- Email: [email protected]
A case-control study of lung cancer among Czech women
Lung Cancer 31 (2001) 111– 122 www.elsevier.nl/locate/lungcan
A case-control study of lung cancer among Czech women Antonı´n Kubı´k a,*, Petr Zatloukal a, Peter Boyle b, Chris Robertson b, Sara Gandini b, Ladislav Toma´sˇek c, Nigel Gray b, Libor Havel a a
Department of Pneumology and Thoracic Surgery, Charles Uni6ersity, 3rd Faculty of Medicine, Uni6ersity Hospital Na Bulo6ce, and Postgraduate Medical Institute, Budı´no6a 2, 18081 Prague, Czech Republic b Di6ision of Epidemiology and Biostatistics, European Institute of Oncology, Via Ripamonti 435, 20141 Milan, Italy c National Radiation Protection Institute, S& roba´ro6a 48, 10042 Prague, Czech Republic Received 16 March 2000; received in revised form 29 March 2000; accepted 30 March 2000
Abstract Few data are available to explain the continuing increase in lung cancer mortality among Czech women. The study was designed to examine the role of active smoking and other known or suspected factors. Data collected by personal interviews during the 15 months of a hospital-based case control study are reported. A total of 140 microscopically confirmed cases and 280 frequency-matched controls were analysed using multiple logistic regression. Cigarette smoking was the most important factor associated with excess risk for lung cancer among women. Significantly increased risk was found both among current smokers (OR = 11.20, 95% CI 5.9– 21.2), and ex-smokers (OR = 10.02, 95% CI 5.5–18.4). Positive dose-response gradients (PB0.001) were observed between lung cancer risk and the daily number of cigarettes, duration of smoking, number of pack-years, inhaling, and grade of nicotine dependence assessed by the Fagerstro¨m test (Heatherton TF, Kozlowski LT, Frecker RC, Fagerstro¨m KO. Br J Addict 1991;86:1119– 1470; Pomerleau OF. In: Bolliger CT, Fagerstro¨m KO, editors. The Tobacco Epidemic. Basle: Karger, 1997: 122–131). Exposure to environmental smoke was associated with elevated lung cancer risk (OR =3.58, for lifetime non-smokers exposed both in childhood and in adult age). Physical exercise and body mass index were inversely associated with lung cancer risk. For the category of physical exercise of more than 5 h per week, the odds ratio was 0.38, compared to subjects admitting no physical exercise. For body mass index, the odds ratio for the highest (compared to the lowest) quartile was 0.50. Chronic cough and phlegm (at least 3 months per year) were associated with excess risk (OR =6.07) only if their duration was less than 2 years before diagnosis of lung cancer, and, therefore, they were suspected of being more likely early symptoms of preclinical lung cancer rather than its cause. Our results support the statement that cigarette smoking is by far the most important cause of the on-going epidemic of lung cancer among Czech women, and are consistent with the concept of a balance between risk and protective factors whose eventual maintenance or alteration determine the development of disease (as suggested by Rylander R, Axelsson G, Andersson L, Liljequist T, Bergman B. Lung Cancer 1996;14(Suppl 1): S75– S83). Concerted control of smoking appears to be an urgent priority in lung cancer prevention among women, including specific approaches targeted on the female population. © 2001 Elsevier Science Ireland Ltd. All rights reserved.
* Corresponding author. Tel.: + 420-2-66082593; fax: + 420-2-6880511. E-mail address: [email protected] (A. Kubı´k). 0169-5002/01/$ - see front matter © 2001 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 1 6 9 - 5 0 0 2 ( 0 0 ) 0 0 1 7 8 - 1
112
A. Kubı´k et al. / Lung Cancer 31 (2001) 111–122
Keywords: Lung cancer; Risk factor; Smoking; Physical activity; Women; Epidemiology
1. Introduction
2.1. Study population and data collection
In the female population of many developed countries increases in lung cancer mortality have been observed [1–6]. While in the male population of these countries, the occurrence of lung cancer is related mainly to the past and current prevalence of active smoking, results of some studies among women suggest a potential role of other factors acting either as independent risk factors or as modifiers of the effect of smoking [7 – 10]. According to a recent estimate [9], more than 20% of this increasing disease among US women has been due to causes other than active smoking. The relative importance and contribution of each factor can vary with geographic area and socio-economic conditions. Consequently, the conclusions of some investigations, e.g. of those done among Japanese, Chinese and Hawaiian women [11 –13], may be poorly applicable for Central European countries. To obtain a better insight into the role of tobacco and some other known or suspected lifestyle and other factors, a case-control study addressing the issue of epidemiology and prevention of lung cancer among Czech women was launched in early 1998. This report presents results based on 140 cases and 280 frequencymatched controls, all interviewed during the initial 15-month period of the study. Our specific objectives were to analyse the role of active smoking in lung cancer risk of women with disease diagnosed in the late 1990s, and to obtain more knowledge about some likely cofactors, such as passive smoking, previous lung disease or cancer, physical activity, and body mass index.
Eligible cases for this hospital-based case-control study comprised all female patients aged 25– 84 years, with microscopically confirmed incident lung cancer presenting at the Department of Pneumology and Thoracic Surgery of the University Hospital Na Bulovce between April 1998 and June 1999. Eligibility was limited to citizens of the Czech Republic. Of the 161 eligible cases, 140 cases (87.0%) completed the interview. The reasons for non-participation included the patient’s inability to cooperate during the interview as a result of severe physical or mental disability (12 patients, 7.4%), or death shortly after admission (nine patients, 5.6%). We did not conduct proxy interviews because a considerable number of questions of our questionnaire (e.g. passive smoking in childhood, dietary history) could not be assessed adequately. Controls were all women, and were spouses, relatives or friends of other patients of the same hospital (mostly at departments of pneumology, thoracic surgery, and internal medicine). Of the 551 women aged 25–84 years who were contacted to participate in the study, interviews were completed for 462 subjects (83.8%). Nonresponse was due to ‘no time for interview’ (48 women, 8.7%), subject refusal to be interviewed (39 women, 7.1%), and a language barrier or mental incompetence (two persons, 0.4%). Out of the series of 462 interviewed women, randomly selected subjects were frequency matched to cases using 5-year age-strata (25 –29, 30–34,…, 80–84 years) at a 2:1 ratio. These selected controls numbered 280 (Table 2). A written consent form was signed by each participant. A detailed interviewer-administered questionnaire was completed for each case and control. The interviewers were trained extensively to standardize data collection and coding techniques and to minimize inter-interviewer variation. The majority of in-person interviews took 20–40 min to complete. The questionnaire in-
2. Materials and methods The study was approved by the Scientific Council and Ethical Committee of the University Hospital Na Bulovce, Prague, and the Advisory Committee of the Internal Grant Agency of the Czech Ministry of Health.
A. Kubı´k et al. / Lung Cancer 31 (2001) 111–122
cluded a basic structured section on demographic characteristics; place of residence; type of house, occupation and workplace; further, a complete active smoking history, including the Fagerstro¨m test of nicotine dependence [14,15]; exposure to environmental tobacco smoke (passive smoking); pre-existing lung disease; family history of cancer among first degree relatives (parents and siblings); physical activity within recent 10 years. After completion of the questionnaire, the trained interviewer took basic anthropometric measures, such as standing height and weight.
2.2. Questionnaire and definitions The tobacco section of our questionnaire included modified questions from the Tobacco Use Questionnaire recommended by the Tobacco or Health Programme of the World Health Organization [16]. Subjects were defined as (current) smokers if they smoked any tobacco product either daily or occasionally at the time of the survey. Since none of the participants used tobacco products other than cigarettes the following definitions are concerned with cigarette smoking only. A daily smoker is someone who smokes at least one cigarette a day for at least 3 months, i.e. a total of approximately 100 cigarettes and over. An occasional smoker is someone who smokes, but not every day. Ex-smokers are people who were formerly smokers but currently have not smoked for at least 6 months. Never smokers either have never smoked at all or have smoked less than 100 cigarettes in their lifetime. To assess passive smoking, the interviewers asked about exposure to environmental tobacco smoke generated by parents, husband, cohabitants, or co-workers. The participants were asked to assess the number of hours per day spent in smoky rooms (at home, at work-place, and elsewhere) in adult age, and exposure to environmental tobacco smoke during childhood (before age 16). Participants were asked whether they had ever suffered from any of several lung diseases (tuberculosis, pneumonia, chronic bronchitis), or malignant tumour. To assess the history of chronic bronchitis, asthma, and related diseases modified
113
questions from the ATS-DLD-78 (American Thoracic Society-Division of Lung Diseases-78) questionnaire [17], and the Medical Research Council Questionnaire on Respiratory Symptoms (MRC, 1976) [18] were used.
2.3. Statistical methods Descriptive statistics were first used to characterize the study population. The data were kept in an Epi-Info database [19]. In preliminary examinations, differences between groups were assessed using 2-tests for independence of categorical variables, and t-tests for continuous variables. Risk analyses were first done using unconditional logistic regression [20] which provides results in the form of both crude and adjusted odds ratios for confounders. Potential interactions of independent variables or confounders, were also evaluated with multiple logistic regressions. The logistic models were conducted in two steps: (1) using all selected variables, and (2) excluding those describing active smoking in order to assess a potential confounding role of smoking. Odds ratios in the usual sense were obtained for categorical variables and odds ratios per unit change for the so called continuous variables. Categorization of continuous variables was based mostly on quartiles, except physical exercise which was divided into three groups. The odds ratios and their 95% confidence intervals (CI) were computed. The Mantel –Haenszel trend test was used, based upon integer category scores, to assess the trend in the risk with increased exposure to smoking. Final calculations were carried out using PROC LOGISTIC in SAS [21].
