Both food preferences and food frequency scores predict fat intakes of women with breast cancer

OBJECTIVE: To determine whether self-reported frequencies of food use were linked to self-reported preferences for the same foods. The hypothesis was that both food frequencies and food preferences can predict nutrient intakes. RESPONDENTS: Participants were adult women patients (n = 339), recruited through the University of Michigan Breast Care Center. The sample included both persons with breast cancer and persons who were cancer-free. DESIGN: All women completed a 98-item food frequency questionnaire and rated preferences for many of the same foods using a 9-point category scale. Percent energy from fat and saturated fat, and intakes of dietary fiber and vitamin C were estimated from analyses of 4-day food records. STATISTICAL ANALYSES: Pearson correlations coefficients were used for data analysis. RESULTS: Dislike was a strong predictor of nonuse. In contrast, the more preferred foods were also reported as more frequently consumed. Significant correlations between preference and frequency scores were obtained for virtually all item pairs. Median Pearson correlation coefficient was 0.30 (range 0.04 to 0.56). Correlations improved when foods were aggregated into the chief dietary sources of fat, saturated fat, and vitamin C. Food frequencies and food preferences showed the same strength of association with percent energy from fat and saturated fat (r = 0.20 to 0.25). Food frequencies showed a stronger association with vitamin C intakes than did preferences for vegetables and fruit. APPLICATIONS: Food preferences may provide a potential alternative to the food frequency approach.