Tobacco Documents Online

142 W ANG ETAL.. TABLE 11-ODDS RAIIOS (OR) AND 95% CONFIDF_YCE INTERVALS (CI) BY EXFOSURE TO ENVIRONMENTAL TOBACCO SMOKE IETS) IN CHILDHOOD AND ADULTHODD FOR NEVER-SMOKING SUBJECTS IORS FOR CHILOHOOD tAUULTH00D1 EXROSCRE ADJUSTED FOR AUULTHOOD tCHILDHOOUi EXPOSURE, REFERENCE AGE- PREFECrURE AND SOCIOECDNOMIC FACTORS. ALL ORS RELATIVE TO NO FXI'OStIRF TO E-r5 NUMBERS VARY DUE TO MISSING DATA.1 E. u ure MJe. F-lcr CumtinW' p a OR 954 CI Cuc, Cun~m/. OR 9]4 CI OR 9YkC1 Lifetime ETS exposure ' No g 42 1.00 20 43 1.00 1.00 Yes 24 72 1.22 0.5-3.3 176 364 1.15 06-2-1 1.19 0.7-2.0 Pack-yearsl ]-9 17 30 200 0 7-5.6 33 I00 0.77 0.4-1.5 1.04 0.6-1.8 10-19 2 17 0.47 0.1-24 43 93 1.14 0.6-2.2 1.13 0.6-2.1 a20 2 14 0,46 0.1-3.2 74 127 L49' 0.8-2.8 1.51' 09-2.7 ETS exposure in childhood No 12 58 1.00 64 160 I.IN) 1.00 Ye-s 20 56 146 0-6-37 132 247 1.51 1.0-2.2 1.52 I.1-2.2 Pack-ycars2 1-9 IS 35 1.42 0.5-3.9 76 768 1.32 0.9-2.0 1.43 1.0-2.1 10-19 1 12 186 0.4-8.5 27 32 228 1.2-4.3 1.81 1.0-3.3 z20 1 1 3.94 0.2-95.0 7 7 2.8Y 0.9-9.0 2.95' 1.0-8.9 ETS exposure in adulthood No 19 57 1 .00 31 70 1.00 1.00 Yes 14 57 0.56 0.2-1.4 169 337 1,03 0.6-L7 0.90 0.6-74 Pack-years' 1-9 11 33 0.76 0.3-2.2 52 122 0.81 0.5-1.4 0.81 0.5-1.3 10-19 2 7 0_85 0.1-5.6 48 98 1.00 0.6-1.8 0.90 0.5-1.6 a20 0 9 58 102 1.03 0.6-1.8 0.86 03-1.5 'ORS additionally adjusted for sex.-2Pack-years is the sum ofciganettes per day and fiang per month from pipe smoking divided by 20 times duration of exposure in childhood or adulthood. The summation was based on a emp!ncal evaluation of relative effects.-'Test of linear trend. 0.01 < p< 0.05.-4Test of linear Irend, p c 0.01. TABLE Itr-ODDS RATX)S (OR) BY PACK-YEARS OF EXPOSURE TO ENVIRONMENTAL T09ACCO SMOKE (ETS) FOR SEVERAL VARIABLES REt.ATED TO INDOOR DwE1LING5 FOR NEVER.SMOKINO 54HJECrS (ORS ADJUSTED FOR REFERENCE AGE. PREFECTVRE. AND. WHERE APPROPRIATE EXPOSURE TO El3 IN CHILDHOOD OR ADULTHOODi C C l ~k-yrs. nr ETS expaxurc _ Peckye.n er E-f5 e.p.urt aae> dxm x 0 1-9 Ip-19 ?30 0 1-9 10..19 2m Type of current dwellin ' Childhood exposure to EfS Adulthood exposure to ETS g Underground 61 182 1.00 0.76 1.11 1.23 1.00 1.97 2.54 3.J0= Standard 137 276 1.00 1.90 2.15 4.992 1.00 0.54 0.59 0.48 Type of longest lived-in dwelling Underground 123 252 1.00 1.14 1.21 1.40' 1.00 0.72 1.23 1.50' Standard 75 207 1.00 1.74 2 62 109' 1.00 0.99 0.71 038 Time weighted radon concentration (Hq/m') for 5-30 years prior to reference age <150 14 92 1.00 0.55 208 1.00 1,89 0.77 1.20 150-249 71 207 1.00 1.37 1.63 3.773 1.00 0.59 0.78 0.76 a250 40 110 1.00 0.84 1.40 3.86' 1.00 0.61 1.17 0.60 Type of fuel used in cunenl dwelling Coal 79 106 1 00 0 1.61 144 5.684 1.00 0.73 0.34 0.71 Frewood 43 141 1.00 1.16 0.80 1.00 0.98 1.04 1,62` Sticks/twigs 71 207 L00 1.57 2.56 3.78' 1.00 0.92 1.50 0.58 Indoor smoky in winter for correm dwelling Very/somewhat 29 93 1.00 1.78 2.99 ' 1.00 0.53 0.13 0-64 Slightly 64 98 1.00 1.16 2.13 5.23' 1.00 0.72 1.87 1.27 Not smoky 104 266 1.00 1.54 1.55 1.62 1.00 0.97 1.01 0.84 'Underground dwelling includes all cave-like housing styles Standard dwelling includes the standard above ground style and apanmenls.- '-Test of trend. p< Q01.-`Ibst of trend, 0.01 < p< 0.0A-4Test of vend, 0,05 < p< 0.10. (0.9,2.3). Our results were similar to other epidemiologic smdies of lung cancer and exposure to ETS. which have consictently shown a 1.2- to 1.5-fold risk of lung cancer (National Research Council, ]986: California Environmental Protection Agenc). 19971. Studies of ETS exposure in China are of particular interest. since lung cancer rates in females, who are mostly non-smokers. are high and ORs for lung cancer for active smokers compared with never-smokers have been consistently found to range from '_ to 4(Liu, 1992). The OR for ever-smokea compared with never- smokers in our complete data was 1.7 (1.2.2.3), and ORs by pack-years of smoking increased monotonieally to an OR of 4.5 for 30 and more pack-years of smoking. While ORs for active smokers in our sludy are low compared with westem studies. it nonetheless appears that subjects who are exposed passively to ETS are at a risk comparable to levels found in studies in western countries. There have been several epidemiologic studies of ETS in China. A meta-unalvsis of 6 studies reported nooverall excess risk for am ETS exposure with OR=O-91, 95% CI (0.8-I.1). although ther.- waa a significant trend of increasing risk with increasing exposurr. as measured bv number of ci¢srenes rer dav nr.d .- ~r ~r,hah- N

