limitations in the use of spousal smoking as an indicator of ETS exposure that need to be considered: 1. Spousal smoking may account for a relatively small propor- tion of lifetime ETS exposure. Janerich et al. (1990) est- imated that spousal exposure accounted for only 30% of lifetime exposure. These authors computed correlation co- efficients of 0.37 and 0.51 between spousal smoking and lifetime ETS exposure for men and women, respectively. In this study childhood exposure was a major source of hf e- time ETS exposure and correlated more highly with lifetime exposure. Likewise, Cummings et al. (1989) found little relationship between childhood, adult home and work place ETS exposure. On the other hand, Thompson and co-workers (1990) found that non-smokers who lived with a smoker re- ported more ETS exposure outside of the home than those who did not live with a smoker. In this way spousal smoking could be a more general indication of ETS exposure than 29 ------- expected on the basis of exposure in the home , but sensitivity for total ETS exposure may vary among different study populations. 2. The results of comparing ETS household exposure to ETS household plus other exposures may vary in different coun- tries and different regions within the U.S. Exposure within the residence depends on size and the type of con- struction of the dwelling, the amount of ventilation, and the proximity of smokers and nonsmokers within the home. Non-domestic background exposure varies with the nature of their workplace exposures, the extent of smoking restric- tions in the work place and public places, the climate and the time of the year. With respect to the latter, exposures as assessed by urinary cotinine concentrations in Buffalo, New York were greater in the winter compared to the sununer, presumably due to more time spent indoors with less ventilation in cold weather (Cummings et al., 1989). Such differences would be expected to be less mar- ked in warmer regions of the country. For non-smoking peo- ple in particular, the extent of exposure outside of the home may depend on whether the woman works and how many other people in the population, who may be friends of non-smoking women, smoke. Thus, in countries such as Japan where fewer women work outside of the home, and fewer women in general smoke, spousal smoking may indicate differential exposure for women who are, and who are not exposed to ETS, than in the U.S. In any case, bias due to concerns (1) and (2) would decrease the difference in true exposure between the “exposed” and “non-exposed” non- smoking spouses, and would favor finding no difference in relative risk. These issues may explain some of the vari- ability found in relative risk for lung cancer with ETS exposure in different countries around the world. 3. As noted previously, a major source of ETS exposure is that incurred in childhood, which could contribute to increased lung cancer risk in an adult. Although not generally spe- cified in quantitating the risk of having a smoking spouse, it is possible that a person whose parent(s) smoked (and therefore who was exposed to ETS as a child) is more li- kely to marry a smoker. In this case the risk of ETS might 30 ------- reflect the risk of combined childhood and spousal expo- sure rather than just exposure to the spouse. 4. The use of spousal smoking as an indicator of exposure may amplify the risk of misclassification of smokers as non—smokers. There appears to be a concordance between spousal smoking and false reporting of current or former smoking status. The misclassification of smoking status would falsely increase the relative risk of lung cancer in non—smokers related to ETS exposure. The misclassifica- tion issue is considered in detail in the report and appro- priate corrections have been made for misclassification. 5. Spousal smoking status could be associated with several sources of potential confounding. For example, it is pos- sible (although not documented by specific studies) that the presence of a smoking spouse is associated with an in- creased likelihood of lower socio—economic class, dietary differences, more alcohol or other drug exposure, more exposure to air pollution, etc. Such factors could possi- bly increase the risk of lung cancer, and published epi- demiologic studies have addressed these factors to varying degrees. The potential sources of confounding based on spousal smoking status should be discussed in the report, with a recommendation that future studies explicitly ad- dress these issues. In summary, considering its various limitations as an indi- cator of ETS exposure, spousal smoking status seems to be a reason- able method of identifying people with greater, versus lesser, ETS exposure. The problems in not accounting for background exposure would, if anything, bias against finding increased risk of lung cancer. Bias related to misclassification associated with smoking status has been addressed and corrected for in the draft report. There are possible confounders related to spousal smoking status, but such confounding concerns are present in other surrogates of exposure as well study. The importance of these confounders has not been determined to be sufficient to alter the conclusion that ETS increases the risk of lung cancer. 5.1.3 United States and Foreign Studies The Committee felt that data from studies conducted overseas as well as in the United States should be utilized in evaluating whether exposure to ETS 31 ------- increases risk of lung cancer. It is appropriate to examine the totality of evidence from all the case-control and cohort studies, regardless of where they were conducted. The Committee commented that the text of Chapter 3 of the report seemed to overemphasize the Japanese cohort study, but felt that this and other non-U.S. investigations were directly relevant to establishing that ETS is a carcinogen for lung tissue. Given the variety of study settings and the potential for differences in exposure to ETS between (and even within) countries, it is not surprising that relative risks vary from study to study. The higher relative risks found in some studies outside the United States may in part be related to differing characteristics of exposure to spousal smoking, differences in background ETS levels, or still other variables. The Committee believes that the report should recognize such potential differences, although adjustment for them may be precluded by lack of detailed ETS exposure data in the various studies. We do not disagree with the draft report’s approach of incorporating data from around the world in estimating the numbers of lung cancer deaths in this country due to ETS, but believe that the estimates should be interpreted cautiously. In this regard, we recommend that the assumptions used, and their accompanying uncertainties in estimating numbers of lung cancer deaths attributable to ETS, be underscored. 