International Journal of Obesity (1998) 22, 885±892
ß 1998 Stockton Press All rights reserved 0307±0565/98 $12.00
http://www.stockton-press.co.uk/ijo
Covert manipulation of energy density of high
carbohydrate diets in `pseudo free-living'
humans
RJ Stubbs1, AM Johnstone1, CG Harbron2 and C Reid2
1
The Rowett Research Institute and 2Biomathematics and Statistics Scotland, The Rowett Research Institute, Greenburn Road,
Bucksburn, Aberdeen AB21 9SB, UK
OBJECTIVE: This study examined the effects of varying the energy density (ED) of high carbohydrate (HC) diets on
food and energy intake (EI), subjective hunger and body weight in humans.
DESIGN: Randomised cross-over design. Subjects were each studied twice during 14 d, throughout which they had
ad libitum access to one of two covertly-manipulated diets.
SUBJECTS AND METHODS: Six healthy men (mean age (s.d.) ˆ 32.17 y s.d. (5.26 y), mean weight ˆ 69.74 kg
s.d. (2.75 kg), mean height ˆ 1.76 m s.d. (0.05 m), body mass index (BMI) ˆ 22.57 (2.2) kg=m2) were studied. The fat,
carbohydrate (CHO) and protein content (as % energy) and ED of each diet were 21 : 66 : 13% and 357 kJ=100 g, (lowenergy density (LED)) or 22 : 66 : 12% and 629 kJ=100 g (high-energy density (HED)). A medium fat diet was provided at
maintenance (1.6 BMR, MF for 2 d) before each ad libitum period. Subjects could alter the amount, but not the
composition of foods eaten.
RESULTS: Mean EI was 8.67 and 14.82 MJ=d on the LED and HED diets, respectively. Subjects felt signi®cantly more
hungry on the LED diet, than on the HED diet (F1,16038.28; P < 0.001) and found the diets to be similarly pleasant
(72.72 mm vs 71.54 mm (F1,3920.31; P ˆ 0.579)). Mean body weight decreased on the LED diet at a rate of 0.1 kg=d and
increased at 0.06 kg=d on the HED diet (F1,13186.60; P < 0.001), giving total weight changes of 71.41 kg and ‡ 0.84 kg,
respectively, both of which were signi®cantly different from zero (P < 0.01).
CONCLUSION: Excess EI is possible on HC, HED diets, at least under conditions where diet selection is precluded.
Comparison of these results with previous studies, which altered ED using fat, suggests that CHO may be a better cue
for hunger than fat.
Keywords: carbohydrate; fat; energy; macronutrients; food intake; appetite; humans; glucostatic theory
Introduction
A large body of literature now implicates high fat
(HF) diets in the aetiology of obesity. Prospective
epidemiological studies suggest that fat consumption
is a risk factor for subsequent weight gain. Crosssectional studies also associate HF (but not always
high energy) intakes with higher levels of body fat.1
Laboratory-based studies have repeatedly shown that
covert changes in the energy density (ED) of the diet,
primarily by using fat, appear to be poorly compensated for when dietary fat content is either surreptitiously altered2 ± 8 or when the fat is substituted by
non-absorbed fat mimetics (for example, see Ref. 9).
These ®ndings have led to a resurgence of interest in
glucostatic and glycogenostatic models of feeding as
potential physiological mechanisms by which HF
Correspondence: R. J. Stubbs, The Rowett Research Institute,
Greenburn Road, Bucksburn, Aberdeen AB21 9SB,UK.
Received 22 August 1997; revised 15 April 1998; accepted
23 April 1998
diets promote excess energy intake (EI).10 ± 12 These
carbohydrate (CHO)-based models of feeding, essentially suggest that CHO status (peripheral utilisation
or stores) is the primary physiological factor which
exerts an unconditioned, negative feedback on feeding
behaviour and EI.
A number of testable predictions about the effects
of the diet on feeding behaviour can be inferred from
these models: (i) that CHO stores or metabolism, exert
negative feedback on EI, (ii) because of such feedback, HF diets (which are low in CHO) will promote
excess EI, due to the purported drive to eat to maintain
CHO status, (iii) that manipulation of CHO status
(oxidation or stores) will reciprocally in¯uence EI and
(iv) voluntary excesses in EI should be dif®cult, if not
impossible to achieve, when subjects are fed on high
CHO (HC) diets, because of the strength of negative
feedback arising from satiety signals related to CHO
status.
