Objectives: To compare validation of reported dietary intakes from weighed records against urinary nitrogen excretion and energy expenditure measured by DLW, and to examine the utility of the Goldberg cut-off for EI:BMR in the identification of under-reporters.
Design: Energy (EI) and nitrogen (protein) intake (NI) were measured by 16 d of weighed diet records collected over 1 y. They were validated against urinary nitrogen excretion in 5-8 (mean 6.0) 24 h urine collections and total energy expenditure (EE) measured by doubly labelled water (DLW). Basal metabolic rate (BMR) as measured by whole body calorimetry in women or bedside ventilated hood (Deltatrac) in men. Individual subjects were identified as under-reporters if Urine N:NI was > 1.00 or if EI:EE was < 0.79. The agreement between the two ratios in detecting under-reporting was examined. The results from the direct validation by DLW were also compared with validation using the Goldberg cut-off for EI:BMR (Goldberg et al, 1991).
Subjects: Eighteen women aged 50-65 y and 27 men aged 55-87 y were selected from participants in two larger dietary surveys as representing the full range of dietary reporting as measured by Urine N:NI. Data from a previous study of 11 post-obese subjects were also included.
Results: The two ratios, Urine N:NI and EI:EE, were significantly related (r = -0.48, P < 0.01). Using the above cut-offs, seven (4F, 3M) subjects were identified as under-reporters by both methods, one (1M) by Urine N:NI only and 8 (3F, 5M) by EI:EE only. There was close agreement in post-obese subjects where 6 subjects showed a substantial degree of under-reporting by both methods (r = -0.87, P < 0.001). The correlation between direct validation by DLW and EI:BMRest was 0.65 (P < 0.001). Some limitations of the Goldberg cut-off for identifying individual under-reporters were demonstrated.
Conclusions: EI:EE provides an estimate of the degree of under-reporting of energy at the group and individual level. Urine N:NI identifies under-reporting of protein intake and the most obvious under-reporters of energy, but is probably of lesser value in estimating the overall degree of under-reporting of energy at group level. Good validation by EI:BMR depends on knowledge of physical activity at both group and individual level. However, the correlation of 0.65 between EI:EE and EI:BMRest suggests that EI:BMR could be usefully incorporated into analysis of data from epidemiological studies. Validation measures consisting of at least predicted EI:BMR ratios and urinary measures should be incorporated into dietary surveys.
Sponsorship: This work was funded by the Ministry of Agriculture Fisheries and Food, the Medical Research Council, the Cancer Research Council and the Swedish Medical Research Council and the Henning and Johan Throne-Holst Foundation.