Evidence that eating frequency is inversely related to body ... - Nature

International Journal of Obesity (1998) 22, 105±112 ß 1998 Stockton Press All rights reserved 0307±0565/98 $12.00

Evidence that eating frequency is inversely related to body weight status in male, but not female, non-obese adults reporting valid dietary intakes SE Drummond, NE Crombie, MC Cursiter and TR Kirk Centre for Food Research and Department of Dietetics and Nutrition, Queen Margaret College, Clerwood Terrace, Edinburgh, EH12 8TS, UK

OBJECTIVE: To investigate the relationships between eating frequency (EF) and body weight status and to determine whether these relationships can be explained in terms of differences in physical activity levels, macronutrient intakes or energy compensation. DESIGN: Cross-sectional design; free-living subjects, 48 men and 47 women (aged 20±55 y, body mass index (BMI) 18±30), recruited in a workplace setting. MEASUREMENTS: Height and weight; skinfold thickness (four sites); EF, energy and macronutrient intakes (food diary, unweighed, recorded for seven consecutive days); physical activity (7 d activity diary and heart rate monitoring over 48 h period). RESULTS: In men there was a signi®cant negative correlation between EF and body weight, and an inverse relationship with body mass index (BMI). EF was positively correlated with % energy from carbohydrate, although not with total energy intake. In women, there was no relationship between EF and body weight status; however, there were signi®cant positive correlations between EF and total energy intake, and between EF and intakes of total carbohydrate and sugars. For both men and women, there were associations between EF and physical activity levels, approaching statistical signi®cance. CONCLUSIONS: In men, the association between increased EF and lower body weight status may have been in¯uenced by increased physical activity levels. As energy intake did not increase with EF, men appear to have compensated by reducing the mean energy consumed per eating episode. Energy compensation did not take place in women, with women who ate most frequently having the highest energy intakes, although this did not lead to higher BMIs. Physical activity, through participation in active leisure pursuits, may have been an important factor in weight control in women. The % contribution of carbohydrate to total energy was positively correlated with EF in both men and women, and further analysis showed that snack foods provided a higher proportion of carbohydrate than did foods eaten as meals. These results indicate that a high EF is likely to lead to a high carbohydrate diet, which may be favourable for weight control. Our ®ndings suggest that in this population, a high EF was associated with leanness in men, and there was no link between EF and body weight status in women. Keywords: obesity; eating frequency; macronutrient composition; energy intake compensation; physical activity levels

Introduction Prevention of obesity is one of the highest priority nutritional targets in the UK,1 as it is associated with increased risks of cardiovascular disease, diabetes, hypertension and cancer.2,3 The prevalence of obesity is increasing, with clinical obesity (body mass index, BMI > 30) increased by 7% in men and by 8% in women since the early 1980s. It is predicted that if this rate of escalation continues, 18% of men and 25% of women would be clinically obese by the year 2005.1 It has been

Correspondence: Terry R Kirk. Received 16 January 1997; revised 24 September 1997; accepted 30 September 1997

suggested that snacking may be a cause of obesity, if the consumption of snack foods and `calori®c' drinks between meals increases total energy intake.4,5 The Booth hypothesis4 suggests that snacks taken more than an hour before meals fail to elicit satiety and compensatory responses at subsequent meals, leading to overeating and obesity. However, epidemiological evidence does not support the theory that snacking adversely affects energy balance and body weight. Some early studies of eating patterns and body weight status found an inverse relationship between adiposity and overall eating frequency (EF),6±8 while more recent studies have reported a gender difference,9 or no relationship between the EF and body weight status.10±12 It is important to note that in those studies which have found a relationship, the relationship is consistently inverse.