ENVIRONMEN[AL TOBACCO SMOKE AND LUNG CANCER itation with a smoker (Wang and Zhou, 1997). The interprctation of this meta-analyais however is problematic, since severxl studies (Gao er af., 1987; Lam a al.. 1987; Geng er of., 1988; Liu er ul., 1993: Wang er af., 1994. 1996). which predated the meta-anaiysis. were not included. Although results fmm individual studies are mixed, the overall evidence supports an increased risk of lung cancer with EI'S exposure (Table IV). Using results in Table IV and a random effects model (Laird and Mosteller, 1990; White- 143 head and Whilehead. 1991), a summary OR with 95% CI for exposure to ETS is 1.3 (1.1.1.7). For 0, <20 and ~20 cigarettes per day of ETS exposure, summary ORs and 95% Cls were 1.0, 1.6 (1.3,2.1), and 2.2 (1.73.0). respectively. with p<0.01 for the test of Irend. Results from ETS studies to date have not exhibited a consistent pattem (or risk from EfS exposures in childhood as compared with TAaLE IV-SUMMARY OF RREVIOUS CASELONrROL STUDIES IN CMIFA OF E(POSURE TO ENVIRONMENTAL TOBACCO SMOKE IETSI Rcrr.arc Lsa~ion - $Wjm. Reu/n Chan and Fung. 1982 Hong Kong Females: cases. 84: controls. 139 Odds ntio (OR) for ETS exposure at home and work: 0.8 (0.5,1 :) Lam. 1985 Hong Kong Females: cases. 60: controls, 144 OR for E7S exposure at home and work: 2.5 (1.5,4.2) Koo et al.. 1987 Hong Kong Female non-smokers: cases. 86: OR for E7S exposure: 1.6 (0.9-3.1) conuols. 136 ORs far husband's cigarettesfd (0, 1-10, 11-20. z21: LO-2.3 (0.9,5.9). l.7 (0.8.3.8), 1.2 (03,3.0): p for trend 0.16 Gao es al.. 1987 Shanghai Female non-smokers: nses, 246; OR for EIS exposure in: childhood: 1. 1 - conuols.375 (0.7.1.7); adulthood: 0.9 (0.6.1.4) ORs far years living with smoking husband (<20, 20-29. 30-39, a•40): L0. 1.1 (0.7-1.8). 1.3 (0$,2.1)- 1.7 Lam et al., 1997 eng et al., 1988 Hong Kong ianjin Female rum-smokers: cases. 199: controls. 335 emales non-smokers: cases. 54; contmis, 93 (1.0,2.9); p for tsend <0.05' OR for ETS uposure: 1.7 ().22.2) ORS for husband's clgamtlWd (0- ]-10, 11-20, >_21): L0. 2.2 (1.1,4.2). 1.9 (1.2.2.9). 2.1 (I.1-4.0); p for trend <0.01 OR for ETS exposure: OR = 2-16 (1.1-3.8) ORs for husband's eigarefies/d (0. 1-9. 30-19, at20): 1.0. 1.4 2.0 (1.4?.7)- 2.8 ( LA4.1): p for trend <0.05 ORs for years living with smoking spouse (0,1-19.20-39,~40): 1.0. 1.5 (1.2,1.9)- 2.2 (1.5-3.2). 3.3 (2.13.2); p value for trend <0.05 Wu-Williams ec al.. 1990 Shenyang and Harbin Femak non-smokers: cases, 417; OR for ETS exposure: 0.8 (0.6,0.9) controls. 602 Liu et al.. 1991 Xuanwei Female non-smokers: cases, 54; OR for ETS exposure: 0.8 (0.3,1.9) controls, 202 Liu et al.. 1993 Cmangthou Female non-smokers: cases. 38; ORs for husband's cigaetre.sld (0. 1-19. controls. 69 z20): 1.0, 0.7 (0.2.2.2). 2.9 (1.2,7.3); p for trend - 0.03 Wang et al.. 1994 Harbin Females: 59 pairs amokers; 55 ORs' for EfS eapow+c in: childhood: 3.2 pairs non-smokers 0.6.6.3); edullhood: 0.85 (0.323) ORs' for household EfS exposvre in gnl/d (<5, 5-14, a15): ehildhoad: 19. 2.4 (L2,4.7), 3.6 (1.5,8.6): p for trend Wang cr aL, 1996 Du et al., 1996 Guangdong Guangzlwu Females: 99 matched pain Fcmale non-smokers: cases 75. controls. 123 <0.01; adulthood: 1.0. I.0(0.6.1.9)- I.1(0.4,2.9) OR for ETS exposlne: 25 (1.3.5.1) OR for EI'S exposure: 1.2 (0.7.2.2) ORS for husbands' cigara¢s/d (0. 1-)9- s20): 1.0, 0.7 (0.3.1.4). 1.6 (0.8.3.2): ORs for years with smoking husband (0, 1-29. z30): 1.0. 1.4 (0.6.32), 1.2 (0.6.2.3) Ko et al.. 1997 Zhong et at.. 1999 Taiwan Shanghai Female non-smokers: 105 masched pairs Female non-smokers: cases. 504: coturols. 601 ORs for EI'S exposed from: parenls0.8 (0.4.L.6): cohabiunts: 1.0 (0.4.2.3): spouses: 1.3 (0.7.2 5) ORs for ETS exposure in: childhood: 0.9 (0.7.7 _^): adulthood: 1.1 (0.8.1.4): Current study Gansu Non-smokers: Fesnalez: cases. 200: controls. 407 occupation: 1.3 (1.0.1.7) OR for ETS exposure: 1.2 (0.7 =.0) ORs for ETS in: childhood: /S (1.1.2.2); adullhood: 0.9 (0.6.1.4); Maks: cases. 33: 1 14 controls ORs for vears accupaswnal ETS (0. <Q0, a20):~1.0. 1.3 (1.0.1.7), 1.8 (0.5,5.6) 'p-value for trend computed using weighted regression.-'ORs for childhood (ages <23) and adulthood lages z23) avcraged over categories of age at exposure.

