24
Printed in Great Britain
Since 1982, cancer has been the leading cause of mor-tality in Taiwan. Among all cancers, the lung is the second most frequently affected site for both men and women. Lung cancer has also been the leading cause of cancer death in women since 1986.1Although cigarette smoking is considered to be the most important cause of lung cancer, smoking behaviour cannot fully explain the epidemiological characteristics of lung cancer in Taiwanese women,2,3who, like Chinese women in the US,4 Australia,5 Hong Kong,6 Singapore,7 Malaysia,8
and in some cities on mainland China;9smoke relat-ively rarely but contract lung cancer relatrelat-ively often.
Epidemiological surveys worldwide have consist-ently shown that the overwhelming majority of male lung cancer patients (more than 97%) have a history of
cigarette smoking. Among female lung cancer patients there are generally fewer smokers and the figures are more variable (6–92%).10 The lower percentages of smokers among Chinese female lung cancer patients were reported from Hong Kong (56%), Singapore (52%), San Francisco (44%), mainland China (35%) and Hawaii (22%), compared with 70–90% in Europe and Amer-ica.10In Taiwan, only 10% of smokers in female lung
cancer patients was observed, whereas 86% male lung cancer patients had a history of smoking.11
Further-more, the prevalence rate of cigarette smoking was 55– 60% in males, 3–4% in females aged over 16 years observed in the nationwide annual investigation.12The trend was steady over the past 30 years but there has been a small decrease in recent years possibly related to health education. If smoking was the sole major cause of lung cancer, then the above rate differences between male and female lung cancer patients and smoking behaviour in the general population of Taiwan appear to be contradictory. Thus, in addition to smoking there are © International Epidemiological Association 1997
* School of Public Health, Kaohsiung Medical College, No. 100 Shih-Chuan 1st Road, Kaohsiung, Taiwan, ROC 807.
** Department of Internal Medical, Kaohsiung Medical College. †Department of Pathology, Kaohsiung Medical College. ‡Department of Ophthalmology, Kaohsiung Medical College.
Risk Factors for Primary Lung
Cancer among Non-Smoking
Women in Taiwan
YING-CHIN KO,* CHIEN-HUNG LEE,* MEI-JU CHEN,* CHI-CHIH HUANG,** WEN-YU CHANG,** HSIANG-JU LIN,†HWEI-ZU WANG‡AND PO-YA CHANG*
Ko Y-C (School of Public Health, Kaohsiung Medical College, No. 100 Shih-Chuan 1st Road, Kaohsiung, Taiwan, ROC 807), Lee C-H, Chen M-J, Huang C-C, Chang W-Y, Lin H-J, Wang H-Z and Chang P-Y. Risk and protective factors for primary lung cancer among non-smoking women in Taiwan. International Journal of Epidemiology 1997; 26: 24–31. Background. Although cigarette smoking is considered to be the most important cause of lung cancer, smoking behaviour cannot fully explain the epidemiological characteristics of lung cancer in Taiwanese women, who rarely smoke but contract lung cancer relatively often. There are other causes of lung cancer that have produced variability in lung cancer incidence.
Methods. A case-control study involving interviews with 117 female patients (including 106 non-smoking) suffering from lung cancer and the same number of individually matched hospital controls was conducted in Kaohsiung, Taiwan between 1992 and 1993. The questionnaire administered to cases and controls collected information on cigarette smok-ing and suspected risk factors for lung cancer. Multivariate logistic regression analysis was applied to assess smoksmok-ing for all women and suspected risk factors for non-smoking women.
Results. The relationship between cigarette smoking and lung cancer was statistically significant although only a small proportion (9.4%) of female patients had smoked. However, the risk of contracting cancer for non-smoking women appears to be associated with certain cooking practices, especially preparing meals in kitchens not equipped with a fume extractor at cooking age of 20–40 years (odds ratio [OR] = 8.3; 95% confidence interval [CI] : 3.1–22.7. These factors and a history of pulmonary tuberculosis plus low consumption of fresh vegetables explained 78% of the summary attributable risks for non-smoking women in a multivariate logistic regression model.
