Passive smoking and lung cancer in Japanese non-smoking women: A prospective study
Although smoking is a major cause of lung cancer, the proportion of lung cancer cases among Japanese women who never smoked is high. As the prevalence of smoking in Japan is relatively high in men but low in women, the development of lung cancer in non-smoking Japanese women may be significantly impacted by passive smoking. We conducted a population-based prospective study established in 1990 for Cohort I and in 1993 for Cohort II. The study population was defined as all residents aged 40–69 years at the baseline survey. 28,414 lifelong non-smoking women provided baseline information on exposure to tobacco smoke from their husband, at the workplace and during childhood. Over 13 years of follow-up, 109 women were newly diagnosed with lung cancer, of whom 82 developed adenocarcinoma. Compared with women married to never smokers, hazard ratio (HR) [95% confidence interval (CI)] for all lung cancer incidence in women who lived with a smoking husband was 1.34 (95% CI 0.81–2.21). An association was clearly identified for adenocarcinoma (HR 2.03, 95% CI 1.07–3.86), for which dose-response relationships were seen for both the intensity (p for trend = 0.02) and amount (p for trend = 0.03) of the husband's smoking. Passive smoking at the workplace also increased the risk of lung cancer (HR 1.32, 95% CI 0.85–2.04). Moreover, a higher risk of adenocarcinoma was seen for combined husband and workplace exposure (HR 1.93, 95% CI 0.88–4.23). These findings confirm that passive smoking is a risk factor for lung cancer, especially for adenocarcinoma among Japanese women. © 2007 Wiley-Liss, Inc.
In Japan, lung cancer has been the second leading cause of cancer death in women since the 1980s.1 Although the majority of lung cancers can be attributed to cigarette smoking, 53% of all women with lung cancer world-wide are never smokers.2, 3 The proportion of Japanese female lung cancer patients who have never smoked is as high as 70%,4 whereas the proportion of Japanese women aged 20 or more who smoke is only around 10%.5 The major risk for lung cancer in Japanese women cannot therefore be attributed to smoking. Given that the urine of non-smokers exposed to passive smoking contains concentrations of carcinogenic N-nitroso compounds, which are specific to tobacco6 and the smoking rate in Japanese men is high, at around 50%,5 passive smoking might be an important risk factor for lung cancer in non-smoking Japanese women.
Since the publication of the first positive findings by Hirayama in Japan, many studies have investigated the relation between passive smoking and lung cancer in non-smoking women.7 Recently, the International Agency for Research on Cancer (IARC) concluded that findings on the risk of lung cancer associated with environmental tobacco smoke were consistent.8 Moreover, a meta-analysis of published studies estimated that the excess risk of lung cancer in non-smokers who lived with a smoker compared to those who lived with a non-smoker was 24%.9 However, relatively few prospective studies have appeared,7, 10, 11, 12, 13, 14, 15, 16 and almost all of previous studies have been case–control studies, for which the limitation of recall bias is controversial.17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37 Additionally, information on spousal smoking status in most of the prospective studies has been obtained from wives. The accuracy of this exposure evaluation is questionable, however, because while high concordance has been shown between information on spousal ever smoking from wives and data from husbands themselves, agreement on the duration and intensity of smoking is lower.38, 39 Further, few prospective studies have considered the importance of multiple sources of exposure to passive smoking.15, 16
Here, we identified married couples among subjects of a large prospective study in Japan, and examined the association between passive smoking from the husband and the risk of lung cancer in the non-smoking wife using smoking status information obtained from the husband himself. Further, we also analyzed the association between passive smoking from other sources (at the workplace or during childhood) in lifelong non-smoking women with lung cancer.
