Abstract
Objective:
To determine if different measures of habitual PA at baseline predict weight change, weight compensation, and changes in energy intake (EI) during a 24-week supervised aerobic exercise intervention.
Methods:
Data from 108 participants [78 women; 48.7 (SD 11.6) years; 31.4 (SD 4.6) kg/m2], randomized to either the moderate-dose exercise group (8 kcal/kg body weight/week; KKW) or the high-dose exercise group (20 KKW) of the E-MECHANIC trial, were analyzed. Moderate-to-vigorous PA (MVPA), steps/day, and PA energy expenditure (PAEE) were measured with SenseWear armbands, and total activity energy expenditure (AEE) and EI were estimated with doubly labeled water, all over 2 weeks, before and towards the end of the intervention. Multiple linear regression models, adjusted for sex, exercise group, and baseline value of the outcome were used.
Results:
Baseline habitual MVPA levels predicted weight change (β=−0.275; P=0.020), weight compensation (β=−0.238; P=0.043), and change in EI (β=−0.318; P=0.001). Associations between baseline PAEE and outcomes were comparable, whereas steps/day and importantly total AEE (via DLW) did not significantly predict change in weight-related outcomes.
Conclusions:
While acknowledging substantial variability in the data, on average, lower baseline habitual MVPA and PAEE levels were associated with less weight loss from exercise, higher compensation, and increased EI.
Keywords: Aerobic exercise, physical activity, energy intake, weight loss, overweight
INTRODUCTION
The prevalence of overweight and obesity has grown into a worldwide epidemic in recent years (1), and excess body weight substantially increases the risk of adverse health conditions (2). Exercise has been shown to support the prevention and management of obesity (3); however, when used for weight loss, exercise interventions consistent with the physical activity (PA) guidelines for weight loss and weight loss maintenance (>225 min of moderate-intensity PA per week) frequently produce less weight loss than expected based on energy expended in exercise (4-7). This discrepancy is called weight compensation (8) and primarily results from exercise-induced increases in appetite and energy intake (EI) as opposed to changes in metabolism or activity (6, 9).
It is unknown if factors pertaining to one’s lifestyle prior to starting an exercise program affect weight compensation and food intake (FI). An individual’s habitual PA level at baseline might be such a determinant of the observed difference between actual weight change and predicted weight loss from the energy balance model. As suggested by Westerterp (10), it is possible that a lower habitual PA level at baseline allows an exercise-induced increase in energy expenditure (EE) without or with less compensatory increase in EI. Conversely, and based on previous research indicating that EI and energy balance are better regulated at higher levels of activity-related EE (11, 12), lower habitual PA levels at baseline might be associated with larger compensatory increases in EI in response to an exercise-induced increase in EE (10).
To further elucidate the mechanisms for weight compensation in response to exercise, the aim of this analysis was to determine if different measures of habitual PA at baseline predict weight change, weight compensation, and changes in EI during a 24-week supervised, controlled, aerobic exercise intervention. Specifically, we aimed to compare the predictive value of (1) minutes spent in moderate-to-vigorous PA (MVPA), (2) steps/day, and (3) physical activity energy expenditure (PAEE), assessed via two validated methods, with regard to these outcomes. Based on previous work (11, 12), we hypothesized that participants with lower PA levels at baseline would show greater weight compensation and larger exercise-induced increases in EI. While PAEE is directly related to the energy balance model and a significant association with our outcomes might be expected, we aimed to additionally assess the association of MVPA and steps/day with our outcomes, as PA recommendations based on these parameters are commonly communicated to patients and a predictive value of these parameters would consequently be of interest to many clinical and research settings.
METHODS
Design and Participants
This report is a secondary analysis of the Examination of Mechanisms of Exercise-induced Weight Compensation (E-MECHANIC) study (ClinicalTrials.gov ID: NCT01264406) that was approved by the Institutional Review Board and conducted between November 2010 and December 2015 at Pennington Biomedical Research Center (Baton Rouge, LA, USA). The complete design, methods, and primary outcomes of the E-MECHANIC study have been previously published (6, 13). In brief, this 24-week randomized controlled trial recruited 198 healthy men and women with overweight/obesity (BMI ≥25 kg/m2 to ≤45 kg/m2) and low levels of PA (≤20 minutes of structured exercise on ≤3 days/week based on self-report; <8,000 steps/day (14) assessed during a 1 week of accelerometer data (SenseWear armband, BodyMedia, Pittsburgh, PA, USA)). Participants were randomly allocated in a 1:1:1 ratio to either a moderate-dose exercise group (8 kcal/kg body weight/week; 8 KKW), a high-dose exercise group (20 KKW), or a non-exercise control group (13). The selected exercise doses reflect recommendations for general health (8 KKW) and for weight loss (20 KKW) (15). Exercise intensity during the supervised exercise sessions was self-selected between 65% and 85% peak oxygen uptake (VO2peak) and sessions varied in length to meet each participant’s EE goal (16).
Participants were excluded if they were currently participating in a weight loss program (and/or ≥4 kg weight change in the past 6 months), were currently pregnant or had been pregnant within the past six months, or were diagnosed with diabetes, cardiovascular disease, or arrhythmia. All participants provided written informed consent prior to inclusion in the study.
The primary aim of the E-MECHANIC study was to identify mechanisms of exercise-induced weight compensation (i.e., less than expected weight loss) by examining the effect of the two different doses of exercise training on EI over the 24-week intervention period. The study found significantly higher weight compensation in the high-dose exercise group compared to the moderate-dose exercise group, which resulted primarily from increased EI and concomitant increases in appetite (6).
In this report, to examine the impact of baseline levels of habitual PA on outcome measures during a supervised exercise intervention, only participants allocated to the two exercise groups (n=110) and who completed the trial per protocol were included in the main analyses. Demographics of those exercisers who did not complete the trial (n=25) did not differ significantly from completers (all P-values ≥ 0.093).
Outcome Variables
Anthropometry and Body Composition
At baseline and follow-up, body weight was assessed under fasting conditions using a Tanita Scale (Arlington Heights, IL, USA) and waist circumference was determined using a non-extensible tape measure (Gulick II, Sammons Preston, Chicago, IL, USA). Dual-energy X-ray absorptiometry (DXA) (Lunar iDXA and Encore software version 13.60; GE Healthcare, Madison, WI, USA) was used to assess fat mass.
Weight Compensation
Weight compensation is the difference between the amount of weight loss predicted from exercise-associated EE and observed weight loss from baseline to follow-up (actual – predicted weight change). Predicted weight loss was calculated utilizing a validated dynamic energy balance model that overcomes the limitations of the conventional assumption that 1 kg of body weight equals 7700 kcal/kg (7, 17, 18), accounting for adaptations that occur when body mass changes, including adaptations to resting metabolic rate, dietary-induced thermogenesis, and non-exercise activity thermogenesis (19).
