Obesity and Physical Activity

KR Westerterp*1

1Department of Human Biology, Maastricht University, The Netherlands

OBJECTIVES: Three aspects of obesity and physical activity are reviewed: whether the obese are inactive; how the activity level can be increased; and which are the effects of an increase in physical activity in combination with a reduction of energy intake. METHOD: The focus is on an objective approach that is, activity associated energy expenditure as measured with doubly labelled water. RESULTS: Activity associated energy expenditure increases with body mass index while the average physical activity level does not change. The majority of obese subjects is moderately active. An increase in the activity level of obese subjects is limited by the ability to perform exercise of higher intensity. Training programs obese subjects can cope with are until now not rewarded by weight loss. A possible loss in fat mass is compensated by a gain in fat-free mass. CONCLUSIONS: Obese subjects can only reach a signi®cant weight loss with an energy restricted diet. Mild energy restriction will already result in very signi®cant weight loss when one complies with the diet. An increase in physical activity is necessary to compensate for the reduction in activity induced energy expenditure and should be facilitated by the lower body mass.

Keywords: energy expenditure; energy intake; doubly-labelled water; fat mass; fat-free mass

Introduction Are the obese inactive?

Obesity is often associated with a low level of The doubly-labelled water method allows accurate physical activity. Many people think that one of the measurement of average daily metabolic rate causes of obesity is that people sit most of the day. (ADMR) under unrestricted conditions over 1 ± 3 The ®rst question we will try to answer is whether week intervals to assess the relation between physical there is evidence for a difference in the level of activity and obesity. In Maastricht, we have applied physical activity between obese and non-obese sub- the method, since 1983, in humans.1 Table 1 shows jects. Obesity can be treated by inducing a negative characteristics of 226 subjects measured since then in energy balance, either by increasing physical activity our laboratory, excluding the following characteris- or be reducing energy intake. The second and third tics: age < 20 y and > 50 y, an intervention in energy question therefore is, respectively, how can the activ- intake, an intervention in physical activity including ity level be increased in obese subjects and what is the athletic performance, pregnancy, lactation and dis- effect of an increase in physical activity in combina- ease. Basal metabolic rate (BMR) was measured tion with a reduction of energy intake. Discussing with a ventilated hood or in a respiration chamber. obesity and physical activity needs a de®nition of both Body composition was measured with isotope dilu- items. Obesity can be quanti®ed as weight corrected tion. Figure 1 shows ADMR as a function of BMI. for height, the body mass index (BMI, kg=m2), or fat The lower limit of ADMR clearly increases with BMI mass corrected for height, the fat mass index, or fat from < 5MJ=d in somebody with the lowest BMI of mass as a fraction of body mass. Physical activity is 12.5 kg=m2 to > 15 MJ=d in the morbid obese, a not easily quanti®ed. Measures range from self report threefold difference. It is obvious from this ®gure to activity-associated energy expenditure. One obvi- that the lower limit of energy expenditure, that is, ously wants an objective approach and therefore can energy expenditure at a minimal physical activity, not rely on self report. Here, the focus is on activity- increases with body size. The upper limit is set by associated energy expenditure as measured with body size and physical capacity. Here, we see a steep doubly-labelled water. increase in the lower range of BMI from 5 MJ=din somebody with the lowest BMI of 12.5 kg=m2 to a value of about 20 MJ=d in somebody with a BMI of 22.5 kg=m2. Higher values can be reached in a selection of the population, such as endurance athletes.2 The energy expenditure associated with physical *Correspondence: Dr KR Westerterp, Department of Human activity was calculated as ADMR minus diet-induced Biology, Maastricht University, PO Box 616, 6200 MD Maastricht, The Netherlands. energy expenditure, assumed to be 10% of ADMR, Email [email protected] and minus BMR. Figure 2 shows activity-associated Obesity and physical activity KR Westerterp 60 Table 1 Characteristics of the subjects