ENVIRONMENTAL TDBACCO SMOKE AND LUNG CANCER 14 ) TADLE r-OISIR1BlR1ON OF NEVER-SMOKING SUa1ELT5 ANO ODDS RATIOS (OR) FOR LUNG CANCER BY CATECARIEb OF OFI.fOGRAVNIC vAR1ABLE5 (ONS ARE AD1USrFD FOR AGE AND PRFIECNRE) My[. Frmncr V_~ C>.n (SI CmuMr ('FI OR Gss. [R) Cwx,..k ['.kl OR Referenee age <45 24.2 17.5 1.00 190 12.5 1.00 45_54 18.2 33.3 0.21' 38.5 38-f 0.70 ' 55- (A 433 29.0 1.06 29.5 31.5 0.56 ' a65 12.1 20.2 0.45 13.0 174 0.51 Plefeeturc ( 69.7 51.g L00 51.0 520 1-DO 2 30.3 48.3 0.34' 49.0 43,0 f I5 Fdu(stion Primary or less 57.6 64.0 1.00 87.0 95.3 1.00 ' Tech/vocation 30.3 33.3 0.91 13.0 4-2 2.78 College and above 12.1 2.6 6.34 0.0 0.5 - Marital status Married 939 93.0 1.00 87.0 867 1.00 Widowed 6.1 6.1 1.42 12.5 13.3 1.13 pivorced 0.0 0.0 - 0.5 0.0 - Nevcr married 0.0 0.9 - 0.0 0.0 - Income <2,000 6.1 26.3 1.00 21.5 24.4 1.00 2.000-2.999 182 26.3 3.25' 16.5 21-2 0.86 3-000-4.399 39.4 22.8 8.95' 28.0 28.2 1.16 z4,000 36.4 24.6 5.61' 34.0 26.2 1.28 Number in household 1,2 6.1 4.4 1.00 8.0 5.2 1.00 3-1 39.4 29.8 0.56 25.6 24A 0.49 5...6 18.2 43.9 0.19 42.7 43.4 0.49 x7 36.4 21.9 0.91 23.6 27.1 0.47 TV - black and whi¢' 18.2 45.1 0.25' 51.8 49.6 1.07 TV - colort 51.5 20.2 3.64' 32.5 17.7 1.97' Tape recorder2 50.0 33.3 1.92 31.0 29.5 1.09 ' Refrigeretor 9.1 0.9 - 73 2.0 3.29 Number large animals 0 54.6 33.3 100 54.5 32.2 1.00 1 21.2 27.2 0.58 23.0 35A 0.44' z2 24.2 39.5 0.51 225 32.4 0.43' Own vehicles (zl ) 12.1 9.7 1.06 10-6 5.7 1.65 Total3 33 114 200 407 'Oks significantly diffen from l. _2f ercentnges indicate subjects with the futor. 'Numbers af subjecu (nr each variable differ due to missing data. ETS exposure of under 20 years and 20 years or more relative to o0 occupational exposure were 1.29 (0.5,3.3) and 1.76 (0.5.5.6), respectively, with p=0.19 for the test of trend with duration. When daa were restricted to the 52 cases and 57 rontrols who ever worked in non-farm jobs, the OR for occupational Ef5 exposure was 1.35 with 959s Cl (0.5.3.4). We evaluated variation in the ORs with pack-years of ETS exposure in childhood and in adulthood within categories of sev- enl variables, including reference age, educational level. marital status, income and a variety of socioeconomic factors. None of the tests of homogeneity of trends in ORs across categories were rejected We also evaluated potential confounding from factors related to indoor air pollution. by computing ORs for paci:-years of ETS exposure by underground or above-ground type of current and longest lived-in dwelling, time weighted mean radon concen- traoion in all dwellings occupied for 2 years or more 5-30 prior to the reference age. type of fuel used for heating and cooking in the currenl home and the subjeci s perceived level of indoor smoki- ness in winter months in the currcnt (rouse (Table III)- Adjustment for these factors, as well as annual amount of coal used- and the degree of indoor smokiness in houses occupied as children (not shown). had little effecr on ORs for pack-years of ETS exposure. Tests of homogeneity of OR trends among categories were not statistically significant. except for the trend in ORs for pack-years of ETS exposure in adulthood by type of current dwelling and type of longest lived-in dwelling. In these two cases, ORs for pack- years of ETS exposure in adulthood inceased among those whose current house or longest lived-in house was an underground de- sign. We also found no confounding of the EPS associations by prior diagnosis by a physician of one or more lung diseases, including asthma, tuberculosis- bronchitis, pneumonia, asthma or emphysema. Finally, the results for 6TS exposure were similar when data from rsextof-kin interviews were omined or only histologically confirmed cases were inchlded. With data restricted to 115 cases and 501 cnnVols who were self-respondents and never smoked, ORs and 95% CIs for ever exposed to ETS in childhood and adulthood were 1.75 (1.12.8) and 0.76 (0.4,1.3). respectively, compared with 1.52 and 0.90 for all nevu-stnokers, tud ORs increased with pad:-years of ETS exposure-ORs forO. 1-9. 10-19 and 20+ pack-years of 1.00, L93- 2-12 and 2.62 for childhood exposure (tesr for vend, p-0-03) and 1.00, 0.67. 0.72 and 0.63 for adulthood exposure. ORs for pack-years of ETS exposure within categories of the variables in Table III were also similar when dara were restricted to self-respondents. Analyzing only histologioally confirmcd cases (71 or 32%). ORs and 95% Cls were 1.55 (0.9,2.8) and 0.99 (0.5,2.0), respectively. DISCUSSIDN Our study found an overall non-significant OR of 1.2 for ever- exposed to residential ETS, with a statistically significant increas- ing trend with pack-,vcarx of EfS exposure. The OR and 95% Cl for any residential or occupalional ETS exposure were 1.4

140 rrANG ET AL Health, reviewed all case diagnoscs The expen panel excluded 277 subjects due to (he absence of supporting evidence or to incorrect diagnosis. leaving 932 ca.es. Of these. 43 cases could not be located and 3 cases moved out of the study area resulting in a total number of 886 cases (656 ma)es and 230 females). Diagnoses of lung cancer were based on clinicaVradiological criteria for 533 cases (60%) and pathological evidence (such as bronchofiberoplic biopsy. sputum cytology and lymphatic node biopsy) for 353 casea (40%). A total of 1.966 controls was randomly selected from 1990 census lists for the 2 prefectures and frequency matched on sge in 1995 to cases in 5-year age groups, within categories of sex and prefecture. The number of controls in each svara was determined by the expected age distribution for lung cancer cases obtained from a search of medical records within the prefectures for 1991 as pan of a feasibility study. Twice as many controls as expected cases were selected. Among controls, 6 refused interview, 23 moved out of the study area, 62 could not be located and 35 could not be interviewed for other reasons. A total of 455 female and 1.310 male convols were included in the article. Interviews were conducted at home or at the hospital for all cases and controls by trained interviewers using a closed-form- struttured questionnaire. We asked detailed questiona on demo- graphic characteristics, smoking habits of the subject, spouse and other cohabitants, diet and cooking practices. as well as detailed occupational. residential and medical histories_ Whenever possible, interviews were conducted with the subjects themselves. If subjects were deceased or too ill, interviews were conducted with next-of-kin, usually the spouse. Next-of-kin pro- vided information for 481 (54%) cases. Nextof-kin interviews were assotiated with year of lung cancer incidence. Among cases identified after 1996, 29% were ncxt-of-kin interviews, while among cases identifirrl in earlier calendar years, 74% were next- of-kin interviews. Next-of-kin provided information for 71 of 1-765 (4.0%) control subjects. For the analysis of ETS exposure, we restricted data to never- smokers- defined as never smoked cigarettes