5.1.4 Use of Meta—Analysis Meta—analysis is an appropriate tool to summarize the epidemiological studies investigating the risk of ETS. However, the priority given the meta-analysis in this report in attempting to demonstrate that ETS is causally associated with lung cancer is not justified. Evidence on the carcinogenic effect of active smoking, the presence of carcinogens in ETS, and pre- dicted lung cancer risk of low dose exposure to tobacco smoke from appropriate models, are an important part of establishing a causal relationship. The meta-analysis could then be interpreted as showing the available epidemiologic evidence is consistent with a small elevated risk. Meta-analysis is a general term applied to a wide range of techniques whose objective is to synthesize findings across related studies. Although, there is still considerable debate over many aspects of conducting a meta-analysis, several criteria are usually considered essential. These include: 1) clear statement of the objective of the meta-analysis; 2) precise definition of criteria 32 ------- used to include (or exclude) studies; 3) critical review of studies included in the analysis; and 4) assessment of the effect of in- dividual studies on the analysis. Many of these points were not adequately addressed in the meta-analysis provided in the EPA document. The authors of the draft report did not provide a precise statement of the role of the meta—analysis. In regard to general methodology, there are several roles a ineta-analysis might play. Bangert-Drowns (1986) distinguishes five different types of meta- analyses depending on the question to be addressed. In the EPA draft report, the consistency of the various studies is addressed, an attempt is made to estimate overall risk, the possibility of heterogeneity of study results is considered, and geographic variation is discussed as a possible source of heterogeneity. Unfortunately, it is not clear which of these issues is the primary target of the analysis. If it was intended to address all four issues, they were inadequately covered. In regard to consistency of findings (which is probably the most important issue), the findings were not presented in the most appropriate way. Estimates with corresponding confidence intervals are the most generally acceptable method of presentation. If the intention was to investigate heterogeneity, then formal tests of heterogeneity should have been provided. If it was intended to address the hypothesized U.S./foreign difference, it would have been useful to test the difference in risk between the two sub-groups of studies. Specific criteria for including studies was not provided. The importance of this was reinforced at the Committee meeting when a reanalysis was presented on a different set of studies than those in the report. This resulted in a change in the overall risk est- imate. Decisions as to study inclusion should be made prior to analysis, based on clearly stated criteria. It is also desirable to evaluate the impact on conclusions of closely related, but ex- cluded, studies. Finally, in testing the hypothesis of an elevated relative risk across studies, the reliance on the measure of “x-nuinber of studies rejecting out of n” as the basis for the p-values seems somewhat arbitrary and inefficient. It would be preferable to use the sum of the S-statistics given in the report as a test statistic. 33 ------- 5.1.5 ConfounderslMisclassification Important potential con- founders of the ETS—lung cancer relationship were addressed in the report mainly by carrying out a separate meta—analysis of those studies which included adjusted analyses. The variables included in these adjusted analyses were age, education, and social class. Comparison of unadjusted and adjusted Rrs in those studies which present both, suggests that these variables are relatively un- important. There is no way to evaluate the importance of occupation, radon exposure, and diet as confounders of the ETS-lung cancer relationship, or to adjust for them, since virtually none of the studies contain information on them. However, they could be mentioned in the text as potential confounders. The issue of misclassification should not be restricted to misclassification of current and ex-srnokers as “never smokers.” It should also be mentioned that non-differential misclassification of diagnosis (diagnoses other than lung cancer being incorrectly classified as lung cancer; or vice versa) will cause a biasing of the RR toward the null. The misclassification of smoking status is differential in that current smokers and (particularly) ex—sutokers are apt to be reported as “never smokers,” whereas the reverse is unlikely. The adjustment for misclassification of smokers as nonsmokers in the Report makes use of the formula used by the National Re- search Council for prospective studies, but no rationale or explan- ation for the formula is given in either Chapter 4 or Appendix B (Note also that several errors have been pointed out in the form- ulae given in Appendix B). Also, no distinction is made between prospective and case-control studies. In the latter, in order for bias due to misclassification of active smoking status to occur, there has to be differential misclassification between cases and controls. Finally, not enough attention is given in the draft report to possible non-differential misclassification of ETS exposure. This is an important issue, since marriage to a smoking spouse is an imperfect proxy for total ETS exposure. In the case of dichotomous exposure, such misclassification would have the effect of biasing the RB estimate toward the null. 34 ------- Other potential biases which deserve mention include recall bias (differential reporting of exposure status by cases compared to controls) and bias due to the use of proxy respondents. 5.1.6 Characterization of Uncertainties Vis-a-vis weight of evidence, the draft document’s conclusion that exposure to ETS sometimes leads to the development of lung cancer in humans rests upon two main arguments: (1) the biological plausibility of such a causal association is high, given the known effects of active smoking and the known composition of ETS; and (2) the accumulating epidemiologic evidence on the relationship between exposure to ETS and lung cancer appears to argue for a positive effect. With exposure levels that are usually quite low, it is not surprising that the association is weak in many studies and in the aggregate, although, given the size of the exposed population, societally important. Because the epidemiologic evidentiary base for drawing conclusions regarding ETS’s carcinogenicity consists mainly of studies of exposure levels produced by spousal smoking, the biological plausibility argument assumes great importance. Each step in that argument should therefore be carefully addressed, with the uncertainties encountered being spelled out explicitly. The biological plausibility argument depends upon establish- ing: (a) cigarette smoking’s known carcinogenic effects; and (b) ETS’s resemblance to mainstream tobacco smoke in