Tobacco Documents Online

I I G- )U 110 4W w This matttigl am be a0loc bY copnight krr ttiW u u,s.coW 5~ Wofl Lung Cancer in Women in the Niagara Region, Ontario: A Case-control Study t ERIC J. HULOWATY, nt.u., F.R.C.P.C. rat sc.,'a HARVEY A. RISCH, nt.D. PH.o.,= ANTHONY B. MILLER, tzt.n., F.R.C.P.,' J. DAVID BURCH, M.A' A case-control study of the etiology of lung cancer in wromen K•as conducted in the Niagara Region of Ontario, licc•ause of loc•al concerns about a high incidence of huig cancer. 51 fenlale patients with lung cancer and 45 matched controls were intervieH•ed. lnformation mas col- lected about acrire and passive s»loking, occffpation and residential hi.ctory* . There wa.c a strong association between active cigarette smoking and ifurg cancer (everlnerer odds ratio 10.0: p < .001) and 85% of the cases of lung cancer were attributed to active cigarette smoking. No other factors were signifc- canNv associated trith lung c•ancer: there tc•as weak evi- dence of uu association hent•eaf urban environment during childhood and lung c•ancer (p = 0.07). Associations between lcutg cancer and air pollution, and residential his- tofy, ivere not deiuonstrated, contrurv to public perception. Thus, a previouslv reported excess of lung cancer in Niagara feonales is urost likely-attributahle to cigarette snloking. In 1984. an exploratory study of cancer mortality in the Niagara Region of Ontario (Figure I) reported a statistically significant excess of fatal lung cancer in Niagara women (but 1. The Diviainn or F.rirkminingy and SWliatict. The Ontario Cancer Trcalmcnl aml Re.earch FiwurLiliixlk and the Deranmelu of Preventive Medicine and Binnlalialicx. tlni.rr,,ilv nf Tnnwnn. Tonrqn.Olnarin. 2. N:ninnat Cancer hntitutc or Canada F.rirkminh.Fy tlnil. Ikpanmem of Precentive Medicine and BlrWall'ltca, tlni\Yrtlly of TnrMNn. TnfMnn. Qllarin. J. Dcprlmenl of 14c.rntive Malicine and Bitwa/iaic.. Unititr.ily nf Tnnnnn. Toronto. Omario. 4. AnrAnr /..r r•..rrr..pnrpfrncr and rrprinl rtqncac Uiviainn or F.pidcmirdnFy and Slaliaica. The Ontario Cancer Treatment and Research Foundalinn. 7 Overlea Bnulcva.d. Tnn.wn. fhnarin. A1JI1 1 Alt. PIM+nc: (416) 42/J2.10 Thin rc.earch vra% wpry+nnl hy furnk frnnm the National Cancer Inailate of Canada and the National Health Reeearch and DevekrnxyN PmEram. .M4 Canadian Journal of Public Health Une Etude de cas contr8lie sur l'Etiologie du cancer du poumon chez la fenune a Nf rfalisEe dans la rfgion de Niagara (Ontario) en raison des prfoccupations locales touchant l'incidence flevcce de cette maladie dans la rfgion. 51 patientes atteintes d'un cancer du pournon et 45 sujets tecntoins ont fail 1'objet d'une entrevue. Des donnfes onr Etec recweillies sur la consommation active et passive de tabac, l'occupation et le lieu de rfsidence des patientes et des sujets tEnnoins. La consonunation active de cigarettes a Erf fortentent assocife aff cancer du pounion (rapport consonunation antfrieure/aucune consonrmation de 10,0; p< 0,001) et 8,576 des cas de cancer du poumon ont ftf attribufs d/a consonrnlation active de tahac. Aucun autre facteur n'a ftf associc' de fayon significative an cancer du pounon; un foible indic•e d'association entre la rEsidence en neilieu urhain pendant 1'enfance et le cancer da pafmton a ltf constatE (p = 0,07). Contrairement d l'opinion gEnfrale r0pandue. aucune association n'a pu ltre ltablie entre le cancer du poumon et la pollution atntosphfrique ou le milieu de rfsidence. La prfvalence llevle du cancer du pouulon chez les fanines de la rfgion de Niagara signalfe par le passf est donc vraisemblablentent attribuable d la consonrnlation de cigarettes. not men), compared with all Ontario women, over the interval 1976 to 1981 ' Because of local concerns about this excess; a case-control study of the etiology of lung cancer in Niagara females was initiated by the Niagara Health Services Dcpanmcnt in 1984. Tobacco smoking has been clearly established as the major cause of lung cancer in both sexes in North America. Over 40,000 reports and papers have been published about this association: " including over 40 case-control studies and at least 8 cohort studies. Most lung cancer in men, and an appre- ciable proportion in women, is attributable to cigarette smok- ing (Table [). Vol. 82, September/October 1991