Conclusions. Exposure to fumes from cooking oils, when not reduced by an extractor, may be an important factor in causing lung cancer in non-smoking Taiwanese women.
other causes of lung cancer that produce variability in lung cancer incidence.
Passive smoking,13,14 occupational exposure to
car-cinogens,15,16 ambient air pollutants,17,18 dietary fac-tors,19,20 cooking fuels21,22 are risk factors that could
independently cause lung cancer. This hospital-based, matched case-control study was designed to clarify the risk factors for lung cancer and their role among non-smoking women in Taiwan.
MATERIALS AND METHODS
Study Population
This study was conducted with a hospital-based, case-control design and the source population was geo-graphically defined from the greater Kaohsiung area, which included a city and some suburban and rural communities.
The cases were drawn from Kaohsiung Medical Col-lege Hospital, which is regarded as a leading teaching hospital in South Taiwan, and is accessible to patients from all socioeconomic categories. All newly hospital-ized women, suspected of suffering from primary carci-noma of the lung (ICD 9th Revision code 162) were verified by the pathologist and from hospital medical records. A total of 128 eligible lung cancer cases were identified between January 1992 and December 1993. Of these, five patients had died or been discharged by the time the interviewers visited the wards, two were too ill to participate, and four refused to be inter-viewed. In all 117 women suffering from lung cancer were interviewed.
The controls were selected from women presenting for a general health check or attending the ophthalmic department in the same hospital with non-smoking re-lated disease. Of a total of 125 age-matched women, who were interviewed concurrently with the cases, 117 (94%) agreed to and completed the interview for the control group (62 female ophthalmic patients and 55 women pre-senting for a health check). A female control was matched to each case by age (± 2 years) and date of interview.
Interviews
Two interviewers completed a questionnaire for all can-cer cases and controls by the cross-over method (un-aware of the hypothesis being explored). They collected information on demographic characteristics, smoking habit, exposure to environmental tobacco smoke, his-tory of lung diseases, occupation, dietary factors, life-time residence, and cooking practices and conditions. A field supervisor checked all completed questionnaires and medical abstracts which were then transferred to coding sheets for analysis.
Data Specification
To minimize errors in quantitative data due to incorrect recall, we used a structured questionnaire in order to collect categorized information. A smoker was defined as someone who had smoked one cigarette or more per day for at least one year. To assess passive smoking, we asked each woman about lifetime exposure to environ-mental tobacco smoke generated by parents, husband, cohabitants or co-workers. Anyone who lived or worked with a smoker and was exposed to tobacco smoke was considered to be a passive smoker. A history of pul-monary tuberculosis was defined as a bout of tubercu-losis confirmed by a chest physician at least 3 or 4 years before the diagnosis of lung cancer. Occupational history involved a lifetime list of jobs which were held for at least one year at a time. According to occupational exposure to airborne contaminants, each main job was classified into one of four categories: administration, hazardous industry, farmer, others, and the category of hazardous industry included iron foundries, and petroleum, chemical and cement factories. Informa-tion on daily diet habits was collected with 12 food items, which included the frequency and quantity of consumption of meat, vegetable, fruit, pulses, pickles, etc. The place of residence where the interviewee had resided for at least 5 years and/or longest, was con-sidered as main residence (average duration is 47.9 ± 17.9 years for cases, and 45.4 ± 18.1 years for con-trols), and if this was within 3 km of a major industrial site, the interviewee was classified as living adjacent to an industrial district. Furthermore, we asked each housewife who had to cook for her family about cook-ing practices and conditions, such as the cookcook-ing fuel used, use of cooking oils and use of fume extractors, amounts, frequencies, personal preferences and dietary habits, etc. Cooking conditions were classified into three cooking ‘age stages’ according to a housewife’s lifetime cooking exposure: cooking before 20 years of age represents early cooking exposure (stage I expos-ure); cooking at 20–40 years of age represents main cooking exposure (stage II exposure); cooking after 40 years of age represents late cooking exposure (stage III exposure), and the age that cooking started was when a housewife cooked regularly on a daily basis.