3. Results Among 140 cases, adenocarcinoma (35.0%) was the most frequent cell type, followed by squamous cell cancer (24.3%), and small cell cancer (23.6%) (Table 1). In the subgroup of 24 never smokers, 15 (or 62.5%) adenocarcinomas and one (or 4.2%) bronchioalveolar cancer were diagnosed. The mean age (61 years) and age distribution were identical among the series of cases and fre-
A. Kubı´k et al. / Lung Cancer 31 (2001) 111–122
114
Table 1 Distribution of cases by cell type and smoking history: Czech Women’s Lung Cancer Study Cell type
Never smokers No.
Ex-smokers %
No.
%
Current smokers
All cases
No.
No.
%
%
Squamous cell Small cell Adenocarcinoma Bronchioloalveolar Large cell Carcinoma NOSa
4 1 15 1 1 2
16.7 4.2 62.5 4.2 4.2 8.2
16 15 21 3 4 2
26.2 24.6 34.4 4.9 6.6 3.3
14 17 13 1 7 3
25.5 30.9 23.6 1.8 12.7 5.5
34 33 49 5 12 7
24.3 23.6 35.0 3.6 8.5 5.0
Total
24
100.0
61
100.0
55
100.0
140
100.0
a
NOS, not otherwise specified.
quency matched controls (Table 2). Using crude odds ratios, risk estimates appeared elevated for rural residence, inversely associated with education, and not significantly associated with marital status. Cigarette smoking was associated directly with lung cancer risk (Table 3). After adjusting for age,
residence and education, the odds ratio was 11.20 for current smokers, and 5.00 for ex-smokers who stopped smoking 10 or more years ago. Among subjects who stopped smoking within the recent 10 years, the odds ratio was as high as 18.63. It cannot be excluded that in some of these cases the motive for stopping might have been related to
Table 2 Distribution of cases and controls by socio-demographic variables: Czech Women’s Lung Cancer Studya Variables
Cases
Controlsb
Population Mean age (S.D.) Age groups (years) 25–34 35–44 45–54 55–64 65–74 75–84 Residence Rural (5100 000) Urban (\100 000) Education Elementary Secondary (ordinary) Secondary (advanced) University Marital status Married Widowed Divorced/separated Never married
140 61.0 (10.5)
280 61.1 (10.5)
a b
Crude OR
95% CI
2 6 31 46 41 14
4 12 62 92 82 28
52 88
65 215
1.00 0.51
0.3–0.8
42 50 44 4
63 92 92 33
1.00 0.82 0.72 0.18
0.5–1.4 0.4–1.2 0.1–0.6
80 38 17 5
163 62 41 14
1.00 1.25 0.84 0.73
0.7–2.0 0.5–1.6 0.3–2.1
OR, odds ratio; CI, confidence interval; S.D., standard deviation. Selected controls frequency matched by age to cases at a 2:1 ratio.
A. Kubı´k et al. / Lung Cancer 31 (2001) 111–122
115
Table 3 Active cigarette smoking, nicotine dependence, and the risk of lung cancer: Czech Women’s Lung Cancer Study Variables Cigarette smoking (acti6e) Never smokers Ever smokers Current smokers All ex-smokers Ex-smokers Quitted]10 years ago QuittedB10 years ago Number of cigarettes/day 1–4 5–14 \14 Test for trend Duration of smoking (years) 1–10 11–20 21–30 31–40 \40 Test for trend Pack-years 1–10 11–20 21–30 \30 Test for trend Inhaling No Yes Test for trend Nicotine dependence c Never- and ex-smokers Current smokers None or very low (0–1)d Low (2–4) Medium (5) High or very high (6–8) Test for trend
Cases
Controls
Adjusted ORa
95% CIb
24 116 55 61
176 104 49 55
1.00 10.52 11.20 10.02
6.0–18.3 5.9–21.2 5.5–18.4
19 42
34 21
5.00 18.63
2.4–10.5 9.0–38.7
8 43 65
17 49 38
4.06 7.92 19.49 PB0.001
1.5–11.1 4.2–15.0 10.0–38.2
9 8 28 38 33
13 23 32 21 15
6.67 2.67 7.26 15.70 21.46 PB0.001
2.4–18.5 1.0–7.1 3.5–15.2 7.5–32.8 9.6–48.2
26 26 26 38
44 32 18 10
4.68 8.72 15.51 37.85 PB0.001
2.3–9.4 4.2–18.3 6.9–35.1 15.8–90.5
17 99
25 79
5.82 12.33 PB0.001
2.5–13.6 6.8–22.5
85
231
10 27 6 12
21 22 2 4
1.00 1.29 4.13 8.19 8.92 PB0.001
0.6–2.9 2.1–8.0 1.6–42.5 2.7–29.4
a
OR, odds ratio, adjusted for age, residence, and education. CI, confidence interval. c Fagerstro¨m test [14,15]. d Levels of nicotine dependence (the scale includes grades 0–10). b
symptoms caused by the cancerous lesion evolving in the lungs. The daily number of cigarettes, duration of smoking, pack years, and inhaling, all were significantly associated with lung cancer risk showing a dose –response relationship as demonstrated by tests for the trend of odds ratios (Table 3). We found a significant association of the risk
of lung cancer with low, medium, and high or very high nicotine dependence, assessed by the Fagerstro¨m test [14,15]. Among women who never actively smoked, the odds ratio for exposure to environmental tobacco smoke in adult age (more than 3 h daily; sum of exposures at home, at work-place, and elsewhere)
A. Kubı´k et al. / Lung Cancer 31 (2001) 111–122
116
was 1.17; for exposures in childhood (before age 16) the odds ratio was 2.02 (Table 4). For never smokers with exposures to passive smoking in both childhood and adult age the odds ratio was 3.58 (95% CI 0.6 –21.9) (not shown in Table 4). Chronic cough and phlegm (at least 3 months per year) were associated with excess risk of lung cancer only if their duration was less than 2 years before diagnosis of lung cancer (OR= 6.07, 95% CI 2.2 –16.7). It cannot be excluded that such symptoms emerging recently before the onset of lung cancer may have been early symptoms of the developing lung cancer rather than risk factors. In contrast, for chronic cough and phlegm observed 2 years or longer before the onset of lung cancer (probably, symptoms of chronic bronchitis) the odds ratio was 0.84 (Table 5). The odds ratio for shortness of breath was OR =1.48 (Table 5). As expected, among subjects with chronic cough and phlegm of less than 2 years duration (hypothesized to be early symptoms of the developing lung cancer) the odds ratio for shortness of breath was 26.74 (95% CI 4.6 – 154.9), as compared to OR=2.50 (95% CI 0.5 – 13.4) in the subgroup of persons with chronic cough and phlegm observed 2 years or longer before the onset of lung cancer. The odds ratios for shortness of breath among persons with chronic cough without chronic phlegm was 2.55
(95% CI 0.5 –12.8), and for the subgroup of subjects admitting no chronic cough and no chronic phlegm OR= 0.88 (95% CI 0.4 –1.8) (not shown in Table 5). The adjusted odds ratio for attacks of shortness of breath with wheezing in the chest (probably, mostly due to asthma) was 0.90 (95% CI 0.4 –2.3) (Table 5). There was a small excess risk for subjects with a history of pneumonia (OR= 1.45), tuberculosis (OR= 1.29), or a personal history of previous cancer (OR= 1.77). No association with lung cancer risk was found for the family history of lung cancer among first degree relatives (parents and siblings, OR= 0.81), and for the family history of other cancers among first degree relatives (parents and siblings, OR= 1.08). Data on the risk of lung cancer associated with physical activity and body mass index are shown in Table 6. Physical exercise (or sport, walking, within the recent 10 years) was inversely associated with lung cancer risk. For the category of physical exercise (or sport, walking) more than 5 h per week (compared to subjects admitting no physical exercise) the odds ratio was 0.38. For body mass index, the odds ratio for the highest quartile (compared to the lowest quartile) was 0.50. A significant trend was found with decreasing risk at higher levels of body mass index (Table 6).
