Dietary patterns and breast cancer risk through multilevel modeling coupled to a sensitivity analysis: an ongoing case-control study in Córdoba, Argentina.

Abstract

Introduction: Several studies have shown the effect of dietary patterns on breast cancer risk but none has been conducted in Argentina. The aim of this study was to extract dietary patterns from FFQ and to estimate their effect on breast cancer occurrence while taking into account aggregation factors (family history of breast cancer),and to explore the sensitivity of the estimates to changes in the assumptions. Methods: A Principal Component Exploratory Factor Analysis was applied to identify dietary patterns, which were then included as covariates in a multilevel logistic regression. Family history of BC was considered as a clustering variable. A multiple probabilistic sensitivity analysis was also performed.Results: The study included 100 cases and 294 controls. Four dietary patterns were identified. Traditional (fat meats, bakery products and vegetable oil and mayonnaise) (OR III tertilevs I 3.13, 95%CI 2.58-3.78), Rural (processed meat) (OR III tertilevs I 2.02, 95%CI 1.21-3.37) and Starchy (refined grains) (OR III tertilevs I 1.82, 95%CI 1.18-2.79) dietary patterns were positively associated with BC risk, whereas the Prudent pattern (fruit and non-starchy vegetables) (OR III tertilevs I 0.56, 95%CI 0.41-0.77) showed a protective effect. For Traditional pattern, the median bias-adjusted ORs (3.52) were higher than the conventional (2.76). Conclusions: While the Prudent pattern was associated with a reduced risk of BC, Traditional, Rural and Starchy patterns showed a promoting effect. Despite the threats to validity, the nature of associations was not strongly affected.