Statistical Analysis
The authors analysed the information by using the statistical software of STATA,23which included χ2test for the similarity of demographic factors and a multi-variate technique designed for matched case-control studies. The odds ratios (OR) of contracting lung cancer through associations with various factors and 95% con-fidence intervals (CI) of OR were estimated from
conditional logistic regression analysis.24 Statistical significance of trends for matching pairs was calculated by categorizing the exposure variable and treating the scored variable as a continuous variable. All estimates of OR were adjusted for educational level (no formal education, 1–9 years, 10 or more years) as well as for residence area (urban, suburban, rural) and socioeco-nomic status (high, medium, low; a ranking score that combines income per family member, place of residence and residence conditions), in order to reduce residual confounding. To focus our regression analysis on the subgroup of non-smoking women, the few interviewees who smoked (11 cases, 3 controls) and their matching pairs were omitted from the computations (leaving 105 non-smoking matched pairs). Interaction effects of any
two suspected risk factors were evaluated by incorporat-ing two binary interaction terms for joint effects of two risk factors in the logistic regression models. The judge-ment of interaction was then performed by comparing observed with expected OR under the assumptions of additive and multiplicative models.25The authors em-ployed Buzzi et al.’s method for estimating the pro-portion of lung cancer cases attributable to one or all risk factors considered (population attributable risks: PAR).26
RESULTS
Cigarette Smoking
Before the smoking matched pairs were deleted in data computation, the effect of cigarette smoking asso-ciated with female lung cancer was assessed. The few female smokers (9.4%) were compared with patients who never smoked. The smokers were adjusted for socioeconomic status, residential area and educational status, and were found to be at a 4.2 fold (95% CI : 1.1–15.6) increased risk of lung cancer (data not shown).
Distribution of Cancer Types
The average age of non-smoking case interviewees (61.3 ± 12.6 years) was somewhat greater than that of controls (60.1 ± 13.2 years), but the age patterns of case-control pairs were closely matched. Non-smoking cancer patients mostly suffered from adenocarcinoma (64.8%), as well as from squamous cell carcinoma (17.1%), small cell carcinoma (15.2%) and large cell carcinoma (2.9%).
Demographic Characteristics
The demographic distribution of cases and controls was tabulated by age and six characteristics for non-smoking females (Table 1). There were no statistical differences between cases and controls observed in this population. The demographic similarity of the two sources of controls was also examined for non-smoking patients, but no statistical differences were apparent.
Suspected Factors
The risk of a non-smoking woman contracting lung cancer, whose husband was a smoker, was 30% higher than for a non-smoking woman whose husband never smoked, but this was not statistically significant. There were no case-control differences for having other cohabitants who smoked or being exposed to environ-mental tobacco smoke in the workplace (Table 2). The risk was significantly elevated for non-smoking women living near industrial districts. A significant dose-response relationship was constructed for risks versus duration of residence near a major industrial
TABLE1 Demographic distributions of lung cancer cases and controls among non-smoking females, Taiwan, 1992–1993
Characteristic/category Cases Controls n = 105 n = 105 % % Age (years) ,51 21.9 22.8 51–60 25.7 24.0 61–70 30.5 33.2 .70 21.9 20.0 Residential area Urban 35.2 41.0 Suburban 15.2 16.2 Rural 49.5 42.9 Marital status Single 1.9 1.9 Married 79.1 80.0 Others 19.1 18.1 Religion Buddhism 41.0 39.1 Taoism 18.1 12.4 Folk religion 32.4 37.1 Others 8.6 11.4 Ethnicity Fukienese 90.5 87.6 Mainland 6.7 5.7 Others 2.9 6.7 Years of education None 62.9 59.1 1–9 29.5 27.6 10+ 7.6 13.3 Socioeconomic status Low 14.3 18.1 Median 68.6 63.8 High 17.1 18.1
complex (P trend = 0.021). Risks of contracting lung cancer were also elevated for women who had had pulmonary tuberculosis, although chronic bronchitis and alcohol consumption appeared to be irrelevant.