terms of particle size distribution and composition of carcinogens, co-carcinogens and tumor promotors. (a) Cigarette smoking’s known effects . The document would benefit from a more complete presentation of the evi- dence concerning mainstream tobacco smoke’s role in causing lung cancer. More detailed consideration of the dose—effect relationship for inhaled tobacco smoke would better set the stage for presenting evidence concerning the biological plausibility that exposure to ETS has similar, albeit lesser, health effects. (b) ETS’s resemblance to mainstream tobacco smoke . The age— ing of tobacco smoke influences its uptake and deposition in the lung and its potential carcinogenicity. Nonethe- less, there are strong similarities in the chemical and n vitro biological activity of ETS and mainstream to- bacco smoke. These similarities should be discussed in 35 ------- the context of other complex mixtures, e.g., coke oven emissions, and diesel exhaust (The work of Lewtas at EPA should be revisited for this purpose). The uncertain- ties surrounding the evidence regarding changes in side- stream smoke composition should be assessed and the im- plications of such findings for the biological plausibil- ity argument should be spelled out more thoroughly. Epidemiologic evidence on the relationship between exposure to ETS and lung cancer should be described more completely, with the deficiencies of individual studies used to weight their contribu- tions to any conclusions that are drawn. The assumptions and un- certainties associated with each step of the risk assessment pro- cess ought to be explicitly stated. Not all the factors that probably contribute to the uncer- tainties surrounding the estimates of deaths attributable to ETS exposure are now considered. For example, it is important to justify the use of the particular biological marker chosen to estimate relative exposures (and, therefore, premature deaths) in passive versus active smokers, since that choice can cause the attributable deaths figure to vary over a twenty-fold range. Consequently, any estimate of the number of deaths to be expected each year from exposure to ETS should be justified more adequately than is now the case. A graphical presentation would clarify the uncertainties associated with each step as well as those inherent in the final estimate of attributable deaths. As noted in Section 3.2, the cigarette-equivalent approach has a great advantage in that it is based on relatively sturdy RR de- terminations in active smokers, which can be used to project the risk (in the form of a percentage of the risk in active smokers) to nonsmokers exposed to ETS. However, the assessment of the ciga- rette-equivalent in non-smokers due to exposure to ETS has a con- siderable level of uncertainty embedded in it, i.e., about an order of magnitude. Neither cotinine nor smoke particulate levels are adequate direct indicators of carcinogenic components. The other type of exposure assessment is based on inferences from the epidemiologic studies. Since spousal smoking is a very important exposure proxy used in many studies, there is concern about how usable this categorical, classification is for quanti- tative exposure assessments. Physical proximity, daily length of 36 ------- exposure, and exposure outside the home to ETS may be quite dif- ferent in different cultures and over decades of time. Misre- porting of smoking status in the cases in some studies also introduces a bias. Various attempts have been made to apply corrections for these biases. Both the relative potency approach and the cigarette equivalent approach share an implicit assumption that particle phase compounds, and polynuclear aromatic hydrocarbons in par- ticular, are the carcinogens of interest. However, other carcin- ogens have been identified in ETS, and many of these are in the vapor phase (refer back to Section 3.7 for a full description). To the degree that vapor phase carcinogens have been ignored, or in- completely collected or extracted for experiments, the potency of ETS has been underestimated. Another consideration is that the relative potency in animals is not necessarily the same as the relative potency in humans. The complexities of tobacco emissions complicate the cigarette equiv- alent approach. The referent mainstream emissions should be those of unfiltered cigarettes, upon which most of the active smoking epidemiologic data is based. The variable ratios in sidestream to mainstream emissions of toxins lead to differences in the calcu- lated cigarette equivalents to which a passive smoker is exposed. These different emission ratios are one source of variability in the ratio of biomarkers in smokers and nonsmokers. For example, cotinine in nonsmokers is typically less than 1% of the level found in smokers, while the median level of 4=aininobiphenyl hemoglobin adducts in nonsmokers was 14% of the median in smokers. Further- more, DNA adducts are subject to repair mechanisms, and the rate of repair may differ in smokers and nonsmokers. 5.1.7 Quantitative Risk Assessment The Committee generally agreed that the quantitative assessment of the risk of lung cancer due to exposures to ETS should be based on the human epidemiology studies and that meta-analysis was a suitable approach to combining the data. This approach is the most direct and makes the fewest assumptions. It should be noted that this approach is fully con- sistent with the risk assessments that have been done for many other carcinogens and that those assessments are generally based on fewer studies. 37 ------- Given that the epidemiology studies should be the basis of the risk assessment, some refinements of the risk assessment are recommended: 1. Criteria for Including Individual Studies in the Meta—analysis Criteria to include or exclude individual studies from the meta—analysis should be determined and explicitly stated (See section 5.1.4, preceding). The effects of individual exclu- ded studies on the quantitative risk assessment should be evaluated and discussed. The power of the individual studies should also be considered and discussed. 2. Adjustment for Smoker Misclassification The rationale for the formula used to adjust for smoker mis- classification should be given. Appropriate distinctions should be made in applying the misclassification formula to the case-control and the cohort studies. Because of the mar- riage aggregation factor -- the greater tendency for smokers to marry smokers -- the misclassification of some smokers as nonsmokers can artificially inflate the relative risk of lung cancer associated with passive smoking in cohort studies. In case control studies, misclassification by itself is not enough to inflate the relative risk. Differential misclassi- fication, with cases mis-reporting more frequently than con- trols, is needed. The assumptions used in adjusting for smo- ker misclassification and their effect on the adjustment should be more fully discussed. If the approach taken is con- servative, then it is noteworthy that the misclassification adjustment only lowers the relative risk estimate from a lit- tle over 1.4 to 1.3. 