Most of the analytic epidemiologic studies of the associa- tion between smoking and lung cancer have been conducted in densely populated urban areas; few have been conducted in less densely populated areas such as the Niagara Region' Evidence supporting a causal association between lung can- cer and occupation is fairly strong for certain occupations and substances, including underground hematite mining, nickel refining, uranium mining, soots and tars, asbestos, arsenic and chromium." Evidence is suggestive for several other sub- stances and occupational groups.' There is also suggestive evidence of an association between lung cancer and passive smoking."" Evidence concerning a causal association between air pollu- tion and lung cancer is weak and conflicting,==" but air pollu- tion remains plausible as a cause of lung cancer'= because: (a) there is an urban faclor for luhg cancer that may not be completely explained by smoking habits and occupational exposures;'l' (b) there is strong evidence from occupational TABLE I Population Allributable Risk l'ercent" for Cigarette Smoking and Lung Cancer fur Women and Men from Case-conlrul and Cohort Studies Sludy potwlalkm t'AR% 1Vanen Aten lyean of rnnAnkm t fidlow•-np) U.S.A.-19SM)cax-mHnil' ?S:; i4dSr U.S.A. - I959/73 cuMxl'" 36'4 83% U.K. physkian. - 1951n3 ovMwiO1: 671a 512°A Japan- IS6S/74 cdNxt" 24~w 72~6 Wesiem t.'un~ - 197NII0 ra.e-.tiwNrul" aK'.~ 1tS ;i• Engl:uxl-IH77/IS2ra.e-c.wund" 7.11; >tN~i t.usAngeles-192{I/It2 ru.c-vt„nnd" 7SC: - Wa.i.m N.Y. Siaic - 198404 ca.e-roNUnd" 71!7 SKSF ing aquantitative relationship between ambient air pollution levels and lung cancer."", We report the findings of a case-conarol study about the possible etiologic association between the above factors (i.e. tobacco smoking, ambient air pollution, passive smoking and oceupational exposures) and lung cancer among female resi- drnts of Niagara Region. M ETHOUS Female residents of Niagara who were newly diagnosed with primary lung cancer between January I, 1983 and March 31, 1985, and who were tunder 75 years of age at diagnusis, were eligible. Cases were identified through chan reviews at 7 of 9 public hospitals in Niagara (two hospitals refused lo panici- paue), and at thc major rel'erral hospitals in Turontu. 72 eligible cases were idenlilied. Of these, auentling physicians refused pennission to contact 3. 6 could not be located or had nwved out of the region, and 12 refused to Ik interviewed. Thus, 51 cases, representing 71~k• ol' eligible cases identilied for this study, were interviewcd. Of Ihosc cases participating in the study, histologic confinnation was available for 88% (45/51). At the time of interview, 53% (27/51) of cases were alive. In the event of death, next-of-kin were interviewed. Female controls were randomly selecled from the municipal assessment lists for 1983 and 1984. They were individually matched to cases on the basis of age (within 4 years), and municipality ol' residence at one year prior tu the date of diag- nosis of the corresponding case. Malching on municipality (i.e. city, town, township) was necessary becaume 2 of 9 public hospitals in Niagara refused to panicip:ne, leading to substan- tial under-enumcration of cases from certain municipalities. In the event that an eligible control could not he located or refused to panicipate, a second eligible control was selected in the same manner as the first. Of 97 potenti:dly eligible controls selected, 2(1 (21%) could not be located or had moved out ol' Niagara Region. Of the 77 studies that pro)onged and heavy exposure to certain airborne eligible controls crNnacted, 45 (5K1;(.) agreed tu panicipcue in the substances (e.g. soot, tar, benzo(a)pyrcne, usbestos, vinyl study. All controls were alive at the timr of interview, but one chloride) may cause lung cancer,'•~ and (c) there is weak evi-. iVas tco ill to Ik inicrvi..wcd, and a proxy was interviewed dence from ecologic and other observational studies suggest-' instead. t C R

