International Journal of Obesity (1999) 23, 1269±1275
ß 1999 Stockton Press All rights reserved 0307±0565/99 $15.00
http://www.stockton-press.co.uk/ijo
Effects of inactivity and diet composition on
human energy balance
PR Murgatroyd1*, GR Goldberg1, FE Leahy1, MB Gilsenan1 and AM Prentice1
1
MRC Dunn Clinical Nutrition Centre, Cambridge, UK
OBJECTIVES: To investigate the in¯uences of inactivity and dietary macronutrient composition on energy and fat
balance and to look for interactions between them.
DESIGN: Two-day measurements of energy expenditure and substrate oxidation on ®ve occasions; ad libitum food
intake from diets of 35% and 60% energy as fat, with and without imposed activity, and a ®xed overfeeding at 35% fat
with free activity.
SUBJECTS: Eight normal-weight male volunteers.
MEASUREMENTS: Energy expenditure and substrate oxidation by indirect whole-body calorimetry, and macronutrient intakes from food consumption on ad libitum regimens.
RESULTS: Subjects consumed the same energy, mean 11.6 MJ=d, regardless of activity level, on the 35% diet.
Subjects consumed more energy on the 60% than the 35% diet, mean 14 vs 11.6 MJ=d. Inactivity induced a strong
positive energy balance: 5.1 (60% diet), and 2.6 MJ=d (35% diet). Energy balance with activity was not signi®cantly
different between diets, nor signi®cantly different from zero: 1.1 MJ=d (60% diet), and 7 0.2 MJ=d (35% diet). When
intentionally overfed, subjects failed to compensate by raising voluntary activity.
CONCLUSION: Energy intake was not regulated over a 2-day period in response to either imposition of inactivity or a
high-fat diet. Activity proved essential to the avoidance of signi®cant positive energy balance.
Keywords: energy balance; fat balance; activity; inactivity; exercise; calorimetry
Introduction
Obesity is the consequence of a long-term failure to
down-regulate energy intake in response to an energy
surplus. The prevalence of obesity continues its inexorable rise in developed countries and has more than
doubled in the UK between 1980 and 1995.1 During
this time the average household's food intake was
falling.2 This has been taken to imply an even more
profound fall in energy expended as physical activity.
In the face of excess intake, exercise is the one
effective stratagem, apart from increased fat mass, by
which fat oxidation can be substantially promoted.3
There is, however, little evidence of close coupling
between energy intake and energy expenditure at high
levels of energy expenditure.4 It is super®cially surprising, therefore, that imposition of exercise alone,
without dietary restriction, seems so ineffective in
combating established obesity within individuals,5
particularly as exercise is undoubtedly important to
the successful maintenance of weight loss through
dieting.6 ± 8 The role of physical activity in the regulation of energy balance is therefore something of a
paradox.
*Correspondence: PR Murgatroyd, MRC Dunn Clinical Nutrition
Centre, Hills Road, Cambridge CB2 2DH, UK.
E-mail: [email protected]
Received 26 October 1998; revised 1 March 1999; accepted
21 June 1999
As long ago as 1954, Mayer et al explored the
relationship between exercise, food intake and body
weight in rats and mice.9 Their results suggested close
tracking of intake and expenditure throughout the
range of moderate activity, failure to eat to the
requirements of high activity, and a tendency to
overeat during enforced inactivity. Later studies in
man suggested that a nadir in energy intake was
associated with light activity levels, below which
energy intake and adiposity increased.10 There is,
therefore, a sound basis for hypothesising that inactivity leads to weight gain.
An association between adiposity and dietary fat
intake has been demonstrated in both epidemiological
and intervention studies, suggesting that the body is
not well adapted to recognising the energy content of
dietary fat.11 ± 15 However, interventions to reduce
dietary fat intake have had variable success in promoting weight loss.16 ± 18 The in¯uence of fat intake
on body fatness has thus been subject to much
debate.19,20
In the light of this rather confused picture, in which
energy regulatory mechanisms seem insensitive to
both physical activity and diet composition, we have
embarked upon a series of studies to explore further
the role which inactivity may play in impairing energy
regulation.
The study described here aimed to examine the
effects of both physical inactivity and diet composition upon energy and macronutrient balance, and to
investigate whether subjects modify their habitual
Inactivity, diet and energy balance
PR Murgatroyd et al
1270
energy intake in response to the imposition of a
sedentary lifestyle. The protocol was designed to
enable interactions between activity and diet composition to be identi®ed, and in particular to address the
question of whether inactivity may be particularly
harmful when associated with a high-fat diet.