A number of laboratory studies have attempted to
test the above predictions (i±iii) in humans: by
examining the day-to-day relationship between nutrient balance=metabolism and subsequent EI of men
Energy density, carbohydrates and energy intake
RJ Stubbs et al
886
housed in a calorimeter over seven days;8 by examining the effect of dietary macronutrients at the same
level of ED on appetite and EI;13 ± 16 and by manipulating CHO stores and assessing their impact on
subsequent EI.17 ± 19 In general, these and other studies
suggest that changes in CHO status alone, may not be
the major unconditioned physiological factor exerting
negative feedback on subsequent EI. Other aspects of
nutrient balance=metabolism may exert important
physiological in¯uences on appetite and energy balance in humans.
A fourth prediction of CHO-based models of feeding, is that excess EI is dif®cult to achieve when
subjects are fed ad libitum on HC diets. Surprisingly
few studies have addressed this issue, although one
study of a culturally-prescribed overfeeding regime
has shown that, under atypical conditions at least,
excess EI and weight gain are possible on HC diets.20
The question therefore remains `how easy is it to
ingest excess energy on HC diets?'.
This study examined the effects of covertly varying
the energy (and hence predominantly (CHO) density
of HC diets in a manner similar to that in which
dietary ED has previously been altered using predominantly fat,7,8 in order to test whether increasing
the ED of HC diets can promote excess EI and weight
gain.
Materials and methods
Subjects and subject characteristics
Six healthy men (mean age (s.d.) ˆ 32.17 y
s.d. (5.26 y), mean weight ˆ 69.74 kg s.d. (2.75 kg),
mean height ˆ 1.76 m s.d. (0.05 m), body mass index
(BMI) ˆ 22.57 s.d. (2.2) kg=m2), were recruited by
advertisement. All had a history of weight stability
and none were taking any medication during this
study. Height was measured to the nearest 0.5 cm
using a stadiometer (Holtain Ltd, Crymych, UK).
Subject were weighed (corrected to nude) each morning of the study, after voiding and before eating, to the
nearest 10 g on a digital scale (DIGI DS-410, CMS
Weighing Equipment Ltd, London).
Study design
Each subject was studied twice in 16 d. The experiments required that they be resident in, but not
con®ned to, the Human Nutrition Unit's metabolic
suite throughout each study period. Each 16 d period
began with 2 d equilibration on a medium fat diet at
1.6 basal metabolic rate (BMR). During the subsequent 14 d, the subjects had continuous ad libitum
access to one of the two covertly manipulated diets
which were low in ED (357 kJ=100 g, LED) or high in
ED (629 kJ=100 g, HED) (Table 1). Each dish was
made in two versions, corresponding to the LED and
HED regimes. The ED across different meal types,
Table 1 Mean energy density and macronutrient composition
per 100 g for each of the diets. Values are given as mean percent
one standard error of the mean (s.e.m.)
Low energy
density
s.e.m.
High energy
density
s.e.m.
% Fat
% CHO
% Protein
Energy density
(kJ=100 g)
21.22
65.40
13.38
357
(0.60)
22.39
(0.91)
65.39
(0.30)
12.22
(14)
629
(0.50)
(0.63)
(0.50)
(18)
CHO ˆ carbohydrate.
within each diet varied, but in each case the ED (per g
of food) of the HED version of that food was approximately 1.5±2.0 times greater than that of the LED
version of the same food. All foods were homogeneous and of a constant measurable composition. Soft
drinks were also available to subjects as diet drinks on
the LED and the regular CHO containing versions of
the same drinks on the HED diet. The CHO density of
the diet was largely altered by using the maltodextrin
CHO supplement MaxiJoule (Scienti®c Hospital Supplies Ltd., Liverpool, UK), which was chosen because
it has a chain length intermediate between monosaccharides and polysaccharides, and because the solubility of MaxiJoule enables large amounts of CHO to be
added to the food without markedly changing the
overall appearance of the food. This type and magnitude of covert manipulation would not have been
possible using common, normal foods. The order of
diets was randomised in a counterbalanced design
across subjects. The energy and macronutrient composition of each item of the diet, the diet recipes,
menus and instructions for preparation of the foods
can be obtained by contacting the authors. Diets were
given on a 3 d rotating menu.
