Abstract

A multifactorial approach to the aetiology of oesophageal cancer was made on the basis of a case-control study in Saitama prefecture, Japan. The joint risks of two factors were calculated directly from joint distributions, following a dichotomous exposure model. Three models of factor combinations were taken into account: two risk enhancing factors, two risk reducing factors, and risk enhancing and reducing factors. We observed remarkable risk elevations in the first model, and the observed joint risks were in the neighbourhood of the multiplicative products of single acting risks of individual factors. The highest odds ratios of about 10 or more were found with combinations of salty foods, excessive intake of rice and alcohol abuse. The second and third models also followed a multiplicative modification of risk. The lowest odds ratios of less than 0.2 were observed in the second model, with combinations of fruits and raw vegetables, fruits and seaweed, and raw vegetables and meat. In the third model, the increased risk caused by an enhancing factor was reduced proportionately to the presence of a risk reducing factor. Finally the dose-response relations of two factors were observed and shown to be categorised into three typical patterns of risk modification, following a three exposure level model. These patterns could be explained by both the dose-response relations of individual factors and the multiplicative modification of risk.