Performance of a new regression equation for predicting energy expenditure and its potential for identifying dietary reference values for energy

Abstract The performance of a prediction equation for total energy expenditure (TEE), derived from a large database of doubly labelled water (DLW) measurements of TEE held at the International Atomic Energy Agency, which has been proposed for estimating population energy requirements (ER) from basic anthropometric data as a replacement for the factorial method, is tested against a dataset of 450 DLW-derived measures of TEE in randomly selected urban US adults aged 40–69 years, and its bias is evaluated. Its performance is also tested by modelling TEE for all ages at the healthy weights identified by the UK Scientific Advisory Committee on Nutrition, (SACN), in its report on dietary reference values (DRV) for energy, and the predicted values for TEE are compared with TEE calculated by the factorial method, as well as with values in a subsequent UK Government report on Dietary Recommendations for Energy, in which values for adolescents and for adults were capped at below their predicted values to help address issues of overweight and obesity. Whilst the prediction equation for TEE is accurate (≤±10 %) for about half of the adults (root mean square of error (RMSE) and %RMSE were 1·67 MJ/d (14·3 %, males) and 1·29 (8·8 %) MJ/d, females), it markedly over-or underpredicts for inactive or active individuals and predicts slightly higher values for males than females than would be expected from their differences in basal energy expenditure. It includes a term for elevation of the subject, which results in additions in TEE which appear unphysiological. These shortcomings raise important questions about its value in improving DRV for energy in the UK.