Eating frequency and body weight SE Drummond et al

106

Some explanation for the con¯icting results from previous studies is offered in a recent study by Summerbell et al12 who found an inverse relationship between feeding frequency and BMI in adolescents, but not in three older age groups. However, the relationship, signi®cant in the adolescent group as a whole, was lost when invalid dietary records from under-reporters were excluded. Summerbell argues that these ®ndings have implications for the interpretation of earlier studies, which did not screen dietary records for valid energy intakes. Dietary under-reporting has been shown to be particularly prevalent in obese subjects;13,14 it may be biased toward snacks, giving the impression that a low EF is positively related to adiposity.15±18 Moreover, selective underreporting of snacks by obese subjects may confuse the interpretation of studies of the macronutrient composition of diets and the control of body weight.16±18 The problem of under-reporting has been discussed in a number of recent studies, which indicate that underreporting may be more widespread and more dif®cult to identify, than was previously thought.19±21 Even when under-reporters have been excluded, the importance of assessing behavioural factors, such as eating restraint, has been highlighted recently by Crawley and Summerbell.22 In a follow-up study of a sample of 731 free-living teenagers, they found an inverse relationship between feeding frequency and BMI, which was signi®cant after exclusion of under-reporters. However, the relationship was negated when dieting behaviour and eating restraint were taken into account, leading to the conclusion that the relationship was an artefact of dieting and dietary restraint in teenagers rather than simply under-reporting.22 This suggests that the contradictory ®ndings from earlier studies of EF and body weight may be partly attributable to differences in eating behaviour in different study populations. These and other methodological issues affecting studies of the periodicity of eating, particularly the lack of consensus on a speci®c de®nition of eating patterns, have been reviewed by Gatenby.23 Recent investigations of obesity and the relationships with food intake have centred on the macronutrient composition of the diet, rather than the overall energy content. Studies indicate that high fat diets are positively associated with a high body weight status whereas high carbohydrate diets are associated with leanness.24±26 An inverse relationship between percentage energy from sugar and body weight status is also well-documented.27,28 Dietary fat is thought to be obesity-promoting due to its high energy density, its palatability,29 its weak effect on satiety30 and the fact that it is more ef®ciently converted to body fat than is carbohydrate.31 By contrast, carbohydrate is less energy dense, more satiating,32 its conversion to body fat is inef®cient and it may induce a higher thermogenic response than fat.31 In terms of eating patterns, evidence from two studies33,34 indicates that, compared with main meals, snacks tend to be higher in carbohydrate and lower in fat; thus an increased EF may be associated with an increased

carbohydrate : fat ratio, which may be bene®cial in body weight control. Recent work by Stubbs et al35 suggests that a key factor in¯uencing appetite is the energy density of foods, which in turn is determined by their macronutrient composition. Increased snacking need not lead to obesity if energy compensation takes place, that is, if the energy consumed at main meals is reduced, although this energy compensation response varies with different population groups and with differing levels of dietary restraint. It appears that normal weight, dietary unrestrained males and children compensate well, by adjusting energy intake at main meals in response to energy de®cit or surfeit at preceding eating occasions.32,36±38 Obese subjects have consistently been found to be poor compensators39,40 and women are more likely than men to be poor compensators. Sedentary lifestyles and lack of physical activity are thought to have at least as important a role as diet in the aetiology of obesity,41,42 and this is now recognised as one of the most modi®able components of the energy balance equation.15 Physical activity increases energy expenditure due to the direct energy cost of the activity, the enhanced dietary induced thermogenesis (DIT) levels found after physical activity43 and by promoting the development of lean body mass inducing a higher basal metabolic rate per kg.44 None of the published studies of EF and energy balance have included a measure of physical activity levels in subjects. Higher levels of physical activity may promote a higher EF to meet increased energy requirements; indeed, it has been suggested that a higher EF may promote spontaneous physical activity.45 In summary, many aspects of the relationship between eating patterns and body weight control remain unclear. Despite the lack of empirical evidence, the view that snacking causes obesity is now widely accepted, both in professional guidelines on obesity prevention46 and by the general public. However, advice to limit snacking on a reducing diet may be inappropriate, since eating snacks between meals may have a bene®cial effect in reducing hunger and preventing subsequent over-compensation.30 There is a need for more information on the role of EF in the control of body weight, if appropriate health promotion strategies for the prevention and management of obesity are to be developed. The aim of this study was to re-examine the Booth hypothesis4 by investigating the relationship between EF and body weight status in free-living men and women, and to evaluate the effects of variations in EF on macronutrient intakes, energy compensation and physical activity levels.

Methods Subjects

Subjects were volunteers recruited from two large manufacturing companies in Edinburgh and Glasgow.


