Comparison of Methods Used to Correct Self-Reported Protein Intake for Systematic Variation in Reported Energy Intake Using Quantitative Biomarkers of Dietary Intake

Abstract

Multiple methods of correcting nutrient intake for misreported energy intake have been proposed but have not been extensively compared. The availability of the Women's Health Initiative (WHI) data set, which includes several objective recovery biomarkers, offers an opportunity to compare these corrections with respect to protein intake. We compared 5 energy-correction methods for self-reported dietary protein against urinary nitrogen-derived protein intake. As part of the WHI Nutritional Biomarkers Study (NBS) 544 participants (50- to 80-y-old women) completed a FFQ and biomarker assessments using doubly labeled water (DLW) for total energy expenditure (TEE) and 24-h urinary nitrogen. Correction methods evaluated were as follows: 1) DLW-TEE; 2) the Institute of Medicine's (IOM's) estimated energy requirement (EER) TEE prediction equation based on sex, height, weight, and age; 3) published NBS total energy TEE prediction (WHI-NBS-TEE) using age, BMI, race, and income; 4) reported protein versus reported energy linear regression-based residual method; and 5) a Goldberg cutoff to exclude subjects reporting energy intakes <1.35 times their basal metabolic rate. Efficacy was evaluated using correlations obtained by regressing corrected protein against biomarker protein (6.25 × urinary nitrogen/0.81). Unadjusted self-reported protein intake from the FFQ (mean=66.7g) correlated weakly (r=0.31) with biomarker protein (mean=74.9g). DLW-TEE-corrected self-reported protein intake (mean=90.7g) had the strongest correlation with biomarker protein (r=0.47). Other energy corrections yielded lower, but still significant correlations: EER, r=0.44 (mean=92.1g); WHI-NBS-TEE, r=0.37 (mean=90.4g); Goldberg cutoff, r=0.36 (mean=88.4g); and residual method, r=0.35 (mean=66.7g). Our data indicate that proportional correction of reported protein intake using a measure of energy requirement from DLW-TEE or IOM-EER performed modestly better than other methods in this cohort. These energy adjustments, however, yielded corrected protein exceeding the biomarker protein, indicating that energy adjustment alone does not eliminate all self-reported protein reporting bias.