• Diogenes;
  • glycaemic index;
  • protein;
  • weight regain


  1. Top of page
  2. Summary
  3. Introduction
  4. Subjects and methods
  5. Results
  6. Discussion
  7. Conflict of Interest Statement
  8. Acknowledgements
  9. References

Diogenes is a Pan-European, randomized, controlled dietary intervention study investigating the effects of dietary protein and glycaemic index on weight (re)gain, metabolic and cardiovascular risk factors in obese and overweight families in eight European centres. The article is methodological in character, and the presentation of ‘results’ will be limited to baseline characteristics of the study populations included. A total of 891 families with at least one overweight/obese parent underwent screening. The parents started an initial 8-week low-calorie diet and families with minimum one parent attaining a weight loss of ≥8%, were randomized to one of five energy ad libitum, low-fat (25–30 E%) diets for 6 or 12 months: low protein/low glycaemic index, low protein/high glycaemic index, high protein/low glycaemic index, high protein/high glycaemic index or control (national dietary guidelines). At two centres the families were provided dietary instruction plus free foods for 6 months followed by 6-month dietary instruction only. At the remaining six centres the families received dietary instruction only for 6 months. The median weight loss during the low-calorie diet was 10.3 kg (inter-quartile range: 8.7–12.8 kg, n = 775). A total of 773 adults and 784 children were randomized to the 6-month weight (re)gain prevention phase. Despite major cultural and dietary regional differences in Europe, interventions addressing effects of dietary factors are feasible with a reasonable attrition.


  1. Top of page
  2. Summary
  3. Introduction
  4. Subjects and methods
  5. Results
  6. Discussion
  7. Conflict of Interest Statement
  8. Acknowledgements
  9. References

According to The International Association for the Study of Obesity the prevalence of overweight or obesity (body mass index, BMI ≥ 25 kg m−2) among men and women has reached a level of over 50% and 40%, respectively in most European countries (1). Current official European dietary guidelines emphasize restriction of dietary fat and added sugar, while recommendations regarding protein are largely based on minimum requirements (2). Ad libitum reduction of dietary fat will lead to an absolute and relative increase in carbohydrate and protein, which is one of the reasons for the growing interest in the effects on weight management caused by changes in the amount and composition of carbohydrate and protein (3). Diets with either a high protein content (4) or a low glycaemic index (GI) (5–8) have been suggested to play a role in body-weight regulation and obesity-related risk factor modification, but their potential benefits for long-term weight (re)gain prevention and possible risks have not been rigorously assessed in large scale studies. Nor has the effectiveness of dietary intervention in whole families been assessed in long-term randomized studies.

The Diogenes programme is a Pan-European project targeting the obesity problem from a dietary perspective. The name Diogenes is an acronym for ‘Diet, Obesity and Genes’. The project consists of five thematic blocks described as ‘Research and Technological Development Lines’ or RTDs. This paper will, together with a paper by Moore et al. (C. S. Moore et al., unpublished data), describe the main aspects of the work conducted in the RTD1 project – a comprehensive, long-term, randomized, dietary intervention study conducted in eight different European countries. Detailed information about the Diogenes project as a whole and the other RTDs is available on the Diogenes homepage ( The core of the RTD1 line of Diogenes is a multi-centre, randomized, parallel, dietary intervention study, examining the relative efficacy of different low-fat diets (25–30 E%), varying in protein content and GI, in preventing weight (re)gain and certain obesity-related risk factors in obese European families. The dietary intervention aimed to enrol approximately 450 obese families in eight European centres: Maastricht (the Netherlands), Copenhagen (Denmark), Cambridge (UK), Heraklion (Greece), Potsdam (Germany), Pamplona (Spain), Sofia (Bulgaria) and Prague (the Czech Republic).

The dietary intervention part of Diogenes is a unique study of the impact of dietary protein and GI and their possible interactions on weight (re)gain in a large number of families, including both adults and children, from eight different European countries, under varying degrees of control.

Subjects and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Subjects and methods
  5. Results
  6. Discussion
  7. Conflict of Interest Statement
  8. Acknowledgements
  9. References