Eligibility requirements for the study were BMI between 18±30, aged between 20±55 y, not pregnant or nursing, not suffering from any known disease or disability and not on any dietary regime, whether medical or voluntary. Of the 150 subjects initially screened, 112 ful®lled the selection criteria and were enrolled in the study. 17 subjects failed to complete the full programme of assessments and a total of 95 volunteers (48 men and 47 women) completed the study. Approval for the study was granted by the Ethical Committee of Queen Margaret College. Procedure and measurements

Anthropometric measurements were made on one occasion only, at the beginning of the study. Subjects were weighed without shoes or jacket on portable scales (Soehnle, up to 127 kg in 0.5 kg intervals). Heights were measured using a portable stadiometer (Leicester height measure) calibrated up to 2 m in 0.1 cm increments. Skinfold thicknesses at 4 sites (biceps, triceps, sub-scapular and supra-iliac) were measured using Harpenden calipers. Percentage body fat was estimated using the method of Durnin and Womersley.47 Food intake was determined using the unweighed diary method. Food diaries, which included a record of the time of eating, were kept for seven consecutive days. The main focus of the study was on EF and the unweighed method was used in an effort to increase compliance and minimise under-reporting of snacks. Subjects were instructed in the accurate completion of diaries by a research assistant (SD). Data from the food diaries were used to estimate dietary intake, to allow the relationships between EF and mean daily intakes of energy and macronutrients to be assessed. Standard food portion sizes were assigned48 and individual food records were analysed for energy and macro-nutrients, using the COMP-EAT 4 nutritional package (Nutrition Systems Ltd, London, UK). Data from the food diaries were also used to calculate EF, the average number of eating occasions/d. An `eating occasion' was de®ned as any occasion when food was taken. If two eating occasions occurred within 15 min of each other, both events were counted as a single eating occasion. The de®nition excluded drinks which were consumed in the absence of food, unless the drink was milk in excess of 0.5 pint, when the nutrient contribution was considered to be equivalent to a food. This de®nition was chosen to avoid the ambiguities of describing eating events as either `meals' or `snacks' and to provide a highly speci®c measure of eating episodes in our study population, which included shift workers. The term EF is used throughout the present report, with one exception, viz in the comparison of our results with published studies of the macronutrient composition of meals and snacks. For this purpose only, data from the food diaries were re-analysed, and each eating occasion was classi®ed as a `meal' or a

`snack', where a `snack' was de®ned as any food or calori®c drink taken outside a regular mealtime (that is, breakfast, lunch and dinner) or a snack item taken in place of such a meal. Snack items eaten instead of a meal included: a piece of fruit, a chocolate bar, a packet of crisps or nuts, biscuits, a yoghurt, a cake/ muf®n/scone/pastry, a glass of milk, a piece of raw vegetable, a piece of cheese, sweets. Two methods of assessing physical activity were used. An activity diary, graduated in 2 min increments, was kept by each subject, over the same period of seven consecutive days as the food diary. The time spent in every activity undertaken throughout the day, together with time asleep, was used to estimate mean daily energy expenditure, using the method set out by COMA (1991), where time spent in each activity is multiplied by the estimated physical activity ratio.49 Energy expenditure was expressed as physical activity level (PAL), a multiple of basal metabolic rate (BMR) which was estimated using Scho®eld's equations.49 For each subject, separate calculations were made for the percentage of total energy spent in manual activities at work and in physically active leisure activities. The combined results gave energy expenditure for all activities. Heart rate was also recorded for two days during this period (one working day and one weekend day), using heart rate monitors (Polar-electro Sports Testers, with a capacity for 33 h continuous recording). The results were expressed as the proportion of the time spent with an elevated heart rate, that is 20±40 beats above resting heart rate and > 40 beats above resting heart rate. Statistical analysis

The principal dependent variable used in this analysis was EF. Relationships between EF and body weight status, activity levels and macronutrient content of diets were investigated with SPSS for Windows, using Pearson correlation coef®cients (two-tailed). Data from male and female subjects were analysed separately, to allow gender differences to be assessed.

Results Individual records of energy intake were scrutinised using the criteria of Goldberg et al50 to screen for possible under-recording. Records from 16 subjects (6 males and 10 females) had energy intake : BMR ratios which were below the cut-off of 1.1 and these subjects were excluded from the ®nal analyses, as suspected under-reporters. The group of under-reporters had a higher mean BMI, 26.5 (s.d 2.3) in males and 25.8 (4.2) in females, compared with subjects with valid records (BMI 25.3 (3.0) in males and 22.8 (5.3) in females). EF was lower in under-reporters, mean 3.1 (0.8) in males and 3.8 (1.0) in females, compared with

