- Email: [email protected]
Effects of energy density of daily food intake on long-term energy intake
Physiology & Behavior 81 (2004) 765 – 771
Effects of energy density of daily food intake on long-term energy intake M.S. Westerterp-Plantenga* Department of Human Biology, Maastricht University, P.O. Box 616, 6200 MD, Maastricht, The Netherlands
Abstract An important question that has been raised recently is whether it is mainly the energy density (ED) of the food consumed or its macronutrient composition that determines daily energy intake (EI). In this scope, the effect of ED on EI has been assessed in short-term as well as long-term experiments. Over the short term, i.e., during a meal, it was found that ED affects EI directly; then subjects mainly monitor the weight of the food ingested. Over the long term, the effects of ED on EI are modulated. Average daily energy intake (ADEI) does not only consist of meals but also includes snacks and drinks. ADEI appears to be related to ED of the food and drinks where ED is at least determined by specific macronutrients (primarily fat and carbohydrate), but not when ED is determined only by the weight of water. With respect to the separate effects of the ED of foods and of drinks on ADEI, only ED from foods had a significant relationship with total EI. Moreover, during daily food intake, subjects seem to adapt their portion sizes to estimated EDs. Long-term studies have shown that dietary restraint compensates for the effect of increases in ED on ADEI, whereas unrestrained eaters compensate for the effect of decreases in ED on ADEI. In conclusion, ED determines short-term EI. This cannot be extrapolated to the long term because only the ED of food, and not the ED of drinks, determines total EI. In addition, over the long term, the short-term effect is modulated by dietary restraint and adapted portion sizes. ED is not a universal concept that determines EI, yet rather a characteristic of the macronutrients: mainly fat and carbohydrate contributing to variation in EI. D 2004 Elsevier Inc. All rights reserved. Keywords: Macronutrient composition; Dietary restraint; Obesity; Energy density; Portion size; Universal eating monitor; Water
1. Introduction Energy intake (EI) is not related, at least in a straightforward fashion, to the weight of food intake. Humans derive their energy from the macronutrients: carbohydrates, lipids, proteins, and alcohol. Energy from food and drink is released during the breakage of chemical bonds, and can be used for energy metabolism or can be converted or stored. Metabolisable energy is the gross energy minus energy in faeces and urine. Knowing the macronutrient composition of foods from chemical analysis, the metabolisable energy can be calculated by multiplying the weight of each nutrient by its metabolisable energy value, the Atwater factors, i.e., 16 kJ/g for carbohydrate and protein, and 37 kJ/g for fat and 29 kJ/g alcohol. The relevant food characteristics that play a role in regulation of EI are energy content, macronutrient
* Tel.: +31-43-381566; fax: +31-43-3670976. E-mail address: [email protected] (M.S. Westerterp-Plantenga). 0031-9384/$ – see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.physbeh.2004.04.030
composition, weight, and energy density (ED) [5,12]. An important question that has recently been raised is whether it is mainly the ED of the food consumed or its macronutrient composition that contributes to the variation in daily EI [1,6 –11,13– 15,17]. In this scope, the effect of ED on EI has been assessed in short-term as well as longterm experiments. ED is defined as follows: ED ¼ kJ ðCarbohydrate þ Protein þ Fat þ AlcÞ=g ðCarbohydrate þ Protein þ Fat þ Alc þ water and undigestible partsÞ ED represents metabolisable energy/gross weight, or kJ/g because the Atwater factors are defined like this, and does not represent energy/volume. If volume is used, then one has to take the specific gravity of the food into account. The aim of this paper is to focus on effects of ED on EI over the long term. This requires a short evaluation of the short-term effects first, before possible extrapolation to the long term can be made.
766
M.S. Westerterp-Plantenga / Physiology & Behavior 81 (2004) 765–771
2. Short-term effects of ED on EI The question with respect to short-term effect of ED on EI is centered around the issue whether it is the amount of food ingested that is monitored by the body, or whether it is the amount of energy. In the case that the amount of energy is monitored, no effect of ED on EI is expected. In the case that the amount of food is monitored, ED may have a profound effect on EI. Food intake during a meal has been studied using cumulative food intake curves [13,16] obtained by monitoring eating from the universal eating monitor [16]: a plate, placed on a scale built into a table, and connected to a digital computer. They describe and integrate parameters of consumption of an ad lib single-course meal, i.e., meal size, meal duration, eating rate, change in eating rate, bite size, and bite frequency [13,16]. EI during a meal was assessed by a covert change from 4.8 to 6.1 kJ/g, not affecting the taste or the appearance of the food. This was done by replacing some of the food by tasteless protein or carbohydrate powders or by a ‘burmanier’ [13,16]. This resulted in a change in macronutrient compositions from 55%:15%:30% of energy from carbohydrate/protein/fat to (i) 50%:10%:40% of energy (high fat), (ii) 70%:10%:20% of energy (high carbohydrate), or (iii) 50%:30%:20% of energy (high protein) [13]. A decrease of ED from 4.8 to 4.0 kJ/g was achieved by adding 20 g fiber (guar gum) to the meals with the basic macronutrient composition (C/P/F = 55:15:30). The covert changes in ED of an otherwise familiar meal did not cause changes in the cumulative intake curve parameters [13]. In addition, the shape of the individual cumulative food intake and satiation curves showed very little variability despite experimentally energy-enriched meals (Fig. 1a and b) [13,16]. Obviously, the weight of the food was monitored rather than the energy from food, and stomach distension was the main satiation signal over this period. This was confirmed by studies where subjects monitored the weight of food rather than the energy of food during a meal [7]. Those studies showed that ED significantly influenced intake, even when both the macronutrient content and palatability of the test foods were matched. For example, when individuals were fed diets varying in ED and could eat as much food as they liked, they ate the same amount of food (by weight) and thereby the EI varied directly with ED. Furthermore, when participants consumed foods of low ED, they felt satisfied, despite reductions in EI [7]. Moreover, with respect to the contributing effect of fat on ED, the influence of ED on EI was assessed when the fat content of entre´e meals which were consumed in the laboratory was changed [1]. The main entre´es, consumed ad libitum, were formulated to vary in fat content (25%, 35%, and 45% of energy) and in ED (5.23 kJ/g, or low ED, and 7.32 kJ/g, or high ED) but to have similar palatability. It was shown that ED influenced EI across all fat conditions in
Fig. 1. Example of cumulative food intake over time in kilojoules (a) and in grams (b) of five similar meals only differing in ED and macronutrient composition. Subjects were normal-weight women. Carbohydrate/protein/ fat: high carbohydrate: 70:10:20; high protein: 50:30:20; high fat: 50:10:40 [11,15].
