A case-control study of diet and cancer of the pancreas.

In a population-based case-control study carried out in Adelaide, South Australia, during the years 1984-1987, the diets of 104 cases of cancer of the pancreas 1 year prior to diagnosis were compared with the diets of 253 community controls. A quantitative food-frequency questionnaire was used to assess usual dietary intakes of 179 food items. Cases were compared with controls in terms of both the amounts of individual food items consumed and the estimated contributions of 48 nutrients to the diet. Food items consumed more by cases than controls included boiled eggs and omelettes as well as a number of items that could be collectively described as sweet and fatty. Food items consumed less by cases than controls included several vegetables and fruits. Conditional logistic regression analysis of nutrient intake adjusted for total energy and for alcohol and tobacco usage yielded an estimate of relative risk of 3.19, with a 95% confidence interval of 1.58-6.47 for the highest quartile of cholesterol intake (relative to the lowest quartile). For the top quartile of refined sugar intake, the estimated relative risk was 2.21 (95% confidence interval 1.07-4.55). Several nutrients derived principally from plant foods were statistically significantly associated with lower risks. Alcohol consumption was significantly lower among cases than controls. Current smokers had a relative risk of 1.76 (95% confidence interval 0.93-3.34) relative to those who had never smoked. There was no association of pancreatic cancer with coffee drinking.