BACKGROUND: Associations of the macronutrient composition of the diet with total energy intake (EI) are uncertain, as are associations of macronutrient composition with self-reported energy underreporting. OBJECTIVES: We aim to estimate associations of biomarker-assessed EI with both biomarker-assessed and self-reported macronutrient component densities in a Women's Health Initiative (WHI) sub-cohort of postmenopausal U.S. females. Secondarily, we examine energy underreporting using food records, recalls and frequencies, for association with macronutrient densities. DESIGN AND METHODS: We used a previously proposed EI biomarker equation based on doubly-labeled water (DLW) and updated biomarker equations for several macronutrient component densities, to estimate EI and macronutrient component densities in a WHI nutritional biomarkers sub-cohort (n=436
2007-2009). We used linear regression of EI biomarker values on biomarker and self-reported macronutrient component densities, and of log-EI underreporting values on biomarker densities, to examine targeted associations. RESULTS: Using biomarker assessments, the geometric mean (95% CI) for EI corresponding to a 20% increment in carbohydrate density was 2.0% (0.1%, 3.9%) higher, and for a 20% protein density increment was 2.1% (0.5%, 3.7%) lower. The former was attributable to added sugars. Similarly, EI values for 20% increments in polyunsaturated (PUFA), and monounsaturated (MUFA) fatty acids densities were respectively 1.4% (0.3%, 2.6%) higher, and 1.5% (0.1%, 2.9%) lower. Pertinent associations were either not detected or were substantially attenuated if instead self-reported macronutrient densities were used. Also, EI underreporting was strongly related to self-reported macronutrient densities using food records, recalls, or frequencies. CONCLUSIONS: Among U.S. postmenopausal females lower EI was associated with diets relatively high in protein or MUFA, and higher EI was associated with diets relatively high in PUFA or added sugars. These associations are of public health importance, but are mostly missed using self-reported dietary density assessments. Self-reported energy underestimation is substantially associated with self-reported macronutrient densities. This study is registered with clinicaltrials.gov identifier: NCT00000611.