both lean and obese women ( P < .0001). Women consumed less energy in the low (7531 kJ) than in the high (9414 kJ) ED condition. Despite this 20% lower EI, there were only small differences in hunger (7%) and fullness (5%). Across all conditions, women consumed a similar volume, but not weight, of food each day. It was concluded that ED affected
M.S. Westerterp-Plantenga / Physiology & Behavior 81 (2004) 765–771
EI across the different fat conditions and at levels of ED comparable with those in typical diets. Cues related to the amount of food consumed had a greater influence on shortterm intake than does the amount of energy consumed [1]. However, clear effects of variation in ED during the intermeal interval was shown when after energy-enriched lunch meals, as well as after energy-diluted meals with fiber, subjects had a longer intermeal interval (6.5 F 0.5 vs. 5.5 F 0.4 h, and a lower EI during dinner, 2.9 F 0.2 vs. 3.4 F 0.2 MJ) [13,17]. These results suggest not only a late effect from ED, but also a late effect from fiber which is in the opposite direction to the ED effect. With respect to effects of ED on EI over the short term, we and many others conclude that food intake and not EI is monitored, and as a consequence, ED affects EI. However, observations during the intermeal interval indicate that possible long-term effects may be different.
3. Analysis of ED effects on ADEI from food and drinks The question then remains, as to how ED affects ADEI. Having observed that during meals, the weight of food and not the energy is monitored by the body, and given that energy balance is achieved over a week, the body might somehow correct for passive over- or underconsumption following high or low energy-dense meals over longer periods of time. Average daily energy intake (ADEI) consists of intake of food and drinks during meals as well as in between meals. This means that studies approaching effects of ED on ADEI assess possible relationships between average daily ED and ADEI. Any data set that contains fully reliable measurements of the components contributing to ED can be used to analyze the determinants of ED and the relationship between ADEI and average ED. This was for instance executed in three data sets [17]. In these data sets, food availability was ad libitum, with respect to choice, amount, and frequency. The first data set was obtained from a study in which 16 dieticians (age: 34 F 9; BMI: 22.1 F 2.3 kg/m2) monitored their food intake during a week using weighed food records. Their recording of EI was accurate, according to a method for the determination of water turnover using deuterium elimination, together with the determination of body weight [2]. The protocol was executed twice with feedback after the first time (during which they lost weight by undereating). For the present analysis, the food records produced the second time were used, during which no undereating or underrecording occurred [2]. The second data set was from a study with female students (mean F S.D.: age 23 F 4, BMI 22.2 F 3.2 kg/m2) in a respiration chamber. The days where they ate ad lib (which were always preceded by a day during which they were fed in energy balance so that they did not have to compensate for the previous day) are used for the present analysis [17].
767
The third data set is similar to the second one, but for male students (age 25 F 6; BMI 22.9 F 3.1 kg/m2) [17]. The contribution of ED to the explained variation in EI was in the experiment with the dietitians: r = .38; P = .0001; in the respiration chamber experiment with the women: r = .93; P = .0001; in the respiration chamber experiment with the men: r = .17; P = .27. Regression analyses showed that the variation in energy from fat (r = .47; P < .0001), carbohydrate (r = .28; P < .001), in the weight of water (r = .64; P < .0001), fat (r = .47; P < .0001), and carbohydrate (r = .28; P < .001) all contributed to the variation in EI [17]. The variation in the relative proportion of protein was so small that it did not contribute significantly to the variation in ED. From the differences between the outcomes of the three data sets, it appears that ED is related to EI when ED is related to the energy content and the weight of macronutrients (Data sets 1 and 2), and possibly but not necessarily to the weight of water (Data set 1). When water is the dominant component determining ED (r = .77; P < .0001), EI is not related to ED (Data set 3) [17]. In other words, when the variation in ED is only determined by WW, ED does not play a role in EI regulation. This means that EI becomes independent of ED when only the range in the weight of water determines the range in ED. When EI was also determined by the weight of water (r = .33; P < .0001), the weight of water correlated positively with EI, but it correlated negatively with ED [17]. This concerns the weight of water in the food, which is included in food intake. It means that the EI from the food cannot take place without the accompanying water. When water is part of the food, it affects stomach emptying, and thus, satiety, as has been shown in preload studies with a homogenous soup vs. a hetergenous soup [4], as well as with preload studies in which the ED of a soup was decreased by adding water to it, which significantly increased fullness and reduced hunger and subsequent EI at lunch, while the equivalent amount of water served as a beverage with a food did not affect satiety [8]. Because the determinants of ED play different roles in EI, we cannot simply substitute the effects of macronutrients on EI regulation by the effect of ED. This means that the characteristics of the macronutrients still play an important role in EI, and that ED can be considered as one of those. The main effect ED has on EI comes from the effect of fat on ED, and subsequently on EI, as shown in the data sets of the dieticians and the women in the respiration chamber [17]. In conclusion, ADEI is related to ED of the food, including water in the food, when ED is determined by specific macronutrients. Conversely, when ED is only determined by the weight of water, it is not related to EI [17]. Thus, this approach shows some similarities to the results of the short-term experiments; for example, ED affects ADEI when the variation in macronutrient composition,
768
M.S. Westerterp-Plantenga / Physiology & Behavior 81 (2004) 765–771
and possibly the water or fiber content, contributes to the variation in ED. A closer assessment of effects of ED from food or drinks on ADEI is warranted because it is observed that the variation in ED is only explained by the variation in water consumed.
