High energy expenditure masks low physical activity in obesity

Abstract

Objective

To investigate energy expenditure in lean and obese individuals, focusing particularly on physical activity and severely obese individuals.

Design

Total daily energy expenditure (TDEE) was assessed using doubly labeled water, resting metabolic rate (RMR) by indirect calorimetry, activity EE (AEE) by difference and time spent in physical activity by multisensor activity monitors.

Subjects

177 lean, Class I and severely obese individuals (age 31–56, BMI 20–64 kg/m²).

Results

All components of EE were elevated in obese individuals. For example, TDEE was 2404±95 kcal/d in lean and 3244 ± 48 kcal/d in Class III obese. After appropriate adjustment, RMR was similar in all groups. Analysis of AEE by body weight and obesity class indicated a lower AEE in the obese. Confirming lower physical activity, obese individuals spent less time engaged in moderate-to-vigorous physical activity (2.7±1.3, 1.8±0.6, 2.0±1.4 and 1.2±1.0 hr/d in lean, Class I, Class II and Class III), and more time in sedentary behaviors.

Conclusions

There was no indication of metabolic efficiency in even the severely obese, as adjusted RMR was similar across all groups. The higher AEE observed in the obese is consistent with a higher cost of activities due to higher body weight. However, the magnitude of the higher AEE (20 to 25% higher in obese) is lower than expected (weight approximately 100% higher in Class III). Confirming a lower volume of physical activity in the obese, the total time spent in moderate-to-vigorous physical activity and average daily MET level were lower with increasing obesity. These findings demonstrate that high body weight in obesity leads to a high TDEE and AEE which masks the fact that they are less physically active, which can be influenced by duration or intensity of activity, than lean individuals.

Introduction

Although the relationship between energy expenditure and obesity has been studied for many years, the role that individual components of energy expenditure, particularly AEE, play in obesity remains unclear ¹. Furthermore, there is a dearth of information regarding the contribution of energy expenditure to severe obesity (BMI > 35), which has increased faster than overall obesity during the past three decades ^2–3. This is particularly troubling because of the increased prevalence of co-morbidities among the severely obese ^4–5.

While a few studies suggest that a low energy expenditure is related to subsequent weight gain ^6–10, TDEE has been shown be to be higher in obese, compared to lean individuals ^11–14. Most cross sectional studies have demonstrated that RMR is greater in the obese, but similar to that observed in lean individuals when appropriately adjusting for differences in body composition ^15–17. Due to the magnitude and variability of physical activity, it is a likely component of energy expenditure involved in the etiology of obesity ^18–20. However, the that physical activity plays in weight control is not clear ¹.

There are few cross sectional studies of energy expenditure that include severely obese individuals. In a study limited to extremely obese women, high total energy expenditure (3415 kcal/d) was observed ²¹. In addition, energy expenditure was significantly higher (3057, 3511 and 3845 kcal/d) with increasing tertiles of BMI (40.8±7, 49.5±1.0 and 60.3±2.2). In a cross sectional study ²², RMR was found to be significantly higher in morbidly obese compared to lean controls (1990±86 vs 1407±52; p<0.0001). These results suggest that severely obese individuals have higher energy expenditure, and that these individuals must consume very high levels of energy intake to maintain their excess weight.

Several methodological issues have hampered the elucidation of the relationship between physical activity and obesity. With the advent of the doubly labeled water (DLW) method, when combined with measurement of RMR, accurate assessments of free living AEE have been achieved. However, the DLW method does not provide any information about activity patterns or time spent in physical activity.

Therefore, to obtain a complete picture of physical activity, we combined multiple methods to assess TDEE, AEE and activity patterns in individuals covering a wide spectrum of obesity form lean, Class I, Class II and Class III obese individuals.

Subjects and Methods

Design

Energy expenditure and physical activity were assessed in lean (BMI 18.5 to 24.9 kg•m⁻²), Class I (BMI 30.0 to 34.9 kg•m⁻²), Class II (BMI of 35 to 39.9 kg•m⁻²) and Class III (BMI ≥ 40.0 kg•m⁻²) obese individuals (Table 1). Participants were between the ages of 30 and 55 years. Most participants (86%) were female. The study was reviewed and approved by the human ethics committee of the University of Pittsburgh.