or pipes regularly for 6 months or longer. There were 200 female and 33 male long cancer cases and 407 female and 114 male controls who never smoked. Numbers for the variables in the tables differ due to nussing data. Statistical annfvs(s We computed odds mtios (ORs). adjusting for the reference age defined as age at disease incidence for cases and age at fnterview for controls. sex and prefecmre, and, where appropriate. other factors using logistic regression models as implemented in the Epicure computer package (Preston er al.. 1996). We calculated 95% Wald confidence intervals (CI) for ORs and used a score statistic for tests of trend. ETS eaposure To estimate ETS exposure, we asked each subject or neA-of-kin about the smoking habits of all cohabitants of the household during the subjeci s childhood, defined as ages 18 years and urder. and during adulthood. The information included number of cigarettes smoked per day (store-bought), and number of I/nng (50 gm) of tobacco per month smoked as (hand-rolled) cigarettes or in Chi- nesc long-srnm pipes. The source of ETS ezposure reflected the changing smoking habits in these areas of China. with pipe smok- ing more prevalent historically and cigarette smoking more pre.- alent recenUy. Among those exposed to ETS in childhood, 9.5. 75.2 and 15 .3'h were exposed from persons who smoked cigarenes only. pipes only or smoked both cigar<nes and pipes- respectively. In childhood, mean ETS exposure from cigarettes was 9.6 and 10.7 cigarettes perday from cigarette only smokers and mixed smokers, respectively, while mean ETS exposure from pipes was 9.4 and 10A /lang per month from pipe only and mixed smokers. respec- tlvely. The sources of ETS exposure changed in adulthood. Among those exposed to ETS in adulthood, 35.3. 27.t and 37.6'k were exposed from persons who smoked cigarettes only, pipes only or both cigarettes and pipes. Means were 14.6 and 12.5 cigarettes per day from cigarette only and mixed smokers. respectively. and 9.8 and 8 7 llang per month from pipe only and mixed smokers. Initial,analyses indicated that the log-linear increases in ORs for ETS exposure per cigarette smoked per day and per hang of tobacco smoked per month in Chinese long-stem pipes were ho- mogeneous (p=0.36 and p=0.44 for tests of homogeneity of trends in childhood and in adulthood, respectively). ETS exposure in cigarettes per day and Gang per rnonth were therefore summed to obtain total ETS exposure in childhood and in adulthood. We refer to this sum as cigarettes per day of ETS exposure. Analyses also revealed ORs for ETS exposure in childhood and in adulthood were statistically homogeneous in females and males. Consequently. unless rated, we show results for females and males combined. with adjustment for sex. RPSULTS Table I shows the distribution of cases and controls for reference age, several demographic factors and ittdicators of socioecorvomic level for never-smokers- By design, cases and controls for the complete data were balanced on the matchingfactors; however, among neversmokers. there was a smaller proportion of older female cases. This difference was primarily due to the selection of controls from the 1990 census list based on their ages in 1995. However, because the mid-point of control enrollment was 1997, the mean age at interview for controls was slightly higher than anticipated. Results also indicate that cases were more likely to have a posoprimary education, have a rolor television or refrig- eralor, and own fewer large animals. While these variables were significantly related to case status, they did not confound the relationship berween lung tancerand ET5-related variables. None- theless, in the remaining analyses, we adjusted for socioeconomic level, which was best characterized by two variables- ownership of a color television and number of cattle, as well as age, sex and prefecture. Table H shows ORs for ever<xposed to Sfs in the household and for categories of amount of exposure, as nreasnred by eigarette pack-years smoked by cohabitants. Oversll- them was a non- significant OR of 1.19 for ever-exposed to ETS, but a significant monotonically increasing trend in ORs with pack-years of expo- surc (test for trend, p=0.05). When exposure was categorized by period of exposure, there was a statistically significant OR of 1.52 with 95% CT (1.1.2.2) for ever-exposed to ETS in childhood and increasing ORs with pack-years of exposure Ip=0.02). 77rere was tso effect on risk of Et5 exposure orcurring in adulthood. Based on a log-linear model for age, sex, prefectum, number of cattle. ownership of a color television and pack-years of exposure in childhood and in adulthood, ORs were 1.53 with 95% CI (1.1,2.1) and 1.02 with 95% CI (0.9.1.2) for each 10 pack-years of ETS exposure in childhood and in adulthood, respectively. The p-value for the test of homogeneity of the two trends was 0.02. The questionnaire also provided information on whether household cohabitants smoked in the pmsence of the subject. Adjustment for smoking in the presence of the subject did not affect the ORs in Table 11 . Table II also shou•, that ORs were similar for males and fe- males_ None of the statistical tests of homogeneity of ORS by sex were rejected- The two prefeaures were predominantly rural with 32.8% of ncver-smoking males and 8.5% of never-smoking females ever employed in non-farrn jobs for more than I-vear. Among never- smokers, there were 21 cases (9.0%) and 17 controls (3.3%) with some ETS cxpocure from non-farm jobs. Afler adjustment for residential CI5 exposure in childhood and in adulthood and th, demographic factors, rhe OR for any occupational ETS exposure was 1.56 with 95'k CI (0.7,3.3). ORs for duration of occupational >

fnr J. Cancer 88, 139-145 (2((10) Pubbshed 2000 Wt1cy-Liss, Inc. 'This atm/t is a US Gnvemmem work and, as such- ie in the pubbc domarn in tha United Surex of Amenca. LUNG CANCER AND ENVIRONMENTAL TOBACCO SMOKE IN A NON-INDUSTRIAL AREA OF CHINA Lnngde WANG', Jay H. LUBINia*, ShU Rong ZHANG , Catherine METAYERr, Ying XIA3, ARna BRFNNEx2, Bing SNANGr, Zaloynan WANG' and Ruth A. KLEINERMAf.-s 'Menisrry of Neahlr, China 2Division of Cancer Epidemiofngtund Genetics. Natianal Canaer, hrsritutq Bethesda, Mai-vland, USA 'Lnbaratory of Industrial Hygiene. Ministn• of Public Health. China We report results from a population-based cue-control study of lung oncer and environmental tobacco smoke (ETS) among never-smokers conducted in 2 rural prefec- ttrres of China, including 200 female and 33 male lung cancer cases, and 401 female and I 14 male rnntrols, matched on age, sex and prefecture of current residence. The odds ratio (OR) for ever-exposed to ETS was 1.19 (95%CI 0.7-LO), with a significant erend (p<0-05) with increasing exposure- ORs were 1.00, 1.04, 1-13 and I.S1 for noncxposed, <I0, 10-19 and 220 pack-years of ETS exposure, respectively. Excess risks were limited to ETS exposures in childhood (518 years of age). The OR for ever-exposed to ETS In childhood, ad-' justing for ETS exposure in adulthood, wa's 1.52 (95% Cl with a sigMficant trend -(p<0.01) with increastng pack-years of childhood estposure, 1.00, i.43, 1.81 and 2,95, respectively. After ad)ustment for ETS in ddldhrwd, there was no excess risk from adult ETS ertposure, The OR for ever-exposed to ETS in adulthood was 0.90 (9S% CI OrL4). These resu/ts were not affected by adjustment for type of residential dwelling, type or arttount of fuel used, perceived indoor smokiness, or meuures of socioeconumic status, or omitUng nexteGkin respondenes Int (. Cancer 88:139-145, 2000. Published 2000 Wilev-L:ss. /nn r Epidemiotogic studies of lung cancer and exposure to environ- menWl tobacco smoke (ETS) have been conducted in countries throughout the world and consistently show a 12 to 1-5-fold risk of lung cancer (National Research Council. 