TABLE II Age-adjusted Odds Ratios for Lung Cancer in Niagara Women in relation to Cigarette Smoking Status Nurnhcr of Cigarette Smnking Statu." (hhl. Ratio t9Y;i CI) Cascs/Gmlrols Never .nKbl.er 1.41 5/27 Evcr urwt.cr 10.01 s.5-.t2.) 46/18 Ex-smnkcr 1.61n.2R-9.(/) 6/I() Currcnt snti+kcr 33,0(7.S-3.J(1.) 41y8 • Status at ~ax ~crr pri~a Io dalc of diag~xhis of casc. Cases and contrals wcrc interviewed in their own homes by an experienced intcrvicwcr, using a standardized qucstion- nairc. Infonnation was collected on lifetime residential and occupational history. lifctimc tobacco smoking history, cxpo- surc to passive tobacco smokc, personal and spousal exposure to occupational substances and suspect industfies, and various socio-dcmographic variables. Statistical analysis was conducted mainly by mullivariatc logistic regression modelling. Subject matching used in the design was preserved in the analysis. Because 6 of the cases could not be individually matched to controls, they were matched to controls within existing matched pairs to create 6 matched triplets. The PECAN computer program." which per- mits analysis where there is a variable number of subjects within each matchcd scl, was used to derive conditional maxi- mum likelihood cstimalcs of the paramctcrs. Because of the relatively small number of subjects in some of the exposure categories, confidence intervals were calculated using the log likclihoaxt-bascd pruccdurc." RESULTS Srx•iu-drmr~,kvaphic 1'trriuhlcs As expected. there was no significant difference in agc, comparing cases with controls. The mean age (at diagnosis) of cases was hO.!i ycar% and the mcan age of controls was 60.4 years. Currcnt annual family incomc. marital status and reli- gion were found to he comparable between cases and controls. TABLE IV Adjusted Odds Ratios for Lung Cancer, in relation to Passive Smoking Number of Exfocurc• OJds Ratiot (95rk Cl) Cascs/Ccmlmis Anyone in samc hixtschuld 3.6 (0.39-38.) 46/35 evcr snw>f;cd Mother ever snx+ked 0.76 (0.(16-8.3) 7/S Father ever smnkcd 0.67 (0.15-2.7) 19/23 Husband ever smokcd# 1.04 (0.27-4.1) 36/25 Ever cxrnsal in workpLsccil 0.37 (0.11-2.7) 14/12 •status at datc or diagnosis of casc. twith adjusuncnl fnr lifetime ccrosumptirm of cigametlcs (pack-ycars). anl uaing `ncvcr" calcgury as hasclinc. tncvcr-marricd suhjccts were excluded. §suhjccts who had never worked outsidc or the house were excluded. TABLE III Trends in Age-adjusted Odds Ratios ror Lung Cancer, in relation to Smoking Exposures considered : as Continuous Variables CixuirnMnis SnNdcing Variahlcs' Odds Ratiot (9S% Cl) Duration smoked (per 10 years) 1.97 0 .4R-3.1) Average frequency (per 20 cigarcncs/day) 23.0 (3.6-190.) Current frcyucncy (per 20 cigarettcs/ilay) 8.1 (3.2-27.) tli Lifetime consumption (per 20 pack-years) 6.7 (2.6-32.0) ~9 Age slarlcd smnking# (per 10 years) 0.54 (0.25-1.10) tp Years since quilting (per H) years) 0.16 (0.028-0.48) rJ •status at one year prior to dale of diagnosis of case. tcorreslxmding to the unit of change specified for each variable. #as adjusted for average frcqucncy (cig./day). Ac•lirt• Cigarclle Smoking Cigarette smoking was strongly associated with risk of lung cancer in Niagara women (Table 11). Strong risk gradients were seen when tobacco smoking was represented by continu- ous variables (Table