Energy expenditure and body composition changes after an isocaloric ketogenic diet in overweight and obese men: A secondary analysis of energy expenditure and physical activity

Background A previously published pilot study assessed energy expenditure (EE) of participants with overweight and obesity after they were switched from a baseline high-carbohydrate diet (BD) to an isocaloric low-carbohydrate ketogenic diet (KD). EE measured using metabolic chambers increased transiently by what was considered a relatively small extent after the switch to the KD, whereas EE measured using doubly labeled water (EEDLW) increased to a greater degree after the response in the chambers had waned. Using a publicly available dataset, we examined the effect of housing conditions on the magnitude of the increase in EEDLW after the switch to the KD and the role of physical activity in that response. Methods The 14-day EEDLW measurement period included 4 days when subjects were confined to chambers instead of living in wards. To determine the effect on EEDLW only for the days subjects were living in the wards, we calculated non-chamber EE (EEnonchamber). To assess the role of physical activity in the response to the KD, we analyzed chamber and non-chamber accelerometer data for the BD and KD EEDLW measurement periods. Results In comparison with the increase in average 14-day EEDLW of 151 kcal/d ± 63 (P = 0.03) after the switch to the KD, EEnonchamber increased by 203 ± 89 kcal/d (P = 0.04) or 283 ± 116 kcal/d (P = 0.03) depending on the analytical approach. Hip accelerometer counts decreased significantly (P = 0.01) after the switch to the KD, whereas wrist and ankle accelerometer counts did not change. Conclusions Switching from the BD to the KD substantially increased EEDLW, but apparently only on days subjects were living in the ward outside the metabolic chamber. Increased physical activity as measured by accelerometry did not appear to account for this effect.

However, five other participants had changes in body weights during the KD period that were incommensurate with the difference between their EE DLW and EI. Three participants (Group 2; Table A) showed more moderate, but substantial, differences between EE DLW and EI. In two of these, body weight decreased despite an EE DLW that was less than their EI, whereas the third gained weight although EE DLW exceeded EI. Two additional participants (Group 3; Table A) exhibited small, negative differences between EE DLW and EI that were associated with a decrease in body weight.
Subjects A and B showed the two greatest increases in EE DLW and nonchamber energy expenditure (EE nonchamber ) after the switch from the BD to KD of all 17 participants. Therefore, exclusion of their expenditure data would be expected to reduce any increase in average expenditures after the switch whether adjusted for energy balance or not. Figure A shows the effect of excluding these two participants (Group 1) on average EE nonchamber , a primary outcome  Table  A either separately or in combination with those in Group 1.
Although EE nonchamber increased significantly after the switch to the KD when all 17 participants are included in the analysis, the increase was not statistically significant when data from Subjects A and B (Group 1) were excluded from the analysis. In contrast, the increase of EE nonchamber was significant despite removal of other groups of putative outliers either alone or in combination with other groups. The effect of excluding Group 2 alone or in combination with Group 1 is especially notable because the energy imbalances reflected by the differences between EE DLW and EI, while less than that in Group 1, were substantial. As reported in this paper (Table 1), the increase in EE nonchamber after the diet switch was statistically significant after exclusion of Subject B's (ID# 04-012) data from the analysis. When only Subject A's data was removed from the analysis, the increase in EE nonchamber after the diet switch (257 + 116 kcal/d) was also statistically significant (P =0.49 by paired t-test).

Asynchronous Measurement Intervals
The reported gain of body weights during the KD period of Subjects A and B was based on body weight measurements taken during two body composition assessments in the latter part of the diet period. Hall et al. [2] claimed that the interval for body composition assessments was coincident with the EE DLW measurement period. However, inspection of dates in the original study's dataset for DLW dosing and body composition assessments shows this was not the case; body composition and EE DLW measurements periods were coincident in only 6 and 4 out of 17 participants in the BD and KD periods, respectively.
Because the dataset includes dates and daily body weight measurements for all subjects throughout the study, it is possible to determine the change in body weight over the EE DLW measurement period independently from the body weight measures taken during body composition assessments.
As discussed in this paper, Hall et al. [1] reported that Subject B gained 0.2 kg during the KD EE DLW measurement period based on body composition assessments (Table B), but daily body weight measurements show a body weight loss of 0.5 kg during the actual EE DLW measurement period (Table B). Similarly, according to the database, Subject A gained 0.6 kg of weight as per  BWBC, body weight for body composition measurement dates using body composition assessment dates; BWCD, body weight for body composition measurement dates using daily body weight measurements; BWDD, body weight for doubly labeled water measurement dates using daily body weight measurements; diff, change in body weight from start to end. body composition assessments, but according to daily body weight measurements, lost 0.5 kg during the EE DLW measurement period.
No other participants showed such a reversal of body weight change from gain to loss (or vice versa) during the KD period when body weight measurements were synchronized with the EE DLW measurement period, although one participant did so during the BD period. The changes in body weight of Subjects A and B over the EE DLW measurement period measured using daily body weight data were well within one standard deviation of that for the group as a whole (-0.9 + 0.6 kg, mean + SD). The differences between body weight measurements from body composition assessments and recorded daily body weights do not appear to be due to differences in the precision of measurement under the two conditions because changes in body weight over the two body composition assessments as determined using daily body weight measurements were consistent with those measured during composition evaluations (Table B).