Cooking Practices and Dietary Factors
Over 96% of non-smoking women participants stated that they regularly cooked the family meals, in accordance
with Taiwanese tradition. The majority of cases (65.0%) and 55.4% of controls began cooking before they were 20 years old (Table 3). The risk of contracting lung cancer for women who began cooking before they were 20 years old was raised 1.6 fold compared with those who started after the age of 20 years. Chinese cooking involves frying ingredients in oil, which produces ample oil fumes to which the cook is exposed. The authors explored the carcinogenic risks to a housewife for three out of four normal cooking techniques involving frying in oil. Cancer risks were found to be independently re-lated to stir frying, the way most Chinese food is pre-pared, and to frying, but not to deep frying (Table 3). After accounting for differences in socioeconomic status, area of residence and educational level, cancer risks for housewives who reported daily consumption of vegetables were found to be somewhat reduced.
TABLE2 Odds ratios (OR) and 95% confidence intervals (CI) of lung cancer associated with suspected factors among non-smoking females, Taiwan, 1992–1993
Factor/category Cases Controls ORa(95% CI)
No. No. Parent smoking No 80 77 1.0 Yes 25 28 0.8 (0.4–1.6) Spouse smoking No 32 37 1.0 Yes 73 68 1.3 (0.7–2.5) Cohabitant smoking No 85 85 1.0 Yes 20 20 1.0 (0.4–2.3)
Passive smoking from workplace No 93 94 1.0 Yes 12 11 1.1 (0.4–3.0) Occupations Administration 54 56 1.0 Hazardous industry 6 5 1.2 (0.3–5.1) Farmer 30 30 0.9 (0.5–1.9) Others 15 14 1.1 (0.5–2.8) Living adjacent to industrial district (years)
None 78 90 1.0 1–20 7 7 1.3 (0.4–3.8) 21+ 20 8 2.8 (1.2–6.5) Chronic bronchitis No 91 96 1.0 Yes 14 9 1.8 (0.7–4.8) Tuberculosis No 89 101 1.0 Yes 16 4 4.7 (1.5–14.7) Alcohol drinking No 98 98 1.0 Daily 7 7 1.0 (0.4–3.1) Tea drinking No 94 85 1.0 Daily 11 20 0.4 (0.2–1.1)
aOdds ratios were derived from matched analyses adjusted for socioeconomic status, residential area and education.
TABLE3 Odds ratios (OR) and 95% confidence intervals (CI) of lung cancer associated with cooking practices and dietary factors among non-smoking females, Taiwan, 1992–1993
Factor/category Cases Controls ORa(95% CI)
No. No.
Age when first cooking (years)b 21+ 36 45 1.0 7–20 67 56 1.6 (0.8–3.0) Stir frying 0–4/week 14 27 1.0 5+/week 91 78 2.4 (1.1–5.2) Frying 0–4/week 29 44 1.0 5+/week 76 61 2.3 (1.2–4.6) Deep frying 0–4/month 82 79 1.0 5+/month 23 26 0.9 (0.5–1.9) Consumption of meat 0–6/week 47 47 1.0 Daily 58 58 0.9 (0.5–1.7) Consumption of vegetables 0–6/week 35 18 1.0 Daily 70 87 0.4 (0.2–0.8) Consumption of fruit 0–6/week 33 28 1.0 Daily 72 77 1.0 (0.5–1.7)
aOdds ratios were derived from matched analyses adjusted for socio-economic status, residential area and education.
bNo cooking housewives (2 cases and 4 controls) were excluded in the analyses.
The Cooking Age
In stage I exposure (cooking before 20 years of age) housewives tended to use wood or charcoal as fuels and animal oils (mainly lard) for frying; in stage II (cooking at 20–40 years of age) gas replaced wood and vegetable oils (mainly peanut or soybean oil) replaced animal oils; and in stage III (cooking after 40 years of age) gas cookers were equipped with fume extractors. After adjusting for demographic factors, housewives exposed in stage II without a fume extractor were 6.4 times more likely to develop cancer than in the other two stages. The protective effects of fume extractors during cook-ing before 20 years of age and after 40 years of age was also observed (Table 4).