3. Misclassification of Exposure Some unexposed women, classified as un-exposed (non-smokers married to non-smokers) may in fact be exposed to relatively high levels of ETS in the workplace or in other settings outside of the home. Some recent of non-smokers’ exposure to nicotine indicate variations in exposures ranging over two orders of magnitude. Correction for “background” exposure does not adequately correct for this misclassification. Fur— thermore, the use of spousal smoking habits to classify ETS 38 ------- exposure status is more likely to misclassify American wo- men’s exposure than Japanese women’s exposure because of dif- ferences in American and Japanese lifestyles. Non-smoking American women married to non-smokers are more likely to be exposed to ETS outside of the home than are Japanese women be- cause more American women work outside the home and have friends who smoke. Some evaluation of the effects of these biases would be appropriate in the risk estimations. 4. Uncertainties in the Estimate of Annual Lung Cancer Deaths Due to Passive Smoking The uncertainty in the relative risk estimate of lung cancer due to passive smoking is based only on statistical consider- ations. There are other uncertainties that influence this estimate. A more critical analysis of the potential for systematic bias should be done. Acknowledging such uncertain- ties would provide greater balance to the report, while not substantially altering its overall message. 5. Dose-Response Estimation of Risk There are many more assumptions and uncertainties in any risk estimation made on the basis of dose-response or dosimetry than for epidemiologic data. Nonetheless, such an estimate may be of value if the assumptions are fully stated and the uncertainties in the estimate are quantitatively estimated. With uncertainty estimates explicitly included, this approach may well be consistent with that based on epidemiology. Ex- posure estimates for ETS should include the exposures from birth to age 15, not only from age 15 on up as is done for mainstream smoking. This can have a substantial impact on the estimated risk. Complex dosimetry models should be con- sidered the subject of research at this point in time, since they require many more assumptions. 6. Dose-Response Model for ETS-Radon Interaction Effects Development of a specific dose-response model for ETS-radon interactions is not recommended. The interactions of ETS with radon are numerous and involve both physical and biological interactions which are not fully understood at present. Fur- 39 ------- thermore, there are no relevant epidemiological data concern- ing such interactions. 5.1.7 Hone vs. Workplace Exposure The Committee recognizes that there is little epidemiologic literature on the health effects of ETS in the workplace, and its importance in relation to total ETS exposure. However, the report should review and comment on the data that do exist, if only to bring out the need for future re- search in this area. The report should also review and comment on the data that exists on exposure to ETS in public places. The Committee also recommends that EPA staff discuss possible approaches for estimating the exposure of children to ETS in homes with one or more smoking parents. This is recommended because of the potentially large public health impact of respiratory disorders in children that may be caused by exposure to ETS. Careful consideration should be given to the differences in the exposure parameters required for lung cancer as opposed to respiratory disorder assessments. For example, cancer assessment may require integrating exposures over longer time intervals than does the assessment of respiratory disorders. Besides developing approaches for estimating average child population exposures, it is also important to establish the shape of the exposure distribution, particularly the tail of the distribution, in order to determine whether a numerically significant subset of children is at high risk. 5.2 RespiratorY Disorders in Children Chapter 5 on respiratory disorders in children was a com- mendable first effort for a very difficult task. Nevertheless, we found that it could be substantively improved and that the conclu- sions drawn could be made much stronger if the chapter is revised in the manner suggested in Section 3.3. The Committee found the evidence for respiratory health effects in children to be stronger and more persuasive than stated in Chapter 5 of the draft ETS Risk Assessment document, and rec- ommends that the new draft contain a chapter devoted to quanti- tative risk assessment, in terms of the number of children at risk for various outcomes. It would be analogous to Chapter 4, which deals only with the evidence for lung cancer risk discussed in chapter 3. The risks are different, but it is possible that the 40 ------- impact of ETS on respiratory health in children may have much greater public health significance than the impacts of ETS on lung cancer in nonsmokers. There will need to be new material in the earlier chapters on lung dosimetry and the physical and chemical factors affecting it. The difference in deposition and retention of ETS components be- tween children and adults need to be recognized and considered in a risk assessment. 5.2.2. Weight of Evidence The scope of Chapter 5 is limited to selected studies published subsequent to the 1986 Surgeon General’s Report and the National Research Council Report. Neither of these reports judged the associations of ETS exposure and children with adverse respiratory effects to be causal; alternative explanations for the associations including confounding and information bias could not be excluded. The additional literature available since 1986 provides a basis for increased concern. Thus, the Committee urges a thorough review of the entire body of evidence. A consid- ered judgment cannot be made concerning causality without assessing the totality of the evidence including studies reviewed in the two 1986 reports and those published subsequently. In reviewing the weight of the evidence, the present Chapter 5 does not establish an appropriate framework for considering the data. The alternative explanations for association of ETS exposure with adverse respiratory effects need to be clearly listed (causality, confounding, information bias) and the individual studies reviewed for the approaches used to address confounding and information bias. The weight of the evidence could then be judged to determine the causality of associations. With regard to including the reviews of the 26 new reports as Appendix E, the scope of the review should be expanded to include all studies not in the 1986 Surgeon General’s Report. A more com- prehensive search is needed since the 26 publications identified by the chapter’s authors do not represent all significant publications on the effects of ETS on children published since 1986. 