Women (n 101) Men (n 125) mean s.d. range mean s.d. range

Age (y) 32 8 20 ± 49 34 7 20 ± 48 Height (m) 1.67 0.06 1.54 ± 1.81 1.80 0.06 1.64 ± 1.97 Body mass (kg) 73 24 38 ± 164 82 20 58 ± 216 Body mass 26.4 8.1 12.5 ± 55.3 25.4 5.6 19.4 ± 61.7 index (kg=m2) Fat mass (kg) 26 17 1 ± 85 20 14 3 ± 129 Fat-free mass (kg) 47 8 33 ± 79 62 8 48 ± 93. BMR (MJ=d) 6.3 1.2 3.6 ± 10.8 7.7 1.3 5.4 ± 12.7 ADMR (MJ=d) 10.9 2.2 4.7 ± 18.4 13.6 2.3 9.7 ± 21.5 ADMR=BMR 1.75 0.22 1.07 ± 2.57 1.79 0.23 1.23 ± 2.57

*BMR, basal metabolic rate; ADMR, average daily metabolic rate.

Figure 2 Activity associated energy expenditure (AEE) as a function of body mass index (BMI) in 226 subjects (Table 1) with the calculated linear regression line (AEE (MJ=d) 0.38 BMI (kg=m2) r 0.17, P< 0.01).

increases with BMI while the PAL for the same population does not is that the latter is corrected for differences in body size by expressing ADMR as a multiple of BMR. Someone with a larger body mass needs more energy for an activity than someone with a lower body mass, especially when the activity involves movement of the body mass. It is impossible to recommend a generalisable coef®cient for adjusting energy expenditure associated with physical activity for body size.4 Dividing by body mass power 1 Figure 1 Average daily metabolic rate as a function of body de®nitely overcorrects for body size as the majority mass index in 226 subjects (Table 1). The dotted lines denote the of the activities for the average subject involve only lower and upper limits. limb movements. Prentice et al 4 calculated an expo- nent of 0.51 for a 24 h activity pattern in a respiration energy expenditure as a function of BMI. As chamber including ADL activities, stepping and expected, there is a wide scatter, the activity-induced cycling. Dividing by BMR, as done in the PAL energy expenditure being the most variable compo- calculation, probably is a justi®ed intermediate. nent of ADMR. It ranges from 1 ± 10 MJ=d and there BMR in the population of the current analysis is a slight but signi®cant increase with BMI was a function of body mass power 0.63 (P < 0.01), overweight subjects spending on average (BMR 0.443 body mass0.63, r 2 0.69). more energy for physical activity. There are at least four meta-analyses of the relation The physical activity level (PAL) of a subject can between body mass or body fatness and physical be calculated by expressing ADMR as a multiple of activity as derived from doubly-labelled water.4 ± 7 BMR (PAL ADMR=BMR) or by adjustment of Westerterp et al 5 analysed 96 existing data sets with ADMR for BMR as suggested by Carpenter et al.8 observations on body composition and PAL of healthy Figure 3 shows the physical activity level expressed in subjects age 19 ± 48 y. Fat-free mass and fat mass both ways and plotted as a function of BMI. Which- were related to PAL. In a regression analysis fat ever of the two methods is used, the physical activity mass explained 53 and 40% of the variation in fat- level is independent of BMI, a result that has been free mass in females and males, respectively. Adding con®rmed with a tri-axial accelerometer for move- PAL to the model raised the explained variation in fat- ment registration in a sub-group of the subjects in free mass in females to 62%. In contrast with females, current analysis.3 An explanation for the fact that there was an independent relationship between PAL energy expenditure associated with physical activity and fat mass in males, such that a higher PAL was Obesity and physical activity KR Westerterp 61 physical activity and it was suggested that a low physical activity is a permissive factor for obesity. Prentice et al 4 analysed 319 data sets of healthy subjects age 18 ± 64 y. Results were analysed accord- ing to four categories: < 25.0, 25.0 ± 29.9, 30.0 ± 35.0 and > 35.0 kg=m2. PAL remained constant across the three lowest BMI groups indicating similar levels of physical activity. There was a non-signi®cant decrease in PAL in the most obese men and women. Westerterp and Goran7 analysed data sets of 290 healthy subjects age 18 ± 49 y. Physical activity was quanti®ed by adjustment of ADMR for BMR as suggested by Carpenter et al. 8 The result was a gender difference of the relation between physical activity and body fatness. In males, there was a signi®cant inverse cross- sectional relationship between activity-energy expenditure and body fatness, whereas no such relationship was apparent in females. Summarizing, activity associated energy expenditure increases with BMI while the average PAL does not change. In contrast, on average, women with a higher PAL have a higher fat-free mass while men with a higher PAL have a lower fat mass.