Energy Intake
EI was estimated with doubly labeled water (DLW) and FI tests at baseline and follow-up. DLW data were collected over a 2-week period at both time points. DLW measures total daily energy expenditure (TDEE), which equals total daily EI during weight stability (20, 21). The DLW period at baseline occurred before participants began exercising. During the DLW period at follow-up, participants exercised at their prescribed dose. During both DLW periods, participants were weight stable (≤0.15 kg change in weight during the 2-week period). Change in EI by DLW was calculated with and without adjusting for change in resting metabolic rate (RMR). For participants who were weight stable or who gained weight during the six-month trial, follow-up TDEE was subtracted from baseline TDEE to quantify change in EI since any changes in RMR from weight gain are reflected in the TDEE value from DLW. For participants who lost weight during the intervention, this calculation fails to consider decreased basal metabolic requirements; therefore the difference between RMR from baseline to follow-up was added to the difference in TDEE for these participants to quantify change in EI during the intervention period.
In addition, at baseline and follow-up, validated laboratory-based FI tests were conducted at lunch and dinner. Following a standard breakfast between 0700 and 0800 consisting of a 190-kcal nutrition bar, participants returned to the center between 1100 and 1200 to complete their test lunch, which consisted of ad libitum sandwiches, potato chips, cookies, water, and choice of an artificially sweetened soda or tea and sugar-sweetened soda or tea. Five and a half hours after the start of their lunch, participants returned to the center again to complete their dinner meal, which consisted of a previously described 18-food item buffet meal (22), presented to the participants all at once within arm’s reach. At both test meals, participants were instructed to eat as much or as little of the presented food items as desired and to avoid distractions (e.g. phone use), focusing completely on the meal. FI testing at follow-up occurred at least 24 h after the last exercise session. We quantified FI at lunch and dinner by covertly weighing food provision and waste and combined EI (kcal) from both meals for the analyses presented in this paper.
Resting Metabolic Rate
We measured RMR with indirect calorimetry over 30 min after a 12-h overnight fast with Max II metabolic carts (AEI Technologies, Pittsburgh, PA, USA) at baseline and follow-up. Change in RMR was calculated as RMR at follow-up minus RMR at baseline. Calculations adjusted for change in body composition (i.e., lean mass measured with DXA), sex and, age did not differ meaningfully from the basic change scores; hence, the basic change scores are reported.
Physical Activity
SenseWear armbands (BodyMedia, Pittsburgh, PA, USA) measured the minutes per day spent in activities of different intensities, steps/day, and PAEE during a 2-week period at baseline and follow-up. In the MVPA-related analyses presented in this paper, only activities ≥3 metabolic equivalents (MET) are included (3), and congruent with the most recent PA guidelines for Americans (23), all MVPA was considered rather than only that accumulated in bouts of at least 10 minutes as recommended previously. The SenseWear software classifies any activity ≤3 MET as sedentary; hence, PAEE only included activities ≥3 MET (24). Participants were instructed to wear the armbands continuously and to only take them off during activities involving water. The SenseWear armbands detect and record wear time and only full days of data, defined as a wear time of ≥95% (equating to 22 h and 48 min or 1368 min), were included in the analyses. During the monitoring period at follow-up, participants exercised at their prescribed dose; therefore, PA data collected by the SenseWear armbands during these sessions was removed before analysis. To account for differing wear times between participants caused by varying durations of the exercise sessions and due to different non-wear times within the 22h and 48 min timeframe, the total number of minutes of daily activity was divided by the total daily wear time (min) and then extrapolated out to a 24 h day.
In addition to the PAEE estimates by the SenseWear armband, we calculated the gold standard of activity energy expenditure (AEE) based on the DLW-estimated TDEE (DLW-AEE = TDEE − [RMR + thermic effect of food (TEF)], which captures all PA-related EE. TEF was estimated as 10% of TDEE.
Questionnaires
Retrospective visual analog scales (VAS) assessed average ratings of appetite during the previous week (25) at baseline and follow-up. The Eating Inventory (26) was used to assess eating behavior, specifically restraint, disinhibition, and hunger at baseline and follow-up. Additional questionnaires included the Multifactorial Assessment of Eating Disorders Symptoms (MAEDS) (27), Food Preference Questionnaire (28), and Food Craving Inventory (29).
Statistical Analyses
The distribution of variables was verified using the Shapiro-Wilk test and by visual inspection of histograms and quantile-quantile plots of the residuals. The influence of outliers was estimated using studentized residuals and multicollinearity was assessed via the variance inflation factor. Exclusion of outliers (≥2 for all models) did not change the results meaningfully; therefore, the models including outliers are reported. Descriptive data are reported as mean and standard deviation (SD). We used multiple linear regression models to estimate the effect of SenseWear-assessed habitual MVPA levels (min/day), steps/day, PAEE, and DLW-estimated AEE at baseline on weight change (kg) and weight compensation (kg) as well as on changes in waist circumference (cm), fat mass (kg), EI (DLW (kcal/day)), EI during FI testing (kcal at a test lunch and test dinner combined), RMR (kcal/day), and habitual MVPA levels (min/day), steps/day, and PAEE, respectively. Covariates in the models were sex, exercise group, and baseline value of the respective outcome. Results of analyses that included age, ethnicity, and baseline BMI did not differ meaningfully; therefore, the models without these additional covariates are reported. Similarly, interaction terms for sex and exercise group were non-significant; therefore, results are reported without the interaction terms in the models. Pearson product-moment correlation analysis was used to assess the association between habitual MVPA levels and questionnaire-assessed eating behaviors at baseline. The analyses were conducted using SPSS Statistics for Windows version 25 (IBM Corp., Armonk, NY, USA) with the significance level set to 0.05 (two-sided).
RESULTS
Two participants were excluded from the analyses because they did not provide baseline accelerometer data. Baseline characteristics of all included 108 participants are shown in Table 1. Baseline characteristics of the control group (not included in main analyses) are provided in Table S1. At baseline, average wear time of the armbands was 1415.2 (SD 9.1) min/day equating to 98.3 (SD 0.6) % and at follow-up, average wear time (excluding study-related exercise sessions) was 1393.1 (SD 31.7) min/day or 97.8 (SD 2.1) %. Baseline habitual MVPA was 61.2 (SD 46.9) min/day on average with an average intensity of 3.7 (SD 0.2) MET (Table 1) and 99.2 (SD 0.2) % of all MVPA below 6 MET. Total habitual PA, measured as steps/day, was 6300 (SD 2301) at baseline. Total duration and intensity of daily habitual MVPA, habitual steps/day, and habitual PAEE (all outside of the structured exercise sessions) did not change significantly from baseline to follow-up (all P-values ≥ 0.094). Average self-chosen exercise intensity during the intervention was 6.9 (SD 1.0) MET with no significant difference between the 20 KKW group and the 8 KKW (P = 0.074). This average exercise intensity corresponds to vigorous PA (3).
Table 1.