Table 1.

Subject Characteristics^*

	lean	Class I	Class II	Class III	P
n	25	22	32	98
Age, y	44.0 ± 6.7	45.4 ± 6.6	46.4 ± 6.1	46.9 ± 6.5	0.22
Wt, kg	63.1 ± 7.8a	88.8 ± 11.0b	102.7 ± 9.1c	124.4 ± 16.3d	0.0001
Ht, cm	167.0 ± 8.7	165.6 ± 8.7	165.2 ± 6.5	165.1 ± 8.0	0.77
BMI, kg/m2	22.6 ± 1.6a	32.3 ± 1.3b	37.6 ± 1.3c	45.6 ± 4.7d	0.0001
Waist, cm	77.0 ± 6.7a	100.5 ± 10.5b	113.5 ± 8.2c	126.2 ± 11.6d	0.0001
FFM, kg	44.6 ± 8.8a	50.3 ± 10.9b	53.1 ± 7.4b	59.2 ± 9.3c	0.0001
Fat, g	18.1 ± 5.5a	37.5 ± 6.1b	48.4 ± 5.4c	63.7 ± 10.6d	0.0001

^*Values are means ± SD. Means with different superscripts are significantly differently.

Body weight, height and waist circumference were measured using standard protocols. Body composition was determined by either dual energy X-Ray absorptiometry (DXA) or by air displacement plethysmography in 24 subjects exceeding the weight capacity of the DXA scanner (>136 kg), and in 1 control subject who refused to be measured by DXA.

Total daily energy expenditure

TDEE was assessed using doubly labeled water ²³. After baseline samples were collected, subjects drank a mixture containing 2 g/kg total body water (TBW) of 10% H₂¹⁸O and 0.12 g/kg TBW of 99+% ²H₂O. Post-dose urine samples were collected 4.5 and 6 hr later, after urine voids at 1.5 and 3 hours were discarded. Two morning urine samples (not first void of the day) were collected 9–13 days after dosing. Urine samples were analyzed for H₂¹⁸O (Finnigan GasBench II) and ²H₂O (Finnigan H/D device) abundances by isotope ratio mass spectrometry using a Thermo Finnigan Delta PlusXP. The ²H and ¹⁸O isotope elimination rates (k_H and k_O) were calculated using linear regression. Total body water was determined from the extrapolated ¹⁸ oxygen enrichment at time zero. The rate of CO₂ (rCO₂) production was calculated using the equation of Schoeller et al. ²⁴ as modified ²⁵: rCO₂ (moles/d) = (N/2.078) (1.007 k_O − 1.041 k_H) − 0.0246 r_GF, where N is total body water calculated from N_O/1.007; and r_GF is the rate of fractionated gaseous (evaporative) water loss, estimated as 1.05N (1.007k_O − 1.041k_H). Total energy expenditure was calculated by multiplying rCO₂ by the energy equivalent of CO₂ at an assumed RQ of 0.86.

Resting Metabolic Rate

RMR was measured with a metabolic cart system (ParvoMedics Inc., TrueOne® 2400 Metabolic Measurement System with Canopy, Sandy, Utah). After an overnight fast (10 h minimum) participants lay down quietly for 30 min (room temperature 22–23 °C) before having the transparent hood placed over their head and neck for a 30 minute measurement of RMR. Metabolic rate was calculated using the Weir equation.²⁶ The first 5 minutes were eliminated, as were outlier measurements during the final 25 minutes not within 5% of the mean RMR.

Physical Activity

Energy expended in physical activity was calculated by subtracting measured RMR and estimated thermic effect of food from total energy expenditure: AEE = TDEE − [RMR + 0.10 (TDEE)].

Multi-sensor physical activity monitors (Sensewear Pro3, BodyMedia, Pittsburgh, PA) were worn during the DLW period. Data were analyzed using SenseWear Pro armband software version 5.1. Parameters examined include time spent in moderate to vigorous physical activity (> 3 METS), time spent in sedentary (0–3 METS), moderate (3–6 METS) vigorous (6–9 METS) and very vigorous (>9 METS) activity, hours of sleep, average daily MET level, and measured steps/day. Only days in which the monitors were worn greater than 85% of the day (mean ± SD, 96±2%) were included in the analyses (average = 8.05±1.68 days).