107


108

valid reporters (mean 4.3 (1.0) in males, 4.4 (1.1) in females) Data from a total of 42 males and 37 females were included in the analysis. The characteristics of subjects are shown on Table 1. Male subjects were slightly older than females, and mean BMI was higher in men than in women. EF was very similar in both groups, with both men and women eating an average of just over four times a day. No correlations were found between EF and age, in either male or female subjects. Men were slightly more active than women, having an average PAL of 1.63 compared with 1.53 for women. Men were more active in their working hours, 65% carrying out some form of manual work while only 11% of women performed manual work. A higher proportion of men, 74% men were physically active in leisure time, compared with 66% of women. The correlation coef®cients between EF and body weight status are shown on Table 2. In men, there was a signi®cant negative correlation between EF and body weight (r 70.3436, P 0.03) and an inverse association between EF and BMI, although this failed to reach statistical signi®cance. In women, there were no signi®cant correlations between EF and either body weight, BMI or % body fat. A gender difference was also seen when EF was compared with dietary intake, shown in Table 3. In men, although there was no relationship between EF and energy intake, a signi®cant positive correlation was found between EF and carbohydrate intake, when carbohydrate was expressed as a percentage of total energy intake. In women, however, EF and energy intake showed a strong positive correlation, which was statistically signi®cant (r 0.4158, P 0.01); women who ate most frequently had the highest energy intakes. Signi®cant positive correlations were also found between EF and % energy from total carbohydrate, with a striking correlation between EF

and carbohydrate intake expressed in grams (r 0.5429, P 0.001). Intake of sugars (g) was also positively correlated with EF (r 0.4170, P 0.01). For both men and women, subjects who ate most frequently had the highest daily intakes of carbohydrate. To investigate this aspect further, the dietary records were re-analysed, with eating occasions classi®ed as either `meals' or `snacks', and their macronutrient composition assessed. The results showed that foods eaten as snacks provided 50% energy as carbohydrate, compared with only 41% in meals. Snacks were also higher in sugar and lower in protein than meals, although the % energy derived from fat in snacks (40%) was similar to the % fat energy from meals (42%). The results of the physical activity assessments are shown on Table 4. In men, there were no correlations between EF and either total energy expenditure (TEE), or energy expenditure at work or during leisure time. However, the heart rate results showed a signi®cant negative correlation (r 70.3362, P 0.04) between EF and time at an elevated heart rate (20±40 beats above resting heart rate, HR 20±40), but not with time spent at higher heart rates (40 beats above resting). Further analysis showed that physically active men were signi®cantly leaner than less active men and there was a negative correlation (r 70.3094, P 0.04) between % body fat and time spent in activities with PAR > 2.5. In women, no correlations were found between EF and either total energy expenditure or physical activity in leisure time (% TEE leisure). However, further inspection of the data identi®ed a clear outlier (This subject (aged 28 y, BMI 22.5) had a sedentary lifestyle and an EF of 9.0, the highest in the study; the second highest EF in women was 6.0.) and when the data for this subject were excluded from the group results, a signi®cant positive correlation was found between EF and %

Table 1 Characteristics of the study population Male (n 42) Mean Age (y) Height (m) Weight (kg) BMI (kg/m2) %Body fat PAL Energy intake (MJ/d) (kcal/d) % energy from: fat carbohydrate of which sugars protein alcohol EF (eating occasions/d, range)

Female (n 37)

s.d.

Mean

s.d.

37.4 1.77 79.6 25.3 20.7 1.63

10.4 0.08 11.7 3.0 5.2 0.15

34 1.61 59.3 22.8 28.2 1.53

8.6 0.06 8.0 2.9 5.3 0.08

11.04 2638

1.64 391

8.15 1948

1.21 290

37.8 41.2

5.4 5.4

40.2 41.4

4.9 4.6

17.9 14.8 6.3 4.3 (2.43±6.29)

5.8 2.2 4.3 1.0

16.6 14.1 4.4 4.4 (2.71±9.00)

5.2 3.0 3.5 1.1

BMI body mass index; PAL physical activity level; EF eating frequency.

Eating frequency and body weight SE Drummond et al Table 2 Correlation coef®cients: eating frequency (EF) and body weight status Males (n 42)

EF and: Body weight BMI % Body fat

Females (n 37)

r

P

r

P

70.3436 70.2631 70.2145

0.03* 0.09 0.17

0.1396 0.1533 0.0113

0.41 0.35 0.95

*Statistical signi®cance, P < 0.05, Pearson correlation coef®cient. BMI body mass index. Table 3 Correlation coef®cients: eating frequency (EF) and dietary intake Males (n 42) r EF and: Energy intake (MJ/d) Energy from: Fat (%) Carbohydrate (%) Sugar (%) Protein (%) Alcohol (%) Fat (g) CHO (g) Sugar (g) Protein (g) Alcohol (g)

0.0835 70.0709 0.3036 70.0208 70.0912 70.2656 0.0238 0.2739 0.0855 70.0217 70.2153

Table 4 Correlation coef®cients: eating frequency (EF) and physical activity levels

P 0.60 0.66 0.05* 0.90 0.57 0.09 0.88 0.08 0.59 0.89 0.17

Females (n 37) r 0.4158 70.0688 0.3791 0.2695 70.1073 70.3096 0.2539 0.5429 0.4170 0.2884 70.2920

P 0.01* 0.69 0.02* 0.11 0.53 0.06 0.13 0.001* 0.01* 0.08 0.08

* Statistical signi®cance, P < 0.05, Pearson correlation coef®cient. CHO carbohydrate.

