Tobacco Documents Online

, ANi1CANCER RESEARCH 10: 971-976 (1490) 'Institute of Public Health, National Taiwan Universiry Collegc, of Medicine. Taipei, Taiwan: 'lnstitute of BiomedicalSciences, Academia Sinica, Taipei. Taiwan; jDepartment of Epidemiology. Pittsburgh University School of Public Health. Pittsburgh. Pennsylania, U.S.A.: `Veteran General Hospital: Taiepi, Taiwan; sBureau of Public Health, Departnunt of Health. Executive Yuan. Taipei: Taiwan; °Chang-Gung Memorial Hospital and Medical College. Taipei, Taiwan; 7Cathay General Hospital. Taipei, Taiwan Epidemiologic Characteristics and Multiple Risk Factors of Lung Cancer in Taiwan* CHIEN-JEN CHEN'=: HSIN-YING WU'. YA-CHIEN CHUANG'. AH-SENG CHANGZ„ KUN-TAI LUH'. HSGO-HSIUNG CHAO4. KUANG-YAW CHEN4. SHIN-GON CHEN'. GI-MING LAI°, HSIH-HSiN HUANG' and HONG-HSIN LEEs Abstract. The specific aim of tleis study was to examine epidcrniologie characteristics and ntultiple risk factors of lung cancer in Taiwan. The age-ad justed rnonality from lung cancer has been increasing since the early 1950s with a constanf male-to-female ratio of around 2.0. lnternational comparison of cumulative moratlity from lung cancer showed a much lower trmle-to-female ratio in Chinese than in other popula- tions. Signif'icantly high moratlhy from lung cancer was observed in highly, urbanized cities and the endemic area of chronic arsenicitnt in Taiwan ~S~Cmt~at3octaROnt°_ t 4E', ",~'t,.~7;r,t7 i • ROTICE This tnaterial maY be protected Of ~Ot bw mtW 17 u.S. Code)° significant socioeconomic impact with a work-year loss of 12,500 annually (1). The one-year survival rate of lung cancer patients was reported as less than 20% in Taiwan (3). The early detection of lung cancer by chest X-ray, sputum cytology and !or fiber optic broncbosrnpy remains ineffective and inefficient (3-5). Other screening methods, including tumor marker levels, still need further evaluation (6). Primary prevention and in- tervention become an important task in the control of the disease. The identification of risk factors for lung cancer is essential for an effective and efficient primary prevention. Both genetic and environmental risk factors have been related to the development of lung cancer (acto. rvhich f hava bee documented include active' a~~4te oecupational and enviionmental exposures to arse- nic, asbestos, chromium, mustard gas, radon and polyaroma- tic hydrocarbons, as well as inadequate consumption of dark green vegetables (7-11)L Populations in different areas may have different risk factors for lung cancer, and the same risk factors may be of different importance in different popula- tions and I or areas. The study of risk factors in various populations is important not only for the disease control program, but also for the elucidation of its etiological mechanism. Although epidemiologic characteristics of lung cancer in Taiwan have been described in two previous reports (12,13), there has never been a case<ontrol study aimed at elucidat- ing possible risk factors for lung cancer in Taiwan. In this report, we updated the analysis of epidemiologic characteris- tics of lung ancer mortality and incidence, and examined multiple risk factors for the disease based on a hospital - based matched case - control study. g (tpaPim.basedccass = controt srudy ca4d rAn Taipei maropolitan` areas. Alcohol drinking, coffee drinking and various types of indoor air pollution were not related to lung cancer after the agaretu smoking habit was adjusted through a multiple logistic regression artalysit. Lung cancer is one of the most important cancers in Taiwan where malignant neoplasm has become the leading cause of death since 1982. The mortality from lung cancer ranked the second in both men and women among various cancer sites in Taiwan (1):. The annual number of deaths from lung cancer was as high as 2,500-3,000 in the 1480a (2). It has resulted in a •Pmented at the Second international Conference of Antiean- cer Research, October 11-15, 1988, Saronis, Greece. Correspondence to: Prof. Chien-Jen Chen. Black Building Room 209. Columbia University Comprehensive Cancer Center. 