The Interaction Effects
The combined effects of any two suspected risk factors were evaluated in the logistic regression equations. The OR thus obtained were compared with the risk es-timates calculated under the assumption of additive and multiplicative effects. There were no significant find-ings for potential interaction effects, except between stir frying and the use of a fume extractor at cooking age of 20–40 years. Moreover, the high risk of stir fry-ing, frying and deep frying using cooking oils was found to be dependent on a kitchen not being equipped with a fume extractor when cooking at 20–40 years of age. An additive interaction model is shown in Table 5.
Multivariate Analyses
Factors indicated by univariate analysis to be involved in the aetiology of lung cancer were evaluated sim-ultaneously within a multivariate conditional logistic model. Cooking conditions and parameters of suspected risk factors for non-smoking women were entered into the regression model (Table 6). Of these, a history of pulmonary tuberculosis and the use of a fume extractor during cooking age of 20–40 years plus consumption of fresh vegetables were found to improve significantly the predictiveness of the regression model, i.e. these factors were significantly and independently associated with lung cancer.
Population Attributable Risks
Following multivariate analysis, the relative attribut-able fraction of a number of risk factors for lung cancer were assessed for non-smoking women (Table 6). The overall effects of risk due to the factors investigated accounted for 78% of summary attributable risk among non-smoking women, even though adjustment was made for other suspected risk factors. Although a history of pulmonary tuberculosis caused a higher OR estimate (5.9 fold), only a small proportion of non-smoking women had had tuberculosis and less than 13% of cases may be attributable to it. Frequent consumption of vegetables may protect 20% of non-smoking women against lung cancer. However, the greatest attributable
TABLE4 Odds ratios (OR) and 95% confidence intervals (CI) of lung cancer associated with cooking conditions and cooking age stages among non-smoking females, Taiwan 1992–1993
Factor/category Before 20 years of age 20–40 years of age After 40 years of age Cases/controls ORa(95% CI) Cases/controls ORa(95% CI) Cases/controls ORa(95% CI)
No. No. No.
Cooking fuels No cooking or gas 40/53 1.0 38/50 1.0 94/90 1.0d Coal or anthracite 9/17 0.5 (0.2–1.6) 14/16 1.1 (0.4–3.0) 2/2 1.1 (0.1–8.0) Wood or charcoal 56/35 2.5 (1.3–5.1) 53/39 2.5 (1.1–5.7) 4/4 1.0 (0.2–3.9) Cooking oils No cooking 38/49 1.0 2/4 –b 2/4 –b Lard 51/44 1.6 (0.8–3.1) 38/42 1.0 7/4 1.0d Vegetable oil 16/12 2.0 (0.8–4.8) 65/59 1.4 (0.8–2.6) 91/88 0.5 (0.1–2.2)
Kitchen with fume extractor
Yesc 7/19 1.0 25/66 1.0 76/82 1.0d
No 60/37 5.3 (1.1–25.6) 78/35 6.4 (2.9–14.1) 22/10 2.3 (1.1–5.1)
aOdds ratios were derived from matched analyses adjusted for socioeconomic status, residential area and education. bBecause of the small number of no cooking housewives in age stage II and III, lard was referred to as reference category.
cHousewives who had kitchens with fume extractors were referred to as reference category, and no cooking housewives (cases/controls number are 38/49, 2/4, 2/4 by age stage of cooking) were excluded from the analyses.
risk for a non-smoking woman was the lack of a fume extractor at cooking age of 20–40 years, when the greatest amount of time was spent cooking.