How can the activity level be increased?

It is very dif®cult to increase the activity level in the obese. First of all, increasing the level of habitual activity requires a change of life style. Secondly, increasing the activity level in the obese does not imply a change from a sedentary life style. Many obese subjects are already moderately active (see section on `Are the obese inactive'). Thirdly, obese subjects are limited in the ability to perform exercise of higher intensity. Finally, exercise is not rewarded by weight loss. Here, controlled exercise programs are Figure 3 Physical activity level as a function of body mass index described without restriction of energy intake. Five in 226 subjects (Table 1), (A) physical activity level calculated as 2 average daily metabolic rate (ADMR) divided by basal metabolic available studies were reviewed recently with respect rate (BMR); (B) physical activity level calculated as the residual of to the effect of exercise training on spontaneous the ADMR ± BMR relation. physical activity and on ADMR. Two studies used jogging as a means of increasing physical activity, the related to a lower fat mass. It was concluded that there other three included indoor exercises, stationary is a relation between PAL and fat-free mass under cycling on a cycle ergometer and weight-training, in normal living conditions though fat mass obscures the a limited number of sessions per week. relation, especially in males. When subjects lose fat Jogging certainly is not the exercise mode of choice mass they also lose fat-free mass, an effect that partly for obese subjects. We studied the effect of an explains why an increase in PAL does not prevent the increase in physical activity on energy balance and loss of fat-free mass when the loss of fat mass is high body composition in subjects not participating in any as in persons on weight-reducing diets (see below, last sport before the start of the experiment and prepared section) Schulz and Schoeller6 analysed 259 existing to run a half-marathon competition after 44 weeks.9 data sets with observations on body fatness and Of 370 respondents to an advertisement in the local physical activity. The relationship between physical media, 16 women and 16 men, aged 28 ± 41 y and with activity, here expressed as non-basal energy expendi- a BMI of 19.4 ± 26.4 kg=m2, were selected. During the ture divided by body mass (ADMR 7 BMR)=kg, and study nine subjects withdrew on being unable to keep body fatness (fat mass divided by body mass) was up with the training program. They all dropped out highly signi®cant. Fatness increased with decreasing within 20 weeks of the start of the training. Reasons Obesity and physical activity KR Westerterp 62 mass. The male participants of the jogging training with the higher percentage body fat were the only subjects losing more fat than gaining fat-free mass. Summarizing, the majority of obese subjects is moderately active. An increase in the activity level of obese subjects is limited by the ability to perform exercise of higher intensity. Training programs with which obese subjects can cope have until now not been rewarded by weight loss. A possible loss in fat mass is compensated by a gain in fat-free mass.

Effects of an increase in physical activity in combination with a reduction of energy intake