4. Analysis of ED effects from food and drinks separately on ADEI To approach the question of the relative importance of the variation in ED of the food and the variation in ED of drinks with respect to affecting EI, we also analysed the data from Data set 1, the dieticians, where ED was determined by the macronutrients as well as by water, separately for food and drinks. Food was defined by the dietician as all the solids and also semifluids that are eaten with a spoon, including pudding, yogurt, soup, whipped cream, sauce, curds, and prepared cereals. Drinks were defined as all the liquids, including coffee, milk in coffee, tea, soft drinks, fruit juice, syrup for lemonade, beer, wine, liqueur, and bouillon. From a total EI of 9.4 F 1.4 MJ, it appeared that 8.2 F 1.2 MJ was derived from food sources and the remaining 1.2 F 0.3 MJ came from drinks. The ED of the food was 6.5 F 1.0 kJ/g and ED of the drinks was 0.7 F 0.2 kJ/g. Total ED was 3.1 F 0.6 kJ/g. In a stepwise regression, it appeared that ED from both food and drinks contributed to total ED (r = .85; partial F = 43.8, P < .0001, for ED from food and partial F = 46.0, P < .0001, for ED from drinks). However, when assessing effects on total EI, only the ED from food (r = .97; P < .0001) and not the ED from drinks contributed significantly to this. We conclude that for ADEI, the ED from food is an important determinant, and not the ED from drinks. This is in line with the finding that when the weight of water is the only determinant of ED, ED does not contribute to the explained variation in EI. The question regarding whether ED affects ADEI as it does in the short term can be answered positively as long as ED is mainly determined by food but negatively when ED is mainly determined by drinks.
The obese women showed a food intake distribution over three classes of ED of foods, i.e., 24% of energy from 0 to 7.5 kJ/g, 52% of energy from 7.5 to 15 kJ/g, and 24% of energy from 15 to 22.5 kJ/g. The macronutrient composition of their food intake was C/P/F = 39%:17%:44% of energy. In the nonobese women, the food intake distribution over these three classes were 38%, 49%, and 13% of energy, with a macronutrient composition of C/P/F = 46%:17%:37% of energy. The distribution in the obese women was significantly different from the values of the nonobese and from the Dutch food guidelines values (Fig. 2a and b) [14]. With respect to portion sizes, it appeared that the obese women took larger portions of food with a high ED than the nonobese did, and larger-than-standard sizes. They took smaller portions of food with a low ED in comparison with the nonobese and in comparison with the standard sizes. In the nonobese, portion sizes were almost standard (Fig. 2a and b) [14]. With respect to the variation of portion sizes in
5. Adjustment of portion sizes to ED categories of daily food intake Another characteristic of ADEI is that not all food is consumed in identical portion sizes. When energy balance is maintained, differences in portion sizes may be related to EDs of the food, to avoid passive consumption. We assessed the effect of the ED of foods in daily food choice and portion size (as determined by the subject) in obese and nonobese women [14]. From 68 subjects [34 obese and 34 nonobese women, matched for age (20 – 50 years)], controlled food intake diaries of two weekdays and one weekend day were analyzed.
Fig. 2. (a) Weight (g) per portion taken from three different ED classes by obese and normal-weight women. Standard portion sizes by guidelines [12]. (b) Percentages of EI from the different ED categories of food. * P < .01, compared to the normal-weight women [12].
M.S. Westerterp-Plantenga / Physiology & Behavior 81 (2004) 765–771
relation to ED of the food, it was also reported that in healthy nonobese males, portion size was inversely related to EDs of snacks of 7.6 –16.5 kJ/g [3]. Thus, in daily life, portion sizes appear to be a learned, or culturally determined modulating factor that may have
769
become a habit, compensating a straightforward effect of ED on EI [14,17]. For individuals with different energy needs (e.g., normal weight vs. overweight and obese), portion size and ED obviously play a role in attempts to achieve energy balance.
Fig. 3. (a) Percentage energy from fat during a 6-month trial. (b) ED of total consumption during the 6-month trial. (c) ADEI during the 6-month trial. (d) Body weight over the 6-month trial. Dietary restrained (5) and unrestrained (w) men and women were kept on a reduced fat (left) or full fat (right) diet [13]. * P < .05, compared to baseline.
770
M.S. Westerterp-Plantenga / Physiology & Behavior 81 (2004) 765–771
The result is a distribution of EI over three distinguishable ED categories. To consume a certain amount of energy during a day, a certain weight of food needs to be consumed, one that cannot be too large because of stomach filling. In the overweight as well as in the normal-weight subjects, this was achieved by a relatively large consumption of food (50% of total daily EI) from the ED category of 7.5 – 15.0 kJ/g. These observations can easily be understood because most of the food items belonging to this ED category are staple foods, like bread, potatoes, rice, and spaghetti, and protein sources, like lean meat, fish, and legumes. Such products are rich in carbohydrate or protein, and with a 50% energy consumption from this category, a large part of the daily requirement will be reached. Thus, EI can be adjusted to ED by varying intakes in the lower and higher ED categories. For the obese women, it might be easier to choose products with a relatively lower weight and high ED because the women consumed relatively large portions of high ED foods and relatively small portions of low ED foods. The nonobese women consumed portion sizes and a macronutrient composition that did not differ significantly from the composition based on the Dutch food guidelines. With respect to the variation of portion sizes in relation to ED of the food, it has been reported in healthy nonobese males that portion size was adapted to a certain extent to EDs (7.6 –16.5 kJ/g) [3]. From these analyses, some suggestions for weight-reduction diets may be made. Firstly, consumption of energydense food could be replaced by consumption of low energydense food by switching from the highest to the lowest category. This would achieve the ED distribution shown by the nonobese as well as by the guidelines. This implies a conscious adjustment to eating more food of low ED and less food of high ED. This may be necessary in all obese subjects, and for subjects who are vulnerable to becoming obese (i.e., overweight) so that they can maintain bodyweight at a healthy level in spite of a sedentary lifestyle. In summary, portion sizes play a role in adapting EI to ED, which may be practiced in normal-weight as well as in overweight women. The overly approach of this by normal-weight women follows the guidelines, whereas the conservative approach by the overweight women leaves room for treatment advices.