5.2.2 Confounders A number of confounders were mentioned by the report, but addressed improperly. These include utero exposure, 41 ------- parental reporting bias, and active smoking. The Committee con- siders the following factors to be critical: Unreported Smoking Other Exposures (outdoor) Other Indoor Pollutants Parental Symptoms Biological Precursors Socio-Economic Factors & Medical Characteristics Other Sources of Reporting Bias Exposure to Biological (includes Annoyance Responses) Agents One must stress the biological precursors important to the effects of ETS in childhood. These include genetic predisposition (physiological, immunological and biochemical), utero exposure, and breast feeding. These also include environmentally-induced atopy and residua of infections. Pre—existing medical conditions, such as cystic fibrosis, congenital defects will also affect responses to ETS. The socio-economic and behavioral factors are important as they relate to nutrition (re: resistance), familial crowding, and other contacts (especially day care), medical attitudes and medical care, etc. Socio-ecoriomic status (SES) and day care have been shown to modify the effects of ETS. Reporting bias is a critical issue for ascertaining exposure, as has been documented by many previously. There are two major components to this, the positive bias and the negative bias. The first is thought to occur associated with parental conditions (e.g., Colley, 1974; Cederlof and Colley, 1974). The second is thought to occur because the respondent becomes annoyed by ETS (e.g., Weber 1984; Hugod 19 84; NCHS 1976; NIOSH 1971), and/or have anger/aggression reactions (e.g., Jones and Bogat 1978). Effects of active/self-smoking interacting with passive smoking should be discussed (Bland et al. 19 78; Lebowitz et al. 1987 and 1988) Other exposures which have similar effects (e.g., wood smoke, other particulate matter, NO 2 , formaldehyde) may be confounding the effects of ETS (Hammer et al. 1976; Anderson 1979; Speizer et al. 1980 (with update); Comstock et al. 1981; Melia et al. 1982; and Koo et al. 1988), or may interact with ETS in producing effects (Lebowitz et al. 1990, and in press). Of course, individuals 42 ------- including children may have multiple micro-environments in which they are exposed, so insufficient information would tend to yield incorrect exposure-response curves. Thus, there are many possible co-variates and confounders which should still be considered (Lebowitz 1990). 5.2.3 Usa of Mata—Analvsis The staff should give serious con- sideration to meta-analysis of those studies of sufficiently sim- ilar design to warrant it. However, it was not clear that there is a body of suitable studies for such an analysis. If one is war- ranted, it should be guided, to the extent possible, by the same considerations outlined in Section 5.1.4. 43 ------- 6.0 BUNXARY AND CONCLtIBIONS In conducting its review of the ETS Risk Assessment document and Policy Guide, the Committee found them to be good faith efforts to address complex and difficult issues affecting public health. The authors attempted to select and interpret the most relevant information from an enormous and diverse scientific data base, most of which was not designed or intended to yield the information needed for this task. Since the task is extremely difficult, it should come as no surprise that the Committee also found the documents to be incomplete in many respects. The situation is analogous to that for the Criteria Air Pollutants, where it has been necessary to prepare and review two or more draft documents prior to their endorsement by the Clean Air Scientific Advisory Committee (CASAC). This Committee has suggested both organiza- tional and specific technical changes and additional analyses that, if followed, can result in improved ETS Risk Assessment and Policy Guide documents, and stands ready to provide further review comments on the revised drafts. The SAB was asked to address the following issues in reviewing the documents: A. Lung Cancer in Adults The Committee noted that Chapters 3 and 4 addressed only the issue of lung cancer risk for non-smoking women due to spousal smoking. The revised document should be expanded to include the full range of cancer impacts of ETS. The Committee also noted a number of areas where substantial improve- ments could be made organizationally, and in terms of content of material that was not adequately covered or not covered at all, and urge the EPA staff to redraft those chapters as well. Comments on specific issues within the broader context of lung cancer follow below. 1. Carcinoaenicity of ETB The Committee concurs with the judgement of EPA that Environmental Tobacco Smoke should be classified as a Class A Carcinogen. The Committee had some difficulty with the use of the Guidelines for Carcinogen Risk Assessment as they are currently formulated (51 FR 33992 August 22, 1986). The strongest evidence for the carcinogenicity of tobacco smoke is that obtained in a large number of epidemiologic studies 44 ------- of smoking and lung cancer. The causality of the connection between direct inhalation of tobacco smoke and excess risk of lung cancer cannot be in doubt. It has been demonstrated that cessation of inhalation of tobacco smoke leads to a reduction of the excess lung cancer risk. The risk has been shown to be proportional to the amount of smoke inhaled. In terms of overall impact it has been shown that a very high proportion of the current lung cancer incidence is due to inhalation of tobacco smoke. The ageing of sidestream tobacco smoke influences its uptake and deposition in the lung and its potential carcinogenicity, but there are strong similarities in the chemical and j vitro biological activity of ETS and mainstream tobacco smoke, and ETS resembles mainstream tobacco smoke in terms of particle size distribution and compo- sition of carcinogens, co-carcinogens and tumor protnotors. The inhalation of ETS by children, by non-smokers or former smokers represents a risk that is much smaller than that experienced by smokers, but it is an involuntary exposure. It is not uncommon to derive quantitative risk assessments of exposures to carcinogens from data obtained in more heavily exposed occu- pational populations, and in that sense smokers represent a more heavily exposed population which can be used for extrapolation to the lower exposures imposed on children and non-smokers. 