Table 4 Exposure to environmental tobacco smoke (passive smoking), active cigarette smoking, and the risk of lung cancer: Czech Women’s Lung Cancer Study Variable: cigarette smoking
Cases
Controls
Adjusted ORa
95% CIb
Active ever
Passive
No No Yes Yes
Noc Yesc Noc Yesc
22 2 87 29
160 16 83 21
1.00 1.17 10.03 14.00
0.2–5.6 5.6–18.1 6.3–31.3
No No Yes Yes
Nod Yesd Nod Yesd
10 14 60 56
105 71 47 57
1.00 2.02 17.39 12.99
0.8–4.9 7.8–38.8 5.9–28.8
a
OR, odds ratio, adjusted for age, residence, and education. CI, confidence interval. c Passive in adulthood: exposure to environmental tobacco smoke \3 h/day, in adult age. d Passive in childhood: exposure to environmental tobacco smoke in childhood (before age 16). b
A. Kubı´k et al. / Lung Cancer 31 (2001) 111–122
117
Table 5 Personal history of previous lung disease or cancer, family history of cancer, and the risk of lung cancer: Czech Women’s Lung Cancer Study Variables
Cases
Controls
Adjusted ORa
95% CIb
No chronicc cough, no chronicc phlegm Chronic cough without chronic phlegm Chronic cough and phlegm,B2 years Chronic cough and phlegm, ]2 years
93 17 18 12
231 22 9 18
1.00 2.19 6.07 0.84
0.9–5.3 2.2–16.7 0.3–2.1
Shortness of breathd Attacks of shortness of breathe Tuberculosis Pneumonia Personal history of previous cancer
43 8 7 49 10
52 26 12 84 12
1.48 0.90 1.29 1.45 1.77
0.8–2.6 0.4–2.3 0.4–4.0 0.9–2.5 0.6–4.9
Family history of lung cancerf Family history of other cancersg
19 48
26 94
0.81 1.08
0.4–1.7 0.6–1.8
a
OR, odds ratio, adjusted for age, residence, education, and pack-years of smoking. CI, confidence interval. c 3 months/year. d Getting short of breath walking with other people at ordinary pace on the level. e Attacks of shortness of breath with wheezing in the chest. f Cancer among first degree relatives (parents and siblings). g Except lung cancer. b
Table 6 Physical activitiesa,b, body mass index, and the risk of lung cancer: Czech Women’s Lung Cancer Study Variables
Cases
Controls
Adjusted ORc
95% CId
Physical exercise a (hours/week) 0 1–5 \5 Test for trend
63 39 38
69 78 133
1.00 0.58 0.38 PB0.001
0.3–1.1 0.2–0.7
Other physical acti6ities b (hours/week) 0 1–5 \5 Test for trend
69 20 51
107 45 128
1.00 0.59 0.65 P=0.61
0.3–1.3 0.4–1.2
Body mass index (BMI, kg/m 2) 13.7–22.9 23.0–25.9 26.0–28.9 29.0–44.9 Test for trend
42 35 35 28
64 62 76 78
1.00 1.04 0.69 0.50 P =0.031
0.5–2.1 0.4–1.4 0.2–1.0
a
Physical exercise (or sport, walking, within recent 10 years). Other non-professional physical activities (e.g. in the garden, house, within recent 10 years). c OR, odds ratio, adjusted for age, residence, education, and pack-years of smoking. d CI, confidence interval. b
118
A. Kubı´k et al. / Lung Cancer 31 (2001) 111–122
Table 7 shows the results of multiple logistic analyses conducted in order to assess potential interactions or confounding among active and passive smoking, chronic cough and phlegm, shortness of breath, physical exercise and body mass index, analysed in the paper. The variables were modelled either as continuous variables, such as pack-years of smoking, physical exercise (hours/week), body mass index (kg/m2); or categorical variables, such as the (active) smoking status, passive smoking, chronic cough and phlegm (at least 3 months/year), and shortness of breath. The comparison of the results of the analyses pointed out the role of active cigarette smoking as the dominant risk factor of lung cancer among Czech women, and suggested a minor role of other factors included in the analyses. Table 7 Multiple logistic analysisa evaluating simultaneously seven selected risk factors: Czech Women’s Lung Cancer Study Variables
ORb
95% CIc
Acti6e smoking Never smokers Ex-smokers ]10 years Ex-smokers B10 years Current smokers Pack-years of smoking (per 1 pack-year)
1.00 4.01 7.39 4.07 1.05
1.8–9.1 2.8–19.9 1.6–10.1 1.02–1.08
1.00 0.85
0.5–2.1 0.5–1.5
Passi6e smoking In adult age (\3 h/day) In childhood (before age 16) No chronicd cough, no chronicd phlegm Chronic cough without chronic phlegm Chronic cough and phlegm,B2 years Chronic cough and phlegm, ]2 years
1.00
Shortness of breathe Physical exercise, or sport, walking (per 1 h/week) BMI, body mass index (per 1 kg/m2)
1.99
0.8–4.9
5.66 0.70
2.0–16.3 0.3–1.8
1.48 0.95
0.8–2.7 0.91–0.99
0.92
0.87–0.98
a All models include adjustment for age, residence, and education. b OR, odds ratio, or odds ratio per unit change in the full model. c CI, confidence interval. d 3 months/year. e Getting short of breath walking with other people at ordinary pace on the level.
4. Discussion Our report presenting results of a hospitalbased case-control study of lung cancer among Czech women has certain potential limitations which should be considered before conclusions are drawn. The exposures of interest were based on self report, therefore, some recall bias is of concern. Small numbers of cases in subgroups of some items (e.g. exposure to environmental tobacco smoke, personal history of cancer) limited the power of analyses and precluded further testing. As expected, adenocarcinoma was the most frequent cell type (35%) found among the 140 cases in the study. Among 24 never smokers, almost two-thirds (62.5%) were adenocarcinomas. Increasing trends of adenocarcinomas have been observed in many developed countries, the ratio between squamous-cell carcinomas and adenocarcinomas was everywhere greater in men than in women [9,22]. Adenocarcinoma has always represented the majority of lung cancers among nonsmokers of both genders, and increased, as a proportion, with increasing duration of smoking cessation [22]. Some authors explained the increases in the proportion of adenocarcinomas among smokers by changes in the design of cigarettes (the use of filter tips, decreasing yields of nicotine and tar) [22]. Results of our case-control study support the claim that cigarette smoking is the most strongly active risk factor for lung cancer among Czech women. The odds ratio for risk associated with lung cancer for ever-smokers compared to neversmokers was 10.52, comparable to observations in western countries (frequently \ 10) while, e.g. in China lower odds ratios (B 5) were reported [23]. Two recently published comprehensive metaanalyses on exposure to environmental tobacco smoke and the risk of lung cancer [24,25] have concluded with the statement that passive smoking is a cause of lung cancer. For a total of 36 case-control studies, Zhong et al. [25] calculated the pooled relative risk 1.19 (95% CI 1.10 –1.29), for five cohort studies 1.29 (95% CI 1.04 –1.62). In our study the adjusted odds ratio was 1.17 for exposure to passive smoking in adult age (more
A. Kubı´k et al. / Lung Cancer 31 (2001) 111–122
than 3 h daily; Table 4), 2.02 for exposure in childhood (before age 16), and 3.58 (95% CI 0.58 –21.91) for exposure both in childhood and in adult age. Many of the controls in our study were relatives, spouses or friends of patients hospitalized at the department of pneumology with diagnosis of smoking related lung disease, therefore, many of the controls were likely to have been largely exposed to passive smoking. This could lead to underestimation of effects of passive smoking in the present analysis. Consequently, any observed adjusted odds ratios greater than 1 in our study should be considered to be under suspicion of signalling a detrimental impact of passive smoking. Trying to control for this confounder we identified a subgroup of controls who were relatives, spouses or friends of patients with a diagnosis of other than smoking-related diseases. For this subgroup, the odds ratio was 3.28 (95% CI 0.20 –53.46), as compared to OR= 0.85 (95% CI 0.16 –4.46) for the subgroup of controls who were relatives, spouses, or friends of patients with smoking-related diseases, and, therefore, were likely to have been much more exposed to environmental tobacco smoke. Hence it follows for investigations into the role of passive smoking in lung carcinogenesis that including spouses, relatives and friends of patients with smoking-related lung diseases is a problematic choice and cannot be recommended, since they are as likely as the cases to have been exposed to environmental tobacco smoke. Previous studies on the role of personal history of chronic respiratory illnesses among non-smokers have been recently reviewed by Brownson et al. [7]. A multicenter population-based case-control study among lifetime non-smoking women in the United States provided evidence of association of asthma, chronic bronchitis, emphysema, and for age groups younger than 55 years tuberculosis and pneumonia with lung cancer risk [26]. In another large case-control study among nonsmoking women [27], each type of lung disease, except chronic bronchitis, showed some elevation in risk of adenocarcinoma, although effect estimates were not always statistically significant. Associations of lung cancer risk with emphysema, chronic bronchitis, and asthma were found among
119
female never and former smokers in a populationbased case-control study in New York State [28]. Some studies have reported an inverse association between asthma or hay fever and lung cancer in women [29]. In our study, respiratory symptoms, such as chronic cough and phlegm (at least 3 months per year) that were observed less than 2 years prior to diagnosis of lung cancer (and showed association with lung cancer risk, OR= 6.07), were more likely to be early symptoms of preclinical lung cancer rather than its cause. Chronic cough and phlegm observed 2 years or longer (probably, symptoms of chronic bronchitis) were not associated with risk of lung cancer (OR = 0.84) (Table 5). Few data are available regarding the association of physical activity with lung cancer risk. A prospective study of 81 516 Norwegian subjects observed that physical activity was inversely related to lung cancer risk in men, but not in women [30]. Data from a cohort study of 13 905 male Harvard University alumni indicated that physical activity was associated with lower risk of lung cancer [31]. The Norwegian investigators [30] hypothesized that the increased pulmonary ventilation and perfusion that occurs with physical activity may lead to reduced interaction time and concentration of carcinogenic agents in the airways. Physical activity, along with reproductive variables and food availability, were recently mentioned by Willett [32] as examples of factors associated with affluence (national wealth), and also related to breast cancer occurrence. Recent evidence from many studies that higher levels of physical activity reduce risk of colon cancer implies that at least part of the high colon cancer rates in affluent countries previously attributed to fat intake are probably due to a sedentary lifestyle [32]. Further studies are needed to understand the essence of the association between the latter variables and lung cancer risk. In our study, an inverse association was found between lung cancer risk and time (hours/week) devoted to physical exercise, within recent 10 years (Table 6). During the ongoing investigation we have realized that physical activity may be subject to changes in time, and lack of physical activity shortly before diagnosis of lung cancer
120
A. Kubı´k et al. / Lung Cancer 31 (2001) 111–122
may more likely be the result, rather than the cause, of the early stage, preclinical disease. Therefore, two additional questions were affixed to the questionnaire, related to physical exercise (sport, walking) 1 year prior to diagnosis, and 20 years prior to diagnosis. The original question on physical exercise (within recent 10 years) was kept in the amended questionnaire. Additional analysis could be done only in a subgroup of 30 cases. Among these cases, the mean time devoted to physical exercise 20 years prior to diagnosis was significantly longer (+1.4 h/week) than 1 year prior to diagnosis. It should be noted that even the time elicited by the original question was significantly longer (+1.0 h/week) than 1 year prior to diagnosis. The difference between the mean time devoted to physical exercise 20 years prior to diagnosis and the mean time elicited by the original question was 0.4 h/week, and was not statistically significant (P= 0.465). Although some part of these differences might be related to ageing, it can be presumed that the information elicited by the original question probably reflected in most cases the time devoted to physical exercise several years prior to diagnosis. Lack of exercise at that time may be suspected of more likely being a risk factor, rather than the result, of early stage lung cancer, however, the opposite cannot be excluded. Further studies are needed to throw more light on the association between physical activity and lung cancer risk. While body mass index is known to be associated with overall mortality [33] and some cancers, e.g. prostate or colon cancer [34], an inverse association with lung cancer was observed in some studies [8,35 –37], including our investigation. A significant inverse gradient between body mass index and the incidence of lung cancer was found in a prospective study of 25 994 men aged 20 – 75 years in Finland [36]. Similar observations were done in case-control studies in the USA: the American Health Foundation (AHF) hospital-based study of tobacco-related cancers [8,35], and the Missouri Women’s Health Study [37]. In the AHF study, the strongest association of leanness was observed in women who never smoked. In a prospective cohort study of 41 837 Iowa women aged 55–69 years, the results of multivariate analy-
ses suggested that the inverse association of body mass index with lung cancer could be explained by smoking status [38]. Even in our study, body mass index was inversely associated with lung cancer risk; body height and weight, used for calculations of body mass index, were taken at the time of interview, and information on previous anthropometric measures was not obtained. Therefore, it should be considered that low body mass index may more likely be the result, rather than the cause, of early stage lung cancer. In conclusion, our results support the statements that cigarette smoking is by far the most important cause of the on-going epidemic of lung cancer among Czech women, and that exposures to environmental tobacco smoke may increase the risk of lung cancer. The dominant role of active smoking among lung cancer factors has generally been recognized [2]. More data are needed to elucidate the role of some cofactors, e.g. some components of diet, tea, coffee and alcohol consumption, events in the personal gynecologic history, the family history of cancer, and others. Our findings support the ‘concept of a balance between risk factors for a disease and protective factors’ [39]. The eventual maintenance or alteration of the balance between these factors have been suggested to determine whether the disease would develop or not. Concerted control of smoking appears to be an urgent priority in lung cancer prevention among women, including specific approaches targeted on the female population, such as efforts to prevent adolescent girls from starting to smoke, and encouraging cessation among established smokers. The issue of preventive recommendations concerning some cofactors continues to be a matter of debate.
