Tobacco Documents Online

714 Statement Professor T. D. Sterling I am Theodor Sterling, University Research Professor at Simon Fraser University, Burnaby, British Columbia. I have also taught at the following universities: Alabama, Cincinnati, Michigan State, Princeton, and Washington University at St. Louis. I have served as an advisor to the Food and Drug Administration, the Federal Trade Commission, the National Science Foundation, the National Security Agency, the National Academy of Sciences, and the Veterans Administration. I am a Fellow of the American Association for the Advancement of Science and of the American Statistical Association. My major professional work concerns the collection and interpretation of scientific data and the design, execution and analysis of experiments and of survey studies. Much of my research has centered on learning about the effects of environmental exposures on the health of animals and men. I wish to comment on the proposed Comprehensive Smoking Prevention Education Act of 1981. I am opposed to this proposed legislation because I believe its far-reaching "findings" do not have adequate scientific support. Further, if it becomes law, this Act could unreasonably divert attention and resources away from crucially important areas of public health research. My comments are based on my analyses of reports of other investigators and my own research efforts in the smoking and health areas. Smoking and Health Studies My interest in the possible health effects of cigarette smoking goes back to the late 1950's. In a number of published articles, I have expressed my concern about many smoking and health studies, primarily because the approaches taken have oversimplified what in reality is an extremely complex problem. I have pointed out specific flaws in the design and execution of many of these studies. The bulk of the data on smoking and health comes from studies of population groups, that is, from epidemiological studies. These data are analyzed using statistical methods to determine whether there are associations between certain factors and disease end-points. As a statistician, I am frequently astonished to see how many people, even many who should know better, treat statistical associations as proof of cause-and- effect relationships. The findings in the proposed Act seem to be a perfect example of this misuse of statistics. Scientists and nonscientists alike mus epidemiologist, who very re Epidemiological s ones . ., canno when the end-poin of a chronic non- In order for atat reliable directions for fur which they are derived must analyzed using appropriate c ::g factors into account. [ scandards frequently have nc of smoking and health. I have evaluated : norcality studies. The lar scudy conducted by the Amer: •4o. Although one would ho7 expensive study would be re: : find that the conclusions .moking and mortality may n, procedures used to select t} The data from thi: b, volunteer workers, and i- !:ad preconceived views an sr .ince the purpose of the at+ and disease, the volunteers .~S,jects who smoked. Indee, ACS population were quite d- tion. For example, the ACS breast cancer deaths and th :ancer deaths, compared to I percent of ACS males who di, ali U.S. males. Yet the ovc tie same as that for U.S. m: lower than that for U.S. vor The ACS populatio: =eneral U.S. population vit: ticnal attainment, racial s- !• clear that these one mil- of the U.S. population. Ce: canoot legitimately be proj, this has been and is still t ACS has recently announced : persons" study, using the s: ~! "dedicated" volunteers: 115-077 0-82-46

-sity Research Professor at _sh Columbia. I have also Alabama, Cincinnati, :on University at St. Louis. I and Drug Administration, )nal Science Foundation, inal Academy of Sciences, a Fellow of the American >nce and of the American 'essioaal work concerns the :ific data and the design, ind of survey studies. irning about the effects of )f animals and men. 3osed Comprehensive Smoking a opposed to this proposed :eaching "findings" do not :her, if it becomes law, :ation and resources away Cc health research. analyses of reports of i efforts in the smoking iealth effects of cigarette In a number of published about many smoking and approaches taken have xtremely complex problem. Sa design and execution of ing and health comes from from epidemiological g statistical methods to as between certain factors cian, I am frequently n many who should know as proof of cause-and- the proposed Act seem to statistics. Scientists 715 and nonscientists alike must heed the words of a well-known epidemiologist, who very receatly said: Epidemiological studies, even prospective ones . ., caanot prove cause-and-effect vhen the end-point, 'effect', is an outcome of a chronic non-communicable condition.1 In order