Each subject had a refrigerator that was ®lled every
day with the items on the menu for that day. The
previous day's food remains were removed, weighed
and recorded. All diets were pre-prepared and only
required heating by the subjects using a microwave. A
separate freezer contained spare meals which volunteers could defrost and cook in the microwave. Extra
snacks could be ordered from the kitchen throughout
the day. All main courses were presented in excess in
large casserole dishes that typically contained 0.8±
1.0 kg of food. Subjects served themselves from these
dishes. They were instructed to replace the remains of
all food items not eaten in one of the refrigerator
compartments. Should subjects wish to consume a
second helping of the same dish at a different time,
they were instructed to defrost a new portion from the
freezer which contained extras (this took only a few
minutes in the microwave). They recorded in a small
diary what they ate and when it was eaten. These
procedures enabled meal time, size, composition and
frequency to be determined. Each subject could select
items from their refrigerator at any time of the day
or night. Four subjects were engaged in full-time
Energy density, carbohydrates and energy intake
RJ Stubbs et al
employment throughout the study. It was ensured that
they could heat and eat their food as they did in the
Unit. Subjects completed subjective hunger, appetite
and palatability ratings 15 min after each meal, as
described by Hill and Blundell.21 There was a minimum washout period of one week between diets,
throughout which subjects had access to free food.
The subjects entirely determined their own daily
activity regimes. Therefore, we did not ensure that
physical activity was strictly controlled across treatments, although subjects reported that they did not
change their normal routines during the course of the
study. The study was approved by the Grampian
Health Board and University of Aberdeen Joint
Ethical Committee.
Statistical analysis
Energy, weight and macronutrient intakes were analysed by ANOVA with subject and menu day as
blocking factors, diet and run number as factors, and
day of run as a covariate.
The analysis of the visual analogue ratings was
complicated by the large number of missing observations for some subjects. As this may bias the results, a
check was conducted for systematic differences in the
times of non-response between diets. A logistic
regression was performed with binary response, observation or missing value. Terms for subject, diet, time
and the interactions between subject and diet, and
subject and time were included in the model. The
interactions between time and diet and subject, time
and diet were then added to the model. After ®nding
no evidence for systematic differences in response,
average ratings were calculated for each day and these
analysed by analysis of variance with factors diet and
blocking factor subject.
Ratings of pleasantness and satisfaction were analysed by ANOVA with subject as a blocking factor
and diet, run, meal and the interactions between them
as factors. All analyses were performed using the
Genstat 5 package (Genstat 5, Rothampstead Experimental Station, Harpenden, UK).
Covert manipulation
The 3 d rotating menu was designed so that there
would be two versions of each dish corresponding to
the two macronutrient manipulations. The success of
the covert manipulation was tested by presenting the
two versions at once, to members of the Rowett's
Human Nutrition Unit staff. They were then told that
one of the two versions of each dish was the LED or
the HED version. The foods were continually developed until it was dif®cult to differentiate between the
two diets. This development necessitated the use of
thickening agents and sweeteners in the LED version,
and various spices and tastants were added to both
diets, in similar proportions, to mask any unusually
sweet tastes. Post-study interview con®rmed that none
of the volunteers were aware that the diets had been
systematically manipulated with reference to the CHO
content of the diet.
Results
Food, energy and macronutrient intakes
Table 2 gives average food, energy and nutrient
intakes for the two dietary treatments, together with
the standard error of the difference (SED), variance
ratios and statistical probability for the main treatment
effects. Average daily food and drink intakes were
Table 2 Average daily food, energy and nutrient intakes for the two dietary treatments, together with
the variance ratios and statistical probability for the main treatment effects
Food and drink (kg)
s.e.m.
Water (kg)
s.e.m.
Solid food (kg)
s.e.m.
Energy (MJ)
s.e.m.
Protein (MJ)
s.e.m.
Fat (MJ)
s.e.m.
CHO (MJ)
s.e.m.
Sugar (g)
s.e.m.
Starch (g)
s.e.m.
NSP (g)
s.e.m.
LED
HED
3.01
(0.08)
2.47
(0.06)
2.56
(61.67)
8.67
(0.22)
1.17
(0.03)
1.83
(0.05)
5.57
(0.14)
146
(5)
191
(6)
18
(1)
2.77
(0.07)
1.93
(0.05)
2.27
(74.34)
14.82
(0.38)
1.67
(0.04)
3.02
(0.08)
9.79
(0.26)
165
(7)
386
(10)
13
(1)
sed
Variance
ratio (F1,5)
Probability
0.125
3.53
0.119
0.06
35.44
0.002
5.87
0.06
0.86
50.74
< 0.001
0.06
77.44
< 0.001
0.12
92.26
< 0.001
0.68
38.46
0.002
30
0.41
0.552
16
163.47
< 0.001
9.16
ˆ 0.029
119.0
1.6
LED ˆ low-energy density; HED ˆ high-energy density; sed ˆ standard error of the difference;
CHO ˆ carbohydrate; NSP ˆ non-starch polysaccharide
887
Energy density, carbohydrates and energy intake
RJ Stubbs et al
888
Figure 1 Mean (s.e.m.) daily intake of food and energy for the
six subjects on the two dietary treatments. ANOVA showed
treatment effects to be signi®cant for energy intake (El,
P < 0.001), but not for food intake (P ˆ 0.119).