As shown above, exercise training can reduce body mass and fat mass but the resultant changes are small. The same seems to hold for the combination of Figure 4 Body mass index of subjects who completed and who exercise training with energy restriction. During withdrew from a training to run a half-marathon. diet-induced weight loss, added exercise results in little further weight loss. The idea of a combination of exercise training with energy restriction is that the exercise will accelerate fat loss and maintain fat-free for giving up were `not enough time to join the mass and thus prevent a decline in resting metabolic training' (n 3), injuries (n 5) and not able to rate. keep up with the training (n 1). Figure 4 gives the Indeed, aerobic exercise provides some conserva- BMI values of the subjects who completed (BMI tion of fat-free mass during weight loss as shown in a 19.4 ± 25.7) and who withdrew (BMI 23.4 ± 26.4). meta-analysis of 21 studies by Garrow and Summer- All dropouts had a BMI above the group mean of bell.12 It was hypothesized that a subject with 10 kg 22.9 kg=m2. Although manifestly obese subjects were excess weight has 2.5 kg excess fat-free mass. Men excluded by the selection criteria, most subjects with a losing 10 kg by dieting alone and by dieting plus BMI > 24 kg=m2 could not cope with the training. exercise lost, respectively, 2.9 and 1.7 kg, and Weight-training is an exercise mode that is suitable women, respectively, 2.2 and 1.7 kg. Part of the for a wide variety of target groups including the conservation of fat-free mass was suggested to occur obese. Van Etten et al 10 did training studies in men through maintaining glycogen and water, implicating with a BMI up to 30 kg=m2 without dropouts. Blaak et no preventative effect on the diet-induced decline in al 11 successfully trained obese boys with more than resting metabolic rate. Ballor and Poehlman13 ana- 30% body fat, with daily one-hour cycling bouts at lysed 33 studies on the extent to which exercise 50 ± 60% of VO2max. Obviously, the increase in PAL training and=or diet-induced weight loss affect resting is a function of the training mode and training volume. metabolic rate. Weight loss programs of the included The jogging training, four sessions of 20 ± 60 min- studies averaged 12 weeks and the overall weight loss =week, resulted in an increase in PAL in the success- was about 12 kg. The weight loss resulted in a con- ful subjects of 1.68Æ 0.18 to 2.13Æ 0.19 (P < 0.0001). sistent decrease in resting metabolic rate of 0.85 MJ=d The weight training, two sessions of 60 min=week, with no exercise training effects. The change in resulted in a PAL increase from 1.76Æ 0.14 to resting metabolic rate was proportional to the magni- 1.92Æ 0.18 (P < 0.01), and the cycling training, ®ve tude of the weight reduction (r 0.66, P < 0.01). sessions of 60 min=week, resulted in a PAL increase Weight loss systematically results in a reduction of from 1.77Æ 0.15 to 2.04Æ 0.15 (P < 0.01). Changes in resting metabolic rate, even when the weight loss body mass were not large in any of the three studies. results from a reduction in fat mass while fat-free The jogging training did not result in a signi®cant mass remains the same or is increased, as shown by a body mass change in women and resulted in a 1.3 kg study of the effect of chronic exercise.9 Body mass body mass loss in men (P < 0.01) at 40 weeks. The decreased during the 40-week training program with weight training resulted in an unaltered body mass an average of (meanÆ s.d.) 1.0Æ 1.7 kg (P < 0.01) after 18 weeks, and the cycling training had no effect while fat-free mass increased by 2.7Æ 1.5 kg on body mass either. There were consistent changes in (P < 0.001) and fat mass decreased by 3.7Æ 2.1 kg body composition, however; losses in fat mass were (P < 0.01). Sleeping metabolic rate decreased from compensated by early equivalent gains in fat-free (meanÆ s.d.) 6.46Æ 0.62 to 6.32Æ 0.61 MJ=d Obesity and physical activity KR Westerterp 63 (P < 0.01). The decrease of body mass explained 38% of the variance in the decrease in sleeping metabolic rate. Thus, an exercise induced increase in fat-free mass did not result in an increase in resting metabolic rate. There was an indication of the opposite effect, a decreasing resting metabolic rate as a defence mechanism in the maintenance of body mass. The addition of exercise to an energy restricted diet results in little further weight loss. Exercise does not reverse the weight loss-induced