2. Spousal Smoking All of the studies cited in the report on ETS and risk of lung cancer have made observations on married women who have been classified as “never-smoking.” Those married to a smoker are assumed to be exposed to greater levels of ETS than those married to a nonsmoker. As noted in the report, this rel- ative risk comparison is implicitly a comparison of women exposed to both spousa]. and other ETS to those exposed to other ETS only. Spousal smoking is believed to be a useful maker for total ETS exposure because (1) it often indicates many years of exposure: this contrasts with biological markers such as urinary cotinine, which indicate exposure at only one point in time; (2) the level of ETS exposure in the home when the spouse smokes appears to be of substantially greater magnitude than the background exposure. There are potential limitations in the use of spousal smoking as an indicator of ETS exposure that need to be considered: 45 ------- 1. Spousal smoking may account for a relatively small pro- portion of lifetime ETS exposure. 2. The difference in ETS exposure comparing household exposure versus household plus background exposure may differ in different countries and different regions within the U.S. 3. A major source of ETS exposure is that incurred in child- hood, which could contribute to increased lung cancer risk in an adult. 4. The use of spousal smoking as an indicator of exposure may amplify the risk of misclassification of smokers as non- smokers. 5. Spousal smoking status could be associated with several sources of confounding, e.g., lower socio-econoinic class, diet, alcohol, drugs, more exposure to air pollution, etc., factors that could possibly increase lung cancer risk. Despite various limitations as an indicator of ETS exposure, spousal smoking status is a reasonable method of identifying people with greater, versus lesser, ETS exposure. 3. United States and Foreiqn Studies The Committee felt that data from studies conducted overseas as well as in the United States should be utilized in evaluating whether exposure to ETS increases risk of lung cancer. It is appropriate to examine the totality of evidence from all the case-control and cohort studies, regardless of where they were conducted. 4. Use of Mata—Anajysis Meta—analysis is an appropriate tool to summarize the epidemiological studies investigating the risk of ETS. However, the priority given the meta-analysis in this report in attempting to demonstrate that ETS is causally associated with lung cancer is not justified. Evidence on the carcinogenic effect of active smoking, the presence of carcinogens in ETS, and predicted lung cancer risk of low dose exposure to tobacco smoke from appropriate models are an important part of establishing a causal relationship. The meta-analysis could then be interpreted as showing the available epidemiologic evidence is consistent with a small elevated risk. 46 ------- S. Confounders/Misclassification Important potential con- founders of the ETS-lung cancer relationship were addressed in the Report mainly by carrying out a separate meta-analysis of those studies which included adjusted analyses. The main confounders included in these adjusted analyses were: age, education, and social class. Comparison of unadjusted and adjusted RRs in those studies which present both, suggests that these variables are not important con founders. As for other potential confounders of the ETS-lung cancer relationship, including occupation, radon exposure, and diet, there is no way to evaluate their importance as confounders or to adjust for them, since virtually none of the studies contains information on them. However, they could be mentioned in the text as potential confounders. The issue of misclassification should not be restricted to misclassification of current and ex-smokers as “never smokers.” It should also be mentioned that misclassification of diagnosis (diagnoses other than lung cancer being incorrectly classified as lung cancer; or vice versa) will cause a biasing of the P.R toward the null. Not enough attention was given to possible non-differential misclassification of ETS exposure. This is an important issue, since marriage to a smoking spouse is an imperfect proxy for total ETS exposure. In the case of dichotomous exposure, such misclas- sification would have the effect of biasing the PR estimate toward the null. Other potential biases which deserve mention include recall bias (differential reporting of exposure status by cases compared to controls) and bias due to the use of proxy respondents. 6. Characterization of uncertainties The draft risk as- sessment document’s findings on the ETS/adult lung cancer relation- ship is based on two main arguments: (1) biological plausibility ; and (2) epidemiologic evidence . With exposure levels that are usually quite low, it is not surprising that the association is likely to be weak although, given the size of the exposed pop- ulation, societally important. Because the epidemiologic eviden- tiary base for drawing conclusions regarding ETS’s carcinogenicity consists mainly of studies of exposure levels produced by spousal smoking, the biological plausibility argument assumes great impor- 47 ------- tance. Each step in that argument should therefore be carefully addressed, with the uncertainties encountered being spelled out explicitly. 7. quantitative Risk Assessment The Conunittee generally agreed that the quantitative assessment of the risk of lung cancer due to exposures to ETS should be based on the human epidemiology studies and that meta—analysis was a suitable approach to combining the data. It is direct and makes few assumptions. It should be noted that this approach is fully consistent with the risk assess- ments that have been done for many other carcinogens and that those assessments are generally based on fewer studies. Given that the epidemiology studies should be the basis of the risk assessment, some refinements of the risk assessment are recommended with respect to: 1. Criteria for Including Individual Studies in the Meta- analysis 2. Adjustment for Smoker Misclassification 3. Misclassification of Exposure 4. Uncertainties in the Estimate of Annual Lung Cancer Deaths Due to Passive Smoking 5. Dose—Response Estimation of Risk 8. Home vs. Workplace Exposure The ommittee recognizes that there is little epidemiologic literature on the health effects of ETS in the workplace. However, the report should review and comment on the data that do exist. B. ResDiratory Disorders in Children Chapter 5 on respiratory disorders in children was a commendable first effort for a very difficult task. Nevertheless, we found that it could be substan- tially improved and that the conclusions drawn in it could be made much stronger if the chapter was revised in the manner suggested in this report. The Committee found the evidence for respiratory health ef- fects in children to be stronger and more persuasive than that stated in Chapter 5 of the draft ETS Risk Assessment document, and 48 ------- recommends that the new draft contain a chapter devoted to quanti- tative risk assessment. It would be analogous to Chapter 4, which deals only with the evidence for lung cancer risk discussed in Chapter 3. The risks are different, but it is possible that the impact of ETS on respiratory health in children may have much greater public health significance than the impact of ETS on lung cancer in nonsmokers. The earlier chapters on lung dosimetry and the