similar across treatments, but approached signi®cance
(P ˆ 0.06) when analysed without soft drinks (that is,
solid food only). EI was thus 1.7 times greater on the
HED diet than on the LED diet (Figure 1). Menu-day
effects were signi®cant but will not be discussed
further. The intakes of fat, CHO (Figure 2) and
starch varied considerably between treatments, re¯ecting the dietary manipulations. It is important to note
that the program used to calculate dietary intakes,
classi®ed MaxiJoule, the maltodextrin supplement, as
pure starch. Sugar intakes were similar across dietary
treatments. The mean daily intake of non starch
polysaccharides (NSP) on the two dietary treatments
varied slightly and signi®cantly between diets (Table
2). These differences were unlikely to have been
suf®ciently great to contaminate the experimental
manipulation.
Auto-regulation of intake
The regression coef®cient for the effect of the previous day's intake on the following day's intake was
70.17 (F1,143 ˆ 4.24; P < 0.041), estimating that on
average, an increase in EI of 1 MJ resulted in a fall in
EI of 0.17 MJ on the subsequent day. There was no
differential between the two diets in terms of the
Figure 2 Mean (s.e.m.) daily protein, carbohydrate (CHO) and
fat intakes for the six subjects on the two dietary treatments.
ANOVA showed treatment effects to be signi®cant (P < 0.002).
strength of this response (F1,142 ˆ 0.53; P ˆ 0.468).
However, the apparent effect of day to day regulation
is confounded with the rotating menu day effect.
When menu day effects were also included in the
regression model, the effect of the previous day's
intake on the subsequent day's intake vanished
(F1,119 ˆ 0.53; P ˆ 0.467). The same regressions for
weight of food eaten yielded a regression co-ef®cient
of 0.08 (F1,143 ˆ 0.93; P ˆ 0.338). Diet and day effects
were also not signi®cant.
Subjectively rated hunger, appetite and perceived
pleasantness of the diets
There was no evidence of any systematic differences
in the timings of completed and uncompleted visual
analogue scales between the two diets (X2 96 ˆ 73:22)
Energy density, carbohydrates and energy intake
RJ Stubbs et al
889
Figure 3 Mean (s.e.m.) daily hunger (expressed in mm along a
100 mm scale) for the six subjects on the two dietary treatments.
ANOVA showed treatment effects to be signi®cant (P < 0.001).
and the mean deviance of the logistic model was
1.1304, suggesting that the model gave a reasonable
®t with no need to model over-dispersion. We therefore proceeded as described above with the calculated
daily averages.
Subjects felt more hungry on the LED diet with a
mean rating of 30 mm compared with 26 mm on the
HED diet [F1,160 ˆ 38.28; P < 0.001]. This difference
was observed in all six subjects (Figure 3). There was
no signi®cant difference in subjectively rated hunger
between the two runs (F1,160 ˆ 2,43; P ˆ 0.121). Similar results were observed for desire to eat, urge to eat
and prospective consumption. Subjects were less full
on the LED diet with a mean rating of 43 mm
compared with 45 mm on the HED diet (F1,160 ˆ
7.62; P ˆ 0.006). This difference was observed in
four of the six subjects. Subjects were also fuller on
the ®rst run, with an average rating of 45 mm compared to an average of 43 mm on the second run
(F1,160 ˆ 5.83; P ˆ 0.017).
The mean ratings for pleasantness were 73 mm and
72 mm and for satisfaction were 78 mm and 78 mm on
the LED and HED diets, respectively. These were not
signi®cantly different at the 5% level (F1,392 ˆ 0.31;
P ˆ 0.579) and (F1,392 ˆ 0.21; P ˆ 0.645), respectively.
