Tobacco Documents Online

828 INTERNATIONAL JOURNAL OP.EPIDEMIOLOGY Finally, most KMIC enrollees might be a selection bias in the study population. KMIC workers tend to be middle-class indi- viduals who may be healthier than the general populationn in Korea, possibly because of their education and employment stams. Overall, the results frum this cohort study should be relevant to other East Asian populations. The age-adjusted mortality rates for lung cancer have been sharply increasing for both men and women in Korea (Figure 2).16 As only a small fraction of Korean women with lung cancer smoke cigarettes, the reasons for their mortality from lung cancer being comparable with men have remained unclear. This study appears to explain why mortality from lung cancer in Korean women is escalating, particularly among wives whose husbands smoke, even though the rate of women's smoking is negligible. In conclusion, the results of this study indicate that the incidence of lung cancer is higher among non-smoking women whose husbands smoke, and a dose-response relationship seems to exist. References t National Statistical Office. Annual Report on the Cause ofDeath Statistics. Republic of Korea, 1980-1996. 2 Prevalence ofSmoking in Korean Adults The Korea Gallup Report. 1994; 12:16. 3 Enviranmental Protection Agency. Respiratory Health Effects of Passive Smoking: Lung Cancerand OtherlMsorders. US EPA office of research and development, Office of Air and Radiation. EPA760016-90, 1992. `' Wu AH. Environmental tobacco smoke II: lung cancer. In: Steenland K. Savitz DA. Topics in Environmental Epidemiology. New York: Oxford University Press, 1997. 5Yuan JM, Ross RK, Wang XL Gao YT, Henderson BE, Yu ^^ _ Morbidity and mortality in relation to agarettc smoking in . China. JAMA 1996;275:1646-50. 6Kiyohaka Y, Ueda K, Fujishima M. Smoking and cardiovascular dis ease in the general population in Japan. JHypertens 199D;8(S•_pp1.C; S9-S15. 7 US Department of Health and Human Services. ReduUng the Consequences of Smoking: 25 Years of Progress. A Report of the f_.~ General. Washington DC: DHHS Publication No. (CDC) 89-841 t 1989. aWhite R3, Proeb FH. Small-airway dysfunction in chronically exposed to tobacco smoke. N Engi J Med 1~ov.3 ' 720-23. 9Jec SH, Appel Lf, Suh II et aL Prevalence of cardiovascular risk ~-.~- in South Korean adults: results from the KMIC study. Ann 1998;8:14-21. to Hirayama T. Non-smoking wives of heavy smokers have a risk of lung cancer: a study from Japan. Br Med J 19?1?a' 183-85. 11 Sandler DP, Everson RB, Wilcox AJ. Passive smoking in and cancer risk. Am J Epidemiol 1985;121:37-48. 12 Dalager NA, Pickle LW, Mason TJ et al. The relation of pascive =_-: _' to lung cancer. Cancer Res 1986;46:4808-11. 13 Zaridzw D. Smoking husband ups wife's lung cancer risk. Intern Cancer 1998;75:355-58. 1iLee PN. Lung cancer and passive smoking: association of an- due to miselassification of smoking habits? 7oxicol Len 1987;75 157-62. 15 Jee SH, Kim IS, Suh H et ai Projec2ed mortality from lung cancer ' South Korea, 1980-2004. Inr J Epidemiol 1998;27:365-69. rs Ohr H. Kim IS, Jee SH. Shon TY. Smoking and female lung cancer' morphological typcs, a case-control study. Korean JEp+demiol 199% 14:151-59. N O O CA N O O 0 OD V

HUSBANDS' SMOKING AND LUNG CANCER 825 1,177,961 KMIC insured workino men or women aged 18 to 65 ~--~ 321,730 working women 856,231 working men ~ 938 ex-smokers 1.756 current smokers 104,923 wives with no questionnaire r Study subjects : 157,436 wives Figure 1 Study samples analysed Materials and Methods The Korean Medical Insurance Corporation (KMIC) provides health insurance to civil service workers, teachexs.and their dependants. For the entire Korean population of approximately 43 :uilIion in 1992, 4 603 361(I0.7%. (10.7%) people were i¢suredby KMLC.Of the 4 603 361, 1 177-961 areinsured workers (pre- dominantly men, $56 231)pand3-425-400 are dependants. All insured workers are required¢o participate in biennial medical examinations performed by KMIG: The examination is optional for dependants. In 1992, 94.4% of workers completed biennial examinations. Aproximately 35% of dependants completed the biennial examinations. Among.856 231 male workers,. 265 053 took medical examinations and were married. A total of 15.$. 