Statistics

Subject characteristics and components of EE were compared using general linear model analysis of variance (GLM: SAS release 9.2 for Windows; SAS Institute Inc, Cary, NC). Data are presented as means ± SD unless indicated otherwise. Various parameters were used in the analysis of variance models of energy expenditure as covariates to adjust for differences in body composition. To compare the relationship between energy expenditure and body weight or FFM between groups, simultaneous tests of slopes and intercepts were conducted using ANOVA. Post hoc tests for differences in group means were accomplished using Fisher’s Least Significant Difference test. Stepwise variable selection regression was used to develop a models to explain the variance in energy expenditure. Race was coded as 0 for Caucasian and 1 for African American, and sex coded was as 1 for male and 2 for female. Relative weight status was coded as 1–4 for lean through Class III obese.

Results

By design, BMI, body weight and body composition of the four groups were significantly different (Table 1). When comparing unadjusted energy expenditure across the 4 groups, all energy expenditure components were higher in the obese compared to lean individuals (Figure 1; Table 2). While TDEE and RMR increased with increasing obesity, AEE did not.

Figure 1.

Energy Expenditure. Data are mean ± SE. AEE: Class I, II and III > lean. RMR: all groups significantly different except for the Class I to Class II comparison. TDEE: all obese significantly different except for the Class I to Class II comparison.

Table 2.

Energy expenditure components^*

	lean	CLI	CLII	CLIII
RMR, kcal/d
Unadjusted	1328 ± 55a	1563 ± 58b	1654 ± 48b	1870 ± 28c
FFM	1571 ± 37a	1672 ± 37b	1696 ± 30b	1770 ± 18c
FFM & FM	1698 ± 59	1724 ± 41	1706 ± 30	1723 ± 25
AEE, kcal/d*
Unadjusted	836 ± 58a	1008 ± 62b	1062 ± 52b	1049 ± 29b
FFM	957 ± 59	1062 ± 59	1083 ± 48	1000 ± 29
Weight	1036 ± 91	1091 ± 68	1083 ± 51	973 ± 40
AEE/kg body wt	13.1 ± 0.5a	11.3 ± 0.6b	10.3 ± 0.5b	8.5 ± 0.3c
TDEE, kcal/d
Unadjusted	2404 ± 95a	2857 ± 102b	3017 ± 84b	3244 ± 48c
FFM	2808 ± 69a	3037 ± 69b	3088 ± 57b	3077 ± 34b
Weight	3306 ± 123a	3234 ± 92a	3113 ± 69a	2898 ± 54b

^*Values are LS Means ± SEM, unadjusted and adjusted (adjustments indicated in first column) for various covariates or EE Means/body weight ± SD.

RMR was significantly different between all groups except for the Class I to Class II comparison. The equation relating RMR to FFM in all participants (RMR (kcal/d) = 26.8 × FFM + 245, r²=0.71) is similar to other published equations in lean and obese individuals ²⁷. Even after adjusting for FFM, RMR was significantly higher in all obese groups compared to lean (Figure 2A and Table 2). However, when adjusting for body fat and FFM, there were no longer differences between the groups (Table 2). The best prediction equation (r² = 0.82, p<0.0001 with each parameter p<0.0002) for RMR (kcal/d) was as follows:

$$\text{RMR}=23.3\times \text{FFM}(\text{kg})+4.9\times \text{body}\text{fat}(\text{kg})-154\times \text{race}-6.4\times \text{age}+525$$

Figure 2.