650 West 168th Street. New York, New York 10032. USA. Key Words: Epidemiology, risk factors.ltmg cancer. 0250-7005N0 S2.O0+.40 Materlak and Metbods .Andysir of ewrmGro md wcidena rver. The data on lun; ancer deaths 971

. ANTICANCER RESEARm 10: 971-976 (1990) Tabk I. Secular rtend of sge-aoward bag cancer ^wrwl+0' +~ per I00.A00'pop+Jocon from 1934 n 1988 ib T.iiwn by sex. Age-adjuped mortatity, Mak-tofemak Year Mak Female ratio 1954-1958 3.88 2.07 1.87 1959-1963 6.49 3.53 1.84 1964-1968 10.35 5.94 1.77 1969-1973' 13.97 6.68 2.09 1974-1978 16.65 8.29 2:01 1979-1983' 21.79' 10.42 2.09 1984-1988 24.91 12.22 2.04 ma, 134 (41.5%) with adeooweinoma, and 9(2:8%) with other minor Wtholo1?cal types. In the univariate analycis, the odds ratio and its statiuical significance of each risk factor were assessed for three major tntbobgical types of lung cancer. Mantel - Haenszel chiquare test (18) was used to assaa the statistical signi5cantt of age-sex-adjusted odds ratios for each risk factor. As several risk factors were inter-correlated.,multipie logistic regression analysis (19) was used to estimate multivariate - adjusted odds ntios. In the regteaion analysis, only • significant risk factors observed in the umvartate analysis were included in the regression equation. BMDP statistical software was used'to estimate regTenioa coefficients through the maximum likelihood method (19). Results from 1954 to 1988 were obtained from the Tai.+an Provincial Department of Health which is in charge of the death oertification system in Taiwan. Population data for the rame period were abstracted from annual demographic statistia (14). As it is mandatory to register any event of birth, death, marriage, employment and education in household registra- tion of6ces, mortality, data are quite complete and tecurate in Taiwan. The inridence of lung cancer in Taiwan was derived from annual' reports of national caneer registry (15). As the completeness and atxuraey, of cancre registry has been sussed in Taipei City only. the incidence data analyzed in this report were limited to tbose of Taipei City. The mortality from lung cancer in 17 selected countries was obtained from the annual vital statistics published by the World Health Organization (16). tnci- dence rates of lung cancer among Chinese males and' females i6 San Francisco. Los Angeles. Hawaii. Hong Kong and Sbanglui.vere ab- stracted from the registration data published by the International Agenry for Research on Cancer (17). In the analysis of mortality and inodena, age-sex-specific rates were first cakvhted' for differenr areas and / or calendar years. The age was stratified into 15 6ve-year groups from less than S to 70 or more. Age-adjusted mortality and incidence rates were calculated using world population in 1976 (17) as the standard population for the study of secvlar trend, migrant comparison and geographical variation in Taiwaa. while cvmulative mortality rates over the age range from 0 to 84 years were for international comparison. Hospfml-bared rnarc6ed'cntesoat.ol study. As mou patients suspected of having lung cancer are referred to teaching hospitals for confirmatory diagnosis and treatment in Taiwan, we recruited serial patients.vitb lung cancer from four major teacfiing hospitals in Taipei City. An patients were newly, diagnosed and pathologically confirmed. A total of 354 new cases were recruited and 332 (93.8%)' agreed to participate in the uudy:. Hospital controls grouPmatcbed with case on bospital, age and sex were recruited from ophthalmic patients of study hospitals with a eontrol-to- ease ratio of 3:1: Among 664 recruited eontrols. 