6. Modulating long-term effects of ED on EI Short-term experiments appear to be adequate to study effects of modulating ED on EI, using various tools. Over the long term, the control over interventions is limited, yet a possible effect on body weight may give evidence for possible long-term effects that outcomes from short-term experiments only can speculate on. Of all the macronutrients, the exceptionally high ED of fat affects the effect of ED on EI, most of all through passive overconsumption. However, a long-term change in percentage of energy from fat causing a change in ED and subsequently in EI and body
weight appeared to be compensated for by dietary restraint [15]. In a multicentre study, the MSFat study, carried out in the Netherlands, subjects received a full fat vs. a reduced fat diet during 6 months from a laboratory supermarket [15]. All the foods the subjects took from the shop were recorded and the leftovers were subtracted. The analyses in relation to dietary restraint showed the following. A 6-month reduced fat diet in combination with an unrestrained eating behaviour (which resulted in positive EI compensation) contributed to weight maintenance [15]. Weight reduction was the consequence of a reduced fat diet in combination with restrained eating behaviour that did not compensate for the reduced EI [15]. A full fat diet combined with unrestrained eating behaviour led to increased EI and body weight [15]. Restrained eating behaviour with a full fat diet prevented such an increase in EI and body weight [15]. Thus, dietary restraint compensated for an increase in ED, whereas lack of dietary restraint compensated for a decrease in ED (Fig. 3a – d) [15]. Therefore, dietary restraint (that may use adapted portion sizes) can be considered as a modulatory factor in the effect of ED on EI on the long term.
7. Conclusions Over the short term, subjects monitor the weight of food ingested and therefore, ED affects EI during a meal in a straightforward fashion. This finding cannot be extrapolated to the long term. When, over the long term, fluids are the only factor determining ED, the relationship of ED with EI disappears. Over the long term, ED from food and not from drinks relates significantly to EI. Thus, ED is not a universal concept that determines EI. More appropriately, it is a characteristic of the macronutrients: mainly fat and carbohydrate contributing to variation in EI. Second, over the long term, portion sizes are used to compensate for EDs resulting in less variation in EI than the short-term effect of ED on EI would imply. Third, over the long term, dietary restraint compensates for the effect of a relatively high ED on EI, whereas unrestrained eaters compensate for the effect of relatively low ED on EI. Taken together, we conclude that with respect to effects of ED on EI over the long term, the short-term effect is modulated through effects from fluids, dietary restraint, and adapted portion sizes. Thus, over the long term, ED is not a universal concept that determines EI. More appropriately, it is a characteristic of the macronutrients: mainly fat and carbohydrate contributing to variation in EI.
References [1] Bell EA, Rolls BJ. Energy density of foods affects energy intake across multiple levels of fat content in lean and obese women. Am J Clin Nutr 2001;73(6):999 – 1000.
M.S. Westerterp-Plantenga / Physiology & Behavior 81 (2004) 765–771 [2] Goris AHC, Westerterp KR. Underreporting of habitual food intake— can it be improved? Br J Nutr 2000;83:363 – 9. [3] Green SM, Burley VJ, Blundell J. Effect of fat- and sucrose-containing foods on the size of eating episodes and energy intake in lean males: potential for causing overconsumption. Eur J Clin Nutr 1994; 48:547 – 55. [4] Himaya A, Louis-Sylvestre J. The effect of soup on satiation. Appetite 1998;30:199 – 210. [5] Melanson KJ, Westerterp-Plantenga MS, Saris WHM, Campfield LA. Short term regulation of food intake in humans. In: Westerterp-Plantenga MS, Steffens AB, Tremblay A, editors. Regulation of food intake and energy expenditure. Milan, Italy: EDRA; 1999. p. 37 – 58. [6] Poppit S, Prentice AM. Energy density and its role in the control of food intake: evidence from metabolic and community studies. Appetite 1996;26:153 – 74. [7] Rolls BJ, Bell EA. Intake of fat and carbohydrate: role of energy density. Eur J Clin Nutr 1999;53:S166 – 73. [8] Rolls BJ, Bell EA, Thorwart ML. Water incorporated into a food but not served with a food decreases energy intake in lean women. Am J Clin Nutr 1999;70(4):448 – 55. [9] Spiegel TA. Caloric regulation of food intake in man. J Comp Physiol Psychol 1973;84:24 – 37. [10] Stubbs RJ, Ritz P, Coward WA, Prentice AM. Covert manipulation of the ratio of dietary fat to carbohydrate and energy density: effect on food intake and energy balance in free living men eating ad libitum. Am J Clin Nutr 1995;62:330 – 7. [11] Stubbs RJ, Johnstone AM, O’Reilly LM, Barton K, Reid C. The effect
[12]
[13]
[14]
[15]
[16]
[17]
771
of covertly manipulating the energy density of mixed diets on ad libitum food intake in ‘pseudo free-living’ humans. Int J Obes Relat Metab Disord 1998;22:980 – 7. Stubbs RJ, Raben A, Westerterp-Plantenga MS. Macronutrient metabolism and appetite. In: Westerterp-Plantenga MS, Steffens AB, Tremblay A, editors. Regulation of food intake and energy expenditure. Milan, Italy: EDRA; 1999. p. 59 – 84. Westerterp-Plantenga MS, Westerterp KR, Nicolson NA, Mordant A, Schoffelen PFM, ten Hoor F. The shape of the cumulative food intake curve in humans, during basic and manipulated meals. Physiol Behav 1990;47:569 – 76. Westerterp-Plantenga MS, Pasman WJ, Yedema MJW, WijckmansDuijsens NEG. Energy intake adaptation of food to extreme energy densities of food by obese and non-obese women. Eur J Clin Nutr 1996;50:401 – 7. Westerterp-Plantenga MS, Wijckmans-Duijsens NEG, Verboeket-van de Venne WPG, de Graaf K, van het Hof KH, Weststrate JA. Energy intake and body weight effects of six months reduced or full fat diets, as a function of dietary restraint. Int J Obes Relat Metab Disord 1998;22:14 – 22. Westerterp-Plantenga MS. Eating behaviour in humans, characterized by cumulative food intake curves—a review. Neurosci Biobehav Rev 2000;24:239 – 48. Westerterp-Plantenga MS. Analysis of energy density of food in relation to energy intake regulation in human subjects. Br J Nutr 2001;85: 351 – 61.