Body weight changes
Subject 6 was omitted from the analysis of the body
weights, as measurements were only reliably
recorded on one of the diets, owing to instrument
error during one run. However, his patterns of intake
and his body weight followed similar general trends
to the other ®ve subjects. There was a decrease in
body weight on the LED of 0.10 kg=d, whilst on the
HED diet, body weight increased at 0.06 kg=d
(F1,131 ˆ 86.60; P < 0.001) (Figure 4), giving total
weight changes of 71.40 kg and ‡ 0.84 kg, respectively. The slopes were signi®cantly different from
zero. The rate of body weight change on the second
run was 0.05 kg=d less than on the ®rst run
(F1,131 ˆ 9.07; P ˆ 0.003).
Figure 4 Mean (s.e.m.) change in body weight, relative to day 1
(set to zero), for ®ve of the six subjects on the two dietary
treatments. ANOVA showed treatment effect to be signi®cant
(P < 0.001).
Discussion
The effect of energy density and diet composition on
energy intake
In previous studies, allowing subjects to feed ad
libitum on low, medium and high fat diets (LF, MF
and HF, respectively) whose ED increased with percentage fat, led to similar ad libitum food and EI that
tended to be proportional to the ED of the diet (for
examples, see Refs 6±8). In the `pseudo free living'
study using those diets, intakes were almost identical
to our calorimeter study at 9.11, 10.32 and 12.78 MJ=d
(c.f. 9.02, 10.2 and 12.35, respectively) while expenditures (estimated by doubly-labelled water) were
12.45, 12.10 and 11.97 on the LF, MF and HF diets
(c.f. 9.48, 9.53 and 9.78 MJ=d in the calorimeter),
respectively. By day 14, mean body weight change
was 70.74 kg, 70.51 kg and 0.09 kg on the LF, MF
and HF diets, respectively (c.f. 0.2, 0.5 and 0.9 kg by
day 7 in the calorimeter, respectively). In the present
study, subjects also ate a similar amount of food on
each diet, consequently overconsuming energy on the
HED diet and underconsuming energy on the LED
diet. This effect was stable, with no compensatory
change in EI across week 1 and 2 of the study, as
demonstrated by the absence of a signi®cant week or
diet-week interaction (P ˆ not signi®cant (NS) at the
5% level of signi®cance).
Variability in food intake
As in our previous studies, food intake was constant
across days and diets, relative to everyday life.7 ± 14
The weight of food eaten was similar on both diets,
but slightly greater on the LED. We believe that it is
likely that had the subjects continued to feed on these
diets for several months then EI would have gradually
increased on the LED diet. Intake may have also
gradually decreased on the HED diet. This effect
has been observed by Kendall et al,22 over 11 weeks
when subjects gradually increased food and EI while
feeding on LF, LED diets. The weight of food eaten in
Energy density, carbohydrates and energy intake
RJ Stubbs et al
890
the present study, on average, was similar to that of
previous studies. The values for food intake, given in
these results, appear slightly higher, owing to the fact
that they also included soft drinks. Without soft
drinks, food intake was very similar to those found
in previous studies.
Evidence of glucostatic or glycogenostatic regulation
Mayer's glucostatic theory10 and Flatt's glycogenostatic model12 both predict that the HED diet would
have led to a decrease in EI, due to potent negative
feedback, which would have increased satiety and
suppressed food intake. In this study a large differential in CHO intake and EI occurred on diets where
EI should have been well matched to requirements
and there was no convergence of EI between the HED
and LED diets as each treatment progressed. Thus,
covertly increasing the ED of high CHO diets, using
maltodextrins, did not have a large effect in limiting
excess EI, as predicted by CHO-based models of
feeding. However, a mean daily subjective hunger
on the HED diet was signi®cantly lower than for the
LED diet in all subjects. They also reported verbally
that they felt more hungry on this diet, but did not
verbally report any corresponding lack of hunger or
elevation of satiety on the HED diet. Thus decreasing
the ED of a HC diet led to signi®cant, detectable
increases in subjective hunger, but not food intake,
over two weeks in these subjects.
This is of interest because it is known that decreases
in prior fat or CHO intake will elicit compensatory
increased in EI at subsequent feeding occasions (for
example, see Ref. 23). Indeed, the fact that food intake
was around 200 g=d greater, albeit non-signi®cantly,
on the LED diet may re¯ect an incipient tendency to
increase food intake in response to elevated hunger. It
is not unusual for the effects of diets to be more
accurately detected (in terms of hunger) than
responsed to (in terms of food intake) in studies of
this nature since subjects are exposed to unfamiliar
diets, the composition of which is covertly manipulated. A detailed discussion of the processes underlying this phenomenon is given by Blundell.24 It may
therefore be concluded from these present data that
increasing the ED of HC diets did not reliably increase
satiety or decrease food intake, but that decreasing the
ED of the diet elevated hunger, did not increase food
intakeÐprobably because of the extent to which the
low ED of this diet precluded larger food, and hence
energy intakes.