Baseline characteristics of the 108 included participants.
| All (N=108) |
8 KKW (n=57) |
20 KKW (n=51) |
|
|---|---|---|---|
| Female, n (%) | 78 (72.2) | 42 (73.7) | 36 (70.6) |
| Ethnicity, n (%) | |||
| Caucasian, n (%) | 74 (68.5) | 37 (64.9) | 37 (72.5) |
| African American, n (%) | 32 (29.6) | 20 (35.1) | 12 (23.5) |
| Hispanic/Other, n (%) | 2 (1.9) | 0 (0.0) | 2 (4.0) |
| Mean (SD) | Mean (SD) | Mean (SD) | |
| Age (years) | 48.7 (11.6) | 48.3 (11.0) | 49.1 (12.4) |
| Height (cm) | 167.1 (8.2) | 167.2 (8.7) | 167.0 (7.6) |
| Weight (kg) | 87.8 (15.5) | 89.0 (16.0) | 86.5 (15.1) |
| Waist Circumference (cm) | 97.8 (12.0) 1 | 98.7 (12.1) 2 | 97.0 (11.9) |
| BMI (kg/m2) | 31.4 (4.6) | 31.8 (4.6) | 30.9 (4.5) |
| Fat Mass (kg) | 36.8 (9.8) | 37.3 (9.7) | 36.2 (9.9) |
| Energy Intake, DLW (kcal/day) | 2497.7 (462.5) | 2530.1 (438.8) | 2461.5 (489.4) |
| Energy Intake – Buffet (kcal at lunch and dinner combined) | 1795.5 (550.7) | 1820.1 (489.7) | 1768.1 (615.7) |
| Resting Metabolic Rate (kcal/day) | 1529.1 (297.1) 1 | 1525.8 (261.2) | 1532.8 (334.7) |
| MVPA (min/day) | 61.2 (46.9) | 63.9 (49.5) | 58.2 (44.0) |
| Average Intensity of MVPA (MET) | 3.7 (0.2) | 3.8 (0.2) | 3.7 (0.2) |
| Steps/day | 6300 (2301) | 6576 (2613) | 5992 (1870) |
| PAEE, SenseWear (kcal/day) | 336.7 (257.8) | 349.1 (257.6) | 322.9 (259.8) |
| AEE, DLW (kcal/day) | 717.5 (216.6) 1 | 749.3 (201.3) 2 | 682.6 (229.2) |
Data are mean (standard deviation) if not stated otherwise. ANOVA (continuous variables) and a Chi-square test (categorical variable) were used to test for baseline differences between the two groups. The 8 KKW and 20 KKW groups did not differ significantly in any of the baseline measures presented in the table.
data available in 107/108 participants
data available in 56/57 participants
Abbreviations: AEE, activity energy expenditure; BMI, body mass index; DLW, doubly labeled water; KKW, kilocalories per kilogram of body weight per week; MVPA, moderate-to-vigorous physical activity; MET, metabolic equivalent; PAEE, physical activity energy expenditure; SD, standard deviation.
Table 2 and Figure 1 (MVPA), Table 3 (steps/day), Table 4 (Sensewear PAEE), and Table 5 (DLW-estimated AEE) show the results of the multiple linear regression analyses. We found significant negative associations between baseline habitual MVPA levels and weight change (P=0.020; Figure 1A), weight compensation (P=0.043; Figure 1B), and change in DLW-estimated EI both with (P=0.001; Figure 1C) and without (P=0.001; not shown in Figure 1) adjustment for change in RMR. The analyses further showed significant negative associations between baseline habitual MVPA levels and changes in waist circumference (P=0.030), fat mass (P=0.025), and habitual MVPA levels (P<0.001; Figure 1D). While there is substantial variability in the data (see Figure 1), these results suggest that, on average, for every 15-minute decrease in habitual MVPA per day at baseline, participants lost 0.23 kg less weight, compensated 0.20 kg more, and increase DLW-estimated daily EI from to baseline to follow-up by 21.5 (adjusted for RMR 23.2) kcal/day.
Table 2.
Multiple linear regression analysis for the association between baseline habitual MVPA levels and changes in body weight, fat mass, energy intake, and MVPA levels.
| R2 | B | SE | β | P | |
|---|---|---|---|---|---|
| Weight Change (kg) | 0.124 | ||||
| Habitual MVPA at Baseline (min/day) | −0.015 | 0.006 | −0.275 | 0.020 | |
| Weight at Baseline (kg) | −0.033 | 0.019 | −0.197 | 0.095 | |
| Sex1 | 1.880 | 0.716 | 0.328 | 0.010 | |
| Group 2 | −1.442 | 0.483 | −0.280 | 0.004 | |
| Waist Circumference Change (cm) | 0.074 | ||||
| Habitual MVPA at Baseline (min/day) | −0.019 | 0.009 | −0.267 | 0.030 | |
| Waist Circumference at Baseline (cm) | −0.034 | 0.034 | −0.122 | 0.317 | |
| Sex 1 | 1.357 | 0.947 | 0.185 | 0.155 | |
| Group 2 | −1.325 | 0.638 | −0.201 | 0.040 | |
| Weight Compensation (kg) | 0.127 | ||||
| Habitual MVPA at Baseline (min/day) | −0.013 | 0.006 | −0.238 | 0.043 | |
| Weight at Baseline (kg) | −0.011 | 0.020 | −0.064 | 0.585 | |
| Sex 1 | 1.826 | 0.723 | 0.315 | 0.013 | |
| Group 2 | 1.049 | 0.488 | 0.201 | 0.034 | |
| Fat Mass Change (kg) | 0.153 | ||||
| Habitual MVPA at Baseline (min/day) | −0.014 | 0.006 | −0.257 | 0.025 | |
| Fat Mass at Baseline (kg) | −0.070 | 0.027 | −0.271 | 0.011 | |
| Sex 1 | 1.173 | 0.562 | 0.209 | 0.039 | |
| Group 2 | −1.478 | 0.462 | −0.293 | 0.002 | |
| Change in Energy Intake, DLW (kcal/day) | 0.220 | ||||
| Habitual MVPA at Baseline (min/day) | −1.546 | 0.442 | −0.336 | 0.001 | |
| Energy Intake, DLW at Baseline (kcal/day) | −0.206 | 0.054 | −0.443 | <0.001 | |
| Sex 1 | 182.400 | 57.888 | 0.381 | 0.002 | |
| Group 2 | −11.680 | 37.824 | −0.027 | 0.758 | |
| Change in Energy Intake, adj. DLW kcal/day)3 | 0.205 | ||||
| Habitual MVPA at Baseline (min/day) | −1.436 | 0.438 | −0.318 | 0.001 | |
| Energy Intake, DLW at Baseline (kcal/day) | −0.195 | 0.054 | −0.427 | <0.001 | |
| Sex 1 | 171.594 | 57.370 | 0.365 | 0.003 | |
| Group 2 | 3.388 | 37.486 | 0.008 | 0.928 | |
| Change Energy Intake – Buffet (kcal at lunch and dinner combined) | 0.158 | ||||
| Habitual MVPA at Baseline (min/day) | −0.938 | 0.951 | −0.098 | 0.326 | |
| Energy Intake – Buffet at Baseline (kcal at lunch and dinner combined) | −0.287 | 0.077 | −0.354 | <0.001 | |
| Sex 1 | 319.655 | 102.282 | 0.322 | 0.002 | |
| Group 2 | 21.191 | 80.993 | 0.024 | 0.794 | |
| Change in Resting Metabolic Rate, indirect calorimetry (kcal/day) | 0.067 | ||||
| Habitual MVPA at Baseline (min/day) | 0.210 | 0.641 | 0.036 | 0.744 | |
| Resting Metabolic Rate at Baseline (kcal/day) | −0.160 | 0.109 | −0.195 | 0.146 | |
| Sex 1 | −44.702 | 77.817 | −0.081 | 0.567 | |
| Group 2 | 41.109 | 50.933 | 0.083 | 0.422 | |
| Change in habitual MVPA (min/day) | 0.223 | ||||
| Habitual MVPA at Baseline (min/day) | −0.274 | 0.058 | −0.450 | <0.001 | |
| Sex 1 | −2.425 | 5.996 | −0.039 | 0.687 | |
| Group 2 | 7.963 | 4.992 | 0.141 | 0.114 |
Note: Bold font indicates statistical significance (P<0.05).