for statistical aasociations to provide reliable directiona for further research, the data from which they are derived must be accurate and these data must be analyzed using appropriate methods, taking all possible confound- ing factors into account. Unfortunately, these scientific standards frequently have not been met in epidemiological studies of smoking and health. I have evaluated in depth two of the major prospective mortality studies. The largest of these was the "million persons" study conducted by the American Cancer Society (ACS) some years ago. Although one would hope that results from such a large and expensive study would be reliable, after detailed consideration, I find that the conclusions about the possible relationship of smoking and mortality may not be valid beca se of the biased procedures used to select the population.2•~ The data from this large study population were gathered by volunteer workers, and it appears that many of the volunteers had preconceived views on smoking and disease. In other words, since the purpose of the study was to show a link between smoking and disease, the volunteers would be more likely to choose ill subjects who smoked. Indeed, some of the disease patterns of the ACS population were quite different from those of the U.S. popula- tion. For example, the ACS women had twice the percentage of breast cancer deaths and three times the percentage of lung cancer deaths, compared to U.S. females generally. Likewise, the percent of ACS males who died of lung cancer was twice that of all U.S. males. Yet the overall death rate for ACS men was about the same as that for U.S. males and that for ACS women was somewhat lower than that for U.S. womea. The ACS population was also very different from the general U.S. population with respect to age distribution, educa- tional attainment, racial structure and place of residence. It is clear that these one million men and women were not representative of the U.S. population. Certainly, results of the ACS study cannot legitimately be projected to the general population, yet this has been and is still being done. And astonishingly, the ACS has recently announced it will soon begin a second "million persons" study, using the same procedure of subject recruitment by "dedicated" volunteers: 2

718 I, : to ask: does smoking kill workers or working kill smokers?8 Evidence in support of the latter has been detailed in a paper I presented at the American Lung Association's Occupational Health Task Force Meeting in April, 1980.9 Here I will briefly summarize that paper. While there are many reasons for the confusion between the possible effects of smoking and other variables, the major complication arises because the prevalence of smoking is highest among those individuals who are exposed to possible deleterious occupational factors.10,11 For instance, 71 percent of construction painters are current smokers while only 27 percent of teachers have that habit.11 In fact, when occupations are arranged by prevalence of smoking, 29 out of 40 occupations in which smoking is most prevalent are also those with high exposure to dusts and fumes, while in only 4 of the 40 occupations with the lowest prevalence of smoking are workers exposed to such hazards.10 There is a strong tendency for a higher proportion of individuals who do dirty work to smoke, compared to individuals who do clean work. Thus, in a statistical sense, the category smoker may be primarily an index for worker exposed to occupational hazards. It follows that any comparison between smokers and non- smokers implies a comparison between groups that probably differ significantly in their exposures to dusts and fumes in the workplace In short, the increased incidence of disease, ascribed to smoking by epidemiological studies that fail to contnol adequately for occupation, could well be due to occupational factors. Obtaining an accurate occupational history is difficult and time consuming, particularly because important exposures may have occurred twenty or more years ago. Studies of Industrial Workers studies of I turn now to a discussion of recent and long overdue 3 lung disease in occupationally exposed workers; some of these have failed to find any significant effect due to cigarette smoking, while others have found even higher disease rates among ,~ non-smokers than among smokers. x; In a study of zinc and lead miners, Axelson, et al.12 report a greater risk for lung cancer among non-smokers compared ~ to smokers. Another study of miners' mortality by Dahlgren, ;d reported in 1979, showed that non-smoking miners had a higher mortality from lung cancer than smoking miners.13 Axelson emphasized that these Swedish studies deal with a lifetime follow-up of ~ miners whereas most other mining populations have been studied by ~ means of cohorts with a follow-up of not more than about 25-30 s,y years or less.14 In other words, the completeness of the follow- up leads to results with added reliability. An inverse relationship between smoking and lung cancer was also found among workers exposed to chloromethyl methyl ether.15 The author opposition to smokin; cigarette smoking eni carcinogenic effects Pinto16 fot his study population general population. "not due to smoking" tSe differences betwe the three groups (smc statistically signifi A study of reported that "Lung c s:.owed a greater incr than did lung cancer tew studies that esti tae workers. ..