These ®ndings are in keeping with a number of data
sources which suggest that decreases in CHO oxidation and=or plasma glucose, increase feeding. A small
(6±12%) drop in plasma glucose, predicts meal initiation in rodents25 and humans.26 A small infusion of
glucose, which blocks this drop, will delay feeding
by up to 3 h. Pharmacological inhibition of glucose
oxidation increases feeding in rats.27 Post-prandial
CHO oxidation correlates negatively with hunger in
humans.28 Parenteral infusions of amino acids, glucose or lipid, in rats have been shown to lead to
caloric compensation. However the degree of caloric
compensation was only complete when amino acids
were infused, around 70% for glucose and less than
50% when fat was infused.29 Similar effects have been
found in one study in humans.30
It is known that pharmacologically-induced
decreases in both CHO and fat metabolism can
in¯uence feeding behaviour in rodents.27 There is
also new evidence which suggests that the central
nervous system (CNS) monitors CHO and fat oxidation separately, but that these signals are integrated in
a way which allows the animals to take account of
overall fuel statusÐat least in relation to glucoprivic
and lipoprivic rodent-based feeding models.31 Thus,
there is evidence that under conditions of lipoprivation and glucoprivation decrements in fat and CHO
intake and oxidation are detected and compensated
for. Increments in fat and CHO intake provide less
than complete compensation.32 ± 33 There is no clear
physiological reason why excess EI cannot occur
when subjects feed on high CHO diets. Humans
appear to compensate more accurately for prior
decreases in EI than for excess EI. The key question
relates to whether diets containing a high density of
readily digested and absorbed CHO can promote
excess EI in normal-weight subjects feeding ad libitum in everyday life. This question as yet remains
unanswered.
Limitations of the present results
As in most studies of feeding behaviour, there were
limitations to the design of this experiment. The
study design (and hence conclusions arising from
it) was subject to the following constraints:
(i)
Subjects had ad libitum access to a diet of nearconstant composition on each treatment. Their
response, in terms of food intake, could only
vary quantitatively, selection of different foods,
which vary in composition and=or ED, was
precluded. It is probable that, if available to
subjects, selection of more energy-dense foods
may have occurred in response to ingesting the
LED diet, since subjects were aware of being
more hungry than on the HED diet. This experimental design may therefore have constrained
compensatory responses.
(ii) The experimental environment of the Human
Nutrition Unit allows great precision and accuracy with respect to dietary intakes and diet
formulation, while maintaining (as far as possible) the naturalistic appearance, taste and texture of foods. However these are not common,
familiar or `real' foods. Furthermore the ED of
the diets range from 357±629 kJ=100, which in
absolute terms is a narrow range relative to foods
available in the free-living environment. It
should not be assumed that the direction and
Energy density, carbohydrates and energy intake
RJ Stubbs et al
magnitude of response that subjects show under
these conditions can be directly extrapolated to
every day life. This statement holds for all such
studies conducted under similar conditions (for
example, see Refs 6±8). Many factors, such as
sensory variety, nutrient interactions and environmental in¯uences usually associated with
eating (for example, social facilitation of
intake), are effectively controlled-out of these
experiments. Equal attention should be given to
studies conducted in more naturalistic environments.
(iii) This study was conducted for a very small
number of subjects and these results should
perhaps be replicated before attempting to extrapolate them to other groups or the general
population.
(iv) Compensatory feeding responses to dietary
manipulations are likely to have a large learned
component (for example, see Refs 34±35).
Experimental protocols, such as that described
in this paper, tend to be designed in a manner
that precludes learning (covert manipulation,
randomised order of diets). The present study
has tested the role of diet induced changes in
CHO status, as an unconditioned signal which
may directionally in¯uence feeding behaviour
and EI. Conditioning cues, which may be important in learning processes that couple feeding
behaviour to the postabsorptive consequences of
ingesting different nutrients, were controlledout of this experiment. This is potentially important because Booth et al34 have demonstrated
that satiety can be associatively conditioned by
pairing appropriate cues to CHO ingestion. It is
likely that subjects would learn to compensate
more rapidly for changes in the ED and composition of manipulated diets, if the manipulations
were overtly paired to appropriate learning cues.
Ideally, the results of overt and covert manipulations should be compared.