635 (95.6%) of them agreed' to participate in the study. A structured questionnaire was vied to obtain sooo-demogr.phic characteristics and' the history of esposure to risk factors ioduding cigarette smoking. akoboi d,inlong. sea and coffee drinking. u.ren as indoor air pollution resulting from buraing incense and mosquito coik. The consumption frequency, quantity and duration were inquired for habits of cigarette smoking. alcohol dridcing, tea drinking and coffee drinking. lo addition to the questions mentioned above, the interview time and interviewer - as.ested rdiabiiity, of the interviewee's response were also included. The average interview time in minutes was 37.4 and 31.3 respectively for cases and hospital ocntrok. There were 9(2.7%) cases and 18 (2:8%) controls whose responses were rated as tmreliabk because of poor memory and / a cooperation. In all, thete were 323 caaes and 617 bospital controls available for the fmal analysis. With regard to the pathological type of tbe 3231ung aacer patients. there wene 133 (41.2%) affected with epidermoid utdnoma.47 (14.6%) with small cell nrcino- 972 Seculcr trend: The secailar, trend of lung cancer mortality from 1954 to 1988 in Taiwan is shown in Table I. The age - adjusted mortality rate of lung cancer increased strikingly during the period for both males and females; it increased significantly from 3.88 per 100,000 in 1954-1958 to 24.91 per 100,000 in 1984-I988 for males, and from 2.07 per 100,00Un 1954-1958 to 12.22 per 100,000 in 19841988 for females. The male-to- female ratio of age-adjusted lung cancer mortality remained around 2.0 during the period from 1954 to 1988. lnternarional comparison, migrant diffennceand geographic- al variation. The international comparison of lung cancer mortality' in Taiwan and 17 selected countries is shown in Table II. Males in Scotland and The Netherlands had the highest mortality from lung cancer, while males in Taiwan and mainland China had the lowest. Females in Hong Kong and Scotland had the highest mortality, while females in the Netherlands and mainfand China had' the lowest. The cumulative mortality rate of lung cancer in Taiwan ranked as the 17th and' 9th, respectively, for males and females. The male-to-female ratio of the cumulative mortality rate of lung cancer varied significantly from greater than 6.0 in the Netherlands, West Germany and Italy to less than 3.0 in Taiwan, China and Hong Kong. The comparison of age-adjusted incidence rate of lung cancer among Chinese in different areas is shown in Table IiIi. The rate for males was highest in Singapore and lowest in Taipei, while that for females was highest in San Francisco and lowest in Los Angeles. The male-to-female ratio in age-adjusted' mortality from lung cancer ranged from 1.33 in Hawaii to 3.43 in Singapore. There was also a striking geographical variation of lung cancer mortality among 361 townships and precincts in Taiwan. Generally speaking, males and females had similar geographica)' variation in age-adjusted mortality of lung can- cer with a correlation coefficient of 0.54. High lung cancer mortality was observed in highly urbanized cities as well as in the endemic area of chronic arsenicism, while low mortality was observed in tural townships where aboriginals and Hakka Taiwanese reside. There was a significant correlation in the geographical variations of lung cancer with cancers of the liver, pancreas, bladder and kidney in males and females as