I11). Ex-smokers (defined as smokers who quit smoking at least one year prior to the datc of diagnosis of the case) had a risk of lung cancer significantly less than current smokers (OR = 0.05; 95% Cl = 0.002-0.30). In comparison to non-smokers, their risk was elcvatcd, but this finding was not statistically significant (OR = 1.6; 95% CI = 0.28-9.0). For ex-smokers, their risk decrca.ced with increasing time since quitting (Table 111). Population attributable risk (PAR) was estimated for cigarette smoking, using lifetime consumption (pack-years) and assuming a monotonic multiplicative model." The esti- mate of PAR was 85%. mostly due to current smoking (80%), rather than cx-smoking (5%). Pas.circ Smoking (Ern•irrriuucrrlal Tolxicco E.)pa.cnrc) Overall. 90% (46/5I ) of cases and 78% (35/45) of controls reported sonie past exposure to passive tobacco smoke in their homes. After adjustment for personal lifetime cigarette con- TABLE V Adjusted Odds Ratios for Lung Cancer in relation to Residential History Number of Resicknlial History• Odds Raliot (95% CI) Cases/Controls Uru-ruinn lircd in Niagara RcRinn <40 years 1.0 22/28 >40 years 6.1 (0.96-102.) 29/17 cont.# (per 10 ycars) 1.22 (0.82-2.4) Chilrlhrxxl rtsidrnc•t . country/small town 1.0 20/29 large lown/city 7.8 (0.62-98.) 31/16 Arlrrlr Re.ridtoH•e country/smalt town 1.0 14/16 large town/eity 2.3 (0.49-13.7) 37/29 •status as of rxrc year prior lo dalc of diagnosis finchKling adjustment for lifetime cigarette eonsumplion ()+ack-years) tcontinuous variable, with OR corresponding to the unit of change speci- fied 306 Canadian .lournal of Public Ilealth Vol. 82

TABLE V1 Exposure to A Priori Suspect Occupations Suspect Mdustries Study Subjects Cases/Controls Spouses of Study Subjects Cnses/Controls Study Subjects C•rxs/Cuntrul. Spouses of Study Suhjecls Cusrs/Controls 1. Asbestos mining. Mfg. or Use O/I 19. Maintenance Work/Sup't 1/1 -1/I 2. Asphalt PYvinf 3l0 20. Ncwsprinlor Il0 3. Beuutician/IIairJrcsser I/I 21. Nickel Refinery tl6 4. Beryllium Mining 22. PP-simer/P~~int Facarcy (I/I 2/1 5. Bricklayer 2/I 23. P{:r.tics F:k'hxy I/I 6. Ceramic/En:mtet Products 7. Chemical Industry 2/3 21. Rubber Mfg. iw Use 25. Shm Meial Work I/0 8. Chromium Mining or Mfg. 1/0 26. Ship Yard 5/II 9. Cobalt Mininp/Relining 10. Coke Oven Worker O/1 27. Suap f•acuwy 28. Steel Industry IA) (NI OJI 4/5 11. Construction 6/6 251. Talc M:utuGkyure 12. Copper Mining/Smelter 1l[) 3(). Textilc Pn>,tuct. 4/I 13. Dental Technician 31. Trsmslx)n Industry S/3 14. F'ire Brick Plant Worker 15. Furnace Boiler Work IHI 32. Uranium Mining 33. Wearing I Ieat ProHectlve Clrnhing IN (NI 3/O 16. Gar.ge/Filling Srrtiwt 4/4 :4t. WooJwort,ing 17. Iron Mining 18. Laundry/Dry Cleaning ()ll 35. Zinc Mining sumption, there was insufficient statistical evidence support- ing an association between tobacco smoke in the household and lung cancer (OR = 3.6; p = 0.24). Further, there was no association between lung cancer and having a mother, father or husband who smoked in the same household (Table IV). Passive smoking in the household was also examined in terms of duration (years) that each subject reported exposure as a child, and as an adult. Neither exposure was significantly associated with risk after adjusting for active smoking. There was no evidence that passive smoking exposure in the work- place was associated with lung cancer. fumher weakened when the moclel wa) al+o adjusted for child- hood residence (OR = 1.4; p = 0.74). Twelve major point sources of air ptillution were identified in or around Niagara, including sources in Hamilton. Buffalo and Niagara Falls, New York.'"" No signilicant association was found (p = 0.65) between lung cancer and distance of res- idence (one year prior to diagnosis) to the nearest point source (3 levels: < 2 km; 2-4 km; > 4 km). When distance was exam- ined as a continuous variable, there was also no statistically significant asscxiation (OR = 1.