Methods
Subjects
The criteria for selection of subjects were that they
should be male, age range 18 ± 55 y, body mass index
(BMI) < 25 kg=m2, in good health, non-smokers and
not habitually taking part in sports or strenuous
muscular activities. Subjects were recruited through
our Register of Research Volunteers, through posters
placed in university buildings and through advertisements in the local press. Prior to consenting to
participation in the study all volunteers received a
verbal and written explanation of the procedures to be
used in the study and were given a guided tour of the
calorimeters and metabolic suite at the Dunn Clinical
Nutrition Centre (DCNC). The study design required
that subjects were not told that food intake would be
measured and that dietary composition would be
manipulated. Each subject was ®nally accepted for
participation after medical examination. The study
was approved by the Dunn Nutrition Unit Ethical
Committee. The physical characteristics of the subjects are presented in Table 1.
Protocol
The protocol required each subject to be studied on
®ve occasions, each with a different treatment. The
®rst four treatments, presented at random, were: 60%
energy from fat, no imposed exercise (60 SED); 60%
fat diet plus imposed exercise (60 ACT); 35% fat diet,
no imposed exercise (35 SED); 35% fat diet plus
imposed exercise (35 ACT). On these treatments
subjects were asked to eat ad libitum from an excess
of food supplied at each meal time. Dietary manipulations were covert. The exercise level on ACT treatments was designed to produce a physical activity
Table 1
Subject
Physical characteristics of the subjects
Age
(y)
Height
(m)
Weight
(kg)
Fat
(%)
1
2
3
4
5
8
9
10
44
27
39
34
38
31
27
44
1.78
1.86
1.80
1.73
1.83
1.95
1.77
1.78
66.75
67.48
70.34
54.30
66.92
82.20
78.43
66.90
15.5
8.1
13.8
10.1
Mean
s.d.
35.50
6.87
1.81
0.07
69.17
8.43
11.96
3.22
8.1
15.5
12.6
level, PAL, of 1.6, corresponding to the modal value
for males aged 18 ± 64 y found by Black et al in their
meta-analysis of doubly labelled water measurements
in free-living subjects.21 On the ®fth occasion subjects
were asked to eat all the food provided and were told
that they were free to undertake as much physical
activity as they wished, including exercise on the
cycle ergometer or aerobics step provided (35
FREE). This activity was not directly quanti®ed.
The 35 FREE contained the average energy previously consumed by that subject on the two 60%
diets, but presented as 35% energy from fat. The
quantity of food supplied appeared to be greater
than that actually consumed on any previous occasion
due to the lower energy density of the 35% meals.
Subjects were intended to recognise this as overfeeding.
Subjects began each visit by spending an equilibration day, day 0, living within the unit's Metabolic
Suite. Its purpose was to control diet and activity,
ensuring a standardised metabolic starting point for
subsequent measurements. Subjects reported to
DCNC in time for breakfast. Activity during this
day was limited and dietary intake was prescribed
with an energy content of 1.35estimated BMR22 and
composition 40% fat, 48% carbohydrate and 12%
protein by energy.
At 20.00 h on the equilibration day the subject
entered the calorimeter. The evening was passed in
sedentary activity (for example reading, watching
television) until bed time at 23.00 h. Subjects
remained in bed until 08.00 h on day 1, the ®rst
manipulation day, when they were woken for a
measurement of basal metabolic rate (BMR). BMR
was measured, with the subjects awake though still in
bed, during the hour between 08.00 and 09.00 h.
Subjects rose at 09.00 h, washed and dressed. Breakfast was consumed between 09.30 and 10.00 h, lunch
between 13.30 and 14.00 h and the evening meal
between 19.00 and 19.30 h. Subjects prepared for
bed between 22.30 and 23.00 h. At all times when
no other activity was scheduled subjects remained
seated and undertook sedentary activities such as
reading or watching the television. Day 2 followed
the same protocol as day 1. Subjects left the calorimeter after the BMR measurement on the morning of
day 3.
During days 1 and 2 of those protocols which
included exercise, subjects were asked to pedal a
cycle ergometer (Monark A.G., Varberg, Sweden or
Seca, Vogel and Halke, Hamburg, Germany) for 40
min from 11.00, 15.00 and 20.30 h. The exercise
imposed an external work rate of 75 W (1.5 kp load,
50 pedal cycles per min).
