Predicting total energy expenditure from self-reported dietary records and physical characteristics in adult and elderly men and women

Abstract

Background: Energy requirements and nutrient intakes are commonly estimated from self-reported dietary records, but such estimation has proven to be unreliable. When energy intakes determined from dietary records are compared with energy expenditures measured with the use of doubly labeled water, the former consistently underestimate energy requirements and have a high degree of variability. Objective: The objective of this study was to reduce the bias and variability of self-reported dietary records through the use of stepwise multiple regression analysis to develop models that relate energy expenditure measured with the use of doubly labeled water to energy intake from dietary records, sex, and fat-free mass (or weight and height). Design: Data from 54 healthy adult men and women were used to develop these models. Results: Fat-free mass, energy intake, and sex accounted for 86% of the variability in energy expenditure, whereas energy intake, sex, height, and weight accounted for 83%. When the model relating fat-free mass, energy intake, and sex to energy expenditure was tested on published data, it reduced the bias and variability of self-reported dietary records from -17 ± 27% to 3 ± 16%, When the model relating energy intake, sex, weight, and height to energy expenditure was tested on published data, it reduced the bias and variability of self-reported dietary records from -19 ± 25% to -0.3 ± 19%. Conclusion: Results from this study indicate that a simple relation can be used to correct self-reported dietary records to estimated energy requirements.