physical and chemical factors affecting it should incorporate new material. The difference in deposition and retention of ETS components between children and adults need to be established and considered in a risk assessment. Comments om specific issues follow: 2.. Weight of Evidence The scope of Chapter 5 is limited to selected studies published subsequent to the 1986 surgeon Gen- eral’s Report and the National Research Council Report. The addi- tional literature available since 1986 provides a basis for in- creased concern. Thus, the Committee urges a thorough review of the entire body of evidence. Judgment cannot be made concerning causality without assessing the totality of the evidence including studies reviewed in the two 1986 reports and those published subsequently. In reviewing the weight of the evidence, the present Chapter 5 does not establish an appropriate framework for considering the data. The alternative explanations for association of ETS exposure with adverse respiratory effects need to be clearly listed (causal- ity, confounding, information bias) and the individual studies re- viewed for the approaches used to address confounding and informa- tion bias. The weight of the evidence could then be judged to de- termine the causality of associations. 2. Cpnfounders A number of confounders were mentioned by the report, but addressed improperly, such as utero exposure, par- ental reporting bias, and active smoking. The biological precursors important to the effects of ETS in childhood include genetic predisposition (physiological, iminuno- logical and biochemical), j utero exposure, and breast feeding. These also include environmentally-induced atopy and residua of 49 ------- infections. Pre-existing medical conditions, such as cystic fibrosis, congenital defects will also affect responses to ETS. The socio-economic and behavioral factors are important as they relate to nutrition (re: resistance), familial crowding and other contacts (especially day care), medical attitudes and medical care, etc. SES and day care have been shown to modify effects of ETS. 3. Use of Meta-Analysis The staff should give serious con- sideration to meta—analysis of those studies of sufficiently sim- ilar design to warrant it. However, it was not clear that there was a body of suitable studies for such an analysis. If one is warranted, it should be guided, to the extent possible, by the same considerations outlined for ineta-analysis for lung cancer. 50 ------- 7.0 REFERENCES CITED American Lung Association. “Maternal Smoking and Health Effects on Fetus and Newborn. An Annotated Bibliography of the Research Data Base.” ALA Report, New York, NY, June 1983 Anderson H.R. “Respiratory Abnormalities, Smoking Habits, and Ventilatory Capacity in a Highland Community in Papua, New Guinea: Prevalence and Effect on Mortality. Irtt J Epidemiol 1979; 8:127- 135 Bangert-Drowns Psychol Bull 1986; pg. 388-399) Bland M., Bewley B.R., Pollard V., and Banks M.H. “Effects of Children’s and Parent’s Smoking on Respiratory Symptoms.” Arch Dis Childhood 1978; 53:100—105. Burrows B., and Martinez F.D. “Bronchial Responsiveness, Atopy, Smoking, and Chronic Obstructive Pulmonary Disease.” Am Rev Respir 1989; 140:1515—1517. Cederlof R., and Colley J. “Epidemiological Investigations on Environmental Tobacco Smoke.” Scan J Respir Dis 1974; (Suppl. 91) :47—49. Colley J.R.T. “Respiratory Symptoms in Children and Parental Smoking and Phlegm Production.” Brit Med J 1974 2:201-204 Comstock G., Meyer M.B., Helsing K.J., and Tockman M.S. “Respiratory Effects of Household Exposures to Tobacco Smoke and Gas Cooking.” Am Rev Respir Dis 1981; 124:143—148 Cuuuuing K.M., Markello S.J., Mahoney M.C., Bhargava A.K., McElroy P.D., Marshall J.R. Measurement of Current Exposure to Environmental Tobacco Smoke. Arch Environ Health , 45:74—79, 1990. Cumxning K.M., Markello S.J., Mahoney M.C., Marshall J.R. Measurement of Lifetime Exposure to Passive Smoke. Am J Epiderniol , 130:122—132, 1989. Dahms T.E., Bolin M.D., and Slavin R.G. “Passive Smoking: Effects of Bronchial Asthma.” Chest 1981; 80:530-534 Freidman G.D. et al. “Prevalence and Correlates of Passive Smoking.” Am J Public Health 1983; 73:401—405 51 ------- Hammer D.I., Miller F.J., Stead A.G., and Hayes C.G. “Air Pollution and Childhood Ventilatory Function. I. Exposure to Particulate Matter in Two Southeastern Cities, 1971-1972.” In A.J. Finkel and W.C. Duel (Eds.) Clinical Implications of Air Pollution Research ” Publishing Sciences Group, Acton MA, pp. 321-37, 1976. Hammond, S. Katharine in, Measurement of Airborne Compounds: Sampling. Analysis, arid Data Interpretation . Edited by Eric D. Winegar. Lewis Publishers. Hammond, S.K., Gann, P.H., Coghlin, J., Tannenbaum, S.R., Skipper, P.L. Tobacco smoke exposure and carcinogen-hemoglobin adducts. In: Indoor Air ‘90, Proceedings of the 5th International Conference on Indoor Air Quality and Climate Toronto, Ontario, Canada, 29 July - 3 August, 1990. Volume 2: Characteristics of Indoor Air, pp. 157—162, 3.990.) Hammond, S. Katharine. The Uses of Markers to Measure Exposures to Complex Mixture. jn, Exposure Assessment for Epidemiology and H&zard Control . edited by Stephen Rappaport and Thomas J. Smith. (Lewis Publishers) 1991. Hammond, S. K., Lewtas, J., Mumford, J. and Henderson, F.W. Exposures to environmental tobacco smoke in homes. Measurement of Toxic and Related Air Pollutants. Environmental Protection Agency/Air and Waste Mariangement Association International Symposium. May 1-5, 1989 Raleigh, North Carolina. Published by the Air and Waste Manarigement Association, Pittsburgh, PA pp. 590—595, 1989. Henderson, R.W., Reid, H.F., Morris, R., Wang, Ou-Li, Hu, P.C., Helms, R.W., Forehand, L., Mumford, J., Lewtas, J., Haley, N.J., and Hammond, S.K. Home air nicotine levels and urinary cotinine excretion in young children. Am Rev Respir Dis , 140: 197-201, 1989. Henderson, R.W., Reid, H.F., Morris, R., Wang, Ou-Li, Hu, P.C., Helms, R.W., Forehand, L., Mumford, J., Lewtas, J., Haley, N.J., and Hammond, S.K. Home air nicotine levels and urinary cotinine excretion in young children. Am Rev Respir Dig , 140: 197-201, 1989. Hiller F.C., McKusker K.T., Mazumber M.K., Wilson J.D., and Bone R.C. “Deposition of Sidestream Cigarette Smoke in the Human Respiratory Tract.” Am Rev Respir Dis 1982; 125:406-408. Hiller F.C. “Deposition of Sidestream Cigarette Smoke in the Human Respiratory Tract.” Preventative Med 1984; 13:602-607. Hugod C. “Indoor Air Pollution with Smoke Constituents - an Experimental Investigation.” E’reventative Med 1984; 13:1435-1438. 