Acknowledgements We wish to thank Professor L. Petruzelka, MD, PhD (Department of Oncology, Charles University, First Faculty of Medicine, General Faculty Hospital, Prague), and the reviewers for valuable comments. The financial support by grant No. 4970-3 of the Internal Grant Agency (IGA) of the Czech Ministry of Health is gratefully acknowledged.
A. Kubı´k et al. / Lung Cancer 31 (2001) 111–122
References [1] Boyle P. Cancer, cigarette smoking and premature death in Europe: a review including the Recommendations of European Cancer Experts Consensus Meeting, Helsinki, October 1996. Lung Cancer 1997;17:1–60. [2] Boyle P, Gandini S, Gray NJ. Epidemiology of lung cancer: A century of great success and ignominious failure. In: Hansen HH, editor. Textbook of Lung Cancer. London: Martin Dunitz, 2000:13–25. [3] Kubı´k AK, Parkin DM, Plesˇko I, et al. Patterns of cigarette sales and lung cancer mortality in some Central and Eastern European countries, 1960–1989. Cancer 1995;75:2452– 60. [4] LaVecchia C, Boyle P. Trends in the tobacco-related cancer epidemic in Europe. Cancer Detect Prev 1993;17:495– 506. [5] Parkin DM. The global burden of cancer. Semin Cancer Biol 1998;8:219– 35. [6] Zatloukal P, Kubı´k A, Petruzelka L. Current and future trends in lung cancer mortality in the Czech Republic. Lung Cancer 1998;21(Suppl 1):S2–3. [7] Brownson RC, Alavanja MCR, Caporaso N, Simoes EJ, Chang JC. Epidemiology and prevention of lung cancer in nonsmokers. Epidemiol Rev 1998;20:218–36. [8] Kabat GC. Aspects of epidemiology of lung cancer in smokers and nonsmokers in the United States. Lung Cancer 1996;15:1– 20. [9] Mason TJ. The descriptive epidemiology of lung cancer. In: Samet JM, editor. Epidemiology of Lung Cancer. New York: Marcel Dekker, 1994:51–69. [10] Saracci R, Boffetta P. Interactions of tobacco smoking with other causes of lung cancer. In: Samet JM, editor. Epidemiology of Lung Cancer. New York: Marcel Dekker, 1994:465– 93. [11] Hinds HW, Stemmerman GN, Yang HY, Kolonel LN, Lee J, Wegner F. Differences in lung cancer risk from smoking among Japanese, Chinese and Hawaiian women in Hawaii. Int J Cancer 1981;27:297–302. [12] Ko YC, Lee CH, Chen MJ, et al. Risk factors for primary lung cancer among non-smoking women in Taiwan. Int J Epidemiol 1997;26:24–31. [13] Seow A, Duffy SW, Ng TP, McGee MA, Lee HP. Lung cancer among Chinese females in Singapore 1968–1992: time trends, dialect group differences and implications for aetiology. Int J Epidemiol 1998;27:167–72. [14] Heatherton TF, Kozlowski LT, Frecker RC, Fagerstro¨m KO. The Fagerstro¨m test of nicotine dependence: a revision of the Fagerstro¨m Tolerance Questionnaire. Br J Addict 1991;86:1119–470. [15] Pomerleau OF. Nicotine dependence. In: Bolliger CT, Fagerstro¨m KO, editors. The Tobacco Epidemic. Basle: Karger, 1997:122– 31. [16] World Health Organization, Tobacco or Health Programme. Guidelines for Controlling and Monitoring the Tobacco Epidemic. Geneva: World Health Organization, 1996.
121
[17] Ferris BG, editor. Epidemiology standardization project. Am Rev Respir Dis 1978; 118(6), part 2 of 2 parts: 1 – 53. [18] Medical Research Council. Questionnaire on Respiratory Symptoms. Approved by the MRC’s Committee on Research into Chronic Bronchitis. London: Medical Research Council, 1976. [19] Dean AG, Dean JA, Coulombier D, et al. Epi Info, Version 6: A Word-Processing, Database, and Statistics Program for Public Health on IBM-compatible Microcomputers. Atlanta, GA: Centers for Disease Control and Prevention, 1995. [20] Preston DL, Lubin JH, Pierce DA, McConney ME. Epicure, Version 1.8w: A Package of 5 Programs for Generalized Risk Modelling and Person-Year Computation. Seattle, WA: HiroSoft International, 1993. [21] SAS Institute, Inc. SAS for Windows version 6.11. Cary, NC: SAS Institute, 1995. [22] Franceschi S, Bidoli E. The epidemiology of lung cancer. Ann Oncol 1999;10(Suppl 5):S3– 6. [23] Wu JM, Du Y-X. Summary of papers and research recommendations presented at the International Symposium on Lifestyle Factors and Human Lung Cancer, Guangzhou, China. Lung Cancer 1996;14(Suppl 1):S223– 34. [24] Hackshaw AK, Law MR, Wald NJ. The accumulated evidence on lung cancer and environmental tobacco smoke. BMJ 1997;315:980– 8. [25] Zhong L, Goldberg MS, Parent M-E, Hanley JA. Exposure to environmental tobacco smoke and the risk of lung cancer: a meta-analysis. Lung Cancer 2000;27:3– 18. [26] Wu AH, Fontham ETH, Reynolds P, et al. Previous lung disease and risk of lung cancer among lifetime nonsmoking women in the United States. Am J Epidemiol 1995;141:1023– 32. [27] Alavanja MCR, Brownson RC, Boice JD, Jr, et al. Preexisting lung disease and lung cancer among nonsmoking women. Am J Epidemiol 1992;136:623– 32. [28] Mayne ST, Buenconsejo J, Janerich DT. Previous lung disease and risk of lung cancer among men and women nonsmokers. Am J Epidemiol 1999;149:13– 20. [29] Osann KE. Lung cancer in women: the importance of smoking, family history of cancer, and medical history of respiratory disease. Cancer Res 1991;51:4893– 7. [30] Thune I, Lund E. The influence of physical activity on lung cancer risk: a prospective study of 81,516 men and women. Int J Cancer 1997;70:57– 62. [31] Lee I-M, Sesso HD, Paffenbarger RS, Jr. Physical activity and risk of lung cancer. Int J Epidemiol 1999;28:620– 5. [32] Willett WC. Dietary fat intake and cancer risk: a controversial and instructive history. Semin Cancer Biol 1998;8:245– 53. [33] Calle EE, Thun MJ, Petrelli JM, Rodriguez C, Heath CW, Jr. Body-mass index and mortality in a prospective cohort of US adults. N Engl J Med 1999;341:1097– 105. [34] Chyou PH, Nomura AM, Stemmermann GN. A prospective study of weight, body mass index and other anthropometric measurements in relation to site-specific cancers. Int J Cancer 1994;57:313– 7.
122
A. Kubı´k et al. / Lung Cancer 31 (2001) 111–122
[35] Kabat GC, Wynder EL. Body mass index and lung cancer risk. Am J Epidemiol 1992;135:769–74. [36] Knekt P, Heliovaara M, Rissanen A, et al. Leanness and lung cancer risk. Int J Cancer 1991;49:208–13. [37] Swanson CA, Brown CC, Sinha R, Kulldorff M, Brownson RC, Alavanja MCR. Dietary fats and lung cancer risk among women: the Missouri Women’s Health Study (United States). Cancer Causes Control 1997;8:883–93.
[38] Drinkard CR, Sellers TA, Potter JD, et al. Association of body mass index and body fat distribution with risk of lung cancer in older women. Am J Epidemiol 1995;142:600– 7. [39] Rylander R, Axelsson G, Andersson L, Liljequist T, Bergman B. Lung cancer, smoking and diet among Swedish men. Lung Cancer 1996;14(Suppl 1):S75– 83.