__ Some scient and cigarette smoking inding (6) of H.R. 4 ::scussed above do no and Weiss separately ;rotective properties The claim o asSestos and smoking, exaa:ned carefully. . vorkers by Hammond an large effect on lung , of smoking and asbest, auccessive reports, a: a2ong non-smoking wor': ;acest report by thest r=sk of lung cancer mc exaosed to asbestos. In addition. dammond-Selikoff stud:. ~:stories available fc vorkers. This promptc large fraction of sub_ uncertain any quantitF .aoking and asbestos.'".ulation workers, tL sites was the same for 'S: of all deaths were 'he proportion of expe IPecific U.S. mortalit a=• These figures st •rPosure for all worke

ag kill smokers?8 detailed in a paper I s Occupational Health will briefly summarize the confusion between -ariables, the major of smoking is highest possible deleterious '1 percent of construction percent of teachers >ns are arranged by :ions in which smoking exposure to dusts and is with the lowest :o such hazards.10 higher proportion of apared to individuals sense, the category ^ exposed to occupational stween smoers and non- s that probably differ and fumes in the workplace. se, ascribed to smoking itrol adequately for aal factors. Obtaining alt and time consuming, ay have occurred twenty cent and long overdue exposed workers; some at effeet due to cigarette er disease rates among rs, Axelson, et al.12 g noa-smokers compared ality by Dahlgren, miners had a higher ners.13 Axelson emphasized 'etime follow-up of na have been studied by Lore than about 25-30 leteness of the follow- smoking and lung cancer ,loromethyl methyl 719 ether.15 The author, W. Weiss, who is known for his militant opposition to smoking, observed: "The data suggest that continued cigarette smoking entailed a factor which partially inhibited the carcinogenic effects of chloromethyl ethers." Pinto16 found elevated lung cancer mortality rates in his study population of arsenic exposed workers, compared to the general population. He commented that these elevated rates were "not due to smoking" in his workers. Further, he reported that the differences between the elevated standard mortality ratios in the three groups (smokers, ex-smokers and non-smokers) were not statistically significant. A study of chrysotile asbestos miners in Canada17 reported that "Lung cancer deaths occurred in non-smokers, and showed a greater increase of incidence with increasing exposure than did lung cancer in smokers . This is one of the very few studies that estimated levels of asbestos dust exposure among the workers. Some scientists have claimed that occupational exposures and cigarette smoking increase the risk of disease. In fact, Finding (6) of H.&. 4957 accepts this view. The recent studies discussed above do not support this hypothesis. In fact, Axelson and Weiss separately raised the possibility that smoking may have protective properties for some types 4 work. The claim of interaction is heard most often regarding asbestos and smoking, but even here the evidence needs to be examined carefully. Although the well known study of insulation workers by Hammond and Selikoff18,19,20 has indicated an apparently large effect on lung cancer rates due to the claimed interaction of smoking and asbestos, that "effect" has become smaller with successive reports, as increasing numbers of lung cancers occur among non-smoking workers. It is important to note that the latest report by these investigators has shown a five-fold relative risk of lung cancer mortality for smoking and non-smoking workers exposed to asbestos. In addition, there are serious problems with the Hammond-Selikoff study. For example, there were no smoking histories available for over 6,000 of the 17,800 insulation workers. This prompted a cancer researcher to note '. . . the large fraction of subjects with unknown smoking habits makes uncertain any quantitative assessment of the joint effect of smoking and asbestos."21 Also, I have pointed out that in these insulation workers, the proportion of deaths due to cancer of all sites was the same for both smokers and non-smokers.8 Approximately 45% of all deaths were due to cancer in these workers. In contrast, the proportion of expected deaths from cancer (based on age- specific U.S. mortality rates for white males) was only about 18Z. These figures strikingly indicate the dangers of asbestos exposure for all workers, whether they smoke or not. 6