DISCUSSION
This study confirms the well-established associa-tion between smoking and lung cancer, even though smoking prevalence is relatively low in the Taiwanese female population (about 3–4%).12 The authors dem-onstrated significant relationships between smoking habit and lung cancer in Chinese women, that was consistent with evidences of studies from mainland China,9Hong Kong6and Singapore.27
Kaohsiung is a heavily industrialized city in south Taiwan, which covers approximately 153 km2 with a population of about 1.4 million inhabitants. Heavy air pollution which prevails on most days of the year in the city is caused by emissions from industry and city traf-fic. The annual average concentration of particles with an aerodiameter of 10 µm or less (PM10) was reported to be as high as 90–100 µg/m3in the past 10 years.28 Industrial factories in this densely populated city are commonly situated in the middle of residential areas. Any emission from a plant or an accidental event may directly affect residents, and an excess of cancer mor-tality at certain sites has been identified.29In this study
residence near a heavy industrial site was used as a proxy measure for exposure to air pollution; iron found-ries, petrochemical installations and cement factories were the main industries in the area. Although we had
TABLE5 The interaction effects of lung cancer associated with cooking practices and the use of a fume extractor among non-smoking females, Taiwan, 1992–1993
Factor/category Presence of fume extractora
Yes No
Cases/controls ORb(95% CI) Cases/controls ORb(95% CI)
No. No. Stir frying 0–4/week 7/20 1.0 5/3 8.6 (1.2–61.3) 5+/week 18/46 2.2 (0.7–7.6) 73/32 13.3 (3.4–52.4) Frying 0–4/week 11/33 1.0 16/17 9.8 (1.9–49.3) 5+/week 14/33 1.8 (0.5–6.3) 62/28 9.2 (2.8–29.9) Deep frying 0–4/month 21/52 1.0 59/23 5.9 (2.6–13.4) 5+/month 4/14 0.5 (0.1–2.3) 19/12 5.9 (1.9–18.2)
aThe use of a fume extractor during cooking at 20–40 years of age.
bOdds ratios were derived from matched analyses adjusted for socioeconomic status, residential area and education.
TABLE6 Adjusted odds ratios (OR) and population attributable risks (PAR) for lung cancer associated with independent factors indicated by univariate analyses among non-smoking females, Taiwan, 1992–1993
Factor/category ORa(95% CI) PARb
%
Living near industrial district (years)
None 1.0 –c
1–20 0.8 (0.2–3.9)
21+ 2.7 (0.9–7.8)
Cooking fuels (20–40 years)
No cooking or gas 1.0 –c Coal or anthracite 1.3 (0.3–5.8) Wood or charcoal 2.7 (0.9–8.9) Tuberculosis No 1.0 12.9 Yes 5.9 (1.3–25.9)
Use of fume extractor (20–40 years)
Yes 1.0 66.7
No 8.3 (3.1–22.7)
Consumption of vegetables
0–6/week 1.0 20.4
Daily 0.4 (0.1–0.9)
Summary attributable risk 77.9
aOdds ratios were derived from multivariate logistic regression model adjusted for covariates, as well as socioeconomic status, residential area and education.
bPopulation attributable risks were derived from case proportion (relative to all cases) and logistic regression model adjusted for covariates. cNot calculated.
no available data for individual air pollution measure-ments, the finding that living within 3 km of an industrial complex independently elevates the risks of contracting lung cancer has also been observed in other studies,18,21 although their measures of pollution exposure were different. Air pollution should not be dismissed as a negligible risk factor for lung cancer in Kaohsiung city. Epidemiological studies on concurrent pulmonary tuberculosis and lung cancer was reviewed by Aoki, who stated in 1993 that patients with active pulmonary tuberculosis had a higher risk of dying from lung cancer or other malignancies, despite the high mortality from tuberculosis.30Our findings support these observations,
although there was insufficient information about tuber-culosis scars and cavities to confirm a suspected rela-tion between tuberculosis and lung cancer. Furthermore, the mortality trends of tuberculosis in Taiwan had been well-described by age-period-cohort analysis; early birth cohorts were seen to have a relatively higher risk of mortality.31 Thus the antagonistic hypothesis be-tween the two diseases may have been based on the fact that tuberculosis patients had mostly died young, i.e. they did not survive long enough to contract cancer before the advent of modern tuberculosis treatments. Additionally, lack of an adequate disease registration system may have contributed to the failure to reveal an association between tuberculosis and lung cancer.
Recent epidemiological investigations revealed a number of sources of indoor air contamination that may explain the elevated risks of lung cancer among non-smoking Chinese women. Smoke and fumes arising from certain cooking practices were evaluated in several case-control studies in selected areas of China,9,21,22,32
as well as the present study in Taiwan. A high risk of contracting lung cancer among Shenyang or Harbin women was related to use of coal-burning stoves (Kang) in north-east China;21,22similarly elevated risks
to women who used wood or charcoal as cooking fuels were observed in early and main cooking exposure stages (stage I and stage II) of this study. Wood or charcoal was the commonest fuel used for cooking and heating in Taiwanese homes before about 1950, and the vast majority of stage I kitchens (89.6% for cases and 66.1% for controls) were without air conditioning or fume extractors. Correspondingly, in north-east China, women living in houses without separate kitchens or effective ventilation were found to be at increased risk of contracting lung cancer.32 The heavy use of coal-burning stoves and regularly coal-burning wood or charcoal contributed part of the increased risk for lung cancer in Shenyang21and in Taiwan women.