.
A case-control study of lung cancer among Czech women Antonı´n Kubı´k a,*, Petr Zatloukal a, Peter Boyle b, Chris Robertson b, Sara Gandini b, Ladislav Toma´sˇek c, Nigel Gray b, Libor Havel a a
Department of Pneumology and Thoracic Surgery, Charles Uni6ersity, 3rd Faculty of Medicine, Uni6ersity Hospital Na Bulo6ce, and Postgraduate Medical Institute, Budı´no6a 2, 18081 Prague, Czech Republic b Di6ision of Epidemiology and Biostatistics, European Institute of Oncology, Via Ripamonti 435, 20141 Milan, Italy c National Radiation Protection Institute, S& roba´ro6a 48, 10042 Prague, Czech Republic Received 16 March 2000; received in revised form 29 March 2000; accepted 30 March 2000
Abstract Few data are available to explain the continuing increase in lung cancer mortality among Czech women. The study was designed to examine the role of active smoking and other known or suspected factors. Data collected by personal interviews during the 15 months of a hospital-based case control study are reported. A total of 140 microscopically confirmed cases and 280 frequency-matched controls were analysed using multiple logistic regression. Cigarette smoking was the most important factor associated with excess risk for lung cancer among women. Significantly increased risk was found both among current smokers (OR = 11.20, 95% CI 5.9– 21.2), and ex-smokers (OR = 10.02, 95% CI 5.5–18.4). Positive dose-response gradients (PB0.001) were observed between lung cancer risk and the daily number of cigarettes, duration of smoking, number of pack-years, inhaling, and grade of nicotine dependence assessed by the Fagerstro¨m test (Heatherton TF, Kozlowski LT, Frecker RC, Fagerstro¨m KO. Br J Addict 1991;86:1119– 1470; Pomerleau OF. In: Bolliger CT, Fagerstro¨m KO, editors. The Tobacco Epidemic. Basle: Karger, 1997: 122–131). Exposure to environmental smoke was associated with elevated lung cancer risk (OR =3.58, for lifetime non-smokers exposed both in childhood and in adult age). Physical exercise and body mass index were inversely associated with lung cancer risk. For the category of physical exercise of more than 5 h per week, the odds ratio was 0.38, compared to subjects admitting no physical exercise. For body mass index, the odds ratio for the highest (compared to the lowest) quartile was 0.50. Chronic cough and phlegm (at least 3 months per year) were associated with excess risk (OR =6.07) only if their duration was less than 2 years before diagnosis of lung cancer, and, therefore, they were suspected of being more likely early symptoms of preclinical lung cancer rather than its cause. Our results support the statement that cigarette smoking is by far the most important cause of the on-going epidemic of lung cancer among Czech women, and are consistent with the concept of a balance between risk and protective factors whose eventual maintenance or alteration determine the development of disease (as suggested by Rylander R, Axelsson G, Andersson L, Liljequist T, Bergman B. Lung Cancer 1996;14(Suppl 1): S75– S83). Concerted control of smoking appears to be an urgent priority in lung cancer prevention among women, including specific approaches targeted on the female population. © 2001 Elsevier Science Ireland Ltd. All rights reserved.
* Corresponding author. Tel.: + 420-2-66082593; fax: + 420-2-6880511. E-mail address: [email protected] (A. Kubı´k). 0169-5002/01/$ - see front matter © 2001 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 1 6 9 - 5 0 0 2 ( 0 0 ) 0 0 1 7 8 - 1
112
A. Kubı´k et al. / Lung Cancer 31 (2001) 111–122
Keywords: Lung cancer; Risk factor; Smoking; Physical activity; Women; Epidemiology
1. Introduction
2.1. Study population and data collection
In the female population of many developed countries increases in lung cancer mortality have been observed [1–6]. While in the male population of these countries, the occurrence of lung cancer is related mainly to the past and current prevalence of active smoking, results of some studies among women suggest a potential role of other factors acting either as independent risk factors or as modifiers of the effect of smoking [7 – 10]. According to a recent estimate [9], more than 20% of this increasing disease among US women has been due to causes other than active smoking. The relative importance and contribution of each factor can vary with geographic area and socio-economic conditions. Consequently, the conclusions of some investigations, e.g. of those done among Japanese, Chinese and Hawaiian women [11 –13], may be poorly applicable for Central European countries. To obtain a better insight into the role of tobacco and some other known or suspected lifestyle and other factors, a case-control study addressing the issue of epidemiology and prevention of lung cancer among Czech women was launched in early 1998. This report presents results based on 140 cases and 280 frequencymatched controls, all interviewed during the initial 15-month period of the study. Our specific objectives were to analyse the role of active smoking in lung cancer risk of women with disease diagnosed in the late 1990s, and to obtain more knowledge about some likely cofactors, such as passive smoking, previous lung disease or cancer, physical activity, and body mass index.
Eligible cases for this hospital-based case-control study comprised all female patients aged 25– 84 years, with microscopically confirmed incident lung cancer presenting at the Department of Pneumology and Thoracic Surgery of the University Hospital Na Bulovce between April 1998 and June 1999. Eligibility was limited to citizens of the Czech Republic. Of the 161 eligible cases, 140 cases (87.0%) completed the interview. The reasons for non-participation included the patient’s inability to cooperate during the interview as a result of severe physical or mental disability (12 patients, 7.4%), or death shortly after admission (nine patients, 5.6%). We did not conduct proxy interviews because a considerable number of questions of our questionnaire (e.g. passive smoking in childhood, dietary history) could not be assessed adequately. Controls were all women, and were spouses, relatives or friends of other patients of the same hospital (mostly at departments of pneumology, thoracic surgery, and internal medicine). Of the 551 women aged 25–84 years who were contacted to participate in the study, interviews were completed for 462 subjects (83.8%). Nonresponse was due to ‘no time for interview’ (48 women, 8.7%), subject refusal to be interviewed (39 women, 7.1%), and a language barrier or mental incompetence (two persons, 0.4%). Out of the series of 462 interviewed women, randomly selected subjects were frequency matched to cases using 5-year age-strata (25 –29, 30–34,…, 80–84 years) at a 2:1 ratio. These selected controls numbered 280 (Table 2). A written consent form was signed by each participant. A detailed interviewer-administered questionnaire was completed for each case and control. The interviewers were trained extensively to standardize data collection and coding techniques and to minimize inter-interviewer variation. The majority of in-person interviews took 20–40 min to complete. The questionnaire in-
2. Materials and methods The study was approved by the Scientific Council and Ethical Committee of the University Hospital Na Bulovce, Prague, and the Advisory Committee of the Internal Grant Agency of the Czech Ministry of Health.
A. Kubı´k et al. / Lung Cancer 31 (2001) 111–122
cluded a basic structured section on demographic characteristics; place of residence; type of house, occupation and workplace; further, a complete active smoking history, including the Fagerstro¨m test of nicotine dependence [14,15]; exposure to environmental tobacco smoke (passive smoking); pre-existing lung disease; family history of cancer among first degree relatives (parents and siblings); physical activity within recent 10 years. After completion of the questionnaire, the trained interviewer took basic anthropometric measures, such as standing height and weight.
2.2. Questionnaire and definitions The tobacco section of our questionnaire included modified questions from the Tobacco Use Questionnaire recommended by the Tobacco or Health Programme of the World Health Organization [16]. Subjects were defined as (current) smokers if they smoked any tobacco product either daily or occasionally at the time of the survey. Since none of the participants used tobacco products other than cigarettes the following definitions are concerned with cigarette smoking only. A daily smoker is someone who smokes at least one cigarette a day for at least 3 months, i.e. a total of approximately 100 cigarettes and over. An occasional smoker is someone who smokes, but not every day. Ex-smokers are people who were formerly smokers but currently have not smoked for at least 6 months. Never smokers either have never smoked at all or have smoked less than 100 cigarettes in their lifetime. To assess passive smoking, the interviewers asked about exposure to environmental tobacco smoke generated by parents, husband, cohabitants, or co-workers. The participants were asked to assess the number of hours per day spent in smoky rooms (at home, at work-place, and elsewhere) in adult age, and exposure to environmental tobacco smoke during childhood (before age 16). Participants were asked whether they had ever suffered from any of several lung diseases (tuberculosis, pneumonia, chronic bronchitis), or malignant tumour. To assess the history of chronic bronchitis, asthma, and related diseases modified
113
questions from the ATS-DLD-78 (American Thoracic Society-Division of Lung Diseases-78) questionnaire [17], and the Medical Research Council Questionnaire on Respiratory Symptoms (MRC, 1976) [18] were used.
2.3. Statistical methods Descriptive statistics were first used to characterize the study population. The data were kept in an Epi-Info database [19]. In preliminary examinations, differences between groups were assessed using 2-tests for independence of categorical variables, and t-tests for continuous variables. Risk analyses were first done using unconditional logistic regression [20] which provides results in the form of both crude and adjusted odds ratios for confounders. Potential interactions of independent variables or confounders, were also evaluated with multiple logistic regressions. The logistic models were conducted in two steps: (1) using all selected variables, and (2) excluding those describing active smoking in order to assess a potential confounding role of smoking. Odds ratios in the usual sense were obtained for categorical variables and odds ratios per unit change for the so called continuous variables. Categorization of continuous variables was based mostly on quartiles, except physical exercise which was divided into three groups. The odds ratios and their 95% confidence intervals (CI) were computed. The Mantel –Haenszel trend test was used, based upon integer category scores, to assess the trend in the risk with increased exposure to smoking. Final calculations were carried out using PROC LOGISTIC in SAS [21].