Chen et al: Epidemiology of Lung Cancer in Taiwan i Table 11. Co,apa,von ofcwmdM,K Mo.mtiry f.oMlwg c,aceff in r.wm the diStribution of age and sez. The age-sex-adjusted odds .ed 17 xrerred cowurfa. ratioS for ci h d I ment of various ve o tt k Male Female Rank Counuy. CMR Country CMR L Scotland 21.22 Hong Kong 5.95 2 Neurerlands 20.04 soodand 4.40 3 England & wakst8.61 Singapore 4-37 4 Hong Kong 13:97 England & wales 3.97 5 Singapore 13:87 USA 3.43 6 Hungary 13.82 Ireland 3.24 7 Austria 13.52 Canada 2.79 8 Canada 13.16 Hungary 2.56 9 USA u.04 Taiwan 2.52 10 West Germany 12.99 Israel 2.39 11 Australia 12,59 Aussralia 2.07 12 In1y tr.47 Japan 2.06 13 treland 11,05 Austria 2•os 14 Japan 7.38 West Germany 1.56 15 Israet 7.14 Itaty 1.48 16 Chile 5.08 Chile 1.45 17 Taiwan 4.98 Netherund: 1.43 18 Cbina 1.95 Ceina 0.93 p gare e smo t ing on t e pathological types of lung cancer are shown in Table VI. j7tere was a significant association between cigareite smoking lind epidertnoid earcinomtr small cell carcinoma and adeno- '6iCidom; AL,ftJltng, with an odds ratio of 6,66, 3.59 and t;pg, respectively: Futhermore, the duration, quantity and inhalation dQ e of ci arette slrlokin were all~~si ifcantl ~ g g gn Y associated with three pathological tvpes of lung cancer in a dose-response relation. Passive smoking was also correlated ~ with the development 5P`epldtfm0ld'tltiicl11on1aL'_SijlalLCeU.~' Onta and adenocarcinoma of the lung•witK`lsigniftcant, j O nta ~ bft ratioa of 4.68, 2.55 and 3.04, respectivelV The age-sex-adjusted risk of developing various patholb- gical types of lung cancer for alcohol drinking, tea drinking. and coffee drinking are shown in Table ViI. Alcohol drinking was significantlY associated with epudtrmoid carcinoma of the lung with an odds ratio of 1.57. Neither smalLcel!carcinoma nor adenocarcinoma was significantly correlated with alcohol drinking. None of the habits of diinking black tea. half- processed tea and green tea was significantly, associated with CMR: Cumulni.e mortality nrcs. 0-84 years (percent) any pathological type of lung cancer. Coffee drinking was found to be associated significantly with epidermoid carcino- Table Ill. Ade-adju.rcd'iuxidcnce nur per 100.000 popaladon of 6rna ma of the lung only. cancea maona Chmoe sn rir acr: N Asia and USA. Table VIII shows associations between various kinds of indoor air pollution and pathological types of lung cancer. Me-adjumed mortality Male-to-female Neither burning incense at home nor type of cooking fuels city Male Female ratio was related' to the development of any type of lung cancer. Singapore 68.0 19:8 3.43 San Francisco 57.8 25.1 2.30 Hong Kong 55.5 23.4 2.37 SAang6ai 51.2 18.1 2.83 Lot Angeles 33.8 13.6 2:49 Ha.vaii' 31.4 23.6 1.33 Taipei' 27.7 14.44 1.92 palhOl0g1 F gegrGSStoal was txrri out foJ, af} t , a ~ C~ ~ , w ~ Table IV. Ecological ro.n/udon beeween oae.Outted nwnalirr rora of~,~ Im'g cancn and ot6e~ cancers in 361 townksiy/u and prorrncn in rawanL a~. ~~b ! 