Women with prolonged exposure to fumes from rapeseed oil in Shanghai,9 soya oil in Shenyang or
Harbin,22and stir frying, frying or deep frying using
cooking oils made from animal fats (mainly lard) or vegetable oils (mainly peanut or soybean oil) within kitchens with no fume extractors in Taiwan, were all found to contract lung cancer at an increased frequency. These observations from different parts of China sug-gest that exposure to fumes from several types of cooking oil may be linked to increased risk of lung cancer. The fact that cooks, whether women or men, are also at higher risk provides further evidence to incrim-inate cooking fumes.33
Our interaction and multivariate analysis have indic-ated that the use of fume extractors in the kitchen, that were wider spread in Taiwan by 1970 but not on main-land China, explained the majority of differences be-tween cases and controls, even if cooking fuels, oils and techniques (stir frying, frying, deep frying) were covered in the multivariate logistic regression model (data not displayed). The authors suspect that substances ab-sorbed by the fume extractors, including some specific toxic compounds derived from all cooking oils at high temperatures associated with frying, were strongly re-lated to the risks of lung cancer for non-smoking women. The nature of these compounds in the hope of revealing the cause and mechanism of lung cancer in non-smoking women requires further study.
In summary, the present matched case-control study has confirmed that cigarette smoking is associated with lung cancer in Taiwanese women. However, the risk of contracting cancer for non-smoking women appears to be associated with certain cooking practices, especially preparing meals in kitchens not equipped with fume extractors when aged 20–40 years. Combined with a history of pulmonary tuberculosis and low consumption of fresh vegetables, these risk factors account for 78% of the attributable risks of contracting lung cancer and provide a rationale for preventive programmes for lung cancer in women in Taiwan.
ACKNOWLEDGEMENTS
This study was supported in part by a grant from National Department of Health, the Executive Yuan, Republic of China (Grant No. DOH-82-DC-001) and National Science Council (Grant No. NSC 83-0412-B-037-049). The authors thank Dr F F Cinkotai for his editorial assistance with this article.
REFERENCES 1
Department of Health, Executive Yuan. Republic of China: Health statistics. Vol. II Vital statistics, 1981–1992. Taipei, Department of Health, 1982–1993.
2
Yang S P, Luh K T, Kuo S H et al. Chronological observation of epidemiological consideration: A 30-year consecutive study. Jpn J Clin Oncol 1984; 14: 7–19.
3Chen C J, Wu H Y, Chuang Y C et al. Epidemiological
characteristics and multiple risk factors of lung cancer in Taiwan. Anticancer Res 1990; 10: 971–76.
4
King H, Locke F B. Cancer mortality among Chinese in the United States. J Natl Cancer Inst 1980; 65: 1141–48.
5McCredie M, Coates M S. Cancer Incidence in Migrants to New
South Wales, 1972 to 1984. NSW Central Cancer Registry. Sydney: NSW Cancer Council, 1989.
6Koo L C, Ho J H C, Lee N. An analysis of some risk factors for
lung cancer in Hong Kong. Int J Cancer 1985; 35: 149–55.
7Maclennan R, Da Costa J, Day N E et al. Risk factors for lung
cancer in Singapore Chinese, a population with high female incidence rates. Int J Cancer 1977; 20: 854–60.
8Menon M A, Saw H S. Lung cancer in Malaysia. Thorax 1979; 34: 269–73.
9
Gao Y T, Blot W J, Zheng W et al. Lung cancer among Chinese women. Int J Cancer 1987; 40: 604–09.
10Koo L C, Ho J H C. Worldwide epidemiological patterns of lung
cancer in nonsmokers. Int J Epidemiol 1990; 19 (Suppl. 1): S14–S23.