3. Results Among 140 cases, adenocarcinoma (35.0%) was the most frequent cell type, followed by squamous cell cancer (24.3%), and small cell cancer (23.6%) (Table 1). In the subgroup of 24 never smokers, 15 (or 62.5%) adenocarcinomas and one (or 4.2%) bronchioalveolar cancer were diagnosed. The mean age (61 years) and age distribution were identical among the series of cases and fre-
A. Kubı´k et al. / Lung Cancer 31 (2001) 111–122
114
Table 1 Distribution of cases by cell type and smoking history: Czech Women’s Lung Cancer Study Cell type
Never smokers No.
Ex-smokers %
No.
%
Current smokers
All cases
No.
No.
%
%
Squamous cell Small cell Adenocarcinoma Bronchioloalveolar Large cell Carcinoma NOSa
4 1 15 1 1 2
16.7 4.2 62.5 4.2 4.2 8.2
16 15 21 3 4 2
26.2 24.6 34.4 4.9 6.6 3.3
14 17 13 1 7 3
25.5 30.9 23.6 1.8 12.7 5.5
34 33 49 5 12 7
24.3 23.6 35.0 3.6 8.5 5.0
Total
24
100.0
61
100.0
55
100.0
140
100.0
a
NOS, not otherwise specified.
quency matched controls (Table 2). Using crude odds ratios, risk estimates appeared elevated for rural residence, inversely associated with education, and not significantly associated with marital status. Cigarette smoking was associated directly with lung cancer risk (Table 3). After adjusting for age,
residence and education, the odds ratio was 11.20 for current smokers, and 5.00 for ex-smokers who stopped smoking 10 or more years ago. Among subjects who stopped smoking within the recent 10 years, the odds ratio was as high as 18.63. It cannot be excluded that in some of these cases the motive for stopping might have been related to
Table 2 Distribution of cases and controls by socio-demographic variables: Czech Women’s Lung Cancer Studya Variables
Cases
Controlsb
Population Mean age (S.D.) Age groups (years) 25–34 35–44 45–54 55–64 65–74 75–84 Residence Rural (5100 000) Urban (\100 000) Education Elementary Secondary (ordinary) Secondary (advanced) University Marital status Married Widowed Divorced/separated Never married
140 61.0 (10.5)
280 61.1 (10.5)
a b
Crude OR
95% CI
2 6 31 46 41 14
4 12 62 92 82 28
52 88
65 215
1.00 0.51
0.3–0.8
42 50 44 4
63 92 92 33
1.00 0.82 0.72 0.18
0.5–1.4 0.4–1.2 0.1–0.6
80 38 17 5
163 62 41 14
1.00 1.25 0.84 0.73
0.7–2.0 0.5–1.6 0.3–2.1
OR, odds ratio; CI, confidence interval; S.D., standard deviation. Selected controls frequency matched by age to cases at a 2:1 ratio.
A. Kubı´k et al. / Lung Cancer 31 (2001) 111–122
115
Table 3 Active cigarette smoking, nicotine dependence, and the risk of lung cancer: Czech Women’s Lung Cancer Study Variables Cigarette smoking (acti6e) Never smokers Ever smokers Current smokers All ex-smokers Ex-smokers Quitted]10 years ago QuittedB10 years ago Number of cigarettes/day 1–4 5–14 \14 Test for trend Duration of smoking (years) 1–10 11–20 21–30 31–40 \40 Test for trend Pack-years 1–10 11–20 21–30 \30 Test for trend Inhaling No Yes Test for trend Nicotine dependence c Never- and ex-smokers Current smokers None or very low (0–1)d Low (2–4) Medium (5) High or very high (6–8) Test for trend
Cases
Controls
Adjusted ORa
95% CIb
24 116 55 61
176 104 49 55
1.00 10.52 11.20 10.02
6.0–18.3 5.9–21.2 5.5–18.4
19 42
34 21
5.00 18.63
2.4–10.5 9.0–38.7
8 43 65
17 49 38
4.06 7.92 19.49 PB0.001
1.5–11.1 4.2–15.0 10.0–38.2
9 8 28 38 33
13 23 32 21 15
6.67 2.67 7.26 15.70 21.46 PB0.001
2.4–18.5 1.0–7.1 3.5–15.2 7.5–32.8 9.6–48.2
26 26 26 38
44 32 18 10
4.68 8.72 15.51 37.85 PB0.001
2.3–9.4 4.2–18.3 6.9–35.1 15.8–90.5
17 99
25 79
5.82 12.33 PB0.001
2.5–13.6 6.8–22.5
85
231
10 27 6 12
21 22 2 4
1.00 1.29 4.13 8.19 8.92 PB0.001
0.6–2.9 2.1–8.0 1.6–42.5 2.7–29.4
a
OR, odds ratio, adjusted for age, residence, and education. CI, confidence interval. c Fagerstro¨m test [14,15]. d Levels of nicotine dependence (the scale includes grades 0–10). b
symptoms caused by the cancerous lesion evolving in the lungs. The daily number of cigarettes, duration of smoking, pack years, and inhaling, all were significantly associated with lung cancer risk showing a dose –response relationship as demonstrated by tests for the trend of odds ratios (Table 3). We found a significant association of the risk
of lung cancer with low, medium, and high or very high nicotine dependence, assessed by the Fagerstro¨m test [14,15]. Among women who never actively smoked, the odds ratio for exposure to environmental tobacco smoke in adult age (more than 3 h daily; sum of exposures at home, at work-place, and elsewhere)
A. Kubı´k et al. / Lung Cancer 31 (2001) 111–122
116
was 1.17; for exposures in childhood (before age 16) the odds ratio was 2.02 (Table 4). For never smokers with exposures to passive smoking in both childhood and adult age the odds ratio was 3.58 (95% CI 0.6 –21.9) (not shown in Table 4). Chronic cough and phlegm (at least 3 months per year) were associated with excess risk of lung cancer only if their duration was less than 2 years before diagnosis of lung cancer (OR= 6.07, 95% CI 2.2 –16.7). It cannot be excluded that such symptoms emerging recently before the onset of lung cancer may have been early symptoms of the developing lung cancer rather than risk factors. In contrast, for chronic cough and phlegm observed 2 years or longer before the onset of lung cancer (probably, symptoms of chronic bronchitis) the odds ratio was 0.84 (Table 5). The odds ratio for shortness of breath was OR =1.48 (Table 5). As expected, among subjects with chronic cough and phlegm of less than 2 years duration (hypothesized to be early symptoms of the developing lung cancer) the odds ratio for shortness of breath was 26.74 (95% CI 4.6 – 154.9), as compared to OR=2.50 (95% CI 0.5 – 13.4) in the subgroup of persons with chronic cough and phlegm observed 2 years or longer before the onset of lung cancer. The odds ratios for shortness of breath among persons with chronic cough without chronic phlegm was 2.55
(95% CI 0.5 –12.8), and for the subgroup of subjects admitting no chronic cough and no chronic phlegm OR= 0.88 (95% CI 0.4 –1.8) (not shown in Table 5). The adjusted odds ratio for attacks of shortness of breath with wheezing in the chest (probably, mostly due to asthma) was 0.90 (95% CI 0.4 –2.3) (Table 5). There was a small excess risk for subjects with a history of pneumonia (OR= 1.45), tuberculosis (OR= 1.29), or a personal history of previous cancer (OR= 1.77). No association with lung cancer risk was found for the family history of lung cancer among first degree relatives (parents and siblings, OR= 0.81), and for the family history of other cancers among first degree relatives (parents and siblings, OR= 1.08). Data on the risk of lung cancer associated with physical activity and body mass index are shown in Table 6. Physical exercise (or sport, walking, within the recent 10 years) was inversely associated with lung cancer risk. For the category of physical exercise (or sport, walking) more than 5 h per week (compared to subjects admitting no physical exercise) the odds ratio was 0.38. For body mass index, the odds ratio for the highest quartile (compared to the lowest quartile) was 0.50. A significant trend was found with decreasing risk at higher levels of body mass index (Table 6).