Samples of urine were collected for measurement
of urinary nitrogen excretion by the Kjeldahl method
(Tecator Kjeltec, HoÈganaÈs, Sweden), from which
protein oxidation rates were calculated.
Body composition was measured using dual X-ray
absorptiometry (DXA, Hologic QDR 1000W).23
Inactivity, diet and energy balance
PR Murgatroyd et al
Diets
Each visit's diet comprised just ®ve homogenous
elements. Breakfast was porridge; on day 1 lunch
was cheese and potato pie followed by carrot cake,
and dinner was ®sh and potato pie followed by fruit
loaf. The lunch and dinner recipes of day 1 were
served as dinner and lunch on day 2. Each dietary
element was formulated, from simple ingredients of
reliably documented composition and digestibility,24
to contain either 35% or 60% energy as fat. Carbohydrate varied reciprocally with fat; protein was kept
constant at 12% of energy. Manipulations were covert
and produced no perceptible difference in appearance,
taste or texture between 35% and 60% formulations.
Decaffeinated tea or coffee and 50 g milk were given
with each meal. All meals on the ®rst four visits were
given as excessively large quantities and subjects
were asked to eat only as much as they wished. We
stressed that we did not expect all the food to be eaten.
By deliberately making the meals homogenous and
somewhat bland and by serving the same items on
both days we aimed to avoid palatability-driven overconsumption. We have previously employed a similar
strategy with success.25 The energy densities of the
meals are given in Table 2.
Calorimetry
Whole-body calorimetry provided non-invasive measurements of energy substrate oxidation and energy
expenditure within an environment which allowed
strict observance of the study protocol to be maintained.
This study used the three calorimeter rooms at
DCNC. From the subject's point of view these were
comfortable bed-sitting rooms approximately 3.5 m
long, 2.8 m wide and 2.1 m high. The rooms were
carpeted and were furnished with a bed, an armchair, a
desk and upright chair, a television, a video and hi-®.
A wash basin with hot and cold water was provided.
Each had a window to the world outside. A telephone
and intercom allowed communication between the
subject and his experimenter or the wider world.
Food entered the room through an air-lock hatch.
Excreta were collected within the room and passed
out through a second hatch. Urine samples were
analysed for nitrogen as an estimate of protein oxidation. Faeces were discarded. Subjects were supervised
around the clock by experimenters and night nurses.
Air within the rooms was recirculated at 20
m3=min, mixing the subject's expired air with the
Table 2
room air to produce uniform room air composition.
The room air temperature for this study was
23 0.5 C. The rooms were ventilated at a rate of
200 l=min with fresh air drawn from outside the
building. The ventilation was measured, as the air
entered the chamber, by a type 2100 Rotameter (KDG
Mowbray, Slough, UK) and a vortex-shedding ¯owmeter type VL512 (Delta Controls, West Molesey,
UK). Samples of fresh air and ventilating air were
drawn for analysis. The moisture content of the
samples was measured (Dewpoint analyser type
1100 ap, General Eastern, Watertown, MA, USA),
then dried by PermaPure membrane dryers (PermaPure Products Inc., Toms River, NJ, USA) before
analysis of oxygen and carbon dioxide (Paramagnetic
O2 analysers types 184 and 1440 and infra-red CO2
analysers type 1510; Servomex, Crowborough, UK).
Data from the analysers was digitised (Systems voltmeter type 7062 with 18 channel scanner, Solartron,
Farnborough, UK) and logged onto a personal computer through a Measurement Co-processor (Hewlett
Packard, Palo Alto, CA, USA). Data were logged
every 5 min throughout the study. Fresh air samples
were analysed every 30 min, while every 3 h analyser
calibrations were checked using O2-free N2 for zeros,
1% CO2 in air for CO2 span and fresh air for O2 span.