depression of resting metabolic rate and weight loss is not different for groups undergoing dietary restriction and dietary restriction plus exercise. The latter implicates that the direct cost of the exercise training is compensated by a reduction of activity associated energy expenditure outside the training sessions. Two studies compared activity-associated energy expenditure before and after dietary restriction and dietary restriction plus exercise, as measured with doubly labelled water. Racette et al 15 designed diets to promote a weight loss of 1 kg=week for 12 weeks by prescribing a diet with 75% of measured resting metabolic rate in the diet-only group and added the calculated energy costs Figure 5 Average daily metabolic rate as a function of sleeping metabolic rate in obese subjects before weight loss (regression of the exercise for the diet plus exercise group to line) and after eight weeks weight loss by an energy restricted create a comparable energy de®cit. They observed a diet only (open symbols) of by the energy restricted diet plus maintenance of ADMR in the exercise group exercise (closed symbols). ( 0.1Æ 1.2 MJ=d, n.s.) while ADMR decreased in the diet-only group ( 7 1.4Æ 1.0 MJ=d, P < 0.01). Kempen et al 16 provided all subjects with an identical mobilized energy from body stores, was 1.1 and 1.5 low-energy formula diet providing 2 MJ=d for 4 times the prescribed intake of about 5 MJ=d in the diet weeks, followed by 4 weeks 1.4 MJ=d of the formula plus exercise and diet only group, respectively, in the diet supplemented with a free choice of food study of Racette et al.15 Kempen et al 16 observed to 3.5 MJ=d. They observed a comparable decrease values of 2.0 and 2.5 times the prescribed intake of of ADMR of 7 0.9Æ 1.0 MJ=d (P < 0.05) and 2.75 MJ=d in the diet plus exercise and diet only 7 1.4Æ 1.3 MJ=d (P < 0.01) in the diet plus exercise group, respectively. Amongst the suggested explana- and the diet-only group, respectively. ADMR dropped tions for the improved compliance in the exercise signi®cantly, and to a similar extent, at both treat- group are an increased contact with the subjects ments suggesting no net effect of the exercise training during the supervised exercise sessions.17 The overall on the activity associated energy expenditure. conclusion is that the size of the exercise intervention Explanations for the relatively minor or non-existent only has a minor or no effect on the activity level of effect of the addition of exercise to an energy- the subjects and consequently does not result in restricted diet are a low compliance to the exercise additional weight loss. Finally there are some studies prescription and=or a negative effect of exercise showing the effect of energy restriction on physical training on dietary compliance. Racette et al 15 calcu- activity. Velthuis-te Wierik et al 18 observed the effect lated that there was an increase of 0.3 MJ=d of of a moderately energy-restricted diet on energy activity-associated energy expenditure over and metabolism in non-obese men (BMI, 24.9Æ above the 0.4 MJ=d of the supervised exercise ses- 1.9 kg=m2). For 10 weeks subjects received a diet sions, that is an increase instead of a decrease in with 67% of their measured ADMR during weight physical activity outside the exercise periods. How- maintenance. The consequent weight loss was ever, before the intervention, subjects in the diet plus 7.4Æ 1.7 kg and the activity level (ADMR=BMR) exercise group had a lower activity induced energy went down from 1.85Æ 0.37 to 1.65Æ 0.29 expenditure than the diet-only group and both groups (P 0.06), that is there was a tendency for a reduction reached the same value, that is activity level, during of physical activity by reducing energy intake. Bouten the intervention. Kempen et al 16 observed no differ- et al 19 measured physical activity in subjects volun- ence in activity-induced energy expenditure between tarily reducing energy intake resulting in underweight, the two groups, starting from the same activity level that is subjects with anorexia nervosa. Subjects with a before the intervention (Figure 5). With respect to BMI 5 17 kg=m2 were equally or more active com- dietary compliance both studies observed that exercise pared with control subjects, while subjects with a enhances compliance. Calculated intake, doubly BMI < 17 kg=m2 were equally or less active compared labelled water measured energy expenditure minus Obesity and physical activity KR Westerterp 64 with control