720 Ongoing Studies In many of these reported studies, conclusions about T the possible relative effects of smoking and occupational exposures ' were almost an afterthought. There have been few studies designed for the express purpose of evaluating the possible interaction of smoking and occupation. In order to fill that void, a number of associates at the Sinai Hospital in Detroit and the Rhode Island a Hospital of Brown University have joined me in a study seeking to untangle the possible lnfluences of the life history of smoking ' and exposure to toxic dust and fumes on patients in lung, cancer, heart, dental and other hospital services. We have developed a thorough questionnaire and a practical interviewing procedure to determine whether an individual has been exposed to occupational hazards, and the extent of aay such exposures. While the analysis of all our data is not yet complete, it is nevertheless clear that the association previously claimed between smoking and respiratory cancers has been greatly magnified because other factors were not considered. The much-publicized "association" between heart disease and smoking may be similarly exaggerated. Indeed, if our preliminary findings are substantiated when data analysis is complete, the statistical association reported by others between smoking and heart disease may have to be completely re-evaluated. Conclusions My purpose here has been to present some results of scientific investigations that are not widely publicized, and to show that an almost exclusive focus on cigarette smoking may have seriously hindered the scientific study of chronic diseases. I believe I have clearly demonstrated that the possible effects of smoking and occupational exposures have not yet been disentangled. As humans living in a frighteningly complex world, we grasp with relief at what appear to be simple answers. Indeed, the readiness with which the existing evidence has been accepted as demonstrating that cigarette smoking is the major antecedent for lung disease is, perhaps, a most striking example of our desire to keep our world simple and orderly. Nature, on the other hand, is not concerned with what we believe causes disease, and the real world is not always simple. In the case of lung disease, in fact, it is highly complex. The factual knowledge about the antecedents of lung disease will remain incomplete if we continue to simplistically blame cigarette smoking and continue to ignore the possible effects of the workplace on the health of workers. I am afraid that the bill under consideration would tend to reinforce what has been the dominant approach of health scientists -- the search for antecedents of disease in the behavior 7 44 ,:,a life-styles of individu .ia.c disease to individual e iack of it, alcohol, an rcc new American Cancer Soc ;ie effects of certain 1. In contrast, the - .x.:ts a major influence on ,.::ough suspected by some, needs only to recall the _ t Spring twenty years E +catement. The fact that mE _ncentrations of volatile f ;-.e orkplace, has not recei _::;ists and public health c In conclusion, I b .:pport my contention that t .. insufficient data. This . .sent public health effort .cate of imbalance and confu : the health consequences o

, conclusions about .d occupational exposures en few studies designed ' ossible interaction of hat void, a number of and the Rhode Island' : in a study seeking to 'e history of smoking -ieats in lung, cancer, We have developed a rviewing procedure to :posed to occupational es. ' a is not yet complete, .on previously claimed been greatly magnified The much-publicized king may be similarly .ndings are substantiated .ical association ' -t disease may have to at some results of ly publicized, and to .rette smoking may have chronic diseases. I :e possible effects of ' : yet been disentangled. a y complex world, we le answers. Indeed, :nce has been accepted the major antecedent ng example of our : concarned with what -1d is not always :ct, it is highly :ntecedeats of lung iue to simplistically tore the possible :orkers. - :onsideration would .t approach of health disease in the behavior z 721 and life-styles of individuals. Attempts are constantly made to link disease to individual habits such as nutrition, exercise or the lack of it, alcohol, and, of course, smoking. For example, the new American Cancer Society study mentioned earlier is targeted on the effects of certain life-style factors on cancer. In contrast, the idea that the day-to-day environment exerts a major influence on the health of large population