(N per kilometer; p = 0.74). Residential distance 10 years prior Ic) diagnosis also showed no evidence of an association with lung cancer. Environmental Etnosnre and Residential History Because of the absence of a more direct measure of expo- sure, residential history was used as a measure of exposure to ambient air pollution, and to the more important industrial point sources of air pollution in Niagara Region. There was some evidence of an association between long duration of residence in Niagara (40 years or more) and risk of lung cancer (OR = 6.1; p = 0.06), but there was insuflicient evi- dence of a risk gradient (p = 0.32) (Table V). There was evi- dence of a statistically significant interaction between smoking (pack-years) and duration lived in Niagara (p = 0.03). When this interaction term (pack-years X years lived in Niagara) was included in the model, there was no change in the estimated effect of the primary variable pack-ycars, but the estimated Occerpulronul Factors No risk was associated with employment outside the home (OR = 0.57; 95% Cl = 0.(/6--I.0S), or with exposure to dust or funles, in the workplace (OR = 0.92; 95% Cl = 0?d-3.-{). Concerning employment in 35 a priori suspect occupations (Table VI), there was no association between any of these and lung cancer. Concerning exposure to any of I I a priori suspect substances (Table VII), no statistically significant associations were found. The occupational history of spouses was also examined, but there were no statistically significant associa- tions with lung cancer. DISCUSSION effect of the primary variable "duration lived in Niaf!ara" A numtxr uf cavecus should he considered tkliwe conclu- reversed (OR = 0.88 per 10 years lived in Niagara; p = 0.68). sions are drawn from this study. First, because of the small There was weak evidence of atf association between urban study size, power considerations set sume limitations on the living in childhood (prior to age 21) and lung cancer, after identification of important differences between cases and con- adjusting for lifetime cigarette consumption (Table V). This trols. For example, in a study of this siie, the power in detect- association did not change appreciably when the model was ing a statistically significant (i.e. a=.05 two-tailed) two-fold adjusted for adult residence or duration lived in Niagara. difference in risk is 38%; the power in detecting a threc-fold A significant association was not found between adult -' +tdifference in risk- is 73%. For this reason, the confidence lim- urban living and risk of lung cancer (Table V). The effect waS its of the estimates of effect are wide. This limitation is likely i m-r

TABLE VII Exposure to A Priori Suspect Substances : Su. pcct Suhctarnr% Sualy Suhj.wtc Caw./C'imtrok Sfrwccc of Study Subjecls CascsK'onaral. Study Suhjects Ca-ec/Conlrmtx Spouses of . Sludy suhjeetc Ca%ec/Conlrok I. Arcrnir /tpl Op) 7. hlcihamol 000 OQI ? AsInsps /1/I 2/2 x. ('idycyclic Anwnalic Ilydnxarlxxts URI n/n 3. BCAIt:K'M1111: Upt 11IU 4). 1(adiaxlive Material (1/ll 0/2 4. Cixil/Charctxil (41 2/5 1/1. Silica 0/1 211 5. CtcJ Tar 11A1 2R 11. Talc 1/1 2/1 b. D.D.T. (NI 1/2 to be a problem in most studies evaluating rather localized conccnrx, and dkxs not ncccs.arily prohibit such xtudiex. Sccoid. hecau.c only 717r of ciigihlc cases and 58% uf located controls participated in this sludy, the results cannot easily be generalized to the total female population of Niagara. However. there were no significant differences in age or histologic type of lung canccr. comparing participating cases with all eligible cascs.' Unfortunately, there was no information about smoking habits (or other possible risk fac- tors) for non-parlicipating cases. Comparison of study controls with women 45 to 74 years of age who were interviewed