Material and methods
The Japan Public Health Center-based Prospective study (JPHC Study) was launched in 1990 for the first population-based cohort (Cohort I) and in 1993 for the second (Cohort II). Cohort I covered 5 public health center (PHC) areas (Iwate, Akita, Nagano, Okinawa and Tokyo) and Cohort II covered 6 (Ibaraki, Niigata, Kochi, Nagasaki, Okinawa and Osaka). All study subjects were residents of Japanese nationality who lived in the study areas at the start of follow-up, and who were aged 40–59 in Cohort I and 40–69 in Cohort II. In the present analysis, we excluded all subjects from the Tokyo and Osaka areas because incidence data in Tokyo were not available and the study population in Osaka included health checkup examinees, and was thus not fully population-based. A population-based cohort of 57,591 men (Cohort I: 26,998; Cohort II: 30,593) and 59,103 women (Cohort I: 27,397; Cohort II: 31,706) was identified using population registries, which were maintained by the respective local governments. Details of the study cohorts have been described elsewhere.40
A self-administered questionnaire, which included smoking history, previous disease history, and other lifestyle factors, was distributed to all eligible registered residents in 1990 for Cohort I and in 1993–1994 for Cohort II. Completed questionnaires were collected from 45,452 men and 49,924 women, giving response rates of 79 and 84%, respectively. To assess the carcinogenic effect of passive smoking exposure from the spouse, we further identified 31,261 pairs as married couples by surname, address, sex and an age difference of less than 16 years. To clarify the effects of passive smoking, we restricted analysis to lifelong non-smoking women only. A further 711 women with a history of cancer at any site were excluded. Thus, 28,414 women were left for analysis. The accuracy of identification was tested in 644 pairs using residence registries; results showed 604 pairs (93.8%) were married couples and 6 (0.9%) were relatives other than a spouse, while the relationship of 34 pairs (5.3%) could not be established.
The questions on smoking habit consisted of current and former smoking status, age at the initiation of smoking, average number of cigarettes smoked per day, and age at the cessation of smoking for former smokers. We determined that a woman who had a husband with a history of smoking had been exposed to passive smoking from her husband. We classified the passive smoking status of a woman by smoking status information from her husband himself (never, former, current), the number of cigarettes smoked per day (<20, ≥20), and the amount of smoking in current smokers (pack years: <30, ≥30). Information on passive smoking at the workplace (or public facilities) was also collected. We considered women who inhaled other people's smoke for more than 1 hr per day and at least 1 day per week as exposed to passive smoking at the workplace. We further asked a subgroup of subjects about passive smoking during childhood, namely whether there had been family members at home with a smoking habit when they were in elementary or junior high school (6–15 years old).
We followed the subjects from the baseline survey until December 31, 2004. Migration and survival status were obtained annually from the residential registry. Among subjects, 1,507 persons (5.3%) moved out of a study area and 74 (0.3%) were lost to follow-up during the study period. The occurrence of cancer was identified by continuous surveillance of hospital records and population-based cancer registries, with permission. Death certificate information was used as a supplementary information source. Cases were coded using the International Classification of Diseases for Oncology, Third Edition (ICD-O-3).41 In our cancer registry system, the proportion of cases for which information was available from death certificates only during the study period was 3.6%. This level of information quality was considered satisfactory for the present study. The earliest date of diagnosis was used in cases with multiple primary cancers at different times. A total of 109 newly diagnosed lung cancer cases were identified among 28,414 lifelong non-smoking women.
Person-years of follow-up were calculated for each subject from the date of questionnaire completion to the date of lung cancer diagnosis, the date of emigration from the study area, or the date of death, whichever came first. If none of these occurred, follow-up was through to the end of the study period (December 31, 2004). Persons who were lost to follow-up were censored at the last confirmed date of presence in the study area. Hazard ratios (HRs) and 95% confidence intervals (CIs) for passive smoking were estimated by the Cox proportional hazards model, according to the SAS procedure (SAS Institute, Cary, NC). Possible confounding factors included as covariates in the model were age (5-year groups), study area (9 PHC areas), menopause (pre/post), alcohol consumption (non-, ex- or occasional drinkers, 1–149 g/week, ≥150 g/week) and family history of lung cancer (yes/no).
p-Values for trends were assessed by assigning ordinal values for categorical variables. All p-values are 2-sided, and statistical significance was determined at the p < 0.05 level.
During the 377,813 person-years of follow-up (average 13.3 years) for 28,414 lifelong non-smoking women, a total of 109 cases of lung cancer were newly diagnosed and included in the analyses. Among all lung cancer cases in this analysis (n = 109), 98 (90%) were histologically confirmed as follows: adenocarcinoma (83.7%), large cell carcinoma (7.2%), squamous cell carcinoma (5.0%), small cell carcinoma (2.1%) and other histological types (1.0%).
Table I shows subject characteristics according to passive smoking from the husband. About half of the women (49.1%) were exposed to passive smoking from husbands who were current smokers. Women with currently smoking husbands were slightly younger and more likely to be regular drinkers than those with husbands who were never or former smokers. The proportion of women who were in menopause was low in women exposed to passive smoking from their husbands. A higher proportion of women exposed to passive smoking from husbands had a family history of lung cancer.