groups, although suspected by some, has not been adequately considered. One needs only to recall the criticisms heaped upon Rachel Carson's Silent Spring twenty years ago to realize the truth of this statement. The fact that many people are exposed to substantial concentrations of volatile fumes and toxic dusts, particularly in the workplace, has not received sufficient attention from epidemi- ologists and public health officials. In conclusion, I believe that the facts I have presented support my contention that the "findings" in the bill are based on insufficient data. This legislation would add little to present public health efforts; it would merely reinforce the state of imbalance and confusion which now exists in investigations of the health consequences of the micro-chemical environment. 0

722 1. 2. 3. 4. 5. 6. 7. 13. Dahlgren, E., "Lung References Smoking in a Grou p o 4814, 1979. 14. Axelson, 0., "Effect Keys, A., at al., "The Diet and All-Causes Death Rate in the from Radon Dau ht Seven Countries Study," Lancet II: 58-61, July 11, 1981. g ers Commission Hearings, Sterling, T., "A Critical Reassessment of the Evidence r.~ 15. Weiss W. "The Ci Bearing on Smoking as the Cause of Health 65(9): 939-953, September, Lung Cancer," Am J Public 1975. , , ga Chloromethyl Ethers, Z 16. Pinto, S., at a1. ": Sterling, T., "Statistical Problems in Multi-variate (Etiological) , Workers " Arch Envi Surveys," Canadian J Stat 7(2): 205-215, 1979. , r, :7. McDonald, J., at al. Sterling, T. and J. Weinkam, "What Happens When Major Errors Chrysotile Minin 1 are Discovered Long After an Important Report has Been Published?," Presented at the American Statistical Associa- g, Hammond E and I S tion Annual Meeting, August 16, 1979. U.S. Public Health Service, "Cigarette Smoking and Health , . . , to Risk of Death of . Insulation Workers i: of Asbestos ad ; P Characteristics; United States -- July 1964-June 1965," +s Vital and Health Statistics, PHS Publication No. 1000, a , . . pp. 312-317. . Series 10, No. 34, Washington, D.C.: U.S. Government Printing Office, 1967. r: L:. Selikoff, I. and E. L Environmental Cancer Sterling, T., "A Review of the Claim that Excess Morbidity J. Fraumeni (New Yori and Disability Can Be Ascribed to Smoking," JN Am Stat Assoc 2J. Hammond, E., et al. 66(334): 251-257, June, 1971. , and Death Rates," Ani U.S. Public Health Service, "Chapter 3. Morbidity," Smoking and Health. A Report of the Surgeon General, Department of .L. Saracci, R., "Asbestc Epidemiological Evid, Health, Education, and Welfare, DHEW Publication No. (PHS) Int J Cancer 20: 32: 79-50066, 1979. 8. Sterling, T., "Does Smoking Kill Workers or Working Kill Smokers?," Int J Health Serv 8(3): 437-452, 1978. 9. Sterling, T., "Smoking, Occupation and Respiratory Disease," Presented at the American Lung Association Occupational Health Task Force Meeting, April 9, 1980. 10. Sterling, T. and J. Weinkam, "Smoking Characteristics by Type of Employment," J Occup Med 18(11): 743-754, 1976. 11. Sterling, T. and J. Weinkam, "Smoking Patterns by Occupa- tion, Industry, Sea, and Race," Arch Environ Health 33(6): 313-317, 1978. 12. Axelson, 0. and L. Sundell, "Mining, Lung Cancer and Smoking," Scand J Work Environ Health 4: 46-52, 1978. 9 ,

Happens When Major Errors tant Report has Been ican Statistical Associa- 79. ette Smoking and Health July 1964-June 1965," - ublication No. 1000, U.S. Government Printing im that Excess Morbidity Smoking," J Am Stat Assoc ar 3. Morbidity," Smoking an General, Department of ZW Publication No. (PHS) 3rkers or Working Kill 437-452, 1978. and Respiratory Disease," iciation Occupational , 1980. ing Characteristics by 3(11): 743-754, 1976. ing Patterns by Occupa- :h Environ Health 33(6): , Lung Cancer and Smoking," -52, 1978. 723 13. Dahlgren, E., "Lung Cancer, Cardiovascular Disease and Smoking in a Group of Miners," Lakartidningen 76: 4811- 4814, 1979. 14. Axelson, 0., "Effects of Low Level and Background Radiation from Radon Daughters," Presented at Vancouver, B.C., Royal Commission Hearings, January, 1980. 15. Weiss, W., "The Cigarette Factor in Lung Cancer Due to Chloromethyl Ethers," J Occup Med 22: 527-529, 1980. 16. Pinto, S., et al., "Mortality Experience of Arsenic-exposed Workers," Arch Environ Health 33: 325-331, 1978. 17. McDonald, J., et al., "Dust Exposure and Mortality in Chrysotile Mining, 1910-75," Br J Ind Med 37: 11-24, 1980. 18. Hammond, E. and I. Selikoff, "Relation of Cigarette Smoking to Risk of Death of Asbestos-associated Disease Among Insulation Workers in the United States," Biological Effects of Asbestos, ed. P. Bogovski (Lyon, France: IARC, 1973), pp. 312-317. 19. Selikoff, I. and E. Hammond, "Multiple Risk Factors in' Environmental Cancer," Persons at High Risk of Cancer, ed. J. Fraumeni (New York: Academic Press, 1975), pp. 467-483. 20. Hammond, E., et al., "Asbestos Exposure, Cigarette Smoking and Death Rates," Ann NY Acad Sci 330: 473-490, 1979. 21. Saracci, R., "Asbestos and Lung Cancer: An Analysis of the Epidemiological Evidence on the Asbestos-Smoking Interaction," Int J Cancer 20: 323-331, 1977. 10