Effects of energy density of daily food intake on long-term energy intake M.S. Westerterp-Plantenga* Department of Human Biology, Maastricht University, P.O. Box 616, 6200 MD, Maastricht, The Netherlands
Abstract An important question that has been raised recently is whether it is mainly the energy density (ED) of the food consumed or its macronutrient composition that determines daily energy intake (EI). In this scope, the effect of ED on EI has been assessed in short-term as well as long-term experiments. Over the short term, i.e., during a meal, it was found that ED affects EI directly; then subjects mainly monitor the weight of the food ingested. Over the long term, the effects of ED on EI are modulated. Average daily energy intake (ADEI) does not only consist of meals but also includes snacks and drinks. ADEI appears to be related to ED of the food and drinks where ED is at least determined by specific macronutrients (primarily fat and carbohydrate), but not when ED is determined only by the weight of water. With respect to the separate effects of the ED of foods and of drinks on ADEI, only ED from foods had a significant relationship with total EI. Moreover, during daily food intake, subjects seem to adapt their portion sizes to estimated EDs. Long-term studies have shown that dietary restraint compensates for the effect of increases in ED on ADEI, whereas unrestrained eaters compensate for the effect of decreases in ED on ADEI. In conclusion, ED determines short-term EI. This cannot be extrapolated to the long term because only the ED of food, and not the ED of drinks, determines total EI. In addition, over the long term, the short-term effect is modulated by dietary restraint and adapted portion sizes. ED is not a universal concept that determines EI, yet rather a characteristic of the macronutrients: mainly fat and carbohydrate contributing to variation in EI. D 2004 Elsevier Inc. All rights reserved. Keywords: Macronutrient composition; Dietary restraint; Obesity; Energy density; Portion size; Universal eating monitor; Water
1. Introduction Energy intake (EI) is not related, at least in a straightforward fashion, to the weight of food intake. Humans derive their energy from the macronutrients: carbohydrates, lipids, proteins, and alcohol. Energy from food and drink is released during the breakage of chemical bonds, and can be used for energy metabolism or can be converted or stored. Metabolisable energy is the gross energy minus energy in faeces and urine. Knowing the macronutrient composition of foods from chemical analysis, the metabolisable energy can be calculated by multiplying the weight of each nutrient by its metabolisable energy value, the Atwater factors, i.e., 16 kJ/g for carbohydrate and protein, and 37 kJ/g for fat and 29 kJ/g alcohol. The relevant food characteristics that play a role in regulation of EI are energy content, macronutrient
* Tel.: +31-43-381566; fax: +31-43-3670976. E-mail address: [email protected] (M.S. Westerterp-Plantenga). 0031-9384/$ – see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.physbeh.2004.04.030
composition, weight, and energy density (ED) [5,12]. An important question that has recently been raised is whether it is mainly the ED of the food consumed or its macronutrient composition that contributes to the variation in daily EI [1,6 –11,13– 15,17]. In this scope, the effect of ED on EI has been assessed in short-term as well as longterm experiments. ED is defined as follows: ED ¼ kJ ðCarbohydrate þ Protein þ Fat þ AlcÞ=g ðCarbohydrate þ Protein þ Fat þ Alc þ water and undigestible partsÞ ED represents metabolisable energy/gross weight, or kJ/g because the Atwater factors are defined like this, and does not represent energy/volume. If volume is used, then one has to take the specific gravity of the food into account. The aim of this paper is to focus on effects of ED on EI over the long term. This requires a short evaluation of the short-term effects first, before possible extrapolation to the long term can be made.
766
M.S. Westerterp-Plantenga / Physiology & Behavior 81 (2004) 765–771
2. Short-term effects of ED on EI The question with respect to short-term effect of ED on EI is centered around the issue whether it is the amount of food ingested that is monitored by the body, or whether it is the amount of energy. In the case that the amount of energy is monitored, no effect of ED on EI is expected. In the case that the amount of food is monitored, ED may have a profound effect on EI. Food intake during a meal has been studied using cumulative food intake curves [13,16] obtained by monitoring eating from the universal eating monitor [16]: a plate, placed on a scale built into a table, and connected to a digital computer. They describe and integrate parameters of consumption of an ad lib single-course meal, i.e., meal size, meal duration, eating rate, change in eating rate, bite size, and bite frequency [13,16]. EI during a meal was assessed by a covert change from 4.8 to 6.1 kJ/g, not affecting the taste or the appearance of the food. This was done by replacing some of the food by tasteless protein or carbohydrate powders or by a ‘burmanier’ [13,16]. This resulted in a change in macronutrient compositions from 55%:15%:30% of energy from carbohydrate/protein/fat to (i) 50%:10%:40% of energy (high fat), (ii) 70%:10%:20% of energy (high carbohydrate), or (iii) 50%:30%:20% of energy (high protein) [13]. A decrease of ED from 4.8 to 4.0 kJ/g was achieved by adding 20 g fiber (guar gum) to the meals with the basic macronutrient composition (C/P/F = 55:15:30). The covert changes in ED of an otherwise familiar meal did not cause changes in the cumulative intake curve parameters [13]. In addition, the shape of the individual cumulative food intake and satiation curves showed very little variability despite experimentally energy-enriched meals (Fig. 1a and b) [13,16]. Obviously, the weight of the food was monitored rather than the energy from food, and stomach distension was the main satiation signal over this period. This was confirmed by studies where subjects monitored the weight of food rather than the energy of food during a meal [7]. Those studies showed that ED significantly influenced intake, even when both the macronutrient content and palatability of the test foods were matched. For example, when individuals were fed diets varying in ED and could eat as much food as they liked, they ate the same amount of food (by weight) and thereby the EI varied directly with ED. Furthermore, when participants consumed foods of low ED, they felt satisfied, despite reductions in EI [7]. Moreover, with respect to the contributing effect of fat on ED, the influence of ED on EI was assessed when the fat content of entre´e meals which were consumed in the laboratory was changed [1]. The main entre´es, consumed ad libitum, were formulated to vary in fat content (25%, 35%, and 45% of energy) and in ED (5.23 kJ/g, or low ED, and 7.32 kJ/g, or high ED) but to have similar palatability. It was shown that ED influenced EI across all fat conditions in
Fig. 1. Example of cumulative food intake over time in kilojoules (a) and in grams (b) of five similar meals only differing in ED and macronutrient composition. Subjects were normal-weight women. Carbohydrate/protein/ fat: high carbohydrate: 70:10:20; high protein: 50:30:20; high fat: 50:10:40 [11,15].