TEE leisure (r 0.3332, P 0.047). There were no signi®cant relationships between EF and heart rate in women.

Discussion This study set out to re-examine the Booth hypothesis,4 which predicts a positive relationship between EF and body weight status in free-living subjects, and to evaluate the effect of EF on energy compensation, macronutrient intakes and physical activity levels. Particular care was taken to ensure that under-reporting did not bias the results. The importance of validating dietary records was highlighted by Summerbell et al,12 who showed that a negative relationship between BMI and feeding frequency in adolescents, signi®cant in the group as a whole, disappeared when under-reporters were excluded from the analysis. The need for screening and the dif®culties of identifying under-reporters have been discussed in recent reports.19±21 In the present study, the dietary assessment method chosen, the unweighed food diary, is thought to be less likely to lead to underreporting of snacks than weighed methods.15 In the selection of subjects, obese volunteers (BMI > 30) were excluded from the study, as were subjects currently on a reducing diet, since previous studies have indicated that obese individuals may selectively

Males (n 42)

EF and: TEE (MJ/d) % TEE leisure % TEE work % TEE all activities HR: 20±40 HR: 40

Females (n 37)

r

P

r

P

70.2099 0.0614 0.1703 0.1877 70.3362 70.0793

0.18 0.70 0.28 0.23 0.04* 0.64

0.1044 0.2517

0.54 0.13a

0.0169 0.0210

0.93 0.92

TEE total energy expenditure. HR 20±40 time spent with heart rate 20±40 beats above resting; HR 40 time spent with heart rate 40 beats above resting heart rate). *Statistical signi®cance, P < 0.05, Pearson correlation coef®cient. a % TEE leisure in females, after removal of an outlier (see text); r 0.3332, P 0.047.

under-report snacks.16,17 Finally, the individual dietary analysis results were screened for under-recording, using statistically validated procedures50 and all suspect dietary records were excluded from the ®nal analyses. Compared to subjects with valid dietary records, the group of suspected under-reporters had a higher mean BMI and lower EF, consistent with other studies which have shown that under-reporting, particularly of snacks, may be characteristic of overweight and obese subjects. In men in the present study, the inverse association between BMI and EF, though not statistically signi®cant, is consistent with research by Fabry et al6 and Metzner et al.7 However, in a more recent study by Summerbell et al,12 when the analysis was con®ned to subjects with valid dietary records, no such relationship was found. The choice of de®nition used to describe eating patterns may have had a bearing on the results. EF, de®ned as the number of eating occasions/d, where consecutive eating occasions are considered separate if they occur more than 14 min apart, was used in the present study. Summerbell et al12 used feeding frequency, where food intake was de®ned in terms of six feeding periods per day (breakfast, mid-morning snacks, lunch, mid-afternoon snacks, evening meal and evening snacks) and feeding occasions which occurred less than an hour apart were treated as a single feeding occasion. Although each of these two methods of assessment may produce very similar results for subjects with a low EF, this is unlikely to be the case for populations with higher EFs. Feeding frequency, by de®ning speci®c feeding periods, may mask the number of actual eating occasions, especially when EF is high.23 The use of EF, as de®ned in the present study, should provide greater discrimination of eating occasions in subjects at the upper end of the range, and this may account for the difference in results. We suggest that the relationship between ER and body weight in men, although weak, is real and is not an artefact of under-reporting, although it might not have been evident if we had chosen an alternative de®nition for example, feeding frequency, for the study.