Implications of these ®ndings
The ®ndings of this study imply that it is possible
to consume excess energy ad libitum (at least over
periods of two weeks) when the ED of the diet is
elevated, by primarily using CHO of a maltodextrin
source. Maltodextrin-based CHO supplements
should be effective in patients who appetites are
not pathologically suppressed. In our opinion, there
is no reason at present to assume that the use of
shorter chain CHO would have produced a different
effect. These results have implications for the use of
`lower fat' and `low-fat' foods. Such foods, if high
in ED may not be as preventative of weight gain as
is often implied. These data do not agree with
models of feeding behaviour which suggest that
altering peripheral physiology by ingesting foods
that are dense in readily absorbed CHO will elevate
unconditioned postabsorptive satiety signals, to the
extent that they will preclude excess EI. On the
other hand, these data do not refute the notion that
decreases in CHO intake and EI produce postingestive consequences that elevate hunger and (under
facilitative conditions) are likely to produce more
accurate caloric compensation than in experiments
which use dietary manipulations where fat content
and ED co-vary.
Conclusion
These data suggest that excess EI can occur on HC,
HED diets, at least under conditions where diet
selection is precluded. Comparison of these results
with previous studies which altered ED using fat,
suggests that CHO may be a better cue for hunger
than fat.
Acknowledgements
We are grateful to Scienti®c Hospital Supplies,
Wavertree Business Park, Liverpool, for donating
the MaxiJoule. This work was supported by the
Scottish Of®ce Department of Agriculture, Environment and Fisheries.
References
1 Lissner L, Heitmann BL. Dietary fat and obesity: evidence
from epidemiology. Eur J Clin Nutr 1995; 49: 79 ± 90.
2 Lissner L, Levitsky DA, Strupp BJ, Kalkwarf HJ, Roe DA.
Dietary fat and the regulation of energy intake in human
subjects. Am J Clin Nutr 1987; 46: 886 ± 892.
3 Duncan KH, Bacon JA, Weinsier RL. The effects of high and
low ED diets on satiety, energy intake, and eating time of
obese and non-obese subjects. Am J Clin Nutr 1987; 46:
886 ± 892.
4 Tremblay A, Lavallee N, Almeras N, Allard L, Despres JP,
Bouchard C. Nutritional determinants of the increase in energy
intake associated with a high-fat diet. Am J Clin Nutr 1991; 53:
1134 ± 1137.
5 Tremblay A, Plourde G, Despres JP, Bouchard C. Impact of
dietary fat content and fat oxidation on energy intake in
humans. Am J Clin Nutr 1989; 49: 799 ± 805.
6 Thomas CD, Peters JC, Reed GW, Abumrad NN, Sun Ming,
Hill JO. Nutrient balance and energy expenditure during ad
libitum feeding of high-fat and high-carbohydrate diets in
humans. Am J Clin Nutr 1992; 55: 934 ± 942.
7 Stubbs RJ, Ritz P, Coward WA, Prentice AM. The effect of
covert manipulation of the dietary fat to carbohydrate ratio and
energy density on food intake and energy balance in `freeliving' men, feeding ad libitum. Am J Clin Nutr 1995; 62:
330 ± 339.
8 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 feeding ad libitum. Am
J Clin Nutr 1995; 62: 316 ± 330.
9 Bray GA, Sparti A, Windhauser MM, York DA. Effect of two
weeks fat replacement by olestra on food intake and energy
metabolism. FASEB 9(3): pp A439.
10 Mayer J. Regulation of energy intake and the body weight.
Ann NY Acad Sci 1955; 63: 15 ± 43.
891
Energy density, carbohydrates and energy intake
RJ Stubbs et al
892
11 Russek M. Current status of the hepatostatic theory of food
intake control. Appetite 1981; 2: 137 ± 143.
12 Flat JP. The difference in storage capacities for carbohydrate
and for fat, and its implications for the regulation of body
weight. Ann NY Acad Sci 1987; 499: 104 ± 123.
13 Van Stratum P, Lussenburg RN, van Wezel LA, Vergroesen
AJ, Cremer HD. The effect of dietary carbohydrate: fat ratio
on energy intake by adult women. Am J Clin Nutr 1978; 31:
206 ± 212.
14 Stubbs RJ, Harbron CG, Prentice AM. The effect of covertly
manipulating the dietary fat to carbohydrate ratio of isoenergetically dense diets on ad libitum food intake in free-living
humans. Int J Obes 1996; 20: 651 ± 660.