8°~g th l i l o og ca ree pa types .w t h t e t h i d h were significantly assouate correiation wtak Female Lung vs. liver 0.17 0.24 Lung vs. , panaca 0.29 011 Lung vi. , Madder 0.33 0.74 Lung vs. kidney 0.24 0.66 Lung vs. prostate 0.29 - well as with cancer of the prostate in males, as indicated in Table IV. Cnre-conrrol sardy: Table V shows frequency distributions of age and sex of 133 epidermoid carcinoma, 47 small cell' carcinoma and 134 adenocarcinoma patients and of 617 ophthalmic hospital controls. They were all comparable in Burning mosquito coils at home was found to be significantly associated with the development of epidermoid carcinoma il and adenocarcinoma of the lung, with an odds ratio of 1.81 and 1.70, respectively. As risk factors significantly associated with various patho• logical types of lung cancer were inter-correlated, a multiple logistic regression analysis was ernpolyed to assess multivari: ate - adjusted odds ratio for various variablesW~~Imte 'of lung Ca4_C iqking,QRffee drinkiag alld btlrning 'inosquito,caih~ hom.ea.were noi significantly associatedwith thological type of lung cancer after cigarette smoking , 0% ~ ~~ ed. . laaeasing secular trend and significant geographical varia- tion are two interesting epidemiologic characteristics of lung cancer. This suggests the importance of environmental fac- tors in the determination of the disease. In this study, we observed an increase in lung cancer mortality in Taiwan since the early 1950s. The result is consistent with those observed in most oountries (20). The increas in age-adjusted lung cancer mortality in Taiwan may be attributable to improved diagnostic methods, increased consumption of cigarettes. 973

i ANTlCANCER RESEARCH 14: 971-976 ()990) Table V. Agt and sex disrribuuo- of 133 epidtrnroid carcinoma. l7sma11 cell carcinoma. and 134 adnacamnoma patients and'617 opthalmic hospital controlr in mrtropolitaA Taipei anar: , Variablc Group Epidcrmoid carcinoma No. (%)' Small cell carcinoma No. (9'0) Adeno- carcinoma Nb: (%) Hospital controls - No. (%) Age < 55 21 (15>8) 17,(36.2) 38 (28.4) 146 (23.7) 55-64 59 (44.4) 14(29.8) 50 (373) 241 (39.5) 65 + 53 (39-8) 16 (34.0) 46 (34.3) 227(36.8) Sex Male I l I($3:5) 36(76.6) 101 (75.4) 48R'(79.1) Femalh 22(16.5) 11 (214') 33 (23.4) 129Y20.9) Table V1. Agr-sxr-adjusred'odds ratios for cigarenr smoking on the development of three pathological types of lung cancer. V i bl G Age-sex-adjustcd odds ratio e ar a roup Epidermoid carcinoma Small cell carcinoma Adeno- carcinoma Habit Yes 6.66'•• 3.59^` 2:08'•• No 1.00 1.00 1.00 Duration (~can) 41+ 8.43"' 5.12'•• 3:79'•' 3140 6.52' 4.30 1.60 21-30 2.76 3.33 1.52 1-20 1.70 2:16 1.23 None 1.00: 1.00 1'.00 Quantit% 31,+ 11.11'•' 8:09'•• 3.61•" (ng./day) 21-30 7.61 . 4.64 2.34 11-20 7.05 3.48 1.74 1-10 ~ 2.59 2.45 1.21 None 1.00 1.00 1.00 Inhalation Deep 723•" 4.37•• '-'1• ShallOw 3:67 3.57 1.92' None 1.00 1',00 1 1.00 Passt.ciiooking °: s* Ye- 4.68`.'i 2.55', 3.04'.. r ~ Nb : . 1.00 "k 1.00 1.0(1 • p < 0.05. "p < 0.01. '•' < 0.001 based'on Mantel - Haensttl dii-square tests. + Table V11. ARr-sez-adjuued oddr raios for &-utragr drinking on the dtvelopmnu of'thrn pathological types of lung cancrr: V bl i Gr Age-sex-adjusted odds ratio ar a o oup Epidennoid pra"Offa Small cell carcinoma Adcno- carcinoma Alcohol drinking Habit Yes 1.ST 1'.36 1.17 No 1.00 1.00 1.110 Frequency 4+ D.72' 1.90 1.50 (da)siwcck) 1-3 11.43 0.90 _ 0.99 None 1.00 1.00 1 00 Tea drinking Black tea Yes 0.20 1.10 0:19' No 1.00 L00 1.00. Half-proeessed tea Yes 1.52 0.99 0.99 No 1.00 1.00 1.00. Green tea Yes 1.48 i.20 1.77 No 1.00 1.00 1.00 Coffee drinking Yes 2.10* 1.44 1.25 No L00 1.00 1.00 • P < 0.05 bued'on Mantel - Haenstel dti-aquare tests 974