both lean and obese women ( P < .0001). Women consumed less energy in the low (7531 kJ) than in the high (9414 kJ) ED condition. Despite this 20% lower EI, there were only small differences in hunger (7%) and fullness (5%). Across all conditions, women consumed a similar volume, but not weight, of food each day. It was concluded that ED affected
M.S. Westerterp-Plantenga / Physiology & Behavior 81 (2004) 765–771
EI across the different fat conditions and at levels of ED comparable with those in typical diets. Cues related to the amount of food consumed had a greater influence on shortterm intake than does the amount of energy consumed [1]. However, clear effects of variation in ED during the intermeal interval was shown when after energy-enriched lunch meals, as well as after energy-diluted meals with fiber, subjects had a longer intermeal interval (6.5 F 0.5 vs. 5.5 F 0.4 h, and a lower EI during dinner, 2.9 F 0.2 vs. 3.4 F 0.2 MJ) [13,17]. These results suggest not only a late effect from ED, but also a late effect from fiber which is in the opposite direction to the ED effect. With respect to effects of ED on EI over the short term, we and many others conclude that food intake and not EI is monitored, and as a consequence, ED affects EI. However, observations during the intermeal interval indicate that possible long-term effects may be different.
3. Analysis of ED effects on ADEI from food and drinks The question then remains, as to how ED affects ADEI. Having observed that during meals, the weight of food and not the energy is monitored by the body, and given that energy balance is achieved over a week, the body might somehow correct for passive over- or underconsumption following high or low energy-dense meals over longer periods of time. Average daily energy intake (ADEI) consists of intake of food and drinks during meals as well as in between meals. This means that studies approaching effects of ED on ADEI assess possible relationships between average daily ED and ADEI. Any data set that contains fully reliable measurements of the components contributing to ED can be used to analyze the determinants of ED and the relationship between ADEI and average ED. This was for instance executed in three data sets [17]. In these data sets, food availability was ad libitum, with respect to choice, amount, and frequency. The first data set was obtained from a study in which 16 dieticians (age: 34 F 9; BMI: 22.1 F 2.3 kg/m2) monitored their food intake during a week using weighed food records. Their recording of EI was accurate, according to a method for the determination of water turnover using deuterium elimination, together with the determination of body weight [2]. The protocol was executed twice with feedback after the first time (during which they lost weight by undereating). For the present analysis, the food records produced the second time were used, during which no undereating or underrecording occurred [2]. The second data set was from a study with female students (mean F S.D.: age 23 F 4, BMI 22.2 F 3.2 kg/m2) in a respiration chamber. The days where they ate ad lib (which were always preceded by a day during which they were fed in energy balance so that they did not have to compensate for the previous day) are used for the present analysis [17].
767
The third data set is similar to the second one, but for male students (age 25 F 6; BMI 22.9 F 3.1 kg/m2) [17]. The contribution of ED to the explained variation in EI was in the experiment with the dietitians: r = .38; P = .0001; in the respiration chamber experiment with the women: r = .93; P = .0001; in the respiration chamber experiment with the men: r = .17; P = .27. Regression analyses showed that the variation in energy from fat (r = .47; P < .0001), carbohydrate (r = .28; P < .001), in the weight of water (r = .64; P < .0001), fat (r = .47; P < .0001), and carbohydrate (r = .28; P < .001) all contributed to the variation in EI [17]. The variation in the relative proportion of protein was so small that it did not contribute significantly to the variation in ED. From the differences between the outcomes of the three data sets, it appears that ED is related to EI when ED is related to the energy content and the weight of macronutrients (Data sets 1 and 2), and possibly but not necessarily to the weight of water (Data set 1). When water is the dominant component determining ED (r = .77; P < .0001), EI is not related to ED (Data set 3) [17]. In other words, when the variation in ED is only determined by WW, ED does not play a role in EI regulation. This means that EI becomes independent of ED when only the range in the weight of water determines the range in ED. When EI was also determined by the weight of water (r = .33; P < .0001), the weight of water correlated positively with EI, but it correlated negatively with ED [17]. This concerns the weight of water in the food, which is included in food intake. It means that the EI from the food cannot take place without the accompanying water. When water is part of the food, it affects stomach emptying, and thus, satiety, as has been shown in preload studies with a homogenous soup vs. a hetergenous soup [4], as well as with preload studies in which the ED of a soup was decreased by adding water to it, which significantly increased fullness and reduced hunger and subsequent EI at lunch, while the equivalent amount of water served as a beverage with a food did not affect satiety [8]. Because the determinants of ED play different roles in EI, we cannot simply substitute the effects of macronutrients on EI regulation by the effect of ED. This means that the characteristics of the macronutrients still play an important role in EI, and that ED can be considered as one of those. The main effect ED has on EI comes from the effect of fat on ED, and subsequently on EI, as shown in the data sets of the dieticians and the women in the respiration chamber [17]. In conclusion, ADEI is related to ED of the food, including water in the food, when ED is determined by specific macronutrients. Conversely, when ED is only determined by the weight of water, it is not related to EI [17]. Thus, this approach shows some similarities to the results of the short-term experiments; for example, ED affects ADEI when the variation in macronutrient composition,
768
M.S. Westerterp-Plantenga / Physiology & Behavior 81 (2004) 765–771
and possibly the water or fiber content, contributes to the variation in ED. A closer assessment of effects of ED from food or drinks on ADEI is warranted because it is observed that the variation in ED is only explained by the variation in water consumed.