in Niagara's Community Health Survey in 1985.' dcmonztrated that thc study controls had less formal cducation. and a higher proportion were married, had U.K. ancestors and were currently working. Also the proportion of ever smokers was somcwhat lower in study controls than in survey females (40% vs. 49%).' This difference could have inflatcd tlrc estimate of Ihc smoking effect somewhat, but it is unlikely to have altered Ihc main conclusions of the study. And third. hccaumc of the rctrospeclive recruitment of most cases. approximately hall' of eligible cases were dead by the time of the eucrvicw. These suhjects were retained in the study, and proxies were interviewed instead. Recent work'- on the reli- ability and validity of respcmdcnt infofmalion in a case-control study of lung cancer suggests that good agreement can be expected between proxies and cases in temis of basic smoking status (i.e.. never, cx-. current). and daily frequency of smoking. In spite of the foregoing considcr.rtions, there is good evi- dencc of a strong association between cigarette smoking and lung cancer in Niagara women. The point estimate of the effect of ever smoking (OR = 10.0) is among the highest reported val- ues as.sociatcd with ever smoking. in women. Similar risks have been found in the much larger companion study of the NCIC Epidemiology Unit. conducted over a similar period;' and in a recently published cohort study" There is evidence that women are now smoking more intensely, and for a longer duration, than in years past; thus, it is not surprising that the risk in smok- ing women is now similar in magnitude to the risk reported in smoking men from earlier studies "-" It is likely that these high- cr risks now being reported for women will continue to be found in the future. The PAR estimate for lung cancer associated with cigarette smoking in Niagara women was 85%. This estimate was higher than that derived from previously published studies (see Table 1). Most of this PAR was associated with current smoking (8(N'~h), rallrcr than ex-smoking (5%). There can be little doubt that most lung cancer in Niagara women is now attributable to active cigarette smoking. There was insufficient statistical evidence from this study to support a strong association between lung cancer and other study variables, including passive smoking, environmental exposure. residential history and occupational factors. The pres- encc of an "urban factor", at least for the childhood years. was suggested in this study. However, it is known that urban living is associated with an earlier age of onset of smoking .*-' It is possible that we were unable to control completely for the smoking effect when studying this "urban factor". Contrary to popular perception; there is little evidence that general ambient air quality is poor in Niagara Region, com- pared with other urban and semi-urban areas in Ontario. In terms of the Air Pollution Index calculated for nine cities in Ontario. St. Catharines and Niagara Falls report the lowest aver.lgc Ievcls" Both St. Catharines and Niagara Falls are con- sidered by the Ontario Ministry of the Environment to have good ambient air quality, although local problems have been identified over the years"•" In tenns of local sources of pollution, distance of residence to the nearest industrial point source of air pollution did not appear to be related to risk. Small study size precluded analysis of indi- vidual point sources. Distance to the nearest point source may not be the best indicator of environmental exposure, either, as it fails to take into account wind direction and other climatic vari- ables. Matching cases and controls on municipality of residence may have resulted in over-matching on distance to the nearest point source of pollution. Atmospheric concentration of particu- lates emitted from industrial point sources decreases logarithmi- cally with distance from the source, with highest concentration within 2 km. of the source, and with levels reaching back- ground values at 4-5 km. from the source." In this study, 38% (37/96) of subjects lived within 2 km. of the nearest industrial point source, 36% (35/96) lived within 2-4 km., and 25% (24/96) lived more than 4 km. from the nearest point source. Thus, there was good variation in the distance of Niagara aub- jects to point sources and no real evidence of over-matching on this variable. 308 - Canadian Journal of Public Health Vol. 82 a!