t Chen r/ a!: Epidemiology of Lung Cancer in Taiwan Table VIII'. Aae:-ses-adjuned odds rctior /or verious rypa oj iuedoo. air pollution oe rlu development oJ dem padholoaicaf rypa o/lun; cancrr. V blr Grou Age-tex-adjiwcd odds ratio ana p Epidermoid carcinoma Small cell carcinoma Adeno- carcinoma Bumin e mccnsc Yes 0.77 1.33 0.99 _ No 1.00 1.00 1.00 Burning mosquho Yhs 1.81' 1.13 1.70' coils No 1.00 1.00 1.00 Cooking fucas Wood & coal 0.85 1.08 1.02 Chatcoal. gas & clectncitv 1.00 1.00 1.00 p< U.1K tiascd on Mantel - Hacnszcl ehi - square tesu.. rapid industrialization and urbanization, and worsened air pollution. International comparison of cumulative mortality from lung cancer revealed a striking difference in the male-to- female ratio among 18 countries studies. The reason for a much lower male-to-female ratio among Chinese in various countries deserves further investigation. As most Chinese women are non-smokers and 60% of female lung cancer patients are affected with adenocarcinoma (12), ir seems reasonable to suspect that risk factors other than active cigarette smoking are involved in the development of adno- carcinoma. The striking geographical variation in lung cancer mortality among 361 townships and precincts in Taiwan also suggests the importance of environmental factors. Heavy air pollution resulting from urbanization and industrialization may at leastpartly contribute to the high mortality from lung cancer in cities. Consumption of high-arsenic artesian well water has been documented to be the major risk factor for lung cancer in the endemic area of chronic arsenicism (21). The significant ecological correlation between mortality rates of lung cancer and pancreas cancer may be explained by the better dignosis of the latter in urbanized areas. However, the similar geographical variation in mortality from cancers of the lung, liver, bladder„ kidney and prostate may be attributable to their associations with arsenic exposure from drinking water.. ~~ ~I.n. our cese-control study, both active and passive cigarette, ~"-"""~i g e 'cantl~ assoctatedwitb,the developmen{ o~t~i°tee~~o>~` t` pl of. lun~ ~cancer.7Almost all epldemiological studies and animal experiments consistently show a significant association between cigarette smoking and lung cancer (7-9. 22). Both epidermoid carcinoma and' small cell carcinoma had' a stronger association with active cigarette smoking than adenocarcinoma in this study. This observation is consistent with those reported previously. However, there .vas no difference in the association with passive cigarette smoking for the three pathological types of lung cancer. Futher investigations are needed to explain such a discre- paricy- Habits of alcohol drinking and coffee drinking were signifi- candy associated with the development of epidermoid carei- noma of the lung in the univariate analysis in this study. However, the association was no longer significant after further adjustment for the effect of cigarette smoking, Indoor air pollution has been documented to cause lung cancer in Yunnan Province of China (10): In this study, an effort was made to assess the effects of various types of indoor air pollution on the development of lung cancer. Although a significant association between lung cancer and burning mos- quito coils at home was observed in the univariate analysis, it became not significant after active and passive cigarette smoking were adjusted for. The reduction of cigarette smok- ing through public education remains the most important task for the primary, , prevention of lung cancer. AcknowkdYemwts This ctndy was supported by a grant from the Department of Health. Executive Yuan. Republic of China. Dr.,C.J. Chen is a Fogarty Research Fellow sponsored by the US National Institutes of Health. Referencrs 1 Chen CJ, Lee SS. Hsu KH, Ttai SF. You SL and Lin TM: Epidemiolbc pc characterisria of malignant neoplasms in Taiwan: I All cancer sites. J' NaU Public Health Ataoc (ROC) 8; 39-21. 