Oxygen consumption and carbon dioxide production were calculated using the expressions derived by
Brown et al.26 Energy substrate oxidation rates were
calculated from gas exchanges using the expressions
of Murgatroyd et al,27 and the values for the respiratory quotients and energy equivalents of oxygen for
each substrate proposed by Elia and Livesey.28
The calorimeter calibrations were veri®ed by infusing a test gas mixture of 20% CO2=80% N2, which
simulated CO2 production directly and O2 consumption by dilution of the room O2.29 The test gas infusion
rate was measured to 0.5% by an oil-®lled wet-type
gas meter (Alexander Wright, Sutton, UK). The
calorimeters yielded results for O2 consumption and
CO2 production within 1% of those predicted from
the test gas infusion rates. From calibration results it
can be inferred that errors in the estimates of energy
expenditure are less than 1%, and that carbohydrate
oxidation can be measured to better than 20 g=d and
fat oxidation to better than 9.5 g=d. The precision of
carbohydrate and fat oxidation measurements has
been shown to be better than 0.55 and 0.27 g,
respectively, for 30 min intervals of analysis, improving further, in proportion to the square root of the
number of samples, over longer analysis intervals.27
Energy densities of meals
Meal
Porridge
Cheese pie
Fish pie
Fruit loaf
Carrot cake
35% Fat
(kJ=g)
60% Fat
(kJ=g)
Ratio
(60%:35%)
5.10
4.32
4.25
8.66
11.01
7.16
6.46
6.35
12.52
14.52
1.40
1.50
1.49
1.45
1.32
Statistical analysis
Treatment effects were analysed in an ANOVA general linear model; post hoc multiple comparisons were
made by the Scheffe method (Unistat statistical package 4.53, Unistat Ltd, London UK). Means are presented with their associated standard deviations in
tabulated data and with standard errors on bar graphs.
1271
Inactivity, diet and energy balance
PR Murgatroyd et al
1272
Results
Table 4
Responses of energy intake and balance to
manipulations of dietary composition and imposed activity
Treatment
Energy expenditure
Energy expenditure was primarily determined by the
activity prescribed in the study protocols. Energy
expenditure on the ACT protocols was signi®cantly
higher than that on the SED protocols (Table 3). The
expenditure due to activity was similar for both dietary treatments, 2.76 MJ=d for 60% and 2.90 MJ=d for
35% an average of 2.83 MJ=d. This resulted from
exercise imposed as three 40 min periods of cycling
at 75 W, equivalent to 0.54 MJ=d external work, plus
the additional energy expended in the non-sedentary
periods preparing for and recovering from the exercise
bouts.
Energy expenditure was also in¯uenced by the diet
composition, through its in¯uence on energy intake.
The excess intake of 3.09 MJ on the 60 SED compared
with the 35 SED resulted in an increase in energy
expenditure of 0.55 MJ. The corresponding excess of
1.65 MJ on the 60 ACT relative to the 35 ACT
resulted in an increase in energy expenditure of
0.41 MJ. Averaged over the two activity levels, expenditure was increased by 0.48 MJ=d on the 60% relative
to the 35% fat diet (P < 0.0001). This represented
20.2% of the excess energy ingested. There were no
interactions between diet and activity in their in¯uence upon total energy expenditure.
The PAL achieved in the ACT treatments was 1.61
for both 35% and 60% diets, close to the target of
1.60. The PAL values for 60 SED and 35 SED
treatments were very much lower by design, 1.27
and 1.22, respectively.
On the 35 FREE treatment, energy expenditure was
signi®cantly lower than on either of the ACT treatments: 2.10 MJ=d lower compared with 35 ACT and
2.50 MJ=d compared with 60 ACT. However, subjects
also expended signi®cantly more energy than on the
35 SED treatment, by 0.81 MJ=d (Table 3). The PAL
for the 35 FREE treatment was 1.31, not signi®cantly
different from that of the 60 SED treatment to which,
by design, it had a comparable level of energy surplus.
Day
35 ACT
35
60
60
35
1
2
Average
SED
1
2
Average
ACT
1
2
Average
SED
1
2
Average
FREE
1
2
Average
Intake (MJ=d) s.d.(MJ) Balance(MJ=d) s.d.(MJ)
11.91
11.48
11.70
11.97
11.10
11.53
13.29
13.42
13.35
14.60cd
14.63cd
14.62cd
14.23
14.15
14.19
3.78
3.33
3.45
3.83
2.99
3.35
4.49
3.83
4.03
5.63
4.91
5.10
3.66
3.73
3.57
0.08
7 0.45
7 0.18
3.02c
2.08c
2.55c
1.01
1.12
1.07
5.05abe
5.13abe
5.09abe
4.55abf
4.26abf
4.41abf
3.45
3.07
3.17
3.80
2.83
3.27
4.32
3.50
3.80
5.09
4.40
4.60
3.24
3.42
3.22
There were no signi®cant differences in energy intake or balance
between the two days of any treatment. Intakes on the 35 FREE
treatment were ®xed. P < 0.0001, avs 35 ACT, bvs 60 ACT.