109


110

The inverse relationship between EF and body weight status, found for men in the present study, does not support the Booth hypothesis.4 The results suggest that men with the highest EFs may also have been the leanest. Part of the explanation may be that high EF was not associated with an overall increase in energy intake; men appear to have compensated for extra eating occasions by reducing the mean energy per eating episode. By contrast, there was no evidence of energy compensation in women, who showed a signi®cant positive correlation between EF and energy intake. Women who eat most frequently had the highest energy intakes, indicating an additive effect of energy from successive eating occasions. These results are consistent with other studies which show that energy intake compensation is poor in women, a factor normally associated with obesity.39,40 Despite this, no relationship was found between EF and BMI or % body fat in women. Women who ate most frequently were no fatter or leaner than others. Taken overall, our results indicate that increased EF was not associated with higher body weight status in subjects in the present study. Increased EF may have altered the macronutrient composition of the diet. EF was positively correlated with mean daily carbohydrate intake, but not with fat intake, in both men and women. Increased carbohydrate intake may have displaced fat in the diet; recent experimental evidence has shown that an increase in carbohydrate intake can in turn cause a passive isoenergetic reduction in fat intake.51 As in previous work,33,34 our study showed that foods eaten as snacks were found to provide a higher proportion of energy from carbohydrate than foods eaten as meals. These results suggest that increased EF may lead to a higher carbohydrate : fat ratio, a change in macronutrient pro®le which is thought likely to favour weight control and the maintenance of a low BMI.24,28 Physical activity has not been assessed in previous studies of EF and energy balance, and this aspect is a novel contribution to the ®eld. In the present study, although there was no correlation between EF and TEE, there were other indications that a high EF may have been associated with increased physical activity. in women, EF was positively correlated with physical activity in leisure time; this increased energy expenditure appears to have been suf®cient to offset the higher energy intakes in women with a higher EF. This may partly explain who women who ate most frequency maintained a normal body weight, despite their lack of energy intake compensation. In men, although the relationships between EF and indices of physical activity were inconclusive, body fat was signi®cantly lower in the more active men in the study, and individuals with a higher proportion of lean tissue are likely to have higher energy requirements per kg for basal metabolism.44 Moreover, although the issue of EF and DIT remains controversial,52±54 physical activity is likely to cause a further

increase in energy expenditure by potentiating the thermic effect of food.5 These results suggest that physical activity may have an important in¯uence on the relationship between EF and body weight status, if subjects with higher EFs are also the most active. Increased EF may be driven by the energy requirements of increased physical activity and subjects taking part in exercise or active leisure pursuits may consciously modify the timing and size of eating events to suit these activities. We suggest that in future studies of the relationships between eating patterns and body weight status, especially in populations which include physically active subjects, the method chosen should be one which includes a direct assessment of the number of eating occasions per day, such as EF.

Conclusion Our ®ndings suggest that in this population, EF was inversely related to body weight status in men and there was no link between EF and body weight status in women. Higher EF was related to leanness in men and this may have been associated with increased physical activity levels. Men appeared to compensate accurately for increased EF by reducing the size of subsequent eating episodes. In women, even in the absence of energy intake compensation, the higher energy intake associated with a high EF appears to have been balanced by greater energy expenditure from physical activity, suf®cient to prevent an increase in body weight. The results indicate that energy expenditure from physical activity may be a key factor in¯uencing weight control in subjects who eat frequently, and this is an aspect which merits further investigation. For both men and women, the results suggest that increased EF may contribute to a dietary intake in line with current national guidelines to increase carbohydrate intake and lower fat intake. The combination of a high EF and increased physical activity may form an effective strategy for the control of body weight status. Acknowledgements

This research was supported by funding from Mars Incorporated. References

1 Department of Health. The Nutrition and Physical Activity Task Forces. Obesity: reversing the increasing problem of obesity in England. HMSO: London, 1995. 2 Royal College of Physicians Obesity Report. J Roy Coll Phys Lond 1983; 17: 3±58. 3 Pi-Sunyer FX. Health implications of obesity. Am J Clin Nutr 1991; 53 (Suppl): 595S±603S. 4 Booth DA. Mechanisms from models±actual effects from real life: the zero calorie drink-break option. Appetite 1988; 11 (Suppl): 94±102.