The previously reporled excess of fatal lung cancer in Niagara females' is most likely attribuwable to active cigarelle smoking. This study failed to demonstrate a strong association between air pollution or residential history and risk of lung cancer. Acknowledgement The assistance of the Rejional Niaj:u. Health Services Depurtment is Erwefully atauwledreJ. REFERENCES 1. Satisrics Canada. 1986 Census of Canada: Final Population and Dwelling Cuunts. Minister of Supply and Services. Ou•rwa. 1987. 2. Mao Y. Semenciw R. Cancer mortality in Niagara county. On1•rrio. 1951 to 1981. Special report no. S. Chronic Diseases in Catwd•r. Hc•rlth and Wcl /'are Canada. Ottawa. 1984. 3. Niagara RegKM>rI Health Scrvtcrs Deparuncnl. Cancer and its risks. The Niagara perspective. 1986. 4. Lroeb LA. Emster VL, Warner KE, Abbott S. Laszlo J. Smoking arrl lung cancer: an overview. Cane rr Rrs 1984; 44: 5940-SS. 5. Holowaty El. A case-control study of lung cancer in Ni•rgarr women (Theais). Toronto. Ontario. University of Toronto. 1987. 257 pp. 6. United States Public Health Service. Smoking and Health. A Report of the Surgeon General. Rockville, MD. United Slates Deparlment of Ileallh. Education and Well'are, Public Ilealth Service. 01'fice on Snwking and ilealth, 1979. 7. Uniled States Public Ilealth Service. The Heallh Consetlucnces of Snwkiag for Wonkn. A Report of tlk Surgeon General. Re>Lkvillc, MD. United States Department uf Ilcalth and Iluman Services. Ollice on Smoking and Health. 1980. 8. United Slates Public Health Service. The Health Conseyuences of Smoking: Cancer. A Report of the Surgeon General. Rnckville, MD. . Depanment of Health and Human Srrvices. Oflice on Smoking and Ikalth 1982. 9. Ilaencel W, Taueber KE. Lung Cancer Mortality as Related to Resieknce and Smoking Hislories. II: White Femalea. JNCI 1964; 32: 803-3lt. 10. Hamnwnd EC. Smoking in Relation to the Death R•rtes of One Million Men and Women. In: I laonzcl W, ed. Epidreuiriluxirul dpprrKerhrs tu rht• Snulv of Cane rr and Other Chronic !)isrusrs. National Cancer Institute Monograph No. 19. U.S. Depanmenl of Ileohh, Fwluraliun and Wrlfare. National Cancer Instilutc. 1966. l t. Dull R. Pelo R. Mortality in Rclatiun lo Smoking: 211 years' Observaliunl on Male British Dorlurs. Br Mrd J 1976; 2: 1325-36. 12. DDII R et al. Mortality in Relalnnl to Smoking: 22 yearl' Ot>aervat{lNls tln Female British Ductors. Br MrdJ 19tl11; 2ttlk 967-71. 13. ilirryarna T. Snwking In Relation to the Death Rates of 265,118 Men and Women in Japan. Tokyo. National Cancer Center. Research Instiwte, Epidemiology Division. 1977. 14. Lubin 1H et al. Patterns of Lung Cancer Risk According to Type of Cigarette Smoked. Int J Cane•rr 1984; 33: 569-76. 15. Aklerson MR et al. Risks of Lung Cancrr, Chronic Bronchitis, Ischaemic Ilean Disrr.ve, and Stroke in Relation to Type of Cigarette Smoked. J £pideauol Connn Ifrulth 1985; 39: 286-93. 16. Wa All ct al. Smoking and Other Risk Factors for Lung Cancer in Wotncn. JNCI 19143; 74: 747-51. 17. Byers TE et al. Diet and Lung Cancer Risk: Findings fnMm the We,trm New York Dict Study. Ani J£pidrwii>l 1987: 125: 351 •63. 18. MacMahon B. Pugh TF. £piJe•miuh~kr: Principles and MrebrKLt. Baston: Little. Brown and Co.. 197(l. 19. lutcrnatiexwl Agency for Research on Cancer. Chemicals and Industrial Processes Assuciats•d with Cancer in I lutn:ms. IARC Mauegrrph No. V I- 29, Suppl. 