P < 0.001, cvs 35 ACT, dvs 35 SED. P < 0.002, evs 35 SED.
P < 0.05, fvs 35 SED.
tended to be suppressed, by 1.26 MJ=d on the 60 ACT
treatment compared to 60 SED (Table 4).
Energy intake was profoundly in¯uenced by the
diet composition on the SED treatments. Intake was
signi®cantly higher on the 60 SED compared to the 35
SED treatment, the difference being 3.09 MJ=d. The
tendency towards suppressed intake on the 60 ACT
treatment (see above) resulted in a substantially
smaller difference, 1.66 MJ=d, between intakes on
the 60 ACT and 35 ACT treatments, which fell just
short of signi®cance.
There were no interactions between intake and day
of study. The small decrease in intake on day 2
compared with day 1 of the 35% treatments (Table
4) was far from signi®cant.
Energy balance
Energy balance was strongly in¯uenced by both
activity and diet (Figure 1). Without imposed activity,
Energy intake
There was no signi®cant difference between ad libitum intakes with and without activity within each of
the 35% or the 60% diets; however, energy intake
Table 3 Responses of energy expenditure to manipulations of
dietary composition and imposed activity
Treatment
35
35
60
60
35
ACT
SED
ACT
SED
FREE
Expenditure (MJ=d)
s.d. (MJ)
11.98
8.98abc
12.29e
9.53abd
9.79ab
0.72
0.78
0.86
0.89
0.99
Values calculated for all subject days. There were no day ±
treatment interactions. P < 0.0001, avs 35 ACT, bvs 60 ACT, cvs
35 FREE. P < 0.002, dvs 35 SED. P < 0.05, evs 35 ACT.
Figure 1 Average daily energy balance on each treatment. Error
bars represent standard errors of the mean. (a) P < 0.0001 vs 35
ACT and 60 ACT. (b) P < 0.001 vs 35 ACT. (c) P < 0.002 vs 35 SED.
(d) P < 0.05 vs 35 SED.
Inactivity, diet and energy balance
PR Murgatroyd et al
1273
energy balance was positive by 5.09 MJ on the 60%
diet (P < 0.0001) and by 2.55 MJ on the 35% diet
(P < 0.005). The lower fat diets improved energy
balance, which was less positive by 2.54 MJ
(P < 0.002) on the 35 SED than the 60 SED treatment.
Activity induced a profound improvement in
energy balance on both 60% and 35% treatments.
Energy balance was not signi®cantly different from
zero with either ACT treatment (Table 4), and the
balances were not different from each other.
On the 35 FREE treatment energy balance was not
signi®cantly different from the 60 SED treatment, but
was signi®cantly more positive than all other treatments. There were no treatment-day interactions.
Energy substrate selection
The dietary availability of carbohydrate exerted a
strong in¯uence on its oxidation. On the 35% fat
(53% CHO) diet, carbohydrate oxidation was higher
than on each corresponding 60% fat (28% CHO)
treatment. Carbohydrate oxidation was 1.19 MJ
greater on the 35 ACT treatment than the 35 SED
(P < 0.002). However, on the 60% fat diet carbohydrate oxidation was not signi®cantly in¯uenced by
activity (Figure 2). The cost of exercise on the 60
ACT treatment was fuelled by higher fat oxidation Ð
2.2 MJ (P < 0.0001; Figure 3). Even on the 35% fat
diet, fat played a major role in fuelling the cost of the
activity Ð 1.58 MJ, 35 ACT vs 35 SED (P < 0.0001).
On the 35 FREE treatment, carbohydrate oxidation
was signi®cantly higher on day 2 than day 1 (0.91 MJ,
P < 0.001), complemented by a signi®cantly lower fat
oxidation (0.71 MJ, P < 0.002).
Substrate balance
Carbohydrate balance re¯ected the combined effects
of energy balance and diet composition (Figure 2).
Carbohydrate balance was signi®cantly more positive
Figure 2 Carbohydrate oxidation (hatched) stacked on balance
(black) for each day of each treatment. The hatched bar height
above zero represents CHO intake. Error bars represent standard
errors of the mean.
Figure 3 Fat oxidation (hatched) stacked on balance (black) for
each day of each treatment. The hatched bar height above zero
represents fat intake. Error bars represent standard errors of the
mean.
on the 35 SED and 35 FREE diets than on any other
treatments (P < 0.005).