1986: California En- vironmental Protection Agency. 1997). Studies of ETS exposure in (]tina however are of particular interest, due to the high lung cancer rate in females who are mostly non-smokers (Gao er at., 1987: Xu er al„ 1989; Wu-Williarns cr al., 1990: Blot and Frau- meni, 1996) and due to the relatively low odds ratios (OR) of lung cancer among acGve smokers, where ORs for ever-smokers corn- pared with never-smokers consistently range from 2 m 4(Liu, 1992; Liu et al.. 19981• markedly lower than in westem countries (U. S. Departarcnt of Heallh and Human Servirxs, 1990). The rates of lung cancer for females in China are high, although nm uni- fortnly so across the country, and may be due to indoor air pollutants, diet, cooking oil fumes, occupational facrors and non- malignant respiratory diseases, as well as ETS (Blot aod Fraumeni. 1996). while the lower overall ORs for active smoking among Chinese adults may be attributable to later age at smoking initia- tion (Yang rr al., 1999), relatively shoner duration of smoking, less imerrse inhalation practices or the greater proponion of pipe. as compared with cigarette, smokers (Liu, 1992: Lubin er al., 1992). Nonetheless. ORs of 20-foid have been reported in those smoking 20 cigarenes per day or morc (Xu et a(., 1989). suggesl- ing that Chinese smokers are not ae an inherently lower risk of lung canccr. Results from studies of residential ETS exposure in China have been mixed (Chan and Fung. 1982: Lam, 1985, 1987: Gao er a/., 1987; Koo et aL, 1987; Geng er al.. 1988: Wu-Williams ef al., 1990; Liu er al.. 1991; 1993; Wang er af., 1994; 1996; Du n al.. 1996; Ko er al.. 1997: ZAong er al., 1999). However, several of the individual studies did not adjust for potentially important cofac- tors, such as indoor afr pollution and occupational exposures. Several outstanding issues concerning ETS exposure and risk of lung cancer remain. Few previuus studies have evaluated factors that modify the OR for ETS exposure. Few studies have evaluated differences in risk from ETS exposure in childhood, as compared with adulthood (Gao er al.. 1987: Janerich et af., 1990; Wang et al., 1994; Zhong er al.. 1999)- One study found a greater risk from childhood E7S exposure than from ETS exposure in adulthood (Wang er al.. 1994). two studies found no difference in risk (Gao et af., 1987; Zhong er al.. 1999) and one found an excess risk from E'IS exposure only in childhood (Janerich et ar_ 1990)- In addi- lion, while there have been studies of lung cancer risk and aclive smoking among Chinese men, the majority of whnmsmoke (Gao cr al., 1988; Qiao er al., )989: Liu er al., 1991: Lubin et al.. 1992t Yao et al., 1994: Do re nL 1996: Yu and Zltao. 1996; Xu et aL, 1996; Lei et al., 1996: Qiao er al.. 1997). there have been appar- ently no smdies of ETS exposure among the few Chinese men who never smoked. To address queslion.s associated with lung cancer and E7S exposure, we analyzed data from a population-based case<onaol study of lung cancer in 2. primarily rural, prefecwres in Gansu Province in north-westem China, These areas are unique because more than half of the population currently live or have lived for an extended period in underground dwellings, where indoor levels of radioactive radon gas are among the highest in the world (Wang er af., 1996). In most rcsidences- brick stoves are used for heating and for cooking, with smoke and fumes vented via enclosed ducts through a bed-like brick platform used for sleeping (call a kang) to the outside. Coal. wood, sticks or other biomass were the principal sources of fuel. MAlPJt/AL AND METHODS Study subjects Beginning in June 1995. we idenli6ed aJl individuals between the ages of 30 and 75 years who were newly diagnosed with lung cancer between January 1994 and April 1998, and were residents of Pingliang or Qingyang. 2 prefectures in Gansv Province. China, with a total adult population of about 4 X 30A persons. Cases were identified from 2 prefecture hospitals, a company hospita) located at a nearby oilficld, 15 county hospitals and local clinics. We also reviewed records from special anti-wberculosis reporting stations in the prefecmres. In addition, hospital records in larger nearby cities, Lanzhou. Xian and Yinchuan, were reviewed for lung cancer patienLs diagnosed in residents of the two prefectures. Based on clinicallradiological symptoms suggestive of lung cancer or pathological evidence, a total of 1,209 possible cases were identified. An expen panel of pathologists, radiologists and clinicians, assemblcd from members of the Gansu Department of •Correspondencc to: Dr. Jay Lubin. Division of Cancer Epidemiology and ('.enetics, National Cancer Instirve. EPSB062. 6120 Executive Bou- levatd. Bcthcsda Maryland 20892-7244 Reccived 22 January 2000. Revised 20 March 2000; Accepted 20 April 2000

144 W nNG Er AL. ETS exposures in adulthood. Risk patterns with tobacco use among active smokers are consistent with tobacco smoke being a complete lung rardnogen. serving both as a initiator and promoter (Brown and Chu. 1987; Darby and Pike, 1988: Moulgavkar er aJ., 1989). The identification of ETS exposure as an initiator. particularly with expo- sures occurring in childhood, would markedly affect estimates of lifetinre lung cancer risk from E7S expcisurc. ORs for EfS exposure in otu study varied sigttificandy by period of uposure, with ORs significantly increased with childhood ETS exposure. This result is at odds with the oonclusians of Brown and Chtr (1987) and Darby and Pike (Darby and Pike, 1988), who fitted the Amutage-Doll multistage model for rarcinogcnecis to smokers and suggested that a low level of tobacco smoke expostue acts as a late stage carcinogen, with duration of exposure having a relatively lesser effsct on risk. Howevcr, this conclusion was disputed by Moolgavkar er a1. (1989) and W hinemore (1988). who suggested that data were too limited for such specific conclusions. Of the three previous China studies that reponed results by period of exposure, I study fnund a grrater effect of ETS expostue in childhood (Wang er al.