4, IARC: Lyon. France. 1982. 20. Wigle DT, Cedlishaw NI:, Kirkluidgc 1, Maa Y. I).aths in Can:Klr fn>tu lung catxer due tu involuntary stned:ing. C'wt .11rd .t.sw• J 192t7;136: 9.35-5 I. 21. BkM WJ, Fraunwni 1F. Passive snoking aixl lung caikcr. JNCI 1986; 77: 993-I(1f1). 22. Shy CM. Air pollution and Iwtg canccr. lo: Mizc1l N. Coxrra P, edx. Lnn,r Canrer: Causes and Prrt•entiwt. New Orleans: Verlag Chemi International. 19l61;p 65-72, 23. Stocks P. Campbell AU. Lung cancer d.ath rates •rnxottg non-snwlxrs and gopc and cigarette snxtkan. An evaluation in nlatiun fu air pollution by pyrrne and other suh.+tancrs. Br Mrd J 1955; 2: 923-39. 24. Dcan O. Lung ccuxer and twtrrchitis in Northern Ircl:uwl 1960-62. Br A/eel J 1966; 1: 1506-1a. 25. Haenszel W. l.ewrland DB, Sirken MG. Lung cancer mortality related to nhidence and snwking history. I: White nwla. JN('1 190; 28: 947-11101. 26. Buell P. Dunn Jli. Relative impa.K M'snKd:ing and air pullutiem on lung cal><rrr. Arch £en•ir lllde 1967; 13: 291-7. 27. Hammond EC, Ikmt D. Snwl:ing and death rrtcs - rcp)rr on 44 months of follow-up ol' 187, 783 men. II: Death r•rtcs by cause. 101.4 195g;165: 1244•3//R. 38. thill R. Alnxrplk ric pmdlutie>rt and lung cats•er. l:'nrireNe /llth Prah 1978; ?t. 2i-31. 29. Winkelaein W. Levin 1-I. Air pollution and cancer. lit: Lilicnfeld Ahf, ed. Rrvirirs in Can •e•r £luJrtnndr~vr. New York: la.rcirr• 1983; Vol. 2. 30. Mills CA, Purtrr Mht. Tobacco snxd.ing. nxMUr e.h:wst fulncs, and gener- al airpwIlutiewt in nl•rtiun w lung cat>ltr u>Li.knce. C'uru•rr Krs 1937. 17: 9g1-eJb. 31. I liutsuji M. Epiekmiolugiaal study of lung cancer with spsci•rl reference to the effect uf air Ixdlutieat and snutLing habit. Bnll hest Pub lllrh 1968; 17: 237-56. 32. Iletxkrwm BE. Gurdwt FU, Mct><k 11. SewheMi 1, Martin SP. Pike MC. (.wt# cancer alxl air Ixdluliun in sexuh•crntral f.us Angcks County. .Int J £pigh•nria! 1973; 1111: a77•ItB. 33. Vena JE. Air~xdlulieat as a risk 1'cklor in lung cancer. Ant J£pidrnrie,l 1982; 116: a2 :56. 34. Malanu,ki GAI, L:uxlau li. Tirtaxia J, cl al. Cancer nwrt•rliry in an indus- trial area of Baltinxrc. 6en•irent Rrx 1'181; 25: 8-26. 35. Grcaves WW, Ram WN. Lyat 1L, ct al. R.lationship tklw.en lung cancer and dislaix•e of n-,idcme I'nxn twMrlcmws stneher slark elllucm. Aat J Arl Aled 198 1; 2: 13-23. 36. Rean WN. Varley G, Lyon JI„ Slxt(tkuw S. Lun g c:uxer mortality •rnumg resieknt% living near 11k I:I P,sn Smelter. Br! b>,1/1lr,l I47i2:3'1: 20-72. 37. Suirrr BI:. Ma>iiaum likclilxxxl littiny, of grt><ral ri.k nuxlrls to stratilistil d•rta..l/qd.Slut 1983; 32: 17?•81. 38. Cox DR.AndJrsisufBitnen•/hnu.l.e>txNMi: hkthurn, 1970. 39. Bruui 1'. Cir.rn 513• Byar 1)1', IiritueMt LA, tirlt.tir. r('. I:.Iinu/inl; the Ix><+- ulaliert allrilwlahlr risk liir tnuhipk risk facarra u.inl .a.r•a>ttued d:ua. Abn J F.)+igh•neie+l 1985;111: eXN• 1-1. .NI. Ontario Ministry of the Iarvironnwtia. 1985 Air Quality t}otr Summary - Regional Municipaliry eH' Niagara. 1 e1K6. 41. Ontario Ministry of the EnvinNmt.m. Air Quality Monitoring Repcut - Ontario 1993-151145. 42. Ptun GE. Reli•rbility and validity of parcy relmut.-d infumwtion in a c•r.e- control study ul' lung cancer. (Thcsis). Tuauuu• Ontario. University of Totxuo. 1987. 43. Risch I IA. Personal Comnmunicauim. National Caixtr Institute of Canada Epidemiology Unit. 191R1. 44. Garlitd;el L and Slellman SD. Snwtking and lung cancer in we>ttt<n: lind- inkS in a~xtnp.•ctivc study. Cunrrr Krr Iy8lt;•Ili: 6951 •5, .15. Doll R. I.ul R. 7Ih• Caaw•s ry'C'enerrr. Oxford: Oxford University Pra•se. 1981. 46. l.andrig:ut 1'1. Bakcr EL. Exphurr of childr.n to heavy metals frum snxlten: epiekmiukVy and luxic .Yar. ryuett<cs. !:'nrirnNe Krxvrrrh 1981; 25: 21u ?a. R%titived: March 111, 1989 AccclNed: f ebruary 5. 1'N10 a i(It)