Carbohydrate balance showed a signi®cant day
effect; it was closer to balance on day 2 than day 1
for all treatments (P < 0.02). The carbohydrate oxidation and balance effects were the only signi®cant
time-dependant outcomes found in this study.
Subjects were in positive fat balance for all but the
35 ACT treatment. The balance on the 60 SED
treatment was signi®cantly more positive than on
any other treatment (Figure 4). There was no signi®cant difference in fat balance between 60 ACT and 35
SED, but each was signi®cantly greater than 35 ACT.
Fat balance on the 35 FREE treatment was more
positive than on the 35 ACT treatment. There were
no signi®cant differences between daily balances on
any treatment.
Figure 4 Average daily fat balance on each treatment. Error
bars represent standard errors of the mean. (a) P < 0.0001 vs 35
ACT, 35 SED and 60 ACT, P < 0.005 vs 35 FREE. (b) P < 0.05 vs 35
ACT. (c) P < 0.0001 vs 35 ACT.
Inactivity, diet and energy balance
PR Murgatroyd et al
1274
Discussion
This study aimed to investigate the in¯uences of
inactivity and dietary macronutrient composition on
energy and fat balance and to look for interactions
between them.
The protocol was carefully designed to ensure that
energy expenditures were typical of free-living levels
on the ACT treatments. PAL values of 1.61 con®rm
that this was achieved. The average value 1.25 on the
imposed SED treatments is well below the levels
derived by Black et al21 and represents a profound
inhibition of any plausible habitual level of activity in
healthy men of this age.
Equal care was taken in the design and presentation
of the diets. The diets were developed using principles
employed in previous studies,14,25 in which we had
successfully avoided the hyperphagia often associated
with excessively palatable meals. This resulted in
consistent intake responses both within and between
subjects. Ad libitum energy intakes on the 35 ACT
treatment closely matched expenditure and by inference were, as intended, close to habitual levels.
The study con®rmed the strong in¯uence of diet
composition on total energy intake. As has been
shown by Stubbs et al 14,15 among others, this arises
because the higher energy density of high fat diets is
not balanced by a reduction in food intake. This effect
persists at least over the time scale of most residential
studies and occurs in spite of fat-induced satiety
signals, a phenomenon discussed by Blundell et
al,30 who describe the effect as passive over-consumption.
Energy intakes on day 2 failed to show any adaptive
response to positive energy balances accrued on day 1.
Indeed, on the 60% treatments the intakes on each pair
of days were virtually identical. This insensitivity to
manipulations of diet composition appears to be
associated with positive energy balance. Manipulations producing negative energy balance have recently
been shown to result in near-perfect compensation.25
This suggests a strong asymmetry to appetite control,
providing powerful defence against negative energy
balance whilst tolerating energy surplus.
The energy balance outcome of the treatments can
be clearly seen in Figure 1. The sedentary treatments
led to substantial positive energy balances, over
5 MJ=d in the case of the 60 SED treatment and half
this for 35 SED. The imposition of a sedentary lifestyle was thus suf®cient to induce a highly positive
energy balance, regardless of dietary composition.
On the ACT treatments energy balance was much
closer to zero. Indeed, there was no signi®cant difference between balances on 35 ACT and 60 ACT
treatments and neither differed signi®cantly from
zero. The effects of dietary composition on energy
balance were thus substantially attenuated by physical
activity. The reduced energy intake on the 60 ACT
protocol, compared with 60 SED, though not quite
signi®cant, strongly suggests that activity may have an
additional bene®t of helping to regulate energy intake
on a high-fat diet.
When subjects were overtly overfed and allowed
free exercise (35 FREE treatment), a small increase in
energy expenditure over that seen on the 35 SED
treatment was observed, however expenditure was not
signi®cantly different from that on the 60 SED treatment whose ad libitum intake was used in the de®nition of the overfeeding energy intake. Moreover, PAL
values on the 35 FREE treatment were not signi®cantly different from the 60 SED treatment. It is likely
that any voluntary activity within the con®nes of the
calorimeter would be consciously undertaken. Whilst
acknowledging this constraint, we conclude that there
was no inclination to increase physical activity or to
exercise in response to the overt overfeeding of the 35
FREE treatment in spite of the fact that the PAL, at
the level of the SED protocol, represented a pattern of
extreme inactivity. Consequently, energy balance on
the 35 FREE treatment was substantially more positive than for all other treatments except 60 SED, from
which it did not differ signi®cantly.