11
Ger L P, Liou S H, Shen C Y et al. Risk factors of lung cancer. J Formosan Med Assoc 1992; 91: S222–31.
12Annual Report of Tobacco, Alcohol Consumption Investigation
in Taiwan Area. Taipei, Taiwan: Bureau of Tobacco and Alcohol Monopoly, Taiwan Provincial Government, 1993.
13Trédaniel J, Boffetta P, Saracci R, Hirsch A. Exposure to
environmental tobacco smoke and risk of lung cancer: the epidemiological–evidence. Eur Respir J 1994; 7: 1877–88.
14US Environmental Protection Agency: Respiratory health effects
of passive smoking: lung cancer and other disorders. Washington DC, EPA/600/6-90/006F, 1992.
15Beckett W S. Epidemiology and etiology of lung cancer. Clinics
Chest Med 1993; 14: 1–15.
16Jöckel K H, Ahrens W, Wichmann H E et al. Occupational and
environmental hazards associated with lung cancer. Int J Epidemiol 1992; 21: 202–13.
17Tomatis L. Outdoor air pollution and lung cancer. Ann Oncol
1991; 21: 265–67.
18
Jedrychowski W, Becher H, Wahrendorf J et al. A case-control study of lung cancer with special reference to the effect of air pollution in Poland. J Epidemiol Community Health 1990; 44: 114–20.
19
Fontham E T H. Protective dietary factors and lung cancer. Int J Epidemiol 1990; 19 (Suppl. 1): S32–42.
20Ziegler R G, Mayne S Y, Swanson C A. Nutrition and lung
can-cer. Cancer Causes Control 1996; 7: 157–77.
21Xu Z Y, Blot W J, Xiao H P et al. Smoking, air pollution, and
the high rates of lung cancer in Shenyang, China. J Natl Cancer Inst 1989; 81: 1800–06.
22Wu-Williams A H, Da X D, Blot W et al. Lung cancer among
women in north-east China. Br J Cancer 1990; 62: 982–87.
23STATA Statistics Software, Release 3.1. Santa Monica, California:
STATA Corporation, USA, 1993.
24Breslow N E, Day N E. Statistical methods in cancer research.
Vol. I. The analysis of case-control studies. Lyon: IARC, 1980.
25Steenland K, Thun M. Interaction between tobacco smoking and
occupational exposures in the causation of lung cancer. J Occup Med 1986; 28: 110–18.
26
Bruzzi P, Green S B, Byar D P et al. Estimating the population attributable risk for multiple risk factors using case-control data. Am J Epidemiol 1985; 122: 904–14.
27MacLennan R, Costa J Da, Day N E, Law C H, Ng Y K,
Shanmugaratnam K. Risk factors for lung cancer in Sing-apore Chinese, a population with high female incidence rates. Int J Cancer 1977; 20: 854–60.
28Ko Y C. Air pollution and its health effects on residents in
Tai-wanese communities. Kaohsiung J Med Sci 1996; 12: 657–70.
29Pan B J, Hong Y J, Chang G C et al. Excess cancer mortality
among children and adolescents in residential districts pol-luted by petrochemical manufacturing plant in Taiwan. J Toxicol Env Health 1994; 43: 117–29.
30Aoki K. Excess incidence of lung cancer among pulmonary
tuberculosis patients. Jpn J Clin Oncol 1993; 23: 205–20.
31Weil J, Chatennir P. Cancer bronchique primitif et tuberculose
pulmonaire. Acta Physiol 1963; 57: 2–29.
32Lee L T, Chen C J, Lee W C et al. Age-period-cohort analysis of
pulmonary tuberculosis mortality in Taiwan, 1961–1990. J Formosan Med Assoc 1994; 93: 129–46.
33Liu Q, Sasco A J, Riboli E et al. Indoor air pollution and lung
cancer in Gauanghou, People’s Republic of China. Am J Epidemiol 1993; 137: 145–54.
34Coggon D, Pannett B, Osmond C et al. A survey of cancer and
occupation in young and middle aged men. I. Cancer of the respiratory tract. Br J Ind Med 1986; 43: 332–38.