927 non-smokirig wives aged 40 or'ovei completed the examination and were thus included in thesmdy (Figure I). Datz collection The KMIC biennial exarninations.are conducted in a standardized fashion by medical staff at local hospitals.. In 1992, examinations were conducted at 416 hospitals. A questionnaire was given to each participant 3-4+days before: examination.9 In the 1992 and 1994 questionnaire for fnsuredworkers, and in the 1993 ques- tionneire for dependants, participants were asked to describe their smokfng habits (including the number of cigarettes smoked per day and the duration of cigarette smoking in years), along with other health habits, induding vegetable consumption and alcohol drinking. The completed questionnaires were reviewed and edited by trained staff. All Korean people have a unique 13-digit fdentification number that identifies their hospital admissions. Also, f:MIC has a computerized system for man- aging discharge data that doctors submitted to KMIC for reimbursement of medical care services. Using data collected in the 1992 examinations, men were classified as current smokers if they had been smoking for at least one year, non-smokers if they had never smoked, and ex-smokers if they had once smoked but had quit. Smoking data in 1994 were used.as a validation check on smoking status. If.the duration and amount smoked were reported for both 1992 and 1994, averages of those reports were used for finall analysis. lf a non-smoker in 1992 became a current smoker in 1994, we checked the duration of smoking and classified him as non-smoker in cases of short duration (<1 year). Current smokers were further classified by the average number of - dgarettes smoked per day (I-19 and >-2o cigarettes/day) and the duration of smoking (1-29 and -30 years). We used the medical care premium as a proxy variable for socioeconomic status because the premium was calculated based on income. Occupations of husbands were classified as blue and white collar. Blue collar was defined as technical and daily jobs with low incomes. Residency was classified as mral or urban. Wives were grouped into three broad vegetable consumption categories based on self-reported diet habits: low, moderate and high intake. Lung cancer occurrence (ICD-9, 162) was ascertained from diagnosis on discharge summaries. For those individuals with more than one discharge event, we used the first event (onset) in our analysis. The follow-up period was from July 1994 to December 1997. ~

HUSBANDS' SMOKING AND LUNG CANCER 827 Table 2 Rate ratiosa (95% confidence interval) for selected causes of morbidity in women according to the smoking habits of husbands Husbands' smoking habit Causes of morbidity Non-smoker Ex-smoker (95% Ci) Current smoker (95% CI) Emphysema asthma (n= 142) 1.0 1.5 (0 9-2.4) 2 1 (0 8-1 9) , ........... . . ......... . . . Cancer of cervix (n = 203) 1-0 0.9 (0.6-1.3) 0.9 (0.6-1.2) Stomach cancer (n = 197) 1.0 1.0 (0.7-1.s) 0.9 (0.! 1.2) Breast cancer (n= 138) Lo 1 2 (O ~LSJ 1 3 (0 9-1 8) . _._...._. . . ._.._.._ . . . Liver cancer (n = 83) 1.0 0.8 (0.5-1.50 0.7 (0.4-1.1) a Rate ratios wexn determined by Poisson regression after adjusting for the age of both husbands and wives, socioeconomic status, residency, husband's vegetable consumption, and husband's ocaupation. their wives' morbidity from lung cancer suggests that these findings were not chance results. Instead, they indicated that the duration of smoking among smokers is a more predictable indicator for exposure to passive smoke than the amount smoked. We found that smoking was the only characteristic of hus- bands which affected their vVives' morbidity from lung cancer. This result was similar to Hirayama's study.1o To determine whether such an effect was limited to lung cancer, a similar analysis was conducted with other causes of morbidity. In relation to breast cancer, it might be prevalent- incident bias. One way to overcome this bias was to perform analysis using past history of smoking (i.e. duration of smok- ing). We found the risk of developing breast cancer by duration of husband's smoking (>30 years) was significant. Although there was a relation between husbands' smoking habits and morbidity from breast cancer in their wives, the effect of passive smoking was strongest for ltmg cancer. Passive smoking did not seem to increase the risk of developing cancer of the stomach, cervix or liver. The evidence on passive smoking and respiratory health was recently reviewed by the USEPA (1992)3 and Wu (1997)4. This review confirmed that environmental tobacco smoke (ETS) is causally-linked to lung cancer. The present findings tend to be in general agreement with previous studies.l0-12 In Hirayama's study, the RR of developing lung cancer was 2.0 for current smokers, which is quite sirttilar to the RR of 1.9 (95% CI : 1.0- 3.5) for current smokers in this study. In other studies, the corresponding RR were 1.6 (P < 0.01) for Sandler (1985 ), t 1 1.5 (95% CI: 0.8-2.8) for Dalager (1986) 12 and 1.5 (P < 0.05) for Zaridzw (1998).13 However, Hirayama's (1981) studyto used a multi-centre prospective design, which did not provide for the duration effect of husbands' smoking. As well, the studies of Sandler (1985)11 and Dalager (1986) tZ used a case-control design, which did not provide the best opportunity for determining whether a relationship existed. Therefore, when we interpreted the results from case-control studies, we had to be concemed about whether an association was true or due to the misclas- si$cation of smoking