The responses of energy substrate oxidation to
changes in store status have been modelled by
Flatt.31 Carbohydrate stores, being much smaller and
more labile than fat stores, in¯uence oxidative fuel
mixture so that departures from carbohydrate balance
are corrected in an auto-regulatory manner.32 This
process can clearly be seen in action in Figure 2, as
displacement of carbohydrate stores feeds through to
increased oxidation and a progressive restoration of
balance at a higher store level. Consequently, carbohydrate balance shows a signi®cant day effect Ð the
only parameter in this study to do so.
Fat complements carbohydrate in the provision of
energy. It is clear from Figure 2 that the close
regulation of carbohydrate balance was facilitated
even by the moderate levels of activity in this study.
Through its effects on carbohydrate, the activity more
effectively and more immediately associated the
energy burden of activity with fat. The fat imbalances
seen in Figure 4 con®rm the bene®ts of an active
lifestyle, particularly in the context of a high-fat diet.
It should, however, be recognised that this ®gure does
not represent the equilibrium picture. Residual carbohydrate imbalance associated with inactivity (Figure
2) can be expected to continue its regression towards
balance, driving fat balance to appear ever more like
the energy balance picture (Figure 1), in which the
bene®ts of activity and the penalties of inactivity are
so clearly evident.
This study took place within a highly controlled
environment. Consequently, one must be cautious
about extrapolating its ®ndings to unrestrained life.
We have surmised that free activity may be inhibited
in this environment. Conversely, the patterns of
energy intake observed here may not re¯ect those
which obtain when subjects have the freedom to select
from a wide range of palatable foods.
Inactivity, diet and energy balance
PR Murgatroyd et al
The study set out to examine the effects of inactivity and dietary composition on energy balance. We
conclude that, within this controlled experimental
environment, energy intake is not down-regulated in
response to either inactivity or a high-fat diet. Indeed,
the adverse in¯uences of inactivity and high fat diet
on energy balance appear to be additive. Activity
proved essential to the avoidance of a signi®cant
positive energy balance.
Acknowledgements
We would like to thank Elaine Collard and Judith
Wills, who prepared the diets; Maxine Durham and
Cathy Baker, our night nurses; Dr Marinos Elia, who
provided medical cover; Heather Davies, who helped
to recruit our volunteers, and of course the volunteers
themselves, who must remain anonymous but without
whom this study could not have been undertaken.
References
1 Prescott-Clarke P, Primatesta P (eds). Health survey for
England 1996. The Stationery Of®ce: London, 1998.
2 Prentice AM, Jebb SA. Obesity in Britain: gluttony or sloth?
Br Med J 1995; 311: 437 ± 439.
3 Hill JO, Peters JC. Exercise and macronutrient balance. Int J
Obes 1995; 19 (Suppl 4): S88 ± S92.
4 King NA, Lluch A, Stubbs RJ, Blundell JE. High dose
exercise does not increase hunger or energy intake in free
living males. Eur J Clin Nutr 1997; 51: 478 ± 483.
5 Garrow JS. Exercise in the treatment of obesity: a marginal
contribution. Int J Obes 1995; 19 (Suppl 4): S126 ± S129.
6 Pavlou K, Krey S, Steffee W. Exercise as an adjunct to weight
loss and maintenance in moderately obese subjects. Am J Clin
Nutr 1989; 49: 1115 ± 1123.
7 Grodstein F, Levine R, Troy L, Spencer T, Colditz GA
Stampfer M. Three-year follow-up of participants in a commercial weight loss program. Arch Intern Med 1996; 156:
1302 ± 1306.
8 Wadden TA, Frey DL. A multicenter evaluation of a proprietary weight loss program for the treatment of marked obesity:
a ®ve year follow-up. Int J Eat Disord 1997; 22: 203 ± 212.
9 Mayer J, Marshall NB, Vitale JJ, Christensen JH, Mashayekh
MB, Stare FJ. Exercise, food intake and body weight in normal
rats and genetically obese adult mice. Am J Physiol 1954; 177:
544 ± 548.
10 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
(2): 169 ± 175.
11 Weststrate JA. Fat and obesity. Int J Obes 1995; 19 (Suppl 5):
S38 ± S43.
12 Lissner L, Heitmann BL. Dietary fat and obesity: evidence
from epidemiology. Eur J Clin Nutr 1995; 49: 79 ± 90.
13 Lawton CL, Burley VJ, Wales JK, Blundell JE. Dietary fat and
appetite control in obese subjects: weak effects on satiation
and satiety. Int J Obes 1993; 17: 630 ± 634.