Table I. Subject Characteristics According to Passive Smoking Status from the Husband in 28,414 Lifelong Non-Smoking Women
|Age, years ± SD||50.5 ± 6.8||52.1 ± 7.4||50.2 ± 7.0||50.8 ± 7.1|
|Regular drinker, yes (%)||7.5||8.7||8.9||8.6|
|Menopause, yes (%)||52.0||59.4||50.5||54.7|
|Family history of lung cancer, yes (%)||19.1||21.2||21.1||20.6|
Table II shows the HRs and 95% CIs for lung cancer in non-smoking women according to passive smoking from the husband. Passive smoking from a husband who was a current smoker slightly increased the risk of all lung cancer (HR 1.34, 95% CI 0.81–2.21). Compared with women without passive smoking, multivariate HRs of all lung cancer for passive smoking from a husband who smoked 20 or more cigarettes per day or had 30 or more pack-years of smoking were 1.47 (95% CI 0.87–2.49) and 1.46 (95% CI 0.85–2.50), respectively. Tests for linear trends were not statistically significant. However, a positive association with passive smoking and lung cancer became clear when our analysis was restricted to adenocarcinoma. Passive smoking from husbands who were current smokers significantly increased the risk of lung adenocarcinoma (never versus current: 2.03, 95% CI 1.07–3.86). The husband's number of cigarettes per day and number of pack-years were both significantly associated with lung adenocarcinoma risk in non-smoking wives, showing a dose-response relationship. Compared with women whose husbands never smoked, multivariable HRs were 2.20 for passive smoking from husbands who smoked 20 or more cigarettes per day (95% CI 1.13–4.28, p for trend = 0.02), and 2.06 for those at 30 or more pack-years (95% CI 1.04–4.10, p for trend = 0.03).
Table II. Association Between Lung Cancer Incidence and Passive Smoking from the Husband in Lifelong Non-Smoking Women (n = 28,414)
|From husband|| || || || || || |
| Former||28||94,427||1.12 (0.63–1.98)||21||94,358||1.50 (0.73–3.09)|
| Current||56||185,919||1.34 (0.81–2.21)||46||185,855||2.03 (1.07–3.86)|
|Number of cigarettes per day|| || || || || || |
| <20||14||52,441||1.02 (0.51–2.04)||13||52,438||1.73 (0.77–3.88)|
| ≥20||41||131,107||1.47 (0.87–2.49)||33||131,055||2.20 (1.13–4.28)|
| p for trend|| || ||0.14|| || ||0.02|
|Pack years of exposure|| || || || || || |
| <30||17||76,125||1.05 (0.55–2.02)||16||76,122||1.86 (0.86–4.01)|
| ≥30||36||104,330||1.46 (0.85–2.50)||28||104,279||2.06 (1.04–4.10)|
| p for trend|| || ||0.17|| || ||0.03|
The association between passive smoking from 2 sources (at the workplace and from the husband) and lung cancer is shown in Table III. Passive smoking at the workplace tended to increase the risk of lung cancer, with HRs for these women of 1.32 for all lung cancer and 1.16 for adenocarcinoma. Concerning passive smoking from 2 sources, no dose-dependent increase in the risk of all lung cancers was found, with HRs in non-smoking women exposed at the workplace only of 2.74 (95% CI 1.11–6.76) but 1.61 (95% CI 0.83–3.11) in those exposed to 2 sources. In contrast, a dose-response relationship was seen for exposure to 2 sources and adenocarcinoma. HR of women exposed to 2 sources was 1.93 (95% CI 0.88–4.23) in adenocarcinoma, but this was not statistically significant.
Table III. Association Between Lung Cancer Incidence and Passive Smoking at the Workplace and from Two Sources in Lifelong Non-Smoking Women (n = 28,414)
|At workplace|| || || || || || |
|≥1 times/week||30||94,652||1.32 (0.85–2.04)||20||94,568||1.16 (0.69–1.97)|
|From two sources|| || || || || || |
|Source of exposure|| || || || || || |
| Almost never1||17||80,428||1||12||80,395||1|
| Workplace only2||8||16,236||2.74 (1.11–6.76)||3||16,195||1.21 (0.26–5.55)|
| Husband only3||60||198,994||1.49 (0.84–2.62)||48||198,904||1.79 (0.90–3.55)|
| Workplace + Husband||22||78,417||1.61 (0.83–3.11)||17||78,373||1.93 (0.88–4.23)|
We also analyzed the association between self-reported passive smoking during childhood and lung cancer in non-smoking women who answered this item (n = 15,467). Passive smoking during childhood was not associated with lung cancer (HR 0.93, 95% CI 0.52–1.66) (data not shown).
