Comparison of self -reported energy and fat intake with objective biomarkers in postmenopausal women

Recent evidence suggests that self-assessment tools traditionally used in the collection of dietary intake data are vulnerable to reporting bias and this bias is thought to be systematic. Evidence is accumulating that suggests macronutrients are differentially reported as well. Inaccurate accounts of fat intake may be limiting the ability to identify associations between dietary fat intake and chronic diseases such as breast cancer. We hypothesized that different groups of postmenopausal women misreport dietary energy and fat intake to different degrees on the Women's Health Initiative (WHI) food frequency questionnaire. Indirect calorimetry, Caltrac(TM) accelerometers, and an estimate of the thermic effect of food (10%) were used to objectively measure total energy expenditure (TEE). We incorporated a novel biomarker of fat intake, the phospholipid fatty acid C18:1(n-7)-cis, to evaluate reporting error differences between fat and energy. We recruited 102 post-menopausal women ages 50--79 years from the greater Seattle area. They completed a series of questionnaires (the Marlowe-Crowne Social Desirability Scale, Sorensen-Stunkard's silhouettes, a health belief questionnaire, and duplicate Women's Health Initiative food frequency questionnaires (FFQ) and collected urine for 24 hours prior to each of 2 visits scheduled one week apart. Reporting error ratios for % energy as fat were compared by participant demographics, adiposity-related factors, and psychosocial factors. Fat reporting significantly decreased with increasing age (p for trend = 0.099) and decreased with increasing adiposity-related factors (body mass index, fat free mass, % body fat, body surface area) but these trends were not significant. Waist-to-hip ratio was significantly inversely associated with fat reporting (p for trend = 0.05). Comparisons of reporting error for fat and energy found significant differences by % body fat (p for trend = 0.04), fat mass (p for trend = 0.05), and fat distribution assessed by waist-to-hip ratio (p for trend = 0.02). The biomarker for dietary fat intake used to identify reporting error differences for fat and energy did not allow defense of absolute values for fat reporting error to energy reporting error ratios. However, relative differences suggest that systematic error exists in self-reported dietary intake data from the WHI food frequency questionnaire and may influence analytical statistical conclusions.