- 1994), while 2 studies found no overall effect of ETS exposure and in particular no difference between ORs for exposures in childhood and adulthood (Gao er al.. 1987; Z7mng er al., 1999) (Table M. Studies in western counoies have similarly shown mixed results (see Table 7.6 in California Environmental Protection Agency, 1997; Boffena er af., 1998; Janerich er aL. 1990). The inconsislcncy observed to date may be due to the difficulty in recalling details of ETS exposures many years in the past artd the relatively small expected effeet Thus, based on published results to date, no definitive conclusions about the relative importance of Fl'$ exposures in childhood and in adulthood are possible. Owing to the historically higher proportion of smoking males. there is a greater number of males than females diagnosed each year with lung cancer (Boring er al., 1994: Travis er al., 1996). While evidence is not conclusive, studies of active smokers have suggested that ORs for lung cancer from exposure to tobacco smoke may be at least as great, if not greater, in females than males, after adjusting (or smoking duration and rate (Lubin and BIoL 1984; Brownson er al., 1992; Riseh er af., 1993; Zang and Wynder, 1996), suggesting females may have a greater sensitivity to tobacco smoke. Most studies of ETS exposure have been carried out in never-smoking females; however, the few studies of ETS exposure that reported ORs separately for never-smoking females and males found no marked differences of ORs in females and males (Kabat and Wynder, 1984; Kabat er al., 1995; Boffetta er aL. 1998), although investigators did nol appear to have conducted formal statistical hypothesis tests. Our study had limited dara on males, but the magnitude of dx ORs and non-significant tests of homogeneity suggested that females were at comparable risk of lung cancer from ET5 exposure. A strength of our population-based study was its locarion in a predominantly rural, non-industrial area of China. Fewer than one third of males and one tenth of females were ever employed in non-farm related jobs. In addition, few of those who did worked off the tarm reporteR any exposure to a list of potential lung carcinogens. ORs for ETS exposure were unchanged when data were restricted to subjects who never worked off the farm. A potential confounding factor in our evaluation of ETS was the level of indoor air pollution, since virtually all subjects Mmted coal. wo9d or sticks in a stove or Anng for cooking and heating. A detailed evaluation of indoor air pollution in 25 houses in the study areat revealed high levels of particle bound polycyclic aromatic hydrocarbons (PAHs), particulate tnaner smaller than 10 µm (PM- 10), CO, NO' and SO_ (Ligman er al„ 1997). Ventilation rates were high, with an average of 1.5 air changes per hr, resulting in levels of indoor air pollutants that were episodic and followed closely the use of indoor stoves. Pollutant levels were low during non<ooking times. Except for CO and PM-10, mean values for integratcd measurements were below U.S. Ambient Air Quality Standards. We did not have measurement data on indoor air pollutants in each house, but in our analysis, OR patterns for ETS exposure were unchanged when adjustment was made for housing type (underground or above-ground dwelling), amount and type of fuel used and the degree of indoor smokiness as reported by the respondent. Values for these potential confounding variables were determined for the current house, the longest lived-in house and for houses in adulthood and in childhood. We relied on next-of-kin interviews when the case patient was deceased or too ill. Next-of-kin, particularly spouses, may be less knowledgeable about events and exposures occurting in childhood, thus raising the possibility that results may have been affected by differential nusclassificalion of ETS exposure, While non-differ- enlial misclassification of an exposure variable most often, al- though not always, results in ORs closer to the null value (Dose- meci er aL, 1990). consequences cannot be predicted if misclassification is differential. In our study, ORs were larger for ETS exposure in childhood than in adulthood. These OR patterns could have been influenced by differential misclassification if subjects or nexo-of-kin of cases were more likely to report ETS exposure than controls. However. this differential recall would had to have differed for exposures in adulthood and in childhood. which seems unlikely. In addition, OR panems for ETS exposure were similar when data were restricted to subject respondents only. In summary, we found an overall increasing risk of lung rancer with pack-years of exposure to 6TS. This increasing risk with greater ETS exposure was primarily due to exposures occurring in childhood. A CK VOU,rtEy('E.IENTS We thank Ms. M. Pacious and Ms. R. Hur, Westat ,Inc.. fordara management and programming support, and S. Lei. J. Cao. S. Lei and W. Wang of the Laboratory of Industrial Hygiene. Ministry of Public Health. China, for data collection support. We are grateful to members of the expen review panel, Drs. J. Shi. Y_ Kang, F. Du, D. Zhao and Y. Wang. for reviewing lung cancer diagnoses. REFERENCFS Bt,oT, WJ. and FRACarr1+L JR. l.F., Cancers of the lung and pleur.. In D. Schortenfeld and l. F. Fraumeni. Xn teds). Cancrr rpldemialogr and prrvrmlon. Ind, pp.637-665, Oxford Un{versity Press, New York ( 1996). Bot•rtTTA, P. and 24 omERi. Multicenar case-control stutly of exposure to envimnmcntal tobacao smoke and lung cancer in Europe (sec comments]- l. narLCancrr/nrL 90. 1440-1450 (J998). Bokmc. C.C., SQUUtrs. T.S., Tonc, T, and MoNrc:oasnr. S., Cancer Statistics. 1994. Ca- Cancer f Clin. 44. 7-26 (19941. Baowa, C.C. and Cau. K.C.. Use of multistage models to infer stage affected by carcinogenic eaposure example of lung canccr and cig+nenc smoking. l. Ch.onic. Dis.. 40 Soppl 2. 171S-179S (1987). Baoa'NSO*. R.C. CnnvG, 1.C. and DA.'n, 1.R_ Gender and hlstologic type variations in smoking-related risk of lung cancer. Epldrmming.. 3.61 -6f (f992)- CAtJFORtl1A EYVIRONMEATAL PROTGCTO> AGEUC). Ilc¢hh Effects f EJ- po.mre rn EnvAronnrentul ToAacco Smokn. Final Repnn, Seprrmber, 199'. Statr of Califomla, Environmental Pmtection Agency, Sacramento (199 %k CHxa. W C. and FuNC. S.C.l-ung cancer in nonsmokers in Hong Kane, h' E. Gmndmann tedJ. Cancer campaign, V. 6 Canerr Epidemfo7ogr- pp199-202, Gustav Fischer Verlag. Stungart, West Germany (1982). Owrtas, S.C. and RRt. ,M C.. Lung cancer and pazzive smoking: predicted effects from a maNematical model for cigarette smoking and lung eancer. Brit. 1. Cancer, 58. 8?5-831 (1988). DosrsaECl M.. WACNCxnER. S. and Luatr<, J.H.. Does nondifferemial .classificarion of eypmre a/wavs biac a true effecr mward ihr ndt vzlud Amrc l Epidrmtnl.. 132, 746-748 (1990). Du, YX.. Cnn. Q.. Cat.. X.W_ CHEn. Y.Z.. HuAN4 LF.. PENa. ZZ.. Wr, X F. erd Wu, J.M_ An epidemiological study of risk factorx for Ivr1 canccr in Guengzhov. Chinu. Lung Concc., 14 Suppl 1. 59-37 (1996i GAO. Y T., &AT. W.1.. ZNENG. W.. E0.SHDW, A.G.. Hsc, C. W. LE,u, L.1.