Table 4 Exposure to environmental tobacco smoke (passive smoking), active cigarette smoking, and the risk of lung cancer: Czech Women’s Lung Cancer Study Variable: cigarette smoking
Cases
Controls
Adjusted ORa
95% CIb
Active ever
Passive
No No Yes Yes
Noc Yesc Noc Yesc
22 2 87 29
160 16 83 21
1.00 1.17 10.03 14.00
0.2–5.6 5.6–18.1 6.3–31.3
No No Yes Yes
Nod Yesd Nod Yesd
10 14 60 56
105 71 47 57
1.00 2.02 17.39 12.99
0.8–4.9 7.8–38.8 5.9–28.8
a
OR, odds ratio, adjusted for age, residence, and education. CI, confidence interval. c Passive in adulthood: exposure to environmental tobacco smoke \3 h/day, in adult age. d Passive in childhood: exposure to environmental tobacco smoke in childhood (before age 16). b
A. Kubı´k et al. / Lung Cancer 31 (2001) 111–122
117
Table 5 Personal history of previous lung disease or cancer, family history of cancer, and the risk of lung cancer: Czech Women’s Lung Cancer Study Variables
Cases
Controls
Adjusted ORa
95% CIb
No chronicc cough, no chronicc phlegm Chronic cough without chronic phlegm Chronic cough and phlegm,B2 years Chronic cough and phlegm, ]2 years
93 17 18 12
231 22 9 18
1.00 2.19 6.07 0.84
0.9–5.3 2.2–16.7 0.3–2.1
Shortness of breathd Attacks of shortness of breathe Tuberculosis Pneumonia Personal history of previous cancer
43 8 7 49 10
52 26 12 84 12
1.48 0.90 1.29 1.45 1.77
0.8–2.6 0.4–2.3 0.4–4.0 0.9–2.5 0.6–4.9
Family history of lung cancerf Family history of other cancersg
19 48
26 94
0.81 1.08
0.4–1.7 0.6–1.8
a
OR, odds ratio, adjusted for age, residence, education, and pack-years of smoking. CI, confidence interval. c 3 months/year. d Getting short of breath walking with other people at ordinary pace on the level. e Attacks of shortness of breath with wheezing in the chest. f Cancer among first degree relatives (parents and siblings). g Except lung cancer. b
Table 6 Physical activitiesa,b, body mass index, and the risk of lung cancer: Czech Women’s Lung Cancer Study Variables
Cases
Controls
Adjusted ORc
95% CId
Physical exercise a (hours/week) 0 1–5 \5 Test for trend
63 39 38
69 78 133
1.00 0.58 0.38 PB0.001
0.3–1.1 0.2–0.7
Other physical acti6ities b (hours/week) 0 1–5 \5 Test for trend
69 20 51
107 45 128
1.00 0.59 0.65 P=0.61
0.3–1.3 0.4–1.2
Body mass index (BMI, kg/m 2) 13.7–22.9 23.0–25.9 26.0–28.9 29.0–44.9 Test for trend
42 35 35 28
64 62 76 78
1.00 1.04 0.69 0.50 P =0.031
0.5–2.1 0.4–1.4 0.2–1.0
a
Physical exercise (or sport, walking, within recent 10 years). Other non-professional physical activities (e.g. in the garden, house, within recent 10 years). c OR, odds ratio, adjusted for age, residence, education, and pack-years of smoking. d CI, confidence interval. b
118
A. Kubı´k et al. / Lung Cancer 31 (2001) 111–122
Table 7 shows the results of multiple logistic analyses conducted in order to assess potential interactions or confounding among active and passive smoking, chronic cough and phlegm, shortness of breath, physical exercise and body mass index, analysed in the paper. The variables were modelled either as continuous variables, such as pack-years of smoking, physical exercise (hours/week), body mass index (kg/m2); or categorical variables, such as the (active) smoking status, passive smoking, chronic cough and phlegm (at least 3 months/year), and shortness of breath. The comparison of the results of the analyses pointed out the role of active cigarette smoking as the dominant risk factor of lung cancer among Czech women, and suggested a minor role of other factors included in the analyses. Table 7 Multiple logistic analysisa evaluating simultaneously seven selected risk factors: Czech Women’s Lung Cancer Study Variables
ORb
95% CIc
Acti6e smoking Never smokers Ex-smokers ]10 years Ex-smokers B10 years Current smokers Pack-years of smoking (per 1 pack-year)
1.00 4.01 7.39 4.07 1.05
1.8–9.1 2.8–19.9 1.6–10.1 1.02–1.08
1.00 0.85
0.5–2.1 0.5–1.5
Passi6e smoking In adult age (\3 h/day) In childhood (before age 16) No chronicd cough, no chronicd phlegm Chronic cough without chronic phlegm Chronic cough and phlegm,B2 years Chronic cough and phlegm, ]2 years
1.00
Shortness of breathe Physical exercise, or sport, walking (per 1 h/week) BMI, body mass index (per 1 kg/m2)
1.99
0.8–4.9
5.66 0.70
2.0–16.3 0.3–1.8
1.48 0.95
0.8–2.7 0.91–0.99
0.92
0.87–0.98
a All models include adjustment for age, residence, and education. b OR, odds ratio, or odds ratio per unit change in the full model. c CI, confidence interval. d 3 months/year. e Getting short of breath walking with other people at ordinary pace on the level.
4. Discussion Our report presenting results of a hospitalbased case-control study of lung cancer among Czech women has certain potential limitations which should be considered before conclusions are drawn. The exposures of interest were based on self report, therefore, some recall bias is of concern. Small numbers of cases in subgroups of some items (e.g. exposure to environmental tobacco smoke, personal history of cancer) limited the power of analyses and precluded further testing. As expected, adenocarcinoma was the most frequent cell type (35%) found among the 140 cases in the study. Among 24 never smokers, almost two-thirds (62.5%) were adenocarcinomas. Increasing trends of adenocarcinomas have been observed in many developed countries, the ratio between squamous-cell carcinomas and adenocarcinomas was everywhere greater in men than in women [9,22]. Adenocarcinoma has always represented the majority of lung cancers among nonsmokers of both genders, and increased, as a proportion, with increasing duration of smoking cessation [22]. Some authors explained the increases in the proportion of adenocarcinomas among smokers by changes in the design of cigarettes (the use of filter tips, decreasing yields of nicotine and tar) [22]. Results of our case-control study support the claim that cigarette smoking is the most strongly active risk factor for lung cancer among Czech women. The odds ratio for risk associated with lung cancer for ever-smokers compared to neversmokers was 10.52, comparable to observations in western countries (frequently \ 10) while, e.g. in China lower odds ratios (B 5) were reported [23]. Two recently published comprehensive metaanalyses on exposure to environmental tobacco smoke and the risk of lung cancer [24,25] have concluded with the statement that passive smoking is a cause of lung cancer. For a total of 36 case-control studies, Zhong et al. [25] calculated the pooled relative risk 1.19 (95% CI 1.10 –1.29), for five cohort studies 1.29 (95% CI 1.04 –1.62). In our study the adjusted odds ratio was 1.17 for exposure to passive smoking in adult age (more
A. Kubı´k et al. / Lung Cancer 31 (2001) 111–122
than 3 h daily; Table 4), 2.02 for exposure in childhood (before age 16), and 3.58 (95% CI 0.58 –21.91) for exposure both in childhood and in adult age. Many of the controls in our study were relatives, spouses or friends of patients hospitalized at the department of pneumology with diagnosis of smoking related lung disease, therefore, many of the controls were likely to have been largely exposed to passive smoking. This could lead to underestimation of effects of passive smoking in the present analysis. Consequently, any observed adjusted odds ratios greater than 1 in our study should be considered to be under suspicion of signalling a detrimental impact of passive smoking. Trying to control for this confounder we identified a subgroup of controls who were relatives, spouses or friends of patients with a diagnosis of other than smoking-related diseases. For this subgroup, the odds ratio was 3.28 (95% CI 0.20 –53.46), as compared to OR= 0.85 (95% CI 0.16 –4.46) for the subgroup of controls who were relatives, spouses, or friends of patients with smoking-related diseases, and, therefore, were likely to have been much more exposed to environmental tobacco smoke. Hence it follows for investigations into the role of passive smoking in lung carcinogenesis that including spouses, relatives and friends of patients with smoking-related lung diseases is a problematic choice and cannot be recommended, since they are as likely as the cases to have been exposed to environmental tobacco smoke. Previous studies on the role of personal history of chronic respiratory illnesses among non-smokers have been recently reviewed by Brownson et al. [7]. A multicenter population-based case-control study among lifetime non-smoking women in the United States provided evidence of association of asthma, chronic bronchitis, emphysema, and for age groups younger than 55 years tuberculosis and pneumonia with lung cancer risk [26]. In another large case-control study among nonsmoking women [27], each type of lung disease, except chronic bronchitis, showed some elevation in risk of adenocarcinoma, although effect estimates were not always statistically significant. Associations of lung cancer risk with emphysema, chronic bronchitis, and asthma were found among
119
female never and former smokers in a populationbased case-control study in New York State [28]. Some studies have reported an inverse association between asthma or hay fever and lung cancer in women [29]. In our study, respiratory symptoms, such as chronic cough and phlegm (at least 3 months per year) that were observed less than 2 years prior to diagnosis of lung cancer (and showed association with lung cancer risk, OR= 6.07), were more likely to be early symptoms of preclinical lung cancer rather than its cause. Chronic cough and phlegm observed 2 years or longer (probably, symptoms of chronic bronchitis) were not associated with risk of lung cancer (OR = 0.84) (Table 5). Few data are available regarding the association of physical activity with lung cancer risk. A prospective study of 81 516 Norwegian subjects observed that physical activity was inversely related to lung cancer risk in men, but not in women [30]. Data from a cohort study of 13 905 male Harvard University alumni indicated that physical activity was associated with lower risk of lung cancer [31]. The Norwegian investigators [30] hypothesized that the increased pulmonary ventilation and perfusion that occurs with physical activity may lead to reduced interaction time and concentration of carcinogenic agents in the airways. Physical activity, along with reproductive variables and food availability, were recently mentioned by Willett [32] as examples of factors associated with affluence (national wealth), and also related to breast cancer occurrence. Recent evidence from many studies that higher levels of physical activity reduce risk of colon cancer implies that at least part of the high colon cancer rates in affluent countries previously attributed to fat intake are probably due to a sedentary lifestyle [32]. Further studies are needed to understand the essence of the association between the latter variables and lung cancer risk. In our study, an inverse association was found between lung cancer risk and time (hours/week) devoted to physical exercise, within recent 10 years (Table 6). During the ongoing investigation we have realized that physical activity may be subject to changes in time, and lack of physical activity shortly before diagnosis of lung cancer
120
A. Kubı´k et al. / Lung Cancer 31 (2001) 111–122
may more likely be the result, rather than the cause, of the early stage, preclinical disease. Therefore, two additional questions were affixed to the questionnaire, related to physical exercise (sport, walking) 1 year prior to diagnosis, and 20 years prior to diagnosis. The original question on physical exercise (within recent 10 years) was kept in the amended questionnaire. Additional analysis could be done only in a subgroup of 30 cases. Among these cases, the mean time devoted to physical exercise 20 years prior to diagnosis was significantly longer (+1.4 h/week) than 1 year prior to diagnosis. It should be noted that even the time elicited by the original question was significantly longer (+1.0 h/week) than 1 year prior to diagnosis. The difference between the mean time devoted to physical exercise 20 years prior to diagnosis and the mean time elicited by the original question was 0.4 h/week, and was not statistically significant (P= 0.465). Although some part of these differences might be related to ageing, it can be presumed that the information elicited by the original question probably reflected in most cases the time devoted to physical exercise several years prior to diagnosis. Lack of exercise at that time may be suspected of more likely being a risk factor, rather than the result, of early stage lung cancer, however, the opposite cannot be excluded. Further studies are needed to throw more light on the association between physical activity and lung cancer risk. While body mass index is known to be associated with overall mortality [33] and some cancers, e.g. prostate or colon cancer [34], an inverse association with lung cancer was observed in some studies [8,35 –37], including our investigation. A significant inverse gradient between body mass index and the incidence of lung cancer was found in a prospective study of 25 994 men aged 20 – 75 years in Finland [36]. Similar observations were done in case-control studies in the USA: the American Health Foundation (AHF) hospital-based study of tobacco-related cancers [8,35], and the Missouri Women’s Health Study [37]. In the AHF study, the strongest association of leanness was observed in women who never smoked. In a prospective cohort study of 41 837 Iowa women aged 55–69 years, the results of multivariate analy-
ses suggested that the inverse association of body mass index with lung cancer could be explained by smoking status [38]. Even in our study, body mass index was inversely associated with lung cancer risk; body height and weight, used for calculations of body mass index, were taken at the time of interview, and information on previous anthropometric measures was not obtained. Therefore, it should be considered that low body mass index may more likely be the result, rather than the cause, of early stage lung cancer. In conclusion, our results support the statements that cigarette smoking is by far the most important cause of the on-going epidemic of lung cancer among Czech women, and that exposures to environmental tobacco smoke may increase the risk of lung cancer. The dominant role of active smoking among lung cancer factors has generally been recognized [2]. More data are needed to elucidate the role of some cofactors, e.g. some components of diet, tea, coffee and alcohol consumption, events in the personal gynecologic history, the family history of cancer, and others. Our findings support the ‘concept of a balance between risk factors for a disease and protective factors’ [39]. The eventual maintenance or alteration of the balance between these factors have been suggested to determine whether the disease would develop or not. Concerted control of smoking appears to be an urgent priority in lung cancer prevention among women, including specific approaches targeted on the female population, such as efforts to prevent adolescent girls from starting to smoke, and encouraging cessation among established smokers. The issue of preventive recommendations concerning some cofactors continues to be a matter of debate.