4. Analysis of ED effects from food and drinks separately on ADEI To approach the question of the relative importance of the variation in ED of the food and the variation in ED of drinks with respect to affecting EI, we also analysed the data from Data set 1, the dieticians, where ED was determined by the macronutrients as well as by water, separately for food and drinks. Food was defined by the dietician as all the solids and also semifluids that are eaten with a spoon, including pudding, yogurt, soup, whipped cream, sauce, curds, and prepared cereals. Drinks were defined as all the liquids, including coffee, milk in coffee, tea, soft drinks, fruit juice, syrup for lemonade, beer, wine, liqueur, and bouillon. From a total EI of 9.4 F 1.4 MJ, it appeared that 8.2 F 1.2 MJ was derived from food sources and the remaining 1.2 F 0.3 MJ came from drinks. The ED of the food was 6.5 F 1.0 kJ/g and ED of the drinks was 0.7 F 0.2 kJ/g. Total ED was 3.1 F 0.6 kJ/g. In a stepwise regression, it appeared that ED from both food and drinks contributed to total ED (r = .85; partial F = 43.8, P < .0001, for ED from food and partial F = 46.0, P < .0001, for ED from drinks). However, when assessing effects on total EI, only the ED from food (r = .97; P < .0001) and not the ED from drinks contributed significantly to this. We conclude that for ADEI, the ED from food is an important determinant, and not the ED from drinks. This is in line with the finding that when the weight of water is the only determinant of ED, ED does not contribute to the explained variation in EI. The question regarding whether ED affects ADEI as it does in the short term can be answered positively as long as ED is mainly determined by food but negatively when ED is mainly determined by drinks.
The obese women showed a food intake distribution over three classes of ED of foods, i.e., 24% of energy from 0 to 7.5 kJ/g, 52% of energy from 7.5 to 15 kJ/g, and 24% of energy from 15 to 22.5 kJ/g. The macronutrient composition of their food intake was C/P/F = 39%:17%:44% of energy. In the nonobese women, the food intake distribution over these three classes were 38%, 49%, and 13% of energy, with a macronutrient composition of C/P/F = 46%:17%:37% of energy. The distribution in the obese women was significantly different from the values of the nonobese and from the Dutch food guidelines values (Fig. 2a and b) [14]. With respect to portion sizes, it appeared that the obese women took larger portions of food with a high ED than the nonobese did, and larger-than-standard sizes. They took smaller portions of food with a low ED in comparison with the nonobese and in comparison with the standard sizes. In the nonobese, portion sizes were almost standard (Fig. 2a and b) [14]. With respect to the variation of portion sizes in
5. Adjustment of portion sizes to ED categories of daily food intake Another characteristic of ADEI is that not all food is consumed in identical portion sizes. When energy balance is maintained, differences in portion sizes may be related to EDs of the food, to avoid passive consumption. We assessed the effect of the ED of foods in daily food choice and portion size (as determined by the subject) in obese and nonobese women [14]. From 68 subjects [34 obese and 34 nonobese women, matched for age (20 – 50 years)], controlled food intake diaries of two weekdays and one weekend day were analyzed.
Fig. 2. (a) Weight (g) per portion taken from three different ED classes by obese and normal-weight women. Standard portion sizes by guidelines [12]. (b) Percentages of EI from the different ED categories of food. * P < .01, compared to the normal-weight women [12].
M.S. Westerterp-Plantenga / Physiology & Behavior 81 (2004) 765–771
relation to ED of the food, it was also reported that in healthy nonobese males, portion size was inversely related to EDs of snacks of 7.6 –16.5 kJ/g [3]. Thus, in daily life, portion sizes appear to be a learned, or culturally determined modulating factor that may have
769
become a habit, compensating a straightforward effect of ED on EI [14,17]. For individuals with different energy needs (e.g., normal weight vs. overweight and obese), portion size and ED obviously play a role in attempts to achieve energy balance.
Fig. 3. (a) Percentage energy from fat during a 6-month trial. (b) ED of total consumption during the 6-month trial. (c) ADEI during the 6-month trial. (d) Body weight over the 6-month trial. Dietary restrained (5) and unrestrained (w) men and women were kept on a reduced fat (left) or full fat (right) diet [13]. * P < .05, compared to baseline.
770
M.S. Westerterp-Plantenga / Physiology & Behavior 81 (2004) 765–771
The result is a distribution of EI over three distinguishable ED categories. To consume a certain amount of energy during a day, a certain weight of food needs to be consumed, one that cannot be too large because of stomach filling. In the overweight as well as in the normal-weight subjects, this was achieved by a relatively large consumption of food (50% of total daily EI) from the ED category of 7.5 – 15.0 kJ/g. These observations can easily be understood because most of the food items belonging to this ED category are staple foods, like bread, potatoes, rice, and spaghetti, and protein sources, like lean meat, fish, and legumes. Such products are rich in carbohydrate or protein, and with a 50% energy consumption from this category, a large part of the daily requirement will be reached. Thus, EI can be adjusted to ED by varying intakes in the lower and higher ED categories. For the obese women, it might be easier to choose products with a relatively lower weight and high ED because the women consumed relatively large portions of high ED foods and relatively small portions of low ED foods. The nonobese women consumed portion sizes and a macronutrient composition that did not differ significantly from the composition based on the Dutch food guidelines. With respect to the variation of portion sizes in relation to ED of the food, it has been reported in healthy nonobese males that portion size was adapted to a certain extent to EDs (7.6 –16.5 kJ/g) [3]. From these analyses, some suggestions for weight-reduction diets may be made. Firstly, consumption of energydense food could be replaced by consumption of low energydense food by switching from the highest to the lowest category. This would achieve the ED distribution shown by the nonobese as well as by the guidelines. This implies a conscious adjustment to eating more food of low ED and less food of high ED. This may be necessary in all obese subjects, and for subjects who are vulnerable to becoming obese (i.e., overweight) so that they can maintain bodyweight at a healthy level in spite of a sedentary lifestyle. In summary, portion sizes play a role in adapting EI to ED, which may be practiced in normal-weight as well as in overweight women. The overly approach of this by normal-weight women follows the guidelines, whereas the conservative approach by the overweight women leaves room for treatment advices.
