Tobacco Documents Online

3 in women have been rising rapidly, from 1.4/100,000 in 1980xo 6.91100,000 in 1996 among women, and at the end they state that "This study appears to explain why mortality from lung cancer in Korean women is escalating, particularly among wives whose husbands smoke, even though the rate of women's smoking is negligible." The difficulty is that the rise in rates has been by almost 5 fold and that such a large rise cannot possibly be explained by their results. From the data in Table 1, 1 estimate that the risk of the women studied is only about 1.5 times higher than had they all been married to nonsmoking husbands, and anyway it can hardly be the case that the proportion of women with smoking husbands was zero in 1980 - probably many women had current smoking husbands then. While the paper is of interest, the results will not have any major effect on the inconclusive nature of the evidence on the relationship of ETS to lung cancer. P N Lee 24.11.99.

2 husbands (RR = 1.9, 95% CI = 1.0-3.5). A significant positive trend (p<0.01) was noted with years of smoking by the husband, with relative risks of 1.6 (0.8-3.0) for 1-29 years and 3.1 (1.4- 6.6) for 30+ years duration. The trend for amount smoked was not significant, with the increase in risk somewhat higher for 1-19 cigs/day (2.0, 1.1-3.9) than for 20+ cigs/day (1.5, 0.7-3.3). The covariate adjusted risks associated with smoking by the husband were only slightly less than the age-adjusted risks (see Table 1), implying that adjustment had little effect. There was no significant association between smoking by the husband (current or ex) and risk of emphysema or asthma, cervix cancer, stomach cancer, breast cancer or liver cancer (see Table 2). The authors note that the results of their study were quite similar to those of Hirayama's study, though surprisingly they only cited the early (1981) paper without citing the latest data, and they incorrectly stated that Hirayama reported a relative risk of 2 for current smoking by the husband. Hirayama's 1981 estimate of about 2 was for husband smoking 20+ cigs/day (and in any case was not adjusted for the age of the wife). The authors refer to a number of weaknesses of the study, including the relatively brief duration of follow-up (and consequently only modest numbers of deaths), the reliance on discharge diagnosis and the possibility of selection bias (only 35% of dependants of the workers participated). Another problem is the lack of objective confirmation of smoking status and of ETS exposure by cotinine. The authors refer briefly to the possibility of bias due to misclassification of smoking habits, but make no reference to the evidence of extremely high misclassification rates in Asian women. In commenting on the stronger evidence of a dose-response for duration than for amount smoked by the husband they suggested that their results showed that duration was a better indicator of ETS exposure. However, they do not note that the overall epidemiological evidence tends to show a significant dose-related trend more often with amount than with duration. A maj or problem with the paper is that, at the start, the authors note that lung cancer rates

I REVIEW 1077 CONFIDENTIAL Subject ref 8b "Effect of husbands' smoking on the incidence of lung cancer in Korean women" SHJeeetal International Journal of Epidemiology (1999), 28, 824-828 Although there have been 50 or so epidemiological studies on ETS and lung cancer, only five have been of prospective design, and two of these (Butler, Hole) have involved very small numbers of deaths (less than 10), though the others (Hirayama, Garfinkel [CPS I] and Cardenas [CPS II]) have involved larger numbers (about 200 each). The prospective study described in the current paper, conducted in Korea, is of intermediate size, based on 79 incident cases of lung cancer occurring in 157,436 women aged 40 or over who reported having never smoked for as long as a year. The study is based on people insured by the Korean Medical Insurance Corporation, and consists of civil service workers and teachers and their dependents. Insured workers (mainly men) completed a questionnaire and were subsequently medically examined in 1992 and again in 1994, while dependants (from whom most of the women were drawn) went through the same routine in 1993. The data were analysed so as to determine who the never smoking women were, to classify them according to their husband's smoking status (never, ex and current) and for current smokers by amount and duration of smoking. Incidences ofnew lung cancer occurring in 1994-1997, based on hospital discharge diagnosis (which usually required pathological confirmation), were compared by the smoking habit of the husband, with adjustment for the age of the woman, the age of the husband, socioeconomic status, occupation ofthe husband (blue/white collar), residency (urban/rnual) and the self-reported vegetable consumption of the woman (low, moderate, high). The adjusted relative risks of lung cancer, all compared with that for women whose husband never smoked, were nonsignificantly higher for women with ex-smoking husbands (RR = 1.3, 95% CI = 0.6-2.7) and marginally significantly higher for women with current smoking