Independent variable in all models: habitual MVPA levels (min/day) at baseline.
female = 0, male = 1
8 KKW = 0, 20 KKW = 1
adjusted for change in resting metabolic rate
Abbreviations: DLW, doubly labeled water; KKW, kcal/kg body weight/week; MVPA, moderate-to-vigorous physical activity; B, unstandardized regression coefficient; β, standardized regression coefficient; SE,
standard error.
Figure 1:
Association between habitual moderate-to-vigorous physical activity (MVPA) at baseline and change in body weight (A), weight compensation (B), change in doubly labeled water (DLW)-measured energy intake, adjusted for change in resting metabolic rate (C), and change in MVPA (D). The regression line (solid line) in each panel represents the relationship for the fully adjusted model with 95% confidence intervals (dotted lines).
Table 3.
Multiple linear regression analysis for the association between habitual steps/day at baseline and changes in body weight, fat mass, energy intake, and steps/day.
| R2 | B | SE | β | P | |
|---|---|---|---|---|---|
| Weight Change (kg) | 0.091 | ||||
| Habitual PA at Baseline (steps/day) | −0.0001 | 0.0001 | −0.1233 | 0.211 | |
| Weight at Baseline (kg) | −0.0156 | 0.0177 | −0.0945 | 0.379 | |
| Sex 1 | 1.0394 | 0.6043 | 0.1812 | 0.088 | |
| Group 2 | −1.3705 | 0.4924 | −0.2664 | 0.006 | |
| Waist Circumference Change (cm) | 0.075 | ||||
| Habitual PA at Baseline (steps/day) | −0.0001 | 0.0001 | −0.2324 | 0.057 | |
| Waist Circumference at Baseline (cm) | −0.0223 | 0.0311 | −0.0807 | 0.476 | |
| Sex 1 | 0.5600 | 0.7905 | 0.0763 | 0.480 | |
| Group 2 | −1.3730 | 0.6409 | −0.2082 | 0.035 | |
| Weight Compensation (kg) | 0.098 | ||||
| Habitual PA at Baseline (steps/day) | −0.0001 | 0.0001 | −0.0818 | 0.404 | |
| Weight at Baseline (kg) | 0.0052 | 0.0179 | 0.0310 | 0.772 | |
| Sex 1 | 1.0611 | 0.6094 | 0.1828 | 0.085 | |
| Group 2 | 1.1322 | 0.4965 | 0.2174 | 0.025 | |
| Fat Mass Change (kg) | 0.124 | ||||
| Habitual PA at Baseline (steps/day) | −0.0001 | 0.0001 | −0.1223 | 0.212 | |
| Fat Mass at Baseline (kg) | −0.0505 | 0.0252 | −0.1954 | 0.048 | |
| Sex 1 | 0.6968 | 0.5256 | 0.1238 | 0.188 | |
| Group 2 | −1.4435 | 0.4715 | −0.2860 | 0.003 | |
| Change in Energy Intake, DLW (kcal/day) | 0.147 | ||||
| Habitual PA at Baseline (steps/day) | −0.0138 | 0.0089 | −0.1475 | 0.128 | |
| Energy Intake, DLW at Baseline (kcal/day) | −0.1923 | 0.0589 | −0.4127 | 0.001 | |
| Sex 1 | 112.1006 | 58.4187 | 0.2340 | 0.058 | |
| Group 2 | −7.8955 | 39.6422 | −0.0183 | 0.843 | |
| Change in Energy Intake, adj. DLW kcal/day)3 | 0.140 | ||||
| Habitual PA at Baseline (steps/day) | −0.0128 | 0.0088 | −0.1397 | 0.150 | |
| Energy Intake, DLW at Baseline (kcal/day) | −0.1823 | 0.0580 | −0.3985 | 0.002 | |
| Sex 1 | 106.2766 | 57.5923 | 0.2260 | 0.068 | |
| Group 2 | 6.8974 | 39.08 | 0.0163 | 0.860 | |
| Change Energy Intake – Buffet (kcal at lunch and dinner combined) | 0.173 | ||||
| Habitual PA at Baseline (steps/day) | −0.0298 | 0.0177 | −0.1537 | 0.095 | |
| Energy Intake – Buffet at Baseline (kcal at lunch and dinner combined) | −0.2771 | 0.0769 | −0.3416 | <0.001 | |
| Sex 1 | 288.0988 | 93.6300 | 0.2901 | 0.003 | |
| Group 2 | 10.5387 | 80.6567 | 0.0118 | 0.896 | |
| Change in Resting Metabolic Rate, indirect calorimetry (kcal/day) | 0.066 | ||||
| Habitual PA at Baseline (steps/day) | −0.0006 | 0.0109 | −0.0057 | 0.956 | |
| Resting Metabolic Rate at Baseline (kcal/day) | −0.1629 | 0.1086 | −0.1984 | 0.137 | |
| Sex 1 | −35.6318 | 72.7619 | −0.0647 | 0.626 | |
| Group 2 | 38.1756 | 51.0429 | 0.0774 | 0.456 | |
| Change in habitual PA (steps/day) | 0.173 | ||||
| Habitual PA at Baseline (steps/day) | −0.3591 | 0.0870 | −0.3823 | <0.001 | |
| Sex 1 | 74.8373 | 438.7446 | 0.0156 | 0.865 | |
| Group 2 | 500.2422 | 399.3502 | 0.1159 | 0.213 |
Note: Bold font indicates statistical significance (P<0.05).
Independent variable in all models: habitual PA levels (steps/day) at baseline.
female = 0, male = 1
8 KKW = 0, 20 KKW = 1
adjusted for change in resting metabolic rate
Abbreviations: DLW, doubly labeled water; KKW, kcal/kg body weight/week; PA, physical activity; B,
unstandardized regression coefficient; β, standardized regression coefficient; SE, standard error.
Table 4.