5 Basdevant A, Carplet C, Guy-Grand B. Snacking patterns in obese French women. Appetite 1993; 21: 17±23. 6 Fabry P, Fodor J, Hejl Z, Braun T, Zvolankova K. The frequency of meals, its relation to overweight, hypercholesterolaemia and decreased glucose tolerance. Lancet 1964; 2: 614±615. 7 Metzner HL, Lamphiear DE, Wheeler NC, Larkin FA. The relationship between frequency of eating and adiposity in adult men and women in the Tecumseh Community Health Study. Am J Clin Nutr 1977; 30: 712±715. 8 Charzewska J, Kulesza W, Brzezinska J, Chwojnowska Z. Relationship between obesity or overweight development and the frequency of meals, their distribution during the day and consumption of atherogenic food products. Zywienie Czlowieka 1981; VIII: 217±227. 9 Kant AK, Schatzkin A, Graubard BI, Ballard Barbash R. Frequency of eating occasions and weight change in the NHANES 1 Epidemiological follow-up study. Int J Obes 1995; 19: 468±474. 10 Dreon DM, Frey Hewitt B, Ellsworth N, Williams PT, Terry RB, Wood PD. Dietary fat : carbohydrate ratio and obesity in middle aged men. Am J Clin Nutr 1988; 47: 995±1000. 11 Edelstein SL, Barrett-Connor EL, Wingard DL, Cohn BA. Increased meal frequency associated with decreased cholesterol concentrations: Rancho Bernando, CA 1984±1987. Am J Clin Nutr 1992; 55: 664±9. 12 Summerbell CD, Moody RC, Shanks J, Stock MJ, Geissler C. Relationship between feeding pattern and body mass index in 220 free-living people in four age groups. Eur J Clin Nut 1996; 50: 513±519. 13 Prentice AM, Black AE, Coward WA, Davies HL, Goldberg GR, Murgatroyd FR, Ashford J, Sawyer M, Whitehead RG. High levels of energy expenditure in obese women. BMJ 1986; 292: 983±987. 14 Black AE, Prentice AM, Goldberg GR, Jebb SA, Bingham SA, Livingstone MBE, Coward WA. Measurements of total energy expenditure provide insights into the validity of dietary measurements of energy intake. J Amer Diet Assoc 1993; 93: 572± 579. 15 Livingstone MBE, Prentice AM, Strain JJ, Coward WA, Black AE, Barker ME, McKenna PG, Whitehead RG. Accuracy of weighed dietary records in studies of diet and health. BMJ 1990; 300: 708±712. 16 Heitmann BL, Lissner L. Dietary under-reporting by obese individuals±is it speci®c or non-speci®c?. BMJ 1995; 311: 986±989. 17 Summerbell CD. Differences in reported macronutrient composition of the diet in overweight and lean individuals dependent on validity of reported energy intakes. Proc Nut Soc 1996; 55: 232A. 18 Poppitt SD, Swan D, Black AE, Prentice AM. Is underreporting of energy intake in obese women macronutrient speci®c? Covert measurements in a metabolic facility. Int J Obes 1995; 19 (Suppl 2): 29. 19 Macdairmid JI, Blundell JE. Dietary under-reporting: what people say about recording their food intake. Eur J Clin Nutr 1997; 51: 199±200. 20 Mela DJ, Aaron JI. Honest but invalid: consumer views on recording their food intake. Proc Nut Soc 1997; 56: 53A. 21 Black AE, Johansson G, Welch A, Bingham SA. Energy intake : basal metabolic rate (EI : BMR) for the evaluation of reported energy intake of individuals. Proc Nut Soc 1997; 56: 122A. 22 Crawley H, Summerbell C. Feeding frequency and BMI among teenagers aged 16±17 years. Int J Obes 1997; 21: 159±161. 23 Gatenby SJ. Eating frequency: methodological and dietary aspects. Brit J Nutr 1997; 77 (Suppl 1): S7±S20. 24 Tucker LA, Kano M. Dietary fat and body fat: a multi-variate study of 205 adult females. Am J Clin Nutr 1992; 56: 616± 622.