FNVIRONMFJ.TAL lO6ACCO SMOKE AND LUNG CANCE% . 145 Zx+ne. R. and FnauMUq 1 F., Jx., Lung cancer.mong Chinese women. .vr-s eulde, HimSoB Imernzrional Carpqztion, Sunle, Washingson lne. J. Cnncer. 40. 665-609 (1987) , (1996). GAO. Y-T., Btnr. 9f J.. 7HSAG. W., FxauMeni, ).F.. )c. and Hstr. C.W.. Quo. Y1... Twvtan, P.R.. Yao, S.X.. Enozsn. Y.5_ t.uo, X.C.. BARnsrr. Lnng en¢er sod smohing in Shanghvi. /ar. J. EP(demiof. 17. 27V-2gf1 M.l.. Yare. Q.Y., Glt~, CA., HcANG. SQ.. MnufR, M.M., J-ropuAn. (1988)- M.R. and ToLw.AR. M.S., Risk facmrt and early dqeclion of lung cancer CENG. G., LiAVC. Z.H. and ZNANq G.L., On the relationship besween in a eohon of Chinese 1}n miners. Ann. Epidemfol. 7, 533-541 (199'!). smoking end femak lung cancer. /n Smoking and Hrnlrh, pp. 483-486. Qreo. Y.L, iwvs.on, P.R.• Ywo, SJ(.. Scsuteqn, A., MAO, B.L.. Lumn. Elsevkr Scienet Publishers, New York (198g). 1.. RAo. J.Y.. McAnevs, M., Xuan•. X.Z. and Lt 1.Y.. Relauon of radon JA.aERKM. D.T.. ThaursuN, W.D.. VacFi w. L.R-, GcEEVwe~. P- upnsure and tobacro use to lung cancer among rin mfners in Yunnan CNOaorr, 5.. Tum, C., Lwwwv, M.B., MptnMm, M.R_. KtELV. M. erd Provinc0. Ch'na Amer. J. fnd Med, 16. 511-521 (1989). MolGuwu.v. M F. Lung cancer and exposure to aobxce smoke in 1he RrseR. H.A.. Howt, G.R, IArN. Ai. Bcaas, 1.D., Hos.on•wtY, EJ. and housrbold. H. Fitgf. J. Med, 323. 632-6M (1990). MIUPA A.B.. Are female smokerx u higher risk for lung cancv Oran male KAewT-G.C., 51UtMnn, S.D. and WvNUwL EL, Rclatlun between enpo- sq°ken? A case-oontml analysis by histologic type. Amrr. J. Eyidemiof. sure to environmenul rubacco.mohe and hm& canceu in lifetime nonsmok- 138- 281-293 (1993). m 1 Wblished ertarum aplT+rs in Am. J. Epidem)uf 1996 Ma. 1:103(5). TRAVIS. W.D., l.uem, 1.H.. KESSUR LG.. RrES, L. and Devav,, 5-5-. 5271- Am J£PrdenuoL. 142. 141-148 (1995) Trenda in ircidence of Wng eascer, bv histologictype. aea and race in Ore KwnA'r. G.C. ard Wv.osv. E.L. LunE eancee in nonamokern. Cancer. 53. ('oA~ ~es. 1969-1991. Canc.r, 77, 2464-2470 (1996). 1214-1221 (1986). US. DF+ArtnssNr os, Hw.TS nno HuM.N StxvsCEs- Thr healeh benejos oJ Ko, Y.C.. LEE. C H.- CHEK. MJ., Huwr:e. C.C.. Cnu+q W.Y.. Lm. HJ_ amOIung cessotion: a repon oJ rhe 9rrBeon Generd.. Public Heahh Scr- nonsnraki g~women in Taiwan~ JmrY Epldemnbt .~2b~2a jjn(j997~g ~~HSPubANO~(PH~ 90.8416. Wuhiogtomc~(( ~)ngand Koa. L.C.. Ho. J.H.. SAw, D. and Ho. C.Y., Measuremcnu of sisx Wa~. FJ... IAV6 EJ.. Lu• N, and D.M. X.D. CTildaod and ado/cseenl g passi~ smoking and the risk of female lung canar /ra. l. enriol., 23, ferrulef 1nr~J rC~r.99,u162-1~69~(1987)among non-smoking inese '13- (1994). ~~ Lwlko. N.M. and Mwr¢sER, F., Some suri.ucal methods for cumbining Wwaq S.Y.. Ho- Yl.- Wu, Y.L.. Lr. X.. Cns, G.B., Cn[x, Y. and DAi. erperimental resuh../nt l. Techwnl. Auesr Heahh Cnre-6, 5-30 (19901 WS., A enmpnmve ssudy nf Ne risk facuns for lung eeruxr in Guang dasg• Chi. faarg Cunrer. 14 SupPI 1. S99-IQC (I996). Lae. T.H.. Ku.ve. I.T., WoNU, C.M., Lsae. W K., Kttrvevs.l.W.- SAw, ys.wxc, 7J~ aM Zxoc, BS., MeL-anal D.. HStA C.. Sf1+EnRwrN4 S, LAM, S.Y. uW La. K.K.- Smokin&pauive ~n ta so rnvironmartal tabss eco of smo Ne ke ~adeoual lun rsnca rtlauoruhin i smoking and histobgical types in lung eancer in Hong Kong C]tinese ~~. ~ wqoen. Lun8 Cnnrrr. 16. 145-I50 fg1997). womcn. Brb. J. Cansrq 56. 673-678 O987). tng LAM. W.X.. A cl(nical and epidemiologicol studv of currinona oJ lung m WaSia. ZY.. Lts9rN, J.H.. WAN4 LD., ConnAix, S., LsAwe. S2• Ktrtx- Hong Kong (docsnrzl thesisl- (19g5). Universlry of Hong Kong. Hong sia'swN• R.. SuwreG. B.. G.o. SK. GAo. P.Y.. Lrs, S W. and Boam J.DJ.. Kong. Radwl mea.vxemenu in undergrourq burllings from two prefenures in Owna. Henhh Phvz- 70. 