Multiple linear regression analysis for the association between baseline habitual PAEE as assessed by the SenseWear armband and changes in body weight, fat mass, energy intake, and SenseWear-assessed PAEE.
| R2 | B | SE | β | P | |
|---|---|---|---|---|---|
| Weight Change (kg) | 0.121 | ||||
| Habitual PAEE at Baseline (kcal/day) | −0.003 | 0.001 | −0.260 | 0.024 | |
| Weight at Baseline (kg) | −0.023 | 0.018 | −0.136 | 0.208 | |
| Sex 1 | 1.856 | 0.718 | 0.324 | 0.011 | |
| Group 2 | −1.400 | 0.481 | −0.272 | 0.004 | |
| Waist Circumference Change (cm) | 0.059 | ||||
| Habitual PAEE at Baseline (kcal/day) | −0.003 | 0.002 | −0.220 | 0.074 | |
| Waist Circumference at Baseline (cm) | −0.019 | 0.032 | −0.068 | 0.553 | |
| Sex 1 | 1.222 | 0.980 | 0.167 | 0.215 | |
| Group 2 | −1.265 | 0.641 | −0.192 | 0.051 | |
| Weight Compensation (kg) | 0.121 | ||||
| Habitual PAEE at Baseline (kcal/day) | −0.002 | 0.001 | −0.212 | 0.065 | |
| Weight at Baseline (kg) | −0.001 | 0.018 | −0.008 | 0.944 | |
| Sex 1 | 1.756 | 0.727 | 0.303 | 0.017 | |
| Group 2 | 1.093 | 0.487 | 0.210 | 0.027 | |
| Fat Mass Change (kg) | 0.156 | ||||
| Habitual PAEE at Baseline (kcal/day) | −0.003 | 0.001 | −0.261 | 0.021 | |
| Fat Mass at Baseline (kg) | −0.059 | 0.025 | −0.230 | 0.019 | |
| Sex 1 | 1.376 | 0.597 | 0.245 | 0.023 | |
| Group 2 | −1.462 | 0.460 | −0.290 | 0.002 | |
| Change in Energy Intake, DLW (kcal/day) | 0.228 | ||||
| Habitual PAEE at Baseline (kcal/day) | −0.322 | 0.088 | −0.385 | <0.001 | |
| Energy Intake, DLW at Baseline (kcal/day) | −0.169 | 0.055 | −0.362 | 0.003 | |
| Sex 1 | 186.132 | 57.668 | 0.389 | 0.002 | |
| Group 2 | −8.899 | 37.567 | −0.021 | 0.813 | |
| Change in Energy Intake, adj. DLW kcal/day)3 | 0.212 | ||||
| Habitual PAEE at Baseline (kcal/day) | −0.298 | 0.087 | −0.363 | 0.001 | |
| Energy Intake, DLW at Baseline (kcal/day) | −0.160 | 0.055 | −0.351 | 0.004 | |
| Sex 1 | 174.869 | 57.211 | 0.372 | 0.003 | |
| Group 2 | 5.993 | 37.269 | 0.014 | 0.873 | |
| Change Energy Intake – Buffet (kcal at lunch and dinner combined) | 0.164 | ||||
| Habitual PAEE at Baseline (kcal/day) | −0.246 | 0.184 | −0.142 | 0.184 | |
| Energy Intake – Buffet at Baseline (kcal at lunch and dinner combined) | −0.284 | 0.077 | −0.350 | <0.001 | |
| Sex 1 | 351.316 | 107.794 | 0.354 | 0.002 | |
| Group 2 | 19.233 | 80.663 | 0.022 | 0.812 | |
| Change in Resting Metabolic Rate, indirect calorimetry (kcal/day) | 0.066 | ||||
| Habitual PAEE at Baseline (kcal/day) | −0.025 | 0.127 | −0.023 | 0.846 | |
| Resting Metabolic Rate at Baseline (kcal/day) | −0.163 | 0.108 | −0.198 | 0.136 | |
| Sex 1 | −29.457 | 79.345 | −0.053 | 0.711 | |
| Group 2 | 37.075 | 51.013 | 0.075 | 0.469 | |
| Change in habitual PAEE (kcal/day) | 0.289 | ||||
| Habitual PAEE at Baseline (kcal/day) | −0.400 | 0.069 | −0.564 | <0.001 | |
| Sex 1 | 52.423 | 38.970 | 0.131 | 0.182 | |
| Group 2 | 43.443 | 30.461 | 0.121 | 0.157 |
Note: Bold font indicates statistical significance (P<0.05).
Independent variable in all models: habitual PAEE (kcal/day) at baseline as assessed by the SenseWear armband.
female = 0, male = 1
8 KKW = 0, 20 KKW = 1
adjusted for change in resting metabolic rate
Abbreviations: KKW, kcal/kg body weight/week; PAEE, physical activity energy expenditure; B, unstandardized regression coefficient; β, standardized regression coefficient; SE, standard error.
Table 5.
Multiple linear regression analysis for the association between baseline habitual AEE as assessed by DLW and changes in body weight, fat mass, and energy intake.
| R2 | B | SE | β | P | |
|---|---|---|---|---|---|
| Weight Change (kg) | 0.076 | ||||
| Habitual AEE at Baseline (kcal/day) | −0.001 | 0.001 | −0.041 | 0.709 | |
| Weight at Baseline (kg) | −0.009 | 0.018 | −0.053 | 0.630 | |
| Sex 1 | 1.009 | 0.620 | 0.177 | 0.107 | |
| Group 2 | −1.250 | 0.496 | −0.244 | 0.013 | |
| Waist Circumference Change (cm) | 0.027 | ||||
| Habitual AEE at Baseline (kcal/day) | 0.001 | 0.002 | 0.035 | 0.749 | |
| Waist Circumference at Baseline (cm) | 0.003 | 0.030 | 0.012 | 0.911 | |
| Sex 1 | 0.098 | 0.829 | 0.014 | 0.906 | |
| Group 2 | −0.999 | 0.654 | −0.153 | 0.130 | |
| Weight Compensation (kg) | 0.098 | ||||
| Habitual AEE at Baseline (kcal/day) | −0.001 | 0.001 | −0.013 | 0.906 | |
| Weight at Baseline (kg) | 0.009 | 0.018 | 0.053 | 0.626 | |
| Sex 1 | 1.031 | 0.624 | 0.178 | 0.101 | |
| Group 2 | 1.237 | 0.499 | 0.237 | 0.015 | |
| Fat Mass Change (kg) | 0.109 | ||||
| Habitual AEE at Baseline (kcal/day) | −0.001 | 0.001 | −0.019 | 0.854 | |
| Fat Mass at Baseline (kg) | −0.040 | 0.025 | −0.155 | 0.111 | |
| Sex 1 | 0.749 | 0.583 | 0.134 | 0.202 | |
| Group 2 | −1.324 | 0.479 | −0.263 | 0.007 | |
| Change in Energy Intake, DLW (kcal/day) | 0.171 | ||||
| Habitual AEE at Baseline (kcal/day) | −0.317 | 0.137 | −0.317 | 0.023 | |
| Energy Intake, DLW at Baseline (kcal/day) | −0.095 | 0.078 | −0.204 | 0.223 | |
| Sex 1 | 99.726 | 58.661 | 0.208 | 0.092 | |
| Group 2 | −13.740 | 39.562 | −0.032 | 0.729 | |
| Change in Energy Intake, adj. DLW kcal/day)3 | 0.167 | ||||
| Habitual AEE at Baseline (kcal/day) | −0.320 | 0.135 | −0.326 | 0.020 | |
| Energy Intake, DLW at Baseline (kcal/day) | −0.082 | 0.076 | −0.179 | 0.287 | |
| Sex 1 | 92.237 | 57.688 | 0.196 | 0.113 | |
| Group 2 | 0.181 | 38.906 | 0.001 | 0.996 | |
| Change Energy Intake – Buffet (kcal at lunch and dinner combined) | 0.149 | ||||
| Habitual AEE at Baseline (kcal/day) | 0.070 | 0.210 | 0.034 | 0.741 | |
| Energy Intake – Buffet at Baseline (kcal at lunch and dinner combined) | −0.295 | 0.079 | −0.366 | <0.001 | |
| Sex 1 | 264.367 | 101.080 | 0.267 | 0.010 | |
| Group 2 | 25.578 | 82.706 | 0.029 | 0.758 | |
| Change in Resting Metabolic Rate, indirect calorimetry (kcal/day) | 0.067 | ||||
| Habitual AEE at Baseline (kcal/day) | 0.043 | 0.128 | 0.038 | 0.736 | |
| Resting Metabolic Rate at Baseline (kcal/day) | −0.165 | 0.108 | −0.201 | 0.132 | |
| Sex 1 | −42.818 | 75.767 | −0.078 | 0.573 | |
| Group 2 | 42.681 | 51.771 | 0.087 | 0.412 |
Note: Bold font indicates statistical significance (P<0.05).