Energy expenditure components by obesity class. To aid in visual inspection, the regression line for the lean group is solid, while those for Class I, Class II, and Class III are single, double and triple dashed. A) RMR vs. FFM: RMR = 9.6 × FFM + 899, R²=0.32 for Lean; RMR = 20.4 × FFM + 535, R²=0.73 for Class I; RMR = 19.3 × FFM + 626, R²=0.46 for Class II; and RMR = 28.2 × FFM + 201, R²=0.69 for Class III; B) AEE vs. body weight by obesity group: AEE = 25.0 × weight − 743, R²=0.42 for Lean; AEE = 12.8 × weight − 125, R²=0.32 for Class I; AEE = 10.3 × weight + 8, R²=0.14 for Class II; and AEE = 2.0 × weight + 805, R²=0.01 for Class III; C) TDEE vs. body weight: TDEE = 41.6 × weight − 223, R²=0.58 for Lean; TDEE = 35.0 × weight − 257, R²=0.62 for Class I; TDEE = 28.6 × weight+ 77, R²=0.48 for Class II; and TDEE = 28.2 × weight+ 201, R²=0.29 for Class III.

Similar to several published equations ^{23, 28–32}, in addition to FFM, body fat, race and age were significant parameters in the prediction equation, while sex was not.

All obese groups had a higher absolute AEE than observed in the lean group (Table 2). No significant differences in AEE were observed between groups when adjusting for FFM or body weight. There was a highly significant, but weak relationship between AEE and body weight in the entire cohort (r²=0.085, p<0.0001):

$$\text{AEE}=3.37\times \text{body}\text{weight}+654$$

When this relationship was examined by obesity group, the relationship became much stronger, particularly for the lean group (r²=0.32), but declining with increasing obesity. In addition, there were significant (p<0.005) differences in the relationship between AEE and body weight between the 4 groups (Figure 2B). This figure demonstrates shallower slopes (coefficient decreases from 25.0 for lean to 2.0 kcal/kg body weight for Class III) and increasing intercepts when going from lean to Class III obese (see Figure 2 legend for regression equations). This relationship was significantly different between the lean group and the Class II (p<0.04) and Class III (p<0.01) groups, but did not quite reach significance in the Class I (p=0.062) group. Figure 2B shows that there was no relationship between AEE and body weight in the Class III group (r²=0.01). When normalizing AEE by dividing by body weight, AEE was lower in all obese groups, and lowest in the Class III group (Table 2).

The lower AEE observed in the obese compared to lean subjects in relation to their higher body weight could be explained by data obtained from the multisensor activity monitors. Obese individuals spent less time in moderate to vigorous activities (>3 METs) and more time in sedentary behaviors, leading to a significantly lower daily MET level (Table 3). Obese individuals spent less time in each intensity activity compared to lean. The lower time spent in activities and increased time in sedentary activities was most apparent in the Class III participants. The average daily number of steps accumulated tended to be lower with increasing obesity, and was significantly lower in the Class III obese group. There was no difference in hours of sleep between the groups.

Table 3.

Physical Activity Monitor Data^*

	lean	Class I	Class II	Class III	p
Activity duration, h/d	2.7 ± 1.3a	1.8 ± 0.6b	2.0 ± 1.4b	1.2 ± 1.0c	0.0001
MOD, h/d	2.5 ± 1.2a	1.7 ± 0.6b	1.9 ± 1.0b	1.2 ± 0.9c	0.0001
VIG, h/d	0.24 ± 0.21a	0.14 ± 0.11bc	0.15 ± 0.17b	0.08 ± 0.13c	0.0001
VVIG, h/d	0.06 ± 0.12a	0.04 ± 0.05ab	0.03 ± 0.05b	0.01 ± 0.04c	0.0001
SED, h/d	20.7 ± 1.5a	21.5 ± 0.8bc	21.1 ± 1.2ab	21.8 ± 1.1c	0.0001
METs	1.6 ± 0.3a	1.3 ± 0.1b	1.3 ± 0.2b	1.1 ± 0.2c	0.0001
Steps	9954 ± 3487a	9457 ± 2585a	8767 ± 2936a	6617 ± 2673b	0.0001
Sleep, h/d	6.3 ± 0.8	6.3 ± 1.1	6.1 ± 1.2	6.0 ± 1.2	0.22

^*Values are means ± SD. Means with different superscripts are significantly differently.