724 CURRICULUM VITAE Theodor D.Sterling EDUCATION B.A. (CUM LAUDE), 1949, M.A., 1952, University of Chicago; Ph.D., 1955, Tulane University. PROFESSIONAL AFFILIATIONS Present: University Research Professor, The Faculty of Ir.terdisciplinary Studies and Department of Computing Science, Simon Fraser University. Prcvious: Visiting Professor, Department of Statistics, Princeton University, (78). Chairman, Department of Computing Science, Simon Fraser University, (72-77). Professor in the Department of Applied Mathematics and Computer Science, Washington University, St. Louis, :•fissouri (66-72). Also Joint appointment as Professor in the Department of Sociology (66-68), and Visiting Professor in Cocputers and Humanities, Hebrew Union College, Cincinnati, Ohio (68-70). Professor of Biostatistics and Director of the Computing Ccnter, College of Medicine, University o~ Cincinnati, Ohio (58-66). Previously instructed in the Department of Statistics, Michigan State University and Department of Mathematics, University of Alabama. fr,r Computing Machinery (72-3, 75-6). Physicists in Medicine (66-69); National Lecturer for Associatioli o t Commlttee on Accre tat on, ssoc a on or mpu g. i (G6-67); Committee on Radiation Dosimetry, American Association of* tin *lachiaery- i f C d i A i Association for Computing Machinery (63-71); Chairman, Ad Yoc Chairman, Committee on Professional Activities of the Blind of thej (70-72); Member of the Panel for Biology, Management, and Social~ , Sciences of the Mathematics Association of America (62-67);- of Kissouri Chapter, American Association of Workers for the Blind`^=' Groups, Humanization of Information Systems (73-80); Chairman, SIGCAS Committee of Information and Public Policy (72- ); President Processing Organization (64-65)• Chairman, CIPS Special Interest.,j President of the Computer Science Association of Canada (75-80);'~r~' Chairman, Ombudsman Committee, Canadian Information Processing- Society (73-80); Chairman, President of Biological Information5 PRO?ESSIOhaI. ACTIVITIES E^ITORIAL ACTIVITIES: W Associate Editor - Canadian Journal of Statistics (73-78);u Editorial Board - International Journal of Biomedical Computiaj (69- ); Computers and Applied Mathematics (73- ), Rumanist ia'_ Canada (73- ).. '~ GOVEI'NMEhZ' AND OTHER SERVIC Consultant/adviser: ( Cormission (B.C.); C Professional and Envir FTC; SRA; U.S. Nation Cour.cil; American (Wisconsin); Environm Citizens Against Toxi Health. P"Or=SSIONAL AND HONORARY S Fellow, Amer. A--er. Scat. Assoc. Assoc.; Assoc. Comp. Scc.; Inst. Math. N.Y. Acad. Sci.; . 0;:thamology; Sigra %i; National Science.

-ling , University of Chicago; Ph.D., Professor, The Faculty of srtment of Computing Science, rtment of Statistics, Princeton ment of Computing Science, Simon sor in the Department of Applied e, Washington University, St. t appointment as Professor in -68), and Visiting Professor in ion College, Cincinnati, Ohio cs and Director of the Computing 3versity of Cincinnati, Ohio in the Department of Statistics, tment of Mathematics, University ~ssociation of Canada (75-80); :anadian Information Processing ident of Biological Information iairman, CIPS Special Interest :ion Systems (73-80); Chairman, 9 Public Policy (72- ); President :iation of Workers for the Blind r Biology, Management, and Social sociation of America (62-67); al Activities of the Blind of the ery (63-71); Chairman, Ad Yoc ociation for Computing Machinery simetry, American Association of National Lecturer for Association 6). arnal of Statistics (73-78); Journal of Biomedical Computing thematics (73- ), Humanist in 725 GOOEFATY.E;'P AND OTHER SERVICES: Consultant/adviser: (In Canada) - Environment Canada, The Royal Cormission (B.C.); Consumers Assoeiation of Canada; Society for Professional and Environmental Control. (In U.S.) - NIH; PHS; EPA; FTC; SRA; U.S. National Academy of Science, National Research Cour.cil; American Lung Association; Natural Resource Board (Wisconsin); Environmental Defense Fund, MECCA (Minnesota); Citizens Against Toxic Sprays (Oregon). (In Kuwait) - Ministry of Health. PROF=SSIOtiaI AND HONORARY SOCIETIES, HONORS, AWARDS Fellow, Amer. Assoc. for Advancement of Science; Fellow, A-mer. Stat. Assoc.; Can. Inf. Proc. Soc.; Can. Comp. Sci. Assoc.; Assoc. Comp. Machinery; Amer. Math. Assoc.; Math. Scc.; Inst. Math. Stat.; Amer. Stat. Assoc.; Biometric Soc.; N.Y. Acad. Sci.; Amer. Assoc. Phys. Med.; Assoc. Res. 0;:thamol.gy; Sigma Xi; Pi Nu Epsilon, Morrison Cressy Award in National Science.