subjects. Thus, undereating may cause a 7 Westerterp KR, Goran M. Relationship between physical decrease in daily physical activity. activity related energy expenditure and body composition: a gender difference. Int J Obes 1997; 21: 184 ± 188. Exercise training in obese subjects does not seem to 8 Carpenter WH, Poehlman ET, O'Connell M, Goran MI. promote weight loss. What are the bene®cial effects of In¯uence of body composition and resting metabolic rate on exercise for obese subjects, being on average moder- variation in total energy expenditure: a meta-analysis. Am J ately active and having a limited ability to perform Clin Nutr 1995; 61: 4 ± 10. exercise? The answer probably is that a high level of 9 Westerterp KR, Meijer GAL, Janssen EME, Saris WHM, Ten Hoor F. Long term effect of physical activity on energy turnover, that is a high level of physical energy balance and body composition. Br J Nutr 1992; 68: activity, promotes weight maintenance. Regulation 21 ± 30. of energy balance is likely to be better at a higher 10 Van Etten LMLA, Westerterp KR, Verstappen FJT, Boon energy turnover. Obviously one can eat more when BJB, Saris WHM. Effect of an 18-wk weight-training program one is physically active, without getting into a positive on energy expenditure and physical activity. J Appl Physiol 1997; 82: 298 ± 304. energy balance. There are speculations that exercise 11 Blaak EE, Westerterp KR, Bar-Or O, Wouters LJM, Saris reduces perceived hunger in obese subjects and there WHM. Effect of training on total energy expenditure and is evidence that exercise increases fat oxidation at a spontaneous activity in obese boys. Am J Clin Nutr 1992; high-fat diet. Obese subjects can only reach a sig- 55: 777 ± 782. ni®cant weight loss with an energy restricted diet. 12 Garrow JS, Summerbell CD. Meta-analysis: effect of exercise, with or without dieting, on body composition of overweight Mild energy restriction will already result in very subjects. Eur J Clin Nutr 1995; 49: 1 ± 10. signi®cant weight loss when one complies with the 13 Ballor DL, Poehlman ET. A meta-analysis of the effects of diet. An increase in physical activity is necessary to exercise and=or dietary restriction on resting metabolic rate. compensate for the reduction in activity induced Eur J Appl Physiol 1995; 71: 535 ± 542. energy expenditure and should be facilitated by the 14 Westerterp KR, Meijer GAL, Schoffelen P, Janssen EME. Body mass, body composition and sleeping metabolic rate lower body mass. before, during and after endurance training. Eur J Appl Physiol 1994; 69: 203 ± 208. 15 Racette SB, Schoeller DA, Kushner RF, Neil KM, Heling- References Iaffaldano K. Effects of aerobic exercise and dietary carbohy- 1 Westerterp KR, De Boer JO, Saris WHM, Schoffelen PFM, drate on energy expenditure and body composition during Ten Hoor F. Measurement of energy expenditure using doubly weight reduction in obese women. Am J Clin Nutr 1995; 61: labelled water. Int J Sports Med 1984; 5: 74 ± 75. 486 ± 494. 2 Westerterp KR. Alterations in energy balance with exercise. 16 Kempen KPG, Saris WHM, Westerterp KR. Energy Am J Clin Nutr 1998; 6 (Suppl): S970 ± S974 balance during 8 weeks energy-restrictive diet with and 3 Westerterp KR, Bouten CVC. Physical activity assessment: without exercise in obese females. Am J Clin Nutr 1995; 62: comparison between movement registration and doubly 722 ± 729. labelled water method. Z ErnaÈhrungswiss 1997; 36: 263 ± 267. 17 Racette SB, Schoeller DA, Kushner RF, Neil KM. Exercise 4 Prentice AM, Black AE, Coward WA, Cole TJ. Energy enhances dietary compliance during moderate energy restric- expenditure in overweight and obese adults in af¯uent socie- tion in obese women. Am J Clin Nutr 1995; 62: 345 ± 349. ties: an analysis of 319 doubly-labelled water measurements. 18 Velthuis-te Wierik EJM, Westerterp KR, Van den Berg H. Int J Obes 1996; 50: 93 ± 97. Impact of a moderately energy-restricted diet on energy 5 Westerterp KR, Meijer GAL, Kester ADM, Wouters L, Ten metabolism and body composition in non-obese men. Int J Hoor F. Fat-free mass as a function of fat mass and habitual Obes 1995; 19: 318 ± 324. activity level. Int J Sports Med 1992; 13: 163 ± 166. 19 Bouten CVC, Van Marken Lichtenbelt WD, Westerterp 6 Schulz LO, Schoeller DA. A compilation of total daily energy KR. In¯uence of body mass index on daily physical expenditures and body weights in healthy adults. Am J Clin activity in anorexia nervosa. Med Sci Sports Exerc 1996; 28: Nutr 1994; 60: 676 ± 681. 967 ± 973.