Diet as a confounder of the association between air pollution and female lung cancer: Hong Kong studies on exposures to environmental tobacco smoke, incense, and cooking fumes as examples

Abstract

Chinese females in Hong Kong, where only about a third of the lung cancer cases can be attributed to a history of active smoking, have a world age-standardized lung cancer incidence rate of 32.6 per 100 000, which is among the highest in the world. Trends in Hong Kong's female lung cancer mortality also indicate a tripling in mortality rates from 1961 to 1990. The characteristically high Chinese female lung cancer incidence among nonsmokers is also found among overseas Chinese communities in Singapore and Hawaii. To help elucidate the role of ingested and inhaled substances in the etiology of lung cancer, four epidemiological studies have been conducted in Hong Kong over the last 15 years: (1) a retrospective study of 200 cases and 200 neighbourhood controls, (2) a cross-sectional study measuring personal exposures to nitrogen dioxide among 362 children and their mothers, (3) a site monitoring study of 33 homes measuring airborne carcinogens, and (4) a telephone survey of 500 women on their dietary habits and exposure to air pollutants. Selected data from each study were drawn to evaluate exposures to three major air pollutants (environmental tobacco smoke, incense, and cooking fumes), their relationship with lung cancer risk, and their association with dietary habits. Generally in this population, nutritionally poorer diets were characterized by higher consumption of alcohol and preserved/cured foods, whereas better diets were characterized by higher intakes of fresh fruits, vegetables, and fish. For environmental tobacco smoke, exposure was only moderately high in Hong Kong (36% have current smokers at home), lung cancer risk was equivocal with exposure, and it was associated with poorer diets among wives with smoking husbands. Incense was identified as a major source of exposure to nitrogen dioxide and airborne carcinogens, but it had no effect on lung cancer risk among nonsmokers and significantly reduced risk (trend, P-value = 0.01) among smokers, even after adjusting for smoking. The last finding may be explained by the relatively better diets among smoking women who burned incense versus those who did not. Although about 94% of the Chinese women cook on a regular basis, and the cooking fires were associated with increased airborne carcinogens, nonsmoking women who cooked for more than 25 years had a 60% reduction in lung cancer risk and the trend was highly significant (P < 0.001). Again, this unexpected finding may be due to the confounding effects of diet. Female controls who cooked for more than 25 years had a poorer diet than those who cooked for shorter durations. These three examples were chosen to illustrate the complexities of assessing air pollution exposure, and understanding the behavioral and dietary dynamics underlying lung cancer risk assessments. Our conclusion is that diet can be an important confounding factor affecting lung cancer risk estimates from air pollution exposures among Chinese women living in an affluent urban environment.