192-198 (1996). lFa, YX.. CAL W.C.. Cnsn, YL and Dq ti.X., Snme lifeslylr fac[ors in WRIP1111(tiaC. A. and W Hrmssau, l.. A m{ a nach to shc human lung wieer:a case-cwnmi soWy of 792 lung rancer cases. Lurg mela-mvlysis of randomimd clinica) t~nals. Srw M~. 10wl665-16T/ Cancer, 14 Suppl 1. 5121~136 (1996). (1991). LRax,N. B.. SsuumwESs., R.. Kafs.vpnb,sN, R.- LuNn. ).H.. Rts1IX• E.. (VHm~~ AS.. Flfa+ of cigarette smoking in eP~dcmioloFrcai srudie. WaNC. ZY.. 7xsue, S ard _ Wwuo, L.Q• Irdoar arc pollution character- of IunE eanorr. Ssm. Med., 7. 223-238 (19881. iaatlon of undergsound dwellings in mfru. fn 1. E. Wonds. D. Gim sndrsnd and N. Boachi (WsJ. Pmttrdinga ufHeolrh Bui/dinBSJJndr»rAir Qealiry Wu-W~. A.H.- D.b X.D- BtnT. W.. Xo, ZY.. Su+, X.W.. Xr•o. 1997, Vul. 3- pp. 51-56, VPI and Sute University Press, Bl.cksburg HJs., Sroh-e, BJ.. Yu• SF.. Fv.G. Y.PR and ErtsHOw- A G.. Lung <anecr (1997). among women in notthrasl Chinea Br. J. Caneer. 62. 982-987 (199(l). Lic, B.Q., Psro- R.. Csn:N, ZM., Bp.&uM, 1.- Wu, Y.P_ 11 1.Y.. Xu, Z.Y_ Bsqr, WJ.- Xlwo. H.P., Wu. A.. FtNC, Y.P.. STOr.e, BJ.- SvN, C-UrBrgL T.CC and CMEU, 1.5_ Emerging tobacco hesards in China. 1. 1. ERbrow. A.G.. HrnmexsoN, H.E. and FRAusawi, 1-FJ., Smoking. air Rdrospective propquqW rranality swdy of one million deatM [see poBUtion, and the high rates of lung caecer in Shenyang. Chiva. 1. nar. eonmtarts]. BNr. Med J., 317, 1411-1422 ((998). Carer Drs[. 81- I800-1806 (1999) . lJu, Q., $A.tfo- A.J.. RteGU• E. and HU. M.X.. Indoor air pollqion and Xsr, ZY., BrtowN. L.. Pan•• G. W.. U. G.. Fer.e. Y.P.. Guan, DX., [.w. IungcancerinGuanrLnu.People'sRepubHcofCMna-Am<cLEpidenuaL. TF_ Lu. LM.. CxAO. RM. Snpae. LH. and Gwa G.C.. Lifestyle. 137. 145-I54 (1993). en..m.u.rnsal polluuon ard kmg csncer in citics of IJaoning in nurtheasr- Lw. Z.. Srswking and lung cJncer in Uina: combincd eeelysis of eigM °u China. C'mE Cancer, 14 Suppl 1. S)49-S160 f 19961. case- conuoi sudies. /ru. l. F{+idemiol.. 21, 197-201 (1992). Yuru. G.. FAS, L. TeN, J.. Qt. G., ZMArw, Y.. Syn~. J.M..-G,vt.on, CS.. Lse, ZY.. HE, X.Z and CHwvru:. RS_ Smaking arsd qhu risk facross for BKxOL K. and Xu- J., Smoking in Cbina: findings of the 1996 National Ivn6 cancer in Xuanwei, Clut¢. Jnr. l. EPidemiof.. 20.26-a1 (1991). Prtvak.pe Snrvey. JAMA. 282 1247-1?53 (1999). Lusm 1.H. and Bs.or, WJ. Asseument of /ugn aancer risk farwrs by Y.w- SX.. Wene- J.N.. QrAn, Y.L. BoxT. 1.DJ.. Lt. 1.Y., CAL S.K. hislolagic casegory. J. Har.Caecer lnrr- 73. 38i-J89 (1984). Zilud~• F.M. and Htnr, W J., Esposure to radon prvgeny, tobacm use and Irng cancrs in s eass- rnmtrol srudy in sauthem China. Radiar. Rr.r.. 138. WaiN.l.H.. L4 LY.. XuAw, X.Z, Cws-S.K.- Luo. p.5. Yw+o. L.F.. Wena. 326-336 (1994). J.Z, YwNO, L and Buvr. W.l.. Risk of lung eancer among cigarette and Yu- SZ. srW 7r,.o, N.. CombinW anelvsis of case-contmi swdies of pipe sumkers in southem Chme. fnv. l. Can<.r, 5i. 390-395 (1992). snwking and lung cancer in Ch{na. l.rng C co. 14 SuPPl 1. S 161-S 170 MOOtCAVKAIt. SJI.. D[wwrat. A. and Ltreatrr:. G.-Cigarene smokingad (19961. lung caq.+er: reuWysn of rla British donorx dau. J nar. Cmcer/nsr.. 81. a15-420 Z,yw;. EAA and WvNp~n. EL. Diffeencesin lung uncer nsk Eerween (1 9691. nw and svomen: examination of the evidencc. lI nae Cancar, fnsr. gg. NaTaVm. RESEAStIX Couticn. Bowxo or: EwtnonMfxr.c StuorES sr.D 18i-192 f1996). ToxKOUOCV.CaNM~ ON PASVVS SMOKtNC.. Ema.onmemul A.barro Zmveo L.. GosnlFno. M.S.. Gwo. Y.T. and JtN. F.. Lung can<er and Smnke. Menssur'vrg Eysnsnrer end Asseuing Heahh Ed ¢e. Nation.l indour eir poiluuon «ising from Chinesosryle cooking among non- Academy Press, Washington. DC (1986). snpking women Hvrag in Shengbal. China. F.fridemiMngr. 10. a88-494 PansioN. D.L, Luaix, LH. Nsne[. D.AA arb McCOnNn. M.E. Epirure /19991.