Tobacco Documents Online

4 cause and effect relationship. The authors consider evidence from other studies in their paper. They correctly note that the results so far "have not exhibited a consistent pattern for risk from ETS exposures in childhood as compared with ETS exposures in adulthood." P N Lee 3.1.2001

1 11`~~~0 t{ REVIEW 1134 CONFIDENTIAL Subj_ect ref 8b "Lung cancer and environmental tobacco smoke in a non-industrial area of China" L Wang et al International Journal of Cancer (2000), 88, 139-145 The authors describe the results of a relatively large case-control study conducted in two prefectures of China (Pingliang and Qingyang) involving a total of 200 female and 33 male never smoking lung cancer cases and 407 female and 114 male population controls. Data were collected for each household member on the number of"store-bought" cigarettes smoked per day and on the quantity of tobacco per month smoked as hand-rolled cigarettes or in Chinese long- stem pipes, the information relating to both the subject's childhood (aged 18 or under) and adulthood. Data were also collected on a range of potential confounding variables including demographic variables, diet and cooking practices and detailed occupational, residential and medical histories. After adjustment for age, sex, prefecture and two socioeconomic indicators (ownership of a colour television and number of cattle owned), the analysis in Table II showed that any lifetime ETS exposure was associated with a non-signifrcantly increased risk of lung cancer of 1.19 (95% CI 0.7-2.0). The estimate was similar for females (1,15, 0.6-2.1) and males (1.22, 0.5- 3.3). However, there was a significant (0.01<p<0.05) trend in risk with pack-years of ETS exposure, with risks estimated as 1.00, 1.04, 1.13 and 1.51 for, respectively, 0, 1-9, 10-19 and 20+ pack-years. Further analysis showed that the association arose because of the contribution of ETS exposure in childhood. After adjusting for the same list of variables and also ETS exposure in adulthood, the relative risk estimate for childhood ETS exposure was 1.52 (1.1-2.2), again similar for females (1.51, 1.0-2.2) and males (1.46, 0.6-3.7). Here the trend was again significant (p<0.01), with risks estimated as 1.00, 1.43, 1.81 and 2.95 for the same four pack-years

2 categories. In contrast, there was no evidence of an association of lung cancer risk with ETS exposure in adulthood. After adjusting for the original list ofpotential confounding variables and also ETS exposure in adulthood, the relative risk estimate for adulthood ETS exposure was 0.90 (0.6-1.4), with no significant increase evident in females (1.03, 0.60-1.7) or in males (0.56, 0.2- 1.4), Nor was any dose-response relationship seen. The authors also reported results of analyses (in Table III) claimed to show no evidence that the magnitude of the associations seen between lung cancer and ETS exposure varied according to the presence or absence of a number of other variables (though this is difficult to see from the analyses presented). They also reported a non-significant relationship between occupational ETS exposure, with the relative risk estimated as 1.56 (0.7-3.3). However, as the two prefectures were predominantly rural, ETS exposure from non-farm jobs was relatively rarely reported. While the study is a relatively large one, and is the first in China (including Hong-Kong) to report results for men as well as for women, there are a number of limitations which should be noted: (1) Although the case diagnoses were reviewed by an expert panel, most were based only on clinical and/or radiological evidence, with only 32% histologically confirmed. However, restricting attention to the confirmed cases, relative risks were estimated as 1.55 (0.9-2.8) for childhood exposure, and 0.99 (0.5-2.0) for adulthood exposure, quite similar to the main findings. (2) Because cases were diagnosed starting in January 1994 but interviewing did not start until at least a year or two later, next-of-kin respondents were used for 54% of them. This proportion was much higher than that, 4%, for the controls. After restricting attention to self-respondents, the relative risks were estimated as 1.75 (1.1-2.8) for childhood exposure and 0.76 (0.4-1.3) for adulthood exposure.

3 (3) Unlike the great majority of previous ETS/lung cancer studies, no results are given for spousal smoking. (4) It is noted that childhood ETS exposure was predominantly from pipes, while ETS exposure in adulthood from pipes and cigarettes was quite similar. However, the results presented do not make it clear whether the association of ETS exposure with lung cancer risk in never smokers arose because of ETS from pipes, ETS from cigarettes, or both. The authors note that the increase in risk per cigarette smoked per day and per liang (50 gms) of tobacco smoked per month in Chinese long-term pipes were similar, and they compute pack-years by equating one cigarette per day to one liang per month in the overall calculation. Whether this is a sensible thing to do is not clear to me, and in some ways I would have liked to see more information about the separate associations of ETS from cigarettes and ETS from pipes in adulthood and in childhood on lung cancer risk. There also seems to be no separation between potential effects of ETS from cigarettes bought in the store and ETS from hand-rolled cigarettes. (5) There is no mention whatsoever of the possibility of misclassification, with some of the subjects reported to be never smokers, and therefore included in the analyses, actually having smoked. (6) The list of potential confounding factors considered is reasonably long (see Tables I and III) but hardly comprehensive. I note that, in Table I, there are some quite striking differences between cases and controls in some factors. Thus in males, cases are about 10 times more likely than controls to have refrigerators and are also very much more likely to have colour rather than black and white TV. While it seems implausible that these factors actually cause lung cancer, it is a concern that cases and controls are so different, implying that the controls are not really representative of the population from which the cases derived. Overall the study is yet another one showing a moderate association of ETS exposure with lung cancer risk in never smokers, that cannot with certainty be considered indicative of a