Independent variable in all models: habitual AEE (kcal/day) at baseline as assessed DLW (total daily energy expenditure − [resting metabolic rate + thermic effect of eating4]).
female = 0, male = 1
8 KKW = 0, 20 KKW = 1
adjusted for change in resting metabolic rate
10% of total daily energy expenditure
Abbreviations: AEE, activity energy expenditure; DLW, doubly labeled water; KKW, kcal/kg body weight/week; B, unstandardized regression coefficient; β, standardized regression coefficient; SE, standard error.
Compared to women, men lost 1.9 kg less weight, compensated 1.8 kg more, and increased DLW-estimated EI by 182.4 (adjusted for RMR 171.6) kcal/day (Table 2). Further, compared to participants in the 8 KKW group, participants in the 20 KKW group lost 1.4 kg more weight but showed 1 kg higher weight compensation (Table 2).
Baseline levels of habitual MVPA were significantly correlated with the disinhibition subscale of the Eating Inventory (r = −0.229; P = 0.018) and with the binge eating subscale of the MEADS (r = −0.230; P = 0.018). No other correlations between baseline levels of habitual MVPA and eating behavior-related constructs, as assessed by questionnaires, were significant.
Baseline PA levels measured as steps/day significantly predicted change in steps/day (β=−0.382; P<0.001); however, no associations between steps/day at baseline and change in any other of the outcome variables were significant (Table 3). Associations between average intensity (MET) of baseline habitual MVPA and all outcomes were non-significant (all P>0.1, data not shown).
Associations between baseline habitual PAEE and outcomes were similar to those of baseline habitual MVPA, albeit slightly attenuated, as indicated by the regression coefficients (Table 4). DLW-estimated AEE only significantly predicted change in DLW-estimated EI (Table 5); all other associations were non-significant (all P-values ≥ 0.709)
As described above, habitual MVPA, steps/day, and PAEE (all outside of structured exercise sessions) did not change significantly from baseline to follow-up on a group level. However, on an individual level, baseline habitual MVPA (Table 2, Figure 1D), steps/day (Table 3), and PAEE (Table 4) were significantly inversely associated with change in the respective measure.
The Tables S2-S5 show the results of the multiple linear regression analyses for the control group. For habitual MVPA (Table S2), steps/day (Table S3), and PAEE (Table S4), only change in each PA measure was significantly associated with the respective baseline value. A Fisher r-to-z transformation revealed that the correlation coefficients for habitual MVPA did not differ between exercisers and the control group (data not shown). For habitual PAEE, the difference between exercisers and control participants was significant with a markedly more pronounced association for the control participants.
DISCUSSION
To our knowledge, this is the first study to determine and compare the effect of prior habitual MVPA, steps/day, and PAEE on changes in weight, EI, RMR, and MVPA, steps/day, and PAEE, respectively, in response to a moderate-to-high dose aerobic exercise intervention. The results show that, on average, lower levels of habitual MVPA/PAEE at baseline are related to less weight loss and greater weight compensation during the exercise intervention, supporting our hypothesis. Importantly, lower levels of habitual MVPA/PAEE at baseline were also associated with greater increases in EI, which likely contributed to the lower weight compensation in those with higher baseline levels of habitual MVPA/PAEE, particularly since changes in RMR were not associated with baseline habitual MVPA/PAEE levels. Interestingly, we found substantial heterogeneity in the weight loss/compensation response, which likely influenced the results of the regression analysis. While many participants across all baseline MVPA levels successfully lost weight during the intervention, some participants with low MVPA at baseline actually gained weight whereas no-one with higher baseline MVPA gained weight.
In line with previous findings (30, 31), participants with lower habitual MVPA levels showed higher tendencies for disinhibition and binge eating at baseline, factors that may have influenced the greater increases in EI and subsequent greater weight compensation in response to the exercise intervention. This assumption is supported by previous findings showing that individuals with lower levels of measured MVPA have weaker appetite control and satiety response to food and thus an impaired regulation of energy balance compared to their more active counterparts (32-34). Consequently, in our study, participants with lower levels of habitual MVPA/PAEE at baseline may have had a more impaired regulation of energy balance than those with higher levels of habitual MVPA/PAEE, becoming (particularly) apparent with the onset of the exercise intervention. While participants were weight stable during the 2-week baseline accelerometer assessment, suggesting an adequately regulated energy balance during that period, the exercise intervention and subsequently the substantial increase in daily EE disrupted this balance. This disruption may have revealed the potentially impaired energy balance regulation in participants with lower baseline levels of habitual MVPA/PAEE, as the intervention-related increases in MVPA/PAEE (i.e. structured exercise sessions) were met by increases in EI, leading to the observed weight compensation. Participants with higher MVPA/PAEE levels at baseline might have already experienced this compensatory effort before the start of the intervention, explaining, at least partially, the observed results. In addition to being driven by homeostatic mechanisms such as the aforementioned changes in appetite and satiety, the observed increases in EI may also be related to hedonic processes such as food reward behaviors (35).
It is noteworthy that, while habitual MVPA/PAEE levels did not change on a group level, on an individual level, these parameters were significantly inversely associated with change in the respective measure, indicating the substitution of habitual PA with prescribed PA (i.e. structured exercise session) in some participants (36, 37). As shown by the results of a Fisher r-to-z transformation, however, the correlation coefficients for habitual MVPA did not differ between exercisers and the control group, suggesting that any substitution was likely not caused by the structured exercise sessions but instead is more likely due to regression to the mean. For habitual PAEE, the difference between exercisers and control participants was significant with a substantially more pronounced association for the control participants, suggesting that the structured exercise sessions actually protected against decreases in habitual PAEE. It is further noteworthy that participants with greater prior habitual MVPA/PAEE remained more active compared to those with lower levels (B=−0.273 (MVPA) and B=−0.400 (PAEE)). Therefore, considering the magnitude of the change in habitual MVPA/PAEE levels and more importantly the opposite directionality compared to weight change, it is unlikely that the decrease in habitual MVPA/PAEE affected participants’ weight compensation. Rather, higher absolute levels of MVPA/PAEE at follow-up, along with the reduced increase in EI during the intervention, contributed to the lower weight compensation in those who were more active at baseline.