In the present study, we found that passive smoking from husbands was associated with a 30% excess risk of lung cancer in non-smoking women. This result is supported by the findings of the IARC8 and many previous papers,10, 11, 13, 18, 21-23, 25, 26, 34 and accords with a summary risk of 1.24 obtained in a meta-analysis.9
Because adenocarcinoma is the predominant lung cancer type in non-smoking women,17, 42, 43, 44 and so the effects of passive smoking may be particularly relevant to it, we also carried out analyses by histological type. Results showed a clear association between adenocarcinoma and an increased risk from passive smoking from the husband. Previous papers have reported a hazard ratio of 1.0–1.6 in adenocarcinoma, ratios smaller than those for other cell types.22, 23, 26, 30-33, 36 Only Fontham et al. similarly reported that passive smoking from the husband increases the risk of adenocarcinoma.21
Nevertheless, our findings might be supported by findings on the mechanism of passage of sidestream smoke components through the nasal passages, which showed that volatile sidestream smoke constituents would be more likely to reach the peripheral portions of the lung than mainstream smoke.45 For example, 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone, a tobacco-specific lung carcinogen, and its metabolites are found in urine of nonsmokers exposed to passive smoking.6 Moreover, 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone predominantly induces adenocarcinoma in animal experiments.46, 47 Particularly in Japan, where room sizes tend to be small and living conditions congested, sidestream smoke may be directly transmitted to non-smoking women before dilution by room air.10, 45
As further support, our use of information from the husband himself likely enhanced the accuracy of exposure to passive smoking, owing to the low agreement between the husband's information on the duration or intensity of his smoking with that from the wife.38, 39 Further, our identification of a statistically significant dose-response relationship between the quantity and intensity of husbands' smoking and their wives' incidence of lung cancer suggest that our findings were not due to chance.
Passive smoking at the workplace tended to increase the risk of lung cancer in non-smoking women. Previous findings on this risk have been inconsistent, with some studies reporting positive results15, 16, 21, 23, 27, 29, 31-33, 35 while others have not.19, 24, 25, 28, 30, 34 This inconsistency may be due to a lack of accuracy assessment of exposure to passive smoking at the workplace. However, our study showed a higher risk of adenocarcinoma for combined husband and workplace exposure, suggesting a modest role of the 2 exposures combined in the incidence of adenocarcinoma.
Our study showed no association between passive smoking during childhood and lung cancer. Previous studies of this risk have also been inconsistent,15, 16, 20-25, 27-37 perhaps unsurprisingly given the difficulty of recall of exposures occurring far in the past.
Our study has several methodological strengths. First, it was a prospective design, which diminishes the probability of the recall bias inherent to case–control studies. Second, passive smoking from the husband could be evaluated with accuracy. We identified married couples and used information on smoking provided by the husbands themselves. Third, the response rate was sufficiently high (approximately 80%), and the proportion of subjects lost to follow-up was negligible (0.3%). Finally, a relatively large number of cases accrued among never-smokers.
Several limitations also warrant mention. First, information was collected only once at baseline. Second, misclassification may have occurred due to errors in identifying married couples. However, even if some men identified as husbands were in fact relatives, time spent with the husband would likely be longer than that with a relative. If present, therefore, such misclassification may have attenuated the true risk. Third, misclassification that women who had in fact smoked were included in category of non-smokers may also have occurred. However, such misclassification might not have substantially affected the increased the risk of adenocarcinoma with passive smoking because the relation between active smoking and adenocarcinoma is weak. Finally, we do not have information on the time spent together in the same room or on the time period when the husband and wife live together. Some misclassification may therefore have occurred.
In conclusion, our study found a positive association between passive smoking from husbands and lung adenocarcinoma in non-smoking women in Japan. The positive results for quantitative indicators (such as the number of cigarette smoked and pack-years) of passive smoking from husbands reinforce this conclusion. We also identified a higher risk of adenocarcinoma for combined husband and workplace exposure. Prohibition of smoking at home and in public places may yield considerable health benefits for non-smoking women.
We thank all staff members in each study area and in the central offices for their cooperation and technical assistance. We also thank the Iwate, Aomori, Ibaraki, Niigata, Osaka, Kochi, Nagasaki and Okinawa Cancer Registries for their provision of incidence data.