Acknowledgements We wish to thank Professor L. Petruzelka, MD, PhD (Department of Oncology, Charles University, First Faculty of Medicine, General Faculty Hospital, Prague), and the reviewers for valuable comments. The financial support by grant No. 4970-3 of the Internal Grant Agency (IGA) of the Czech Ministry of Health is gratefully acknowledged.
A. Kubı´k et al. / Lung Cancer 31 (2001) 111–122
References [1] Boyle P. Cancer, cigarette smoking and premature death in Europe: a review including the Recommendations of European Cancer Experts Consensus Meeting, Helsinki, October 1996. Lung Cancer 1997;17:1–60. [2] Boyle P, Gandini S, Gray NJ. Epidemiology of lung cancer: A century of great success and ignominious failure. In: Hansen HH, editor. Textbook of Lung Cancer. London: Martin Dunitz, 2000:13–25. [3] Kubı´k AK, Parkin DM, Plesˇko I, et al. Patterns of cigarette sales and lung cancer mortality in some Central and Eastern European countries, 1960–1989. Cancer 1995;75:2452– 60. [4] LaVecchia C, Boyle P. Trends in the tobacco-related cancer epidemic in Europe. Cancer Detect Prev 1993;17:495– 506. [5] Parkin DM. The global burden of cancer. Semin Cancer Biol 1998;8:219– 35. [6] Zatloukal P, Kubı´k A, Petruzelka L. Current and future trends in lung cancer mortality in the Czech Republic. Lung Cancer 1998;21(Suppl 1):S2–3. [7] Brownson RC, Alavanja MCR, Caporaso N, Simoes EJ, Chang JC. Epidemiology and prevention of lung cancer in nonsmokers. Epidemiol Rev 1998;20:218–36. [8] Kabat GC. Aspects of epidemiology of lung cancer in smokers and nonsmokers in the United States. Lung Cancer 1996;15:1– 20. [9] Mason TJ. The descriptive epidemiology of lung cancer. In: Samet JM, editor. Epidemiology of Lung Cancer. New York: Marcel Dekker, 1994:51–69. [10] Saracci R, Boffetta P. Interactions of tobacco smoking with other causes of lung cancer. In: Samet JM, editor. Epidemiology of Lung Cancer. New York: Marcel Dekker, 1994:465– 93. [11] Hinds HW, Stemmerman GN, Yang HY, Kolonel LN, Lee J, Wegner F. Differences in lung cancer risk from smoking among Japanese, Chinese and Hawaiian women in Hawaii. Int J Cancer 1981;27:297–302. [12] Ko YC, Lee CH, Chen MJ, et al. Risk factors for primary lung cancer among non-smoking women in Taiwan. Int J Epidemiol 1997;26:24–31. [13] Seow A, Duffy SW, Ng TP, McGee MA, Lee HP. Lung cancer among Chinese females in Singapore 1968–1992: time trends, dialect group differences and implications for aetiology. Int J Epidemiol 1998;27:167–72. [14] Heatherton TF, Kozlowski LT, Frecker RC, Fagerstro¨m KO. The Fagerstro¨m test of nicotine dependence: a revision of the Fagerstro¨m Tolerance Questionnaire. Br J Addict 1991;86:1119–470. [15] Pomerleau OF. Nicotine dependence. In: Bolliger CT, Fagerstro¨m KO, editors. The Tobacco Epidemic. Basle: Karger, 1997:122– 31. [16] World Health Organization, Tobacco or Health Programme. Guidelines for Controlling and Monitoring the Tobacco Epidemic. Geneva: World Health Organization, 1996.
121
[17] Ferris BG, editor. Epidemiology standardization project. Am Rev Respir Dis 1978; 118(6), part 2 of 2 parts: 1 – 53. [18] Medical Research Council. Questionnaire on Respiratory Symptoms. Approved by the MRC’s Committee on Research into Chronic Bronchitis. London: Medical Research Council, 1976. [19] Dean AG, Dean JA, Coulombier D, et al. Epi Info, Version 6: A Word-Processing, Database, and Statistics Program for Public Health on IBM-compatible Microcomputers. Atlanta, GA: Centers for Disease Control and Prevention, 1995. [20] Preston DL, Lubin JH, Pierce DA, McConney ME. Epicure, Version 1.8w: A Package of 5 Programs for Generalized Risk Modelling and Person-Year Computation. Seattle, WA: HiroSoft International, 1993. [21] SAS Institute, Inc. SAS for Windows version 6.11. Cary, NC: SAS Institute, 1995. [22] Franceschi S, Bidoli E. The epidemiology of lung cancer. Ann Oncol 1999;10(Suppl 5):S3– 6. [23] Wu JM, Du Y-X. Summary of papers and research recommendations presented at the International Symposium on Lifestyle Factors and Human Lung Cancer, Guangzhou, China. Lung Cancer 1996;14(Suppl 1):S223– 34. [24] Hackshaw AK, Law MR, Wald NJ. The accumulated evidence on lung cancer and environmental tobacco smoke. BMJ 1997;315:980– 8. [25] Zhong L, Goldberg MS, Parent M-E, Hanley JA. Exposure to environmental tobacco smoke and the risk of lung cancer: a meta-analysis. Lung Cancer 2000;27:3– 18. [26] Wu AH, Fontham ETH, Reynolds P, et al. Previous lung disease and risk of lung cancer among lifetime nonsmoking women in the United States. Am J Epidemiol 1995;141:1023– 32. [27] Alavanja MCR, Brownson RC, Boice JD, Jr, et al. Preexisting lung disease and lung cancer among nonsmoking women. Am J Epidemiol 1992;136:623– 32. [28] Mayne ST, Buenconsejo J, Janerich DT. Previous lung disease and risk of lung cancer among men and women nonsmokers. Am J Epidemiol 1999;149:13– 20. [29] Osann KE. Lung cancer in women: the importance of smoking, family history of cancer, and medical history of respiratory disease. Cancer Res 1991;51:4893– 7. [30] Thune I, Lund E. The influence of physical activity on lung cancer risk: a prospective study of 81,516 men and women. Int J Cancer 1997;70:57– 62. [31] Lee I-M, Sesso HD, Paffenbarger RS, Jr. Physical activity and risk of lung cancer. Int J Epidemiol 1999;28:620– 5. [32] Willett WC. Dietary fat intake and cancer risk: a controversial and instructive history. Semin Cancer Biol 1998;8:245– 53. [33] Calle EE, Thun MJ, Petrelli JM, Rodriguez C, Heath CW, Jr. Body-mass index and mortality in a prospective cohort of US adults. N Engl J Med 1999;341:1097– 105. [34] Chyou PH, Nomura AM, Stemmermann GN. A prospective study of weight, body mass index and other anthropometric measurements in relation to site-specific cancers. Int J Cancer 1994;57:313– 7.
122
A. Kubı´k et al. / Lung Cancer 31 (2001) 111–122
[35] Kabat GC, Wynder EL. Body mass index and lung cancer risk. Am J Epidemiol 1992;135:769–74. [36] Knekt P, Heliovaara M, Rissanen A, et al. Leanness and lung cancer risk. Int J Cancer 1991;49:208–13. [37] Swanson CA, Brown CC, Sinha R, Kulldorff M, Brownson RC, Alavanja MCR. Dietary fats and lung cancer risk among women: the Missouri Women’s Health Study (United States). Cancer Causes Control 1997;8:883–93.
[38] Drinkard CR, Sellers TA, Potter JD, et al. Association of body mass index and body fat distribution with risk of lung cancer in older women. Am J Epidemiol 1995;142:600– 7. [39] Rylander R, Axelsson G, Andersson L, Liljequist T, Bergman B. Lung cancer, smoking and diet among Swedish men. Lung Cancer 1996;14(Suppl 1):S75– 83.
.