1988. 2 Department of Health, Executive Yuan, Republic of China: Health statistia. Vol. 11 ' Vital uatistiet, 1981-1968. Taipei, Department of Health, 1982-1989. 3 Yang SP. Lin CC. iVu MC. Lub KT, KuoSH. Wu YT and Li TS: Prevaknce survey of primary lung cancer in Taiwan. J Formosan Med Aswc 75: 429-434, 1976. 4 Weiss W, Bouea KR and Ssidmrn H: The Philadelphia pulmonary neoplasm research project. Clin Cbeeu Med 3: 24}236. 1982. 5 WoolnerLB. Foetana RS, Sanderson DR. Miller WE. Taylor WF and Uhknhopp MA: Mayo lung projcct: evaluation of lung cancer screen- in8 through December 1979. Mayo Ctin Proc 16: , 544-547. 1981. 6 Rosen ST and Rido.evicb JA: Biobay of lung canttr: In: Bitruf ID. Gokanb HM. littk AG and Weichselbaum RR (edi). Lung cancerr a comprehensive ueatire. Grune d Strattoo. Inc.. 1988, pp 35-54. 7 Doll R: EpidemiobW. In: Wyoder EL and Hedit S(eds)., Lung cancer. UICC Tectinital Report Series 25. 1976, pp 3-41. 8 Fnumeni IF Jr and Blot WI: Lung and pleura. In: Schonenfeld D and Fnumeui JF Jr (eds). Caxcr epidemiology and pnwention. Phi- ladelphia, WB Saundera. 1982, pp 536353. 9 Spiro SG: Epide>miob8y. In: Hooptnten B. Addia B1. Hansen HH. Marvtu N and Sp'uo SG (eds), Lung tumota. BerLn. Springer- VeriaB. 1988. pp 3-8.. 10 Chen G, Kuo TL aad Wu MM: Arsenic and cancers. 11 ncet l: 414413, 1988. 975

c ANTICANCER RESEARCH 10: 971-976 (1990) 11 M'untford JL, He XZ, Cbapman RS; Cao SR. Harris DB. Li XM, Xian YL. Jiang WZ, Xu CW, Chuang 1C, Wilsom WE and Cooke Mi Lung cancer and indoor air pollution io Xuan Wei. China. Soence 233: 217-220, 1997, 12 Luh KT. Kuo SH. Lin CC, Yang SP and Chen KP: Primary lung cancer in Taiwan: part 1_ Cbronological observation of epidemiologic- al charactertistia with etiological consideration. 1 Fottnosatt Med Assoc 73, 129-146- 1974. 13 Tay SC. Tsai'SF. Lee SS, Hsu KH, Lin TM and Chen: G: Epidemiolo- gie characteristics of malignant neoplasms in Taiwan: IV. Lung cancer. 1 Nat1 Public Health Assoc (ROC) 8: 189-201. 1988. 14 Ministry of Interior. Republic of China: Demographic f.eta, 195¢ 19g8. Taipei! Ministry of Interior, 1955-1989. 15 Department of Helath. Executive Yuan. Republic of China: Cancer registry annual!rcpon in Taiwan arca„1983-19BG, Taipei. Department of Health. 1985-1988. 16 World Health Organization: World health statistia: annual vital statistics and causes of death. Cxneva„Wotid Health Organization, 1985. 17 International Agency for Research on Cancsr and International Association of Cancsr Registries: Cattocr iacidencs in five continents. Lyon, International Agency for Research on Cancer, 1983: 19 Mantel IN: Chi-square tests with one degree of freedom: extensions of the Mantel-Haenszel procedure. 1 Am Stat Assoc 58: 690700, 1963. 19 Breslow NE and Day NE: Statistical methods io cancer researdt. Vol. 1. The analysis of case-control studies. Lyon, lnternational Agency for Research on Cancer, 1980, pp 192=247: 20 Doll R: Cancenof the larynx and lung. ln: Magnus K(ed), Trends in cancer incidence: causes and practical imPlieations. Washington, Hemisphere Publishing Corporation, 1982, pp 183-184. 21 Chen CJ, Cbuang YC. Lin TM and Wu HY: Malignant neoplasms among residents of a blacktoot diseaseendemic area in Taiwan: high-arsenic artesian well water and cancers. Cancer Res 45: 5895, 5699. 1995, 22 Loeb LA, Ernster VL, Warner KE, Abbotts 1 and Laszlo ]: Smoking and lung nncer: an overview. Cancer Res N:,5940-5438. 1984. Received February 28. 1990 Accepted April 18, 1990 N C N Cj CJ W N O); ~~ 976