6. Modulating long-term effects of ED on EI Short-term experiments appear to be adequate to study effects of modulating ED on EI, using various tools. Over the long term, the control over interventions is limited, yet a possible effect on body weight may give evidence for possible long-term effects that outcomes from short-term experiments only can speculate on. Of all the macronutrients, the exceptionally high ED of fat affects the effect of ED on EI, most of all through passive overconsumption. However, a long-term change in percentage of energy from fat causing a change in ED and subsequently in EI and body
weight appeared to be compensated for by dietary restraint [15]. In a multicentre study, the MSFat study, carried out in the Netherlands, subjects received a full fat vs. a reduced fat diet during 6 months from a laboratory supermarket [15]. All the foods the subjects took from the shop were recorded and the leftovers were subtracted. The analyses in relation to dietary restraint showed the following. A 6-month reduced fat diet in combination with an unrestrained eating behaviour (which resulted in positive EI compensation) contributed to weight maintenance [15]. Weight reduction was the consequence of a reduced fat diet in combination with restrained eating behaviour that did not compensate for the reduced EI [15]. A full fat diet combined with unrestrained eating behaviour led to increased EI and body weight [15]. Restrained eating behaviour with a full fat diet prevented such an increase in EI and body weight [15]. Thus, dietary restraint compensated for an increase in ED, whereas lack of dietary restraint compensated for a decrease in ED (Fig. 3a – d) [15]. Therefore, dietary restraint (that may use adapted portion sizes) can be considered as a modulatory factor in the effect of ED on EI on the long term.
7. Conclusions Over the short term, subjects monitor the weight of food ingested and therefore, ED affects EI during a meal in a straightforward fashion. This finding cannot be extrapolated to the long term. When, over the long term, fluids are the only factor determining ED, the relationship of ED with EI disappears. Over the long term, ED from food and not from drinks relates significantly to EI. Thus, ED is not a universal concept that determines EI. More appropriately, it is a characteristic of the macronutrients: mainly fat and carbohydrate contributing to variation in EI. Second, over the long term, portion sizes are used to compensate for EDs resulting in less variation in EI than the short-term effect of ED on EI would imply. Third, over the long term, dietary restraint compensates for the effect of a relatively high ED on EI, whereas unrestrained eaters compensate for the effect of relatively low ED on EI. Taken together, we conclude that with respect to effects of ED on EI over the long term, the short-term effect is modulated through effects from fluids, dietary restraint, and adapted portion sizes. Thus, over the long term, ED is not a universal concept that determines EI. More appropriately, it is a characteristic of the macronutrients: mainly fat and carbohydrate contributing to variation in EI.
References [1] Bell EA, Rolls BJ. Energy density of foods affects energy intake across multiple levels of fat content in lean and obese women. Am J Clin Nutr 2001;73(6):999 – 1000.
M.S. Westerterp-Plantenga / Physiology & Behavior 81 (2004) 765–771 [2] Goris AHC, Westerterp KR. Underreporting of habitual food intake— can it be improved? Br J Nutr 2000;83:363 – 9. [3] Green SM, Burley VJ, Blundell J. Effect of fat- and sucrose-containing foods on the size of eating episodes and energy intake in lean males: potential for causing overconsumption. Eur J Clin Nutr 1994; 48:547 – 55. [4] Himaya A, Louis-Sylvestre J. The effect of soup on satiation. Appetite 1998;30:199 – 210. [5] Melanson KJ, Westerterp-Plantenga MS, Saris WHM, Campfield LA. Short term regulation of food intake in humans. In: Westerterp-Plantenga MS, Steffens AB, Tremblay A, editors. Regulation of food intake and energy expenditure. Milan, Italy: EDRA; 1999. p. 37 – 58. [6] Poppit S, Prentice AM. Energy density and its role in the control of food intake: evidence from metabolic and community studies. Appetite 1996;26:153 – 74. [7] Rolls BJ, Bell EA. Intake of fat and carbohydrate: role of energy density. Eur J Clin Nutr 1999;53:S166 – 73. [8] Rolls BJ, Bell EA, Thorwart ML. Water incorporated into a food but not served with a food decreases energy intake in lean women. Am J Clin Nutr 1999;70(4):448 – 55. [9] Spiegel TA. Caloric regulation of food intake in man. J Comp Physiol Psychol 1973;84:24 – 37. [10] Stubbs RJ, Ritz P, Coward WA, Prentice AM. Covert manipulation of the ratio of dietary fat to carbohydrate and energy density: effect on food intake and energy balance in free living men eating ad libitum. Am J Clin Nutr 1995;62:330 – 7. [11] Stubbs RJ, Johnstone AM, O’Reilly LM, Barton K, Reid C. The effect
[12]
[13]
[14]
[15]
[16]
[17]
771
of covertly manipulating the energy density of mixed diets on ad libitum food intake in ‘pseudo free-living’ humans. Int J Obes Relat Metab Disord 1998;22:980 – 7. Stubbs RJ, Raben A, Westerterp-Plantenga MS. Macronutrient metabolism and appetite. In: Westerterp-Plantenga MS, Steffens AB, Tremblay A, editors. Regulation of food intake and energy expenditure. Milan, Italy: EDRA; 1999. p. 59 – 84. Westerterp-Plantenga MS, Westerterp KR, Nicolson NA, Mordant A, Schoffelen PFM, ten Hoor F. The shape of the cumulative food intake curve in humans, during basic and manipulated meals. Physiol Behav 1990;47:569 – 76. Westerterp-Plantenga MS, Pasman WJ, Yedema MJW, WijckmansDuijsens NEG. Energy intake adaptation of food to extreme energy densities of food by obese and non-obese women. Eur J Clin Nutr 1996;50:401 – 7. Westerterp-Plantenga MS, Wijckmans-Duijsens NEG, Verboeket-van de Venne WPG, de Graaf K, van het Hof KH, Weststrate JA. Energy intake and body weight effects of six months reduced or full fat diets, as a function of dietary restraint. Int J Obes Relat Metab Disord 1998;22:14 – 22. Westerterp-Plantenga MS. Eating behaviour in humans, characterized by cumulative food intake curves—a review. Neurosci Biobehav Rev 2000;24:239 – 48. Westerterp-Plantenga MS. Analysis of energy density of food in relation to energy intake regulation in human subjects. Br J Nutr 2001;85: 351 – 61.