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

Chinese females in Hong Kong, where only about a third of the lung can cer cases can be attributed to a history of active smoking, have a wor ld age-standardized lung cancer incidence rate of 32.6 per 100 000, wh ich is among the highest in the world. Trends in Hong Kong's female lu ng cancer mortality also indicate a tripling in mortality rates from 1 961 to 1990. The characteristically high Chinese female lung cancer in cidence among nonsmokers is also found among overseas Chinese communit ies in Singapore and Hawaii. To help elucidate the role of ingested an d inhaled substances in the etiology of lung cancer, four epidemiologi cal 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 d ioxide among 362 children and their mothers, (3) a site monitoring stu dy of 33 homes measuring airborne carcinogens, and (4) a telephone sur vey of 500 women on their dietary habits and exposure to air pollutant s. Selected data from each study were drawn to evaluate exposures to t hree major air pollutants (environmental tobacco smoke, incense, and c ooking fumes), their relationship with lung cancer risk, and their ass ociation with dietary habits. Generally in this population, nutritiona lly poorer diets were characterized by higher consumption of alcohol a nd preserved/cured foods, whereas better diets were characterized by h igher intakes of fresh fruits, vegetables, and fish. For environmental tobacco smoke, exposure was only moderately high in Hong Kong (36% ha ve current smokers at home), lung cancer risk was equivocal with expos ure, and it was associated with poorer diets among wives with smoking husbands. Incense was identified as a major source of exposure to nitr ogen dioxide and airborne carcinogens, but it had no effect on lung ca ncer risk among nonsmokers and significantly reduced risk (trend, P-va lue = 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 we re associated with increased airborne carcinogens, nonsmoking women wh o cooked for more than 25 years had a 60% reduction in lung cancer ris k and the trend was highly significant (P < 0.001). Again, this unexpe cted finding may be due to the confounding effects of diet. Female con trols who cooked for more than 25 years had a poorer diet than those w ho cooked for shorter durations. These three examples were chosen to i llustrate the complexities of assessing air pollution exposure, and un derstanding the behavioral and dietary dynamics underlying lung cancer risk assessments. Our conclusion is that diet can be an important con founding factor affecting lung cancer risk estimates from air pollutio n exposures among Chinese women living in an affluent urban environmen t.