52 ------- Janerich D.T., et al. Lung Cancer and Exposure to Tobacco Smoke in the Household. N Enal J Med , 323:632—6, 1990. Jarvis M.J., Tunstall-Pedoe H., Feyerabend C., Vesey C., and Saloojee Y. “Comparison of Test Used to Distinguish Smokers from Nonsmokers.” Am J Public Health 1987; 77:1435-1438 Jones J.W., and Bogat A. “Air Pollution and Human Aggression.” Psychol Rep 1978; 43:721—722. Koo L.C., Ho J.H.C., Matsuki H., Mon T., and Tominaga S. “A Comparison of the Prevalence of Respiratory Illnesses Among Nonsmoking Mothers and their Children in Japan and Hong Kong.” Rev Respir Djs 1987’ 136:290—295. Leaderer, B.P. and Hammond, S.K. An evaluation of vapor phase nicotine and respirable suspended particle mass as markers for environmental tobacco smoke. Environmental Science and Technology , in press. Lebowitz M.D. and Burrows B. “Respiratory Symptoms Related to Smoking Habits of Family Adults.” Chest 1976; 69:48—50. Lebowitz M.D., Holberg C.J., Knudson R.J., and Burrows B. “Longitudinal Study of Pulmonary Function in Children and Adolescents: Analysis of Two Longitudinal Population Studies.” J Epidemiol 1987; 136:69—75. Lebowitz M.D.and Holberg C.J. “Effects of Parental Smoking and Other Risk Factors on the Development of Pulmonary Function in Children and Adolescents: Analysis of Two Longitudinal Population Studies.” Am J Epidemiol 1988; 128:589—596. Lebowitz M.D. “Environmental Tobacco Smoke: A Public Health Issue.” Environ Intn’l 1989; 15—11—18. Lebowitz M.D., Quackenboss J.J., and Krzyzanowski M. “Multi- pollutant Exposures and Health Responses: Epidemiological Aspects of Particulate Matter.” In: D. Walkinsahw (Ed.) Indoor Air ‘90 Ottawa: CHMC, 1990; Vol I, pp. 375-380. Lebowitz M.D. “Methodological Issues in the Epidemiological Investigations of Lung Cancer Related to Low Level Risk.” Intnt’l J Epidemjol 1990; 19 (Suppi. 1): S43—S47. Lebowitz M.D. “Environmental Contaminants and Low Level Cancer Risks: Perceptions and Scientific Strategies.” Intnt’l J Epidemiol 1990; 19 )Suppl. 1): S11—S13. Lebowitz M.D. “Introduction: Implications of Exposure and Dose in Health Effects Studies.” In: E. Pellexarro Total Exposure Assessment Methodology 1990; pp. 232-234, Pittsburgh:AWMA. 53 ------- Lebowitz M.D. and Quackemboss J.J. “The Effects of Environmental Tobacco Smoke on Pulmonary Function.” Intn’l Arch 0cc Environ Hith 1990; (Suppi. ) 157—152. Lebowitz M.D. “The Use of Peak Expiratory Flow Rate Measurements in Respiratory Disease (Clinical and Epideiniological).” Ped Puim 1991; in press. Martinez F.D., Morgan W.J., Wright A.L., Taussig L.M., and GHI4A Pediatricians. “Pre-iliness Lung Function as Predisposing Factor for Lower Respiratory tract Illness (LRIs) and Symptoms in Infants (abstract).” Ant Rev Respir Dis 1988; 137:405. Melia R.J.W, Florey C. du V., Ahinan D.G., and Swan A.V. “The Association Between Respiratory Illness and NO 2 , Temperature and relative Humidity.” mt J Epidemiol 1982; 11:164—169. National center for Health statistics: Health, United States, 1986. DHHS Pub. No. (PHS) 87-1232, Public Health Service. Washington D.C.: Government Printing Office, december 1986. National Clearing House for Smoking and Health. “Adult Use of Tobacco.” U.S. Department of Health and Human Services, Washington D.C., p. 23; 1976. National Research Council (NRC). “Indoor Pollutants.” National Academy Press, Washington D.C. p. 149-168, 369-382; 1981. National Research Council (NRC). “Environmental Tobacco Smoke.” National Academy Press, Washington D.C. 1986. Quackenboss J.J., Lebowitz M.D., and Hayes C. “Epidemiological Study of Respiratory Responses to Indoor/Outdoor Air Quality.” Environ Intn’l 1989; 15:493—502. Rantakallio P. “The Effect of Maternal Smoking on Birth Weight and the Subsequent Health of the Child.” Early Human 0ev 1978; 2:371- 381. Rantakallio P. “Social Background of Mothers Who Smoke During Pregnancy and Influence of These Factors on the Offspring.” Sci and Med 1979; 13A:423—429. Sandier D.P., Helsing K.J., Comstock G.W., and Shore D.L. “Factors Associated with Past Household Exposure to Tobacco Smoke.” Am J Epjdemiol 1989; 129:380—387. Schilling R.S.F., Letai A.D., Hui S.L., Beck G.H., Schoenbart J.B., and Bouhuys A. “Lung Function, Respiratory Disease, and Smoking in Families.” Ant J Epidemiol 1977; 106:274—283. 54 ------- Schwartz J. and Zeger S. “Passive Smoking, Air Pollution, and Acute Respiratory Symptoms in a Diary Study of Student Nurses.” Rev Respir Dis 190; 141:62—67 Shepard R.J., Collins R., and Silverman F. “Passive Exposure of Asthmatic Subjects to Cigarette Smoke.” Environ Res 1979; 20:329- 402. Sherrill D.L., Holberg C.J., and Lebowitz M.D. “Differential rates of Lung Growth as Measured Longitudinally by Pulmonary Function in Children and Adolescents.” Ped Pulmonol . 1990; 8:145—154. Speizer F.E., Ferris B. Jr., Bishop Y.M.M., and Spengler J. “Respiratory Disease Rates and Pulmonary Function in Children Associated with NO 2 Exposure.” Am Rev Respir Dis 1980; 121:3-10. Surgeon General Report. “The Health Consequences of Smoking - Cancer.” Superintendent of Documents, U.S. Government Printing Office, Washington D.C. 1982. Surgeon General Report. “The Health Consequences of Smoking - Cardiovascular Disease.” Superintendent of Documents, U.S. Government Printing Office, Washington D.C. 1983. Surgeon General Report. “The Health Consequences of Smoking - Chronic Obstructive Lung Disease.” Superintendent of Documents, U.S. Government Printing Office, Washington D.C. 1984. Surgeon General Report. “The Health Consequences of Smoking - Chronic Obstructive Lung Disease.” Superintendent of Documents, U.S. Government Printing Office, Washington D.C. 1984. Surgeon General Report. “The Health Consequences of Smoking - Cancer and Chronic Lung Disease in the Workplace.” Superintendent of Documents, U.S. Government Printing Office, Washington D.C. 1985. Surgeon General Report. “The Health Consequences of Involuntary Smoking.” Superintendent of Documents, U.S. Government Printing Office, Washington D.C. 1986. Tager I.B., Segal M.R., Munoz A., Weiss S.T., and Speizer F.E. “The Effect of Maternal Cigarette Smoking on the Pulmonary Function of Children and Adolescents.” Am Rev Respir Dis 1987; 136:1366- 1370. Tager I.B. “Passive Smoking — Bronchial Responsiveness and Atopy.” Am Rev Respir Dis 1988; 138:507-509. Thompson S.E.., Stone R, Nanchabal K, and Wald N.J. Relation of Urinary cotinine Concentrations to Cigarette Smoking and to Exposure to Other People’s Smoke. Thorox , 45:356—361, 1990. 55 ------- Yarnell J.W.G. and St. Leger A.S. “Respiratory Illness, Maternal Smoking Habit and Lunq Function in Children.” Brit 3 Dis Chest 1979; 73:230—236. Wall M., Brooks J., and Redding C. “Health Effects of Smoking on Children.” Ant Rev ResDir Dis 1985; 132:1137—1138. Weiss S.T., Tager I.B., Munoz A., and Speizer F.E. “The Relationship of Respiratory Infections in early Childhood to the Occurrence of Increased Levels of Bronchial Responsiveness and Atopy.” ? nt Rev Respir Dis 1985; 131:573—578. Weiss S.T., Tager I.B., Speizer F.E., and Rosner B. “Persistent Wheeze: Its Relation to Respiratory Illness, Cigarette Smoking, and Level of Pulmonary Function in a Population Sample of Children.” Ant Rev Respir Dis 1980’ 122:697—707. World Health Organization (WHO). “Estimating Human Exposure to Air Pollutants.” WHO, Copenhagen/Geneva; 1982. Weber A. “Annoyance and Irritation by Passive Smoking.” Preventive Med 1984; 13:618—625. 56 ------- |