The identification of baseline habitual MVPA/PAEE levels as predictors of weight loss, weight compensation, and changes in EI in this study may have important ramifications for future exercise interventions targeting weight loss. Less than expected weight loss from exercise likely leads to frustration and possibly causes discontinuation of the newly started exercise regimen due to the perceived lack of benefit. Assessing prior habitual PA levels may help determine when the exercise prescription should be combined with a concomitant lifestyle, dietary, or possibly pharmacological intervention to counteract weight compensation and increase the weight loss intervention-related health benefits.
While habitual MVPA and PAEE predicted our outcomes quite comparably, daily step counts at baseline did not have the same predictive value with regard to weight loss, weight compensation, or EI during the intervention as habitual MVPA/PAEE. The better predictive value of PAEE compared to steps/day was expected due to the fact that PAEE is directly related to the energy balance model. The better predictive value of habitual MVPA compared to steps/day is likely due to the fact that MVPA includes an intensity component whereas steps/day does not. Therefore, to identify individuals with a higher risk for exercise-induced weight compensation, baseline levels of habitual MVPA or PAEE should be considered. It should be noted that AEE, as estimated by DLW, did not predict most of our outcomes, with a substantial discrepancy compared to the associations from SenseWear-assessed PAEE. This suggests that the intensity component included in PAEE (and MVPA) makes these parameters better predictors with regard to our outcomes. Therefore, total AEE seems to be less important than EE at an intensity ≥3 MET, which is different from Mayer’s original suggestion (11, 12). The use of MVPA as a predictor offers the advantage of being accurately assessable via most current accelerometers and accelerometer data being more straightforward compared to PAEE-data such as that of the SenseWear armband, which is based on a complex pattern-recognition algorithm consisting of heat flux, skin temperature, near-body ambient temperature, and galvanic skin response in addition to the accelerometer-recorded activity counts.
The present study has several strengths. E-MECHANIC was a large randomized controlled trial, in which exercise dose was strictly monitored and supervised. Habitual PA was measured with validated accelerometers that allow an estimation of the intensity and EE of habitual PA. Additionally, energy expenditure/intake (via DLW) and RMR (via indirect calorimetry) were measured with the gold-standard methods to comprehensively assess all aspects of energy balance. The assessment of EI via validated laboratory-based FI tests and via DLW over 2 weeks particularly are a major strength as self-reported EI, which is still commonly used in many trials today, has been found to be fundamentally inaccurate (38, 39). A limitation of this analysis is that, although PA assessment at follow-up was performed while participants were still exercising at their prescribed dose, we did not measure habitual MVPA, steps/day, and PAEE continuously throughout the intervention period and have thus no record of the effect of the exercise training on these outcomes over the course of the intervention.
In conclusion, taking into account the substantial variability in the data, our results show that habitual MVPA/PAEE levels before engaging in a structured exercise intervention predict weight loss, weight compensation, and changes in EI during that intervention. Importantly, habitual MVPA/PAEE (≥3 MET) at baseline showed a superior predictive value with regard to these outcome measures compared to steps/day and total AEE, suggesting that time spent and energy expended during MVPA rather than total activity-related EE before an exercise intervention targeting weight loss are protective against weight compensation. In this regard, habitual MVPA may be the preferable parameter compared to PAEE due to its easier, more economical and likely more accurate assessment. Future studies are needed to elucidate the observed heterogenic relationship between baseline habitual MVPA/PAEE levels and weight loss/compensation to develop individualized strategies to mitigate the detrimental compensatory increase in EI in response to an exercise-induced increase in EE in some individuals.
Supplementary Material
What is already known about this subject?
Exercise is recommended for weight management.
Exercise-induced weight loss often is less than expected based on measured energy expenditure.
This is called weight compensation and results primarily from increased energy intake.
What are the new findings in your manuscript?
MVPA and PAEE (≥3 MET) levels prior to engaging in a moderate-to-high dose aerobic exercise intervention predict weight loss, weight compensation, and changes in energy intake during the intervention.
Prior MVPA and PAEE have a superior predictive value compared to steps/day and total activity-related EE, as estimated by DLW, regarding these outcomes.
How might your results change the direction of research or the focus of clinical practice?
Further research is needed to understand why participants with lower baseline habitual MVPA and PAEE levels lose less weight from structured exercise, show higher weight compensation, and increase their energy intake more than those who are more active at baseline to develop strategies to mitigate this detrimental effect.
Acknowledgments:
The authors would like to thank participants for their time and commitment to the study.
Access to individual de-identified data that underlie the results reported in this article is possible via Pennington Biomedical Research Center’s Data Sharing policy.
Funding:
Research reported in this publication was supported by the National Institutes of Health via the National Heart, Lung, and Blood Institute with the Multiple Principal Investigators being C. Martin and T. Church (R01 HL102166); NORC Center Grant P30 DK072476, entitled “Nutritional Programming: Environmental and Molecular Interactions” sponsored by NIDDK; and the National Institute of General Medical Sciences, which funds the Louisiana Clinical and Translational Science Center (U54 GM104940). C. Höchsmann is funded by an NIH NIDDK National Research Service Award (T32 DK064584).
Footnotes
Competing Interests:
CKM gives lectures and presentations on the topic to lay and educational groups including the Academy of Nutrition and Dietetics. The authors report no other competing interests related to this study.
