[Relationship between smoking and risk of esophageal cancer in 103 areas in China: a large-scale case-control study incorporated into a nationwide survey of mortality].
Liu BQ., Jiang JM., Chen ZM., Chen JS., Zhang KL., Zeng XJ., Zhao P., Boreham J., Wu YP., Li JY., Peto R.
OBJECTIVE: To examine the relationship between smoking and risk of esophageal cancer (EC), and present a theoretical framework of control selection in population-based case-control study which was incorporated into a nationwide retrospective survey of mortality in China. METHODS: A large-scale population-based case-control study was incorporated into the nationwide retrospective survey of mortality conducted 1989 - 1991 in 24 urban cities selected by non-random sampling and 79 rural counties selected from 3000 counties included in the 1973 - 1975 cancer distribution survey by random sampling during. A questionnaire survey was conducted by home visit to investigate the death causes and smoking history of 19 734 deceased male adults who died of esophageal cancer during 1986 - 1988 at the age >or= 35. Two control groups were set up to undergo questionnaire survey by home visit to investigate the smoking history of the deceased persons and the informants. Control group I included the surviving spouses or other informants of 31 989 male adults who died of non-malignant digestive diseases during 1986 - 1988 at the age >or= 35, and control group II included 104 846 male spouses of the deceased female adults who died of different causes during 1986 - 1989 at the age >or= 35. The relative risks and population smoking attributable risks for EC were calculated using non-conditional logistic model, and the results were compared for consistency between the analyses using two different control groups. RESULTS: The EC absolute death rates were higher in the smokers than in the non-smokers in all urban and rural area groups. The total EC absolute death rate per 1000 among the non-smokers vs. smokers was 0.37:0.65 in the urban areas, 0.99:1.29 in the inland rural areas, and 1.09:1.62 in the coastal rural areas in the control group I, and there was a similar trend in the control group II. There was a significant dose-response relation between the period of smoking and the death risk of EC and between the daily cigarette consumption and the death risk of EC. The risk ratios, for example, for cigarette per day < 10, 10-, and 20- in the urban men were 1.42, 1.82, 2.22 in the control group I (trend test P < 0.01), and 1.57, 1.95, and 3.18 in the control group II (trend test P < 0.01). CONCLUSION: Smoking is an important risk factor for mortality from EC in China. Investigating the surviving spouses of the deceased patients is a creative, effective, and feasible trial, with the prerequisite of whole population-based survey, in study of the main types of death and the relevant risk factors.