15 Rolls BJ, Kim-Harris S, Fischman MW, Foltin RW, Moran
TH, Stoner SA. Satiety after preloads with different amounts
of fat and carbohydrate: implications for obesity. Am J Clin
Nutr 1994; 60: 476 ± 487.
16 Johnstone AM, Harbron CG, Stubbs RJ. Macronutrients,
appetite and day-to-day food intake in humans. Eur J Clin
Nutr 1996; 50: 418 ± 430.
17 Stubbs RJ, Goldberg G, Murgatroyd PR, Prentice AM. Carbohydrate balance and day-to-day food intake in man. Am J
Clin Nutr 1993; 57: 897 ± 903.
18 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 man. Am J Clin
Nutr 1994; 60: 534 ± 543.
19 Snitker S, Larson DE, Tatarinni PA, Ravussin E. Ad libitum
food intake in humans after manipulation of glycogen stores.
Am J Clin Nutr 1997; 65: 941 ± 946.
20 Pasquet P, Apfelbaum R. Recovery of initial body weight and
composition after long-term massive overfeeding in man. Am J
Clin Nutr 1994; 60: 861 ± 863.
21 Hill AJ, Blundell JE. Nutrients and behaviour: research
strategies for the investigation of taste characteristics food
preferences, hunger sensations and eating patterns in man.
J Psychol Res 1982; 17: 203 ± 212.
22 Kendall A, Levitsky DA, Strupp BJ, Lissner L. Weight-loss on
a low fat diet: consequence of the impression of the control of
food intake in humans. Am J Clin Nutr 1991; 53: 1124 ± 1129.
23 Heavey PM, McKenna APM, Goldberg GR, Murgatroyd PR,
Prentice AM. Underfeeding by reduction in fat or carbohydrate intake: effects on energy expenditure, macronutrient
oxidation and subsequent food intake in lean men. Proc Nutr
Soc 1995; 54: 160A.
24 Blundell JE. Food intake and body weight regulation. In:
Bouchard C, Bray G (eds), Regulation of Body Weight:
Biological and Behavioural Mechanisms. Dahlem workshop
reports, Life Sciences Research Report 57, John Wiley and
Sons Ltd.: Chichester.
25 Camp®eld LA, Smith FJ. Transient declines in blood glucose
signal meal initiation. Int J Obes 1990; 14(suppl. 3): 15 ± 33.
26 Camp®eld LA, Smith FJ, Rosenbaum M, Geary N. Human
hunger: is there a role for blood glucose dynamics? (letter)
Appetite 1992; 8: 244.
27 Friedman MI, Tordoff MG. Fatty acid oxidation and glucose
utilisation interact to control food intake in rats. Am J Physiol
1986; 251: R840 ± R845.
28 Thompson DA, Campbell RG. Hunger in humans induced by
2-Deoxy-D-glucose: Glucoprivic control of taste preference
and food intake. Science 1977; 198: 1065 ± 1068.
29 Walls EK, Koopmans HS. Differential effects of intravenous
glucose, amino acids and lipid on daily food intake in rats. Am
J Physiol 1992; 262: R225 ± R234.
30 Gil K, Skeie B, Kvetan V, Askanazi J, Friedman MI. Parenteral nutrition and oral intake: Effect of glucose and fat
infusion. J Parenter Enteral Nutr 1991; 15: 426 ± 432.
31 Ritter S, Calingasan NY. Neural substrates for metabolic
controls of feeding. In: Fernstrom, J.D. and Miller, G.D.
(eds). Appetite and body weight regulation: Sugar, fat and
macronutrient substitutes. CRC Press Inc.: Florida, 1994.
32 Foltin RW, Fischman MW, Moran TH, Rolls BJ, Kelly TH.
Caloric compensation for lunches varying in fat and carbohydrate contents by humans in a residential laboratory. Am J Clin
Nutr 1990; 52: 969 ± 980.
33 Foltin RW, Rolls BJ, Moran TH, Kelly TH, McNelis AL,
Fischman MW. Caloric, but not macronutrient compensation
by humans for required eating occasions with meals and
snacks varying in fat and carbohydrate. Am J Clin Nutr
1992; 55: 331 ± 342.
34 Booth DA, Mather D, Fuller J. Starch content of associatively
conditioned human appetite and satiation, indexed by intake
and eating pleasantness of starch- paired ¯avours. Appetite
1982; 3: 163 ± 184.
35 Kyriazakis I, Emmans GC. Selection of a diet by growing pigs
given choices between foods differing in contents protein and
rapeseed meal. Appetite 1992; 19: 121 ± 132.