14 Stubbs RJ, Harbron CG, Murgatroyd PR, Prentice AM. Covert
manipulation of dietary fat and energy density: effect on
substrate ¯ux and food intake in men eating ad libitum. Am
J Clin Nutr 1995; 62: 316 ± 329.
15 Stubbs RJ, Harbron CG, Prentice AM. Covert manipulation of
the dietary fat to carbohydrate ratio of isoenergetically dense
diets: effect on food intake in feeding men ad libitum. Int J
Obes 1996; 20: 651 ± 660.
16 Kendall A, Levitsky DA, Strupp BJ, Lissner L. Weight loss on
a low-fat diet: consequences of imprecision of the control of
food intake in humans. Am J Clin Nutr 1991; 53: 1124 ± 1129.
17 Schundt D, Hill J, Pope-Cordle J, Arnold D, Virts K, Katahn
M. Randomised evaluation of a low fat ad libitum carbohydrate diet for weight reduction. Int J Obes 1993; 17: 623 ± 629.
18 Shah M, McGovern P, French S, Baxter J. Comparison of a
low-fat, ad libitum complex-carbohydrate diet with a lowenergy diet in moderately obese women. Am J Clin Nutr 1994;
59: 980 ± 984.
19 Willett WC. Is dietary fat a major determinant of body fat? Am
J Clin Nutr 1998; 67: 556s ± 562s.
20 Bray GA, Popkin BM. Dietary fat intake does affect obesity!
Am J Clin Nutr 1998; 68: 1157 ± 1173.
21 Black AE, Coward WA, Cole TJ, Prentice AM. Human energy
expenditure in af¯uent societies: an analysis of 574 doublylabelled water measurements. Eur J Clin Nutr 1996; 50:
72 ± 92.
22 Scho®eld WN, Scho®eld C, James WPT. Basal metabolic
rate Ð review and prediction, together with an annotated
bibliography of source material. Hum Nutr: Clin Nutr 1985;
39C(Suppl 1): 1 ± 96.
23 Jebb SA, Elia M. Techniques for the measurement of
body composition: a practical guide. Int J Obes 1993; 17:
611 ± 621.
24 Holland B Welch AA, Unwin ID, Buss DH, Paul AA, Southgate DAT (eds). McCance and Widdowson's the composition
of foods, 5th ed. Royal Society of Chemistry: Cambridge,
1991.
25 Goldberg GR, Murgatroyd PR, McKenna APM, Heavey PM,
Prentice AM. Dietary compensation in response to covert
imposition of negative energy balance by removal of fat or
carbohydrate. Br J Nutr 1998; 80: 141 ± 147.
26 Brown D, Cole TJ, Dauncey MJ, Marrs RW, Murgatroyd PR.
Analysis of gaseous exchange in open circuit indirect calorimetry. Med Biol Engng Comput 1984; 22: 333 ± 338.
27 Murgatroyd PR, Sonko BJ, Wittekind A, Goldberg GR,
Ceesay SM, Prentice AM. Non-invasive techniques for assessing carbohydrate ¯ux. I. Measurement of depletion by indirect calorimetry. Acta Physiol Scand 1993; 147: 91 ± 98.
28 Elia M, Livesey G. Energy expenditure and fuel selection on
biological systems: the theory and practice of calculations
based on indirect calorimetry and tracer methods, In: Simopoulos AP (ed). Control of eating, energy expenditure and the
bioenergetics of obesity. Karger: Basel, 1992; pp68 ± 131.
29 Murgatroyd PR, Davies HL, Prentice AM. Intra-individual
variability and measurement noise in estimates of energy
expenditure by whole body indirect calorimetry. Br J Nutr
1987; 58: 347 ± 356.
30 Blundell JE, Cotton JR, Delargy H, Green S, Greenough A,
King NA, Lawton CL. The fat paradox: fat-induced satiety
signals versus high fat overconsumption. Int J Obes 1995; 19:
832 ± 835.
31 Flatt JP. The difference in storage capacities for carbohydrate
and for fat, and its implications in the regulation of body
weight. Ann NY Acad Sci 1987; 499: 104 ± 123.
32 Shetty PS, Prentice AM, Goldberg GR, Murgatroyd PR,
McKenna APM, Stubbs RJ, Volschenk PA. Alterations in
fuel selection and voluntary food intake in response to isoenergetic manipulation of glycogen stores in humans. Am J
Clin Nutr 1994; 60: 534 ± 543.
1275