Reply to DM Thomas and K Westerterp

Dear Editor:

Thomas and Westerterp raised 3 points with regard to our article (1). First, they noted that the operationalization of energy balance, which was based on doubly labeled water–estimated habitual energy intake over a 2-wk baseline observation period minus estimated habitual energy expenditure over the same period, would be highly correlated with changes in weight over the 2-wk observational period. We would have assumed that participants who were in a positive energy balance would gain weight over the 2-wk observational period, whereas those who were in a negative energy balance would lose weight. Our data confirmed that our estimate of energy balance correlated with weight change over the 2-wk interval (r = 0.97, P < 0.0001), statistically controlling for body composition (kcal · kg fat-free mass⁻¹ · d⁻¹). It is important to note, however, that the calculations that Thomas and Westerterp used assumed that change in weight was between −1.5 and 1.5 kg (i.e., that participants were largely weight stable), which was not the case. The data indicated that the weight change over the 2-wk doubly labeled water observation period actually ranged from −4.9 to 2.55 kg for the entire sample, as well as the “in flux” subsample that was used in some of our analyses.

The second point made by Thomas and Westerterp was that our definition of energy flux [total energy intake + total energy expenditure (TEI + TEE)] was equivalent to 2TEE; however, that calculation again assumed that the sample did not show marked weight change over the 2-wk observational period, which, as noted, was not correct. In addition, controlling for baseline body fat, 2TEE at baseline correlated with future body fat change in study 1 (r = 0.18, P = 0.03) but not in study 2 (r = 0.06, P = 0.62), whereas TEI + TEE predicted future body fat gain in both samples after baseline body fat was controlled for (1).

Third, Thomas and Westerterp state that “examining the true influence of low energy turnover on fat gain requires adjusting TEE by resting metabolic rate (RMR).” We did not do this because of experimental evidence that increasing TEE results in a subsequent increase in RMR (2). If increasing energy expenditure results in increased RMR, statistically controlling for RMR would needlessly remove variance from TEE, which would attenuate predictive effects. This did not seem to be useful from our perspective. Thomas and Westerterp also assert that “individuals with higher baseline FM [fat mass] gain more FM than do their lean counterparts.” Curiously, however, our Tables 1 and 2 indicate that baseline body fat did not significantly predict future change in body fat over follow-up in either study.

Last, we should clarify that we did not argue or test whether high physical activity level would predict future body fat loss. Rather, our findings suggested that TEI + TEE did significantly predict future body fat change over 2–3 y of follow-up (1). In addition, TEE did not predict future change in body fat in either of our samples after baseline body fat was controlled for, which agrees with previous research that uses a much smaller sample (3). We hope that this clarifies the points raised by Thomas and Westerterp with regard to our article on the predictive effects of TEI + TEE on future body fat gain.