Trial Registration: ClinicalTrials.gov ID: NCT01264406
REFERENCES
- 1.NCD Risk Factor Collaboration (NCD-RisC). Trends in adult body-mass index in 200 countries from 1975 to 2014: a pooled analysis of 1698 population-based measurement studies with 19·2 million participants. Lancet Lond Engl 2016;387:1377–1396. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Heymsfield SB, Wadden TA. Mechanisms, Pathophysiology, and Management of Obesity. N Engl J Med 2017;376:254–266. [DOI] [PubMed] [Google Scholar]
- 3.Garber CE, Blissmer B, Deschenes MR, et al. American College of Sports Medicine position stand. Quantity and quality of exercise for developing and maintaining cardiorespiratory, musculoskeletal, and neuromotor fitness in apparently healthy adults: guidance for prescribing exercise. Med Sci Sports Exerc 2011;43:1334–1359. [DOI] [PubMed] [Google Scholar]
- 4.Swift DL, McGee JE, Earnest CP, Carlisle E, Nygard M, Johannsen NM. The Effects of Exercise and Physical Activity on Weight Loss and Maintenance. Prog Cardiovasc Dis 2018;61:206–213. [DOI] [PubMed] [Google Scholar]
- 5.Church TS, Martin CK, Thompson AM, Earnest CP, Mikus CR, Blair SN. Changes in Weight, Waist Circumference and Compensatory Responses with Different Doses of Exercise among Sedentary, Overweight Postmenopausal Women. PLoS ONE 2009;4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Martin CK, Johnson WD, Myers CA, et al. Effect of different doses of supervised exercise on food intake, metabolism, and non-exercise physical activity: The E-MECHANIC randomized controlled trial. Am J Clin Nutr 2019;110:583–592. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Thomas DM, Bouchard C, Church T, et al. Why do individuals not lose more weight from an exercise intervention at a defined dose? An energy balance analysis. Obes Rev Off J Int Assoc Study Obes 2012;13:835–847. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Blundell JE, Stubbs RJ, Hughes DA, Whybrow S, King NA. Cross talk between physical activity and appetite control: does physical activity stimulate appetite? Proc Nutr Soc 2003;62:651–661. [DOI] [PubMed] [Google Scholar]
- 9.King NA, Hopkins M, Caudwell P, Stubbs RJ, Blundell JE. Individual variability following 12 weeks of supervised exercise: identification and characterization of compensation for exercise-induced weight loss. Int J Obes 2005 2008;32:177–184. [DOI] [PubMed] [Google Scholar]
- 10.Westerterp KR. Exercise for weight loss. Am J Clin Nutr 2019;110:540–541. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Mayer J, Roy P, Mitra KP. Relation between caloric intake, body weight, and physical work: studies in an industrial male population in West Bengal. Am J Clin Nutr 1956;4:169–175. [DOI] [PubMed] [Google Scholar]
- 12.Church T, Martin CK. The Obesity Epidemic: A Consequence of Reduced Energy Expenditure and the Uncoupling of Energy Intake? Obes Silver Spring Md 2018;26:14–16. [DOI] [PubMed] [Google Scholar]
- 13.Myers CA, Johnson WD, Earnest CP, et al. Examination of mechanisms (E-MECHANIC) of exercise-induced weight compensation: study protocol for a randomized controlled trial. Trials 2014;15:212. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Tudor-Locke C, Leonardi C, Johnson WD, Katzmarzyk PT, Church TS. Accelerometer steps/day translation of moderate-to-vigorous activity. Prev Med 2011;53:31–33. [DOI] [PubMed] [Google Scholar]
- 15.West Suitor C, Kraak V, National Research Council. Adequacy of Evidence for Physical Activity Guidelines Development: Workshop Summary. Washington, DC: The National Academic Press; 2007. [Google Scholar]
- 16.American College of Sports Medicine. ACSM’s guidelines for exercise testing and prescription. 9th Ed. Philadelphia, PA: Wolters Kluwer/Lippincott Williams & Wilkins Health; 2014. [Google Scholar]
- 17.Thomas DM, Martin CK, Lettieri S, et al. Can a weight loss of one pound a week be achieved with a 3500-kcal deficit? Commentary on a commonly accepted rule. Int J Obes 2005 2013;37:1611–1613. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Heymsfield SB, Thomas D, Nguyen AM, et al. Voluntary weight loss: systematic review of early phase body composition changes. Obes Rev Off J Int Assoc Study Obes 2011;12:e348–361. [DOI] [PubMed] [Google Scholar]
- 19.Thomas DM, Ciesla A, Levine JA, Stevens JG, Martin CK. A mathematical model of weight change with adaptation. Math Biosci Eng MBE 2009;6:873–887. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Livingstone MBE, Black AE. Markers of the Validity of Reported Energy Intake. J Nutr 2003;133:895S–920S. [DOI] [PubMed] [Google Scholar]
- 21.Schoeller DA. How accurate is self-reported dietary energy intake? Nutr Rev 1990;48:373–379. [DOI] [PubMed] [Google Scholar]
- 22.Martin CK, Coulon SM, Markward N, Greenway FL, Anton SD. Association between energy intake and viewing television, distractibility, and memory for advertisements12345. Am J Clin Nutr 2009;89:37–44. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Piercy KL, Troiano RP, Ballard RM, et al. The Physical Activity Guidelines for Americans. JAMA 2018;320:2020–2028. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Jones V, Bults R, de Wijk R, Widya I, Batista R, Hermens H. Experience with Using the Sensewear BMS Sensor System in the Context of a Health and Wellbeing Application. Int J Telemed Appl 2011;2011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Womble LG, Wadden TA, Chandler JM, Martin AR. Agreement between weekly vs. daily assessment of appetite. Appetite 2003;40:131–135. [DOI] [PubMed] [Google Scholar]
- 26.Stunkard AJ, Messick S. Eating Inventory Manual (The Psychological Corporation). San Antonio (TX): Harcourt Brace; 1988. [Google Scholar]
- 27.Anderson DA, Williamson DA, Duchmann EG, Gleaves DH, Barbin JM. Development and validation of a multifactorial treatment outcome measure for eating disorders. Assessment 1999;6:7–20. [DOI] [PubMed] [Google Scholar]
- 28.Geiselman PJ, Anderson AM, Dowdy ML, West DB, Redmann SM, Smith SR. Reliability and validity of a macronutrient self-selection paradigm and a food preference questionnaire. Physiol Behav 1998;63:919–928. [DOI] [PubMed] [Google Scholar]
- 29.White MA, Whisenhunt BL, Williamson DA, Greenway FL, Netemeyer RG. Development and validation of the food-craving inventory. Obes Res 2002;10:107–114. [DOI] [PubMed] [Google Scholar]
- 30.Shook RP, Hand GA, Drenowatz C, et al. Low levels of physical activity are associated with dysregulation of energy intake and fat mass gain over 1 year. Am J Clin Nutr 2015;102:1332–1338. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Myers A, Gibbons C, Finlayson G, Blundell J. Associations among sedentary and active behaviours, body fat and appetite dysregulation: investigating the myth of physical inactivity and obesity. Br J Sports Med 2017;51:1540–1544. [DOI] [PubMed] [Google Scholar]
- 32.Beaulieu K, Hopkins M, Long C, Blundell J, Finlayson G. High Habitual Physical Activity Improves Acute Energy Compensation in Nonobese Adults. Med Sci Sports Exerc 2017;49:2268. [DOI] [PubMed] [Google Scholar]
- 33.Beaulieu K, Hopkins M, Blundell J, Finlayson G. Homeostatic and non-homeostatic appetite control along the spectrum of physical activity levels: An updated perspective. Physiol Behav 2018;192:23–29. [DOI] [PubMed] [Google Scholar]
- 34.Beaulieu K, Hopkins M, Blundell J, Finlayson G. Does Habitual Physical Activity Increase the Sensitivity of the Appetite Control System? A Systematic Review. Sports Med Auckl Nz 2016;46:1897–1919. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Dalton M, Finlayson G, Esdaile E, King N. Appetite, Satiety, and Food Reward in Obese Individuals: A Behavioral Phenotype Approach. Curr Nutr Rep 2013;2:207–215. [Google Scholar]
- 36.Thompson D, Peacock OJ, Betts JA. Substitution and compensation Erode the energy deficit from exercise interventions. Med Sci Sports Exerc 2014;46:423. [DOI] [PubMed] [Google Scholar]
- 37.Thompson D, Karpe F, Lafontan M, Frayn K. Physical activity and exercise in the regulation of human adipose tissue physiology. Physiol Rev 2012;92:157–191. [DOI] [PubMed] [Google Scholar]
- 38.Dhurandhar NV, Schoeller D, Brown AW, et al. Energy balance measurement: when something is not better than nothing. Int J Obes 2005 2015;39:1109–1113. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Dorling J, Broom DR, Burns SF, et al. Acute and Chronic Effects of Exercise on Appetite, Energy Intake, and Appetite-Related Hormones: The Modulating Effect of Adiposity, Sex, and Habitual Physical Activity. Nutrients 2018;10. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.