habits.14 Even prospective studies could be influenced by such misclassification bias. However, it is unlikely that this prospective, observational study was affected by relevant biases because exposure, includfng smoking habits, was questioned in husbands and wives independently. Because of the low prevalence of smoking among Korean women, we could not determine the risk association between lung cancer and smoking in women. The strengths of this study included high follow-up rates, large sample size, and repeated measures of smoking habits, leading to high precision of the exposure estimates. To increase the generality of the study results, nationwide representative data sets were used. The potential limitations of the study included the relatively brief duration of follow-up, inclusion of individuals with pre- valent lung cancer in the cohort, and reliance on diagnoses from discharge summaries. Although the duration of follow-up in our analyses was just 3.5 years, the large size of the cohon (>150000 participants) provided sufficient statistical power, even in subgroup and dose-response analyses. The inclusion of ~ people with antecedent lung cancer events could potentially lead to biased estimates. However, the impact of prevalent lung cancer was diminished because individuals who experienced cancers between June 1992 and June 1994, the years of baseline data collection, were excluded. Reliance on diagnoses from hospitalizations may have introduced random and system- atic errors. Random error would tend to diminish the study's power to detect associations. Systematic error could alter the distribution of events and perhaps risk factor-disease relation- ships if the errors were related to exposure status. However, the consistency of our findings suggests that major systematic errors related to the coding of lung cancer were unlikely. In relation to the validity of diagnosis, most hospitals required pathological examination to cnnfirm cancer diagnosis. One earlier study reported distribution of morphological types of non-smoking Korean women lung cancer patients: adenocarcinoma, 50.3%; squamous cell, 27.2%; small cell carcinoma, 11.3%; and other and mixed types, 11.2%.t5 30 20 3 21983 1985 1987 1989 1991 1993 1995 Calendar year Bigure 2 Age-adjusted mortality for lung cancer in Korea (I983-1996)

0Intemational Epidemiological Association 1999 Printed in Great Britain IntematonalJOUrnaloJBpidemiology 1999;28:824-828 Effects of husbands' smoking on the incidence of lung cancer in Korean women Sun Ha Jee,a Heechoul Ohrrb and 11 Soon Kimb .................._:...........__ .-._.__.............._.._._.._.._.._........_..._..__ ._..........._....._....._. .__.._._..._....................... Background Although smoking remains tmcommon among Korean women, lung cancer mortality is rapidly escalating.~ ~ ................................ ....................... . ...................... _.................. ........ .... _......... .... Methods Weinvestigated the effects of spousal smoking in-160 130 Korean ivomen, aged 40-88, who received health insurance from the Korea Medical Insurance Cor- poration poration (KMIC). Exposure data were collected.during medical examinations conducted between April 1992 and June 1994. The primary outcome variable was the incidence of lung cancer defined by hospital admissions between July 1994 and December 1997. Standardized rates_ for the incidence of lung cancer were assessed according to the smoking habits of their husbands. . ..........................._._......................................._........_......._........__.__ ........_....._...................................................... _. Resulis At baseline (n = 160 130),-53.9% of ~husbands were smokers and 23,3% were ex-smokers, while 1.1% of wives {n =. 1756) were current smokers and 0.6% (n = 938) were ex-smokers..Dtuing follow-up, 79 cases of ltmg~cancer occurred among non-smoking wives (n = 157 436):-Wives of heavy smokers were found to have a higher risk of developing lung cancer..The husbands' smoking habits did not affect their wives' risk of developing other cancers such as those di the stomach, liver and cervix, but they did affect breast cancer, which has a signific- antly higher risk in relation to the longer duration of husbands' smoking. In Poissfon regression models, adjusting for the age of both husband and wife, socioeconomic status, occupation, residencyy and vegetable intake, the rate ratio (RR) of lung cancer in non-smoking wives was 1.9 (95% CI : 1.0-3.5) in current smokers and 1.3 (95 % CI : 0.6-2.7)) in.ex-smokers. The RR of lung cancer was 3.1 (95 % CI : L4- 6.6) in wives of husbands who had smokedfor30-years or more compared with . wives of non-smokfng-husbands. .................. ' ............. :............................................ ::. .............................................. . . . . . ................... ... . .... ... ........ Conclusion InKorea the inadence.of-lung cancer is higher among non-smoking women ~ whose husbands smoke,'and a dose-response relationship seems to exist. ................ ............................ _..___..... ._-............................. _._..___........................... ._....__.._...._..___.._..---. Keywords Husbands' smoking, lung cancer, non-smoking wives ... ......................................... --..... .......................................... .......................... . ......... .................. .... Accepted 11 February 1999 ~ -- _ .................................:...:.................. ........................................ ....... ................................ ....._........ ........... ........._................ Lung cancer mortality has been reported as the most rapidly increasing cause of death among Koreans. Rates increased from 2.11100 000 in 1980 to 28.0/100 000 in 1996 among men and from 1.4/100 000 in 1980 to 6.91100 000 in1996 among women.t This increase in lung cancer mortality is persistent in men and women despite the fact that few women smoke, while the prevalence of smoking among Korean adult'men is 72 % Z Other risk factors for lung cancer such as radon and asbestos are a Departments of Epidemiology and Disease Control, GraduateSchool of Health sdence and Management Yoasei Univeisity,'Seou1; Korea. b DeParnnent of Preventive Medicine and Public Health,Yonsel University 'CoRege of Medidne, Seoul Koxea. .. Reprint requests to: Sua Ha Jee, Depaxtment of Epidemiology and D'sease Connol Graduate school of Health Sdence and Managemenq Yonsei Univer- sity, PO Box 8044, Beoul 120-749, Korea. E-mail: jsunha&yumc.yonsei.ac.kr uncommon. Thus, passive smoke is a probably a cause of lung cancer in women. . . . Cigarette smoke, in particular passive smoke, is widely recog- nized as a major risk factor for iung cancer in Westem countrles.3-4 Nonetheless, few-studies~have examined the relationship be- tween passive smokfng and lung cancer in East Asian countries, where the prevalence of smoking is reportedly among the high- est in the~world.5-6 The possible~.health consequences of long- term exposure to cigarette smoke should be studied thoroughly among non-smoking wives of smokers because the side-stream . and second-hand smoke fmm dgarettes contain various toxic substances, including carcinogens.7-8 In this report, the effect of passive smoking on lung cancer was studied by following 157-436 non-smoking wives aged 40 and over and measuring,their risk of developing lung cancer according to the smoking habits of their husbands. 824 2505586083

826 INTERNATIONAL JOURNAL OP EPIDEMIOLOGY Table 1 Age-adjusted rate of lung cancer per 100 000 person-years and adjusted rate ratio of lung cancer, by smoking habits of husbands: KMIC 5[udy, 1992-1997 . . . . . . . . Cases of Rate/100 000 Age-adjitsted Multivariate- Husbands' smoking ~ No. lung cancer person-yeara RRb•° 95% CIb adjusted RRd 95% CI Smoking status .............. ............... _.................. ................................... ..._..................... .._....................................... ......... .... _........_........._..................._. Non-smoker 36 109 12 4.4 1.0 1.0 .. __....._ .............. ....... .__................... ... _........ ......................................... ........................................ ................. .... 6-2.7 Ex-smoker 36 802 16 6.5 1.3 0.6-2.7 1.3 07-1-1 ......... . . . . ... . ............... . . ...... . ........... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . : . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .................. . ....... . .............................. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cunentsmoker 84525 51 12.5 2.0 1.1-3.8 1.9 1.0-3.5 ........._...................................._....... ..................... ._._...................... ...._......................... ....................... ........................ No. of cigarettes among current smokers . ............. ....... ............ _.................................... ......................... .............. ........ _............._................... ...... ___.............................. ......... 1-19 72 254 35 12.5 2.1 1.1-4.0 2.0 1.1-3.9 .................................................................................................... ... ............................... ................. :...... ....... ....... ......... ............ ..................... ........... ...................... 320 12 271 16 13.3 1.6 0.8-3.5 1.5 0.7-3.3 ................................... ..... . ......... . Pfornend P<0.1 P<0.1 -.. ..................... .:....._ . . :: . .___..._ _ . _ ..::.... ....._..__.._ _..._.__ -___.... . ...._._..._ .__.__ __._.... Period (year) among current smokers ......... ......... ............ . ............... ........... 