25 Gazzaniga JM, Burns TL. Relationship between diet composition and body fatness, with adjustment for resting energy expenditure and physical activity in preadolescent children. Am J Clin Nutr 1993; 58: 21±28. 26 Miller WC, Linderman AK, Wallace J, Niederpruem. Diet composition, energy intake and exercise in relation to body fat in men and women. Am J Clin Nutr 1990; 52: 426±430. 27 Gibney MJ. Dietary guidelines: a critical appraisal. J Hum Nutr Diet 1990; 3: 245±254. 28 Bolton-Smith C, Woodward M. Dietary composition and fat to sugar ratios in relation to obesity. Int J Obes 1994; 18: 820± 828. 29 Mela DJ, Sacchetti DA. Sensory preferences for fats: relationships with diets and body composition. Am J Clin Nutr 1991; 53: 908±915. 30 Blundell JE, Burley VJ, Cotton JR, Lawton CL. Dietary fat and the control of energy intake: evaluating the effects of fat on meal size and post-meal satiety. Am J Clin Nutr 1993; 57 (Suppl): 772S±778S. 31 Acheson KJ, Flatt JP, Jecquier E. Glycogen synthesis versus lipogenesis after a 500 gram carbohydrate meal in man. Metabolism 1982; 31: 1234±1240. 32 Rolls BJ, Kim-Harris S, Fischmann 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. 33 Summerbell CD, Moody RC, Shanks J, Stock MJ, Geissler C. Sources of energy from meals versus snacks in 220 people in 4 age groups. Eur J Clin Nutr 1995; 49: 33±41. 34 Ballard-Barbash R, Thompson FE, Graubard BI, Krebs-Smith SM. Variability in percent energy from fat throughout the day: implications for application of total diet goals. J Nutr Ed 1994; 26: 278±283. 35 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. 36 Foltin RW, Fischman MW, Moran TH, Rolls BJ, Kelly TH. Caloric compensation for lunches varying in fat and carbohydrate content by humans in a residential laboratory. Am J Clin Nutr 1990; 52: 969±980. 37 Birch LL, McPhee LS, Bryant JL, Johnson SL. Children's lunch intake: effects of midmorning snacks varying in energy density and fat content. Appetite 1993; 20: 83± 94. 38 Ruxton CHS, Kirk TR, Belton NR. The contribution of speci®c dietary patterns to energy and nutrient intakes in 7± 8 year old Scottish schoolchildren. J Hum Nutr Diet 1996; 9: 23±31. 39 Lissner L, Levitsky DA, Strupp BJ, Kalkwarf HJ, Roe DA. Dietary fat intake and regulation of energy intake in human subjects. Am J Clin Nutr 1987; 46: 886±892. 40 Spiegel TA, Shrager EE, Stellar E. Responses of lean and obese subjects to preloads, deprivation and palatability. Appetite 1989; 13: 45±69. 41 Gortmaker SL, Dietz WH, Cheung LW. Inactivity, diet and the fattening of America. J Am Diet Assoc 1990; 90: 1247± 1252. 42 Prentice AM, Jebb SA. Obesity in Britain: gluttony or sloth?. BMJ 1995; 311: 437±439. 43 Van Zant RS. In¯uence of diet and exercise on energy expenditure±a review. Int J Sports Nutr 1992; 2: 1±19. 44 Drummond S, Kirk T, de Looy A. Snacking: implications in body composition and energy balance. Br Food J 1995; 97: 12±15. 45 Dallosso HM, Murgatroyd PR, James WPT. Feeding frequency and energy balance in adult men. Hum Nutr Clin Nutr 1982; 36C: 25±39. 46 Scottish Intercollegiate Guidelines Network (SIGN) Report. Obesity in Scotland. Integrating Prevention with Weight Management. SIGN, Edinburgh: 1996.

111


112

47 Durnin JVGA, Womersley J. Body fat assessed from total body density and its estimation from skinfold thickness; measurements on 481 men and women aged from 16±72 years. Br J Nutr 1974; 32: 77±97. 48 Ministry of Agriculture Fisheries and Food. Food Portion Sizes (2nd edn). HMSO: London, 1993. 49 Committee on Medical Aspects of Food Policy (COMA). Dietary reference value for food energy and nutrients for the United Kingdom. Report on Health and Social Subjects 41. HMSO: London, 1991. 50 Goldberg GR, Black ED, Jebb SA, Cole PR, Murgatroyd PR, Coward WA, Prentice AM. Critical evaluation of energy intake data using fundamental principles of energy physiology: 1. Derivation of cut-off limits to identify under-recording. Eur J Clin Nutr 1991; 45: 569±581.

51 Kirk TR, Burkill S, Cursiter M. Dietary fat reduction achieved by increasing consumption of a starchy food±an intervention study. Eur J Clin Nut 1997; 51: 455±461. 52 Leblanc J, Mercier I, Nadeau A. Components of postprandial thermogenesis in relation to meal frequency in humans. Can J Physiol Pharm 1993; 71: 879±883. 53 Belko AZ, Barbieri TE. The effect of meal size and frequency on thermic effect of food. Nutr Res 1987; 7: 237±242. 54 Drummond S, Crombie N, Kirk T. A critique of the effects of snacking on body weight status. Eur J Clin Nutr 1996; 50: 779±783. 55 Wirth A. The Role of Exercise in Weight Control. In: Body Weight Control: the physiology, clinical treatment and prevention of obesity. AE Bender, LJ Brookes (eds). Churchill Livingstone: Edinburgh, 1987.