TDEE was higher with increasing obesity and higher in the Class III obese compared to all other groups (Table 2). Similar differences in TDEE between groups were still observed after adjusting for FFM, but when adjusting for body weight, TDEE was lowest in Class III participants (Table 2). There was a reasonable relationship (r²=0.49, p<0.0001) between TDEE and body weight in the entire cohort. However, the relationship between TDEE and body weight was significantly different between the 4 groups (p<0.01). There was a relatively strong relation between TDEE and body weight in the lean group (r²=0.58; Figure 2C). While this relationship was significant in the obese groups, the 3 obese groups were significantly different from the lean group (p<0.03). There was also a significant difference between the class I and class III groups (p<0.01). The comparisons between Class I and II (p=0.08) and Class II and III (p=0.07) approached significance. Figure 2C illustrates the shallower slopes (coefficient decreases from 41.6 for lean to 28.2 kcal/kg body weight for Class III) and increasing intercepts when going from lean to Class III obese. The best overall models of TDEE including body weight (r²=0.55, p<0.0001) or body composition parameters (r²=0.70, p<0.0001) were as follows:

$$\begin{array}{c}\text{TDEE}=21.3\times \text{body}\text{weight}-225\times \text{Race}-151\times \text{obesity}\text{class}+1298\\ \text{TDEE}=39.9\times \text{FFM}-227\times \text{Race}+4.6\times \text{body}\text{fat}+684\end{array}$$

There was a tendency for TDEE to be lower with increasing age in the body composition model (−6.2 kcal/y), but this parameter did not quite reach significance (p=0.08) and explained less than 1% of the variance. Section on TDEE/wt removed.

African-American participants (n=59) were heavier than Caucasians (113.8±3.3 vs. 104.2±2.4 kg; p<0.02) and had higher body fat (55.5±2.4 vs. 49.1±1.7 kg; p<0.03) and FFM (57.1±1.4 vs. 53.8±1.0 kg; p=0.054). After adjusting for FFM and FM, TDEE was 227 kcal/d lower in AA compared to Caucasian participants (see TDEE equations above) with race explaining 3.2% of the variance. The majority of the lower TDEE was due to a 154 kcal/d lower RMR in African-Americans (see RMR equation above). There was no indication of a racial difference in AEE.

Discussion

A primary finding of this study was that energy expenditure was higher with increasing obesity, with very high levels observed in the severely obese. To illustrate the magnitude of the elevated energy expenditure, TDEE was 840 kcal/d higher in the Class III obese compared to lean individuals. Our finding of high TDEE in the Class III obese (3244 kcal/d) confirms previous reports of high TDEE (3415 kcal/d ²¹, 3310 kcal/d ³³, 3226 kcal/d¹³,4036 kcal/d³⁴) in severely obese individuals. Based on the elevated TDEE, the Class III obese individuals would need to consume 35% more calories per day than lean individuals to maintain body weight. While it is clear that the obese have a higher energy intake, we do not know whether such a high energy intake led to increased weight or is a consequence of the increased weight.

Another finding in this study was that absolute RMR was higher in obese groups even after normalizing for FFM, but was similar across all groups after normalizing for both FFM and fat mass. High RMR has previously been reported in morbidly obese individuals.^{21, 35–36} Therefore, these data do not support a low RMR being involved in the maintenance of even severe obesity. However, our cross-sectional findings do not preclude a role for a low RMR in the development of obesity.

The higher absolute AEE we observed in the obese is consistent with a higher cost of activities in obese individuals due to higher body weight ^37–39. While gross energy cost (J/kg) of walking has been shown to be similar in obese and normal weight subjects, net energy cost (gross − standing metabolic rate) has been shown to be higher in obese subjects.^38–40 Preferred walking speed of obese and normal weight individuals (1.41 m/s vs. 1.47 m/s) has been shown to be similar.³⁹ Furthermore, the gross energy cost of walking at the preferred walking speed was shown to be similar for lean and obese, men and women (between 2.81 and 3.04 J/kg/m). Comparing the energy cost of walking for one hour at 1.4 m/s in a typical lean (63 kg) and Class III (124 kg) individual, using a gross energy cost of 3 J/kg/m, shows that the energy cost is nearly twice as high in the Class III obese individual (448 kcal vs. 228 kcal). Therefore, the magnitude of the difference in AEE we observed in obese individuals (20 to 25% higher than lean) is lower than would be expected in light of the increased body weight (approximately 100% higher in Class III compared to lean) if the obese were as active as the lean.

Our analysis of AEE vs. body weight by obesity class indicated a lower AEE in relation to body weight in the obese (Figure 2B). This is in contrast to the comparisons when using FFM or body weight as simple covariates, which indicated that AEE was similar in the 4 groups (Table 2). However, when dividing AEE by body weight (Table 2) we obtain results similar to our analysis of slopes and intercepts of AEE and body weight by obesity class (Figure 2B), indicating that this may be an appropriate normalizing procedure when comparing AEE between groups. The procedure of calculating a physical activity index by dividing AEE by body weight, has previously been shown to be appropriate for comparing the volume (intensity x time) of physical activity between individuals ^41–42. The energy expenditure (above resting) of several activities were shown to be proportional to body weight, with intercepts that did not differ from zero. ⁴² This approach has also been used to compare AEE in individuals with Prader-Willi syndrome⁴³ and when comparing AEE in a wide variety of individuals across multiple studies.⁴⁴ Further support for this normalizing procedure comes from studies showing similar energy cost per kg body weight in lean and obese adults when walking at their preferred walking speed, as described above ³⁹ and lean and obese children walking at the same speed on a treadmill.⁴⁵

Confirming a lower volume of physical activity in the obese, the total time spent in moderate-to-vigorous physical activity, and average daily MET level were lower with increasing obesity, and was lowest in the Class III obese individuals. These data demonstrate that although TDEE and AEE are quite high in Class III obese due to high body weight, the lower than expected AEE in the obese groups was due to less time spent in activity and more time in sedentary behaviors. These results are consistent with the findings of other who have shown a relationship between obesity and sedentary behavior. Banks et al. reported an association between sedentary behavior, measured by screen time, and prevalence of obesity in adults.⁴⁶ Buchowski et al. reported a direct relationship between BMI and sedentary behavior, with women in the highest quartile of sedentary behavior having a higher odds ratio of being at a BMI of 30–39 kg/m² (OR = 2.3) or ≥40 kg/m² (OR = 4.0) when compared to women in the lowest quartile of sedentary behavior.⁴⁷ Whether this lower activity level plays a role in increased obesity, or is a consequence of the increased weight is unknown.

When adjusting for body weight, TDEE was lowest in Class III obese individuals, indicating that their activity level was not proportional to their increased body weight when compared to less obese individuals. In support of this observation, examination of the relationship between TDEE and body weight by obesity class (Figure 2C) indicates that TDEE is lower in obese groups compared to the lean group in relation to their increased body weight. Section on TDEE/wt removed.

As has been previously reported in lean and obese individuals, ^{30–31, 48–53} TDEE and RMR were significantly lower in the African-American participants than in Caucasians when adjusting for body composition. A 227 kcal/d lower TDEE observed in the African-Americans was primarily due to a 154 kcal/d lower RMR. Our data confirms that the lower energy expenditure observed in African-Americans, is also observed in severely obese individuals. However, there was no racial difference in AEE.

In conclusion, the strengths of this study include the assessment of TDEE, RMR, AEE and physical activity patterns in a large number of individuals (n=177) covering a wide spectrum of obesity, including a substantial number of Class III obese (n=98). We confirm that absolute TDEE is elevated in obese individuals, and is very high in severely obese individuals. However, after adjusting for differences in body composition, RMR was similar in all subjects. By combining objective measures of total and activity related free living energy expenditure, we demonstrate that obese individuals have a lower AEE than lean individuals in relation to their increased body weight. Furthermore, the lower AEE could be explained by less time spent in all intensities of physical activity and more time spent in sedentary behaviors. Our findings clearly demonstrate that a high TDEE in obese individuals masks a low level of physical activity. The low level of physical activity observed in the obese, and particularly the severely obese, emphasizes the need for interventions targeting increased physical activity in these individuals.

Abbreviations

EE

energy expenditure

TDEE

total daily energy expenditure

RMR

resting metabolic rate by indirect calorimeter

AEE

activity energy expenditure

DLW

doubly labeled water

FFM

fat free mass

DXA

dual energy X-Ray absorptiometry

TBW

total body water