1-29 53 881 36 11.0 1.7 0.9-3.3 1.6 0.8-3.0 . . . . . . . . ... . . . ... . ........... . . . . . . . . . . . . . . . . . . ........... . ....... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ............. . ...... . . . . ........... . . . . . . .......... . . . . . . . . . . . . . . . .......................... . ......... 330 30 644 15 21.1 3.3 LG-7.0 3.1 1.4-6.6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .................... . ....................... . ............ . . . . . . . . . . . . . . . . . . . . ....... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ....... . . . . . . . . . . . . . . . ..... Pfortrend . P<0.01 P<0.01 ° Age-adjusted rate by directed method using 1995 Korean National Census Population. b RR = rate mtio; CI = cnnfidence interval. . .. ` Adjusted for the age of both husbands and vrives: . d Adjusted for the age of both husbands and vrives, socioeconomic status. residency, husband's vegetable consumption, and husband's occupation. Statisrical analysis .. . of wives, such as socioeconomic status and residency were not Standardized morbidity rates for lung canrer among non- associated with wives' morbidity from lung cancer, smoking women calculated by-smokfng habits of husbands. The independent effects of smoking on lung cancer wele ex- Direct standardization was employed to obtain age-adjusted amined in Poisson regression models that simultaneouslp con- rates per 100 0000 person-years of observation by 5-year age trolled for the age of husbands and wives, sodoeconomic status, intervals, using the age distribution of the 1995 Koreann census residency occupation, and vegetable consumption. Compared as the standard. In univariate analysis, ordinal variables were with wives of non-smoking husbands, the RR for developing tested individually for risk trends across categories. A Poisson lung cancer in non-smoking wives were 1.9 (95% CI : 1.0-3.5) regression model was used to assess the independent relation- in curient smokers and 1.3 (95% CI : 0.6-2.7) in ex-smokers. ship between smoking.and lung cancer, controlling for other Compared with wives of non-smoking husbands, the wives of risk factors. . . ... husbands who had smoked for a30 years had an increased risk of lung cancer (RR = 3.1, 95 %CL: 1.4-6.6). The husbands' smoking habits seemed to-have no effect on Results their wives' risk of developing other major cancers, such as Among 160 130 spouses, . 53.9% of husbands . were current cancers. of the cervix (n.= 203), stomach (n = 197), and liver smokers and 23.3%.were ex-smokers; whileL1% of.wives .(n=83);buttheydidaffectbreastcancer(n=138).Theriskof (n = 1756) were currenrsmokers and 0.6% (n = 938) were ex- developing emphysema and asthma (n = 142) also seemed to be smokers. A total of 79 cases of lung cancer in 157 436 non- higher among.the wives of smokers, and the effect was statistic- smoking wives were recorded during 3.5 years of follow-up ally borderline significant for ex-smokers (P = 0.08), but not (July 1994 to December 1997). . forcurrent smokers (P = 0.33) (Table 2). Compared to non- Age-adjusted morbidity rates-~ for lung cancer-r were..4.4/ . smoking husbands,_there was an increased risk of breast cancer 100 000 person-years when husbands were non-smokers, 6.5 among the wives of current smokers (>30 years) (RR = 1.7, when husbands were ex-smokers, and 12.5 when husbands 95% CI : 1.0-2.8, P = 0.035) (data not shown). were current smokers (Table 1). Compared with wives of non- smoking husbands, the wives of husbands who had smoked for ~30 years had 230% increased morbidity after adjusting for the Discussion age of husbands and wives (age-adjusted rate Yatio.[RR[.=..3:3, In a population where smoking is uncommon among women, 95% CI : 1.6-7.0).,Tn terms of the amounrof husbands'~smok- this prospective, observational study demonstrated that hus- ing, wives whose husbands.smoked 320 cigatettes/day.seemed: bands'.smoking.wasan independent risk:factor in their wives' to have aninereased risk for.lung cancer(age-adjusted RR = 1.6, lung cancer. Compared to non-smokers, the RR of developing 95% CI : 0.8-3.5), however tests for trend were not statistically lung cancer in non-smoking wives were 1.9 (95% CI : 1.0-3.5) significant, probably because of the small number of cases. Other in smokers and1.3 (95% CI : 0.6-2.7) in ex-smokers. Con- characteristics of.husbands, such as occupat'ton,.alcohol and tinued exposure (>30 years) to their husbands' smoking vegetable consumption, did not affect morbidity from lung cancer increased morbidity from lung cancer in non-smoking wives up in their wives. The RR of discharge from lung cancer were 1.0 to about threefold. . and 1.4 (95% CI : 0.8-2.6), respectively, when husbands' occu- . The fact that there was a statisflcally significant dose-response pations were white collar or blue collar. Common characteristics relationship between the duration of husbands' smoking and 2505586085