Dietary Patterns and Cancer

Randall, D

Abstract

The goal of the research proposed is to examine food patterns for their ability to support more accurate estimations of diet and cancer risk. Estimates of cancer risk associated with diet emerging from case-control and cohort studies are small in comparison both to estimates based on international comparisons and to other risk factors, notably smoking. Nutrient intakes from frequency of food use data may underestimate risk. The combinations in which foods are consumed more accurately and realistically reflect actual exposures to risk factors. We intend to analyze data collected previously in the western New York Diet Study (1975-86), case-control studies of diet and cancer at various sites. The proposed study has several objectives; 1) to identify major dietary patterns existing in the study population, 2) to ascertain how these patterns correlate with demographic traits and 30 to determine whether diet patterns are associated with risk of cancer at various sites. Diet patterns will be identified for the 2255 control subjects (1475 males, 780 females) by applying factor analyses to data collected with a 110-item food frequency interview. We will apply regression analysis using factor scores as dependent variables to identify demographic variables that are known to relate to cancer risk (age, marital status, occupation, education, ethnicity, smoking status) that also correlate with the diet patterns. Logistic regression analyses will then be completed to estimate risk of cancer of the colon, rectum, stomach and lung, controlling for the influences of these demographic variables identified in the previous stem pa s being related to both cancer risk and diet patterns. The relevant demographic variables will be entered simultaneously into the logistic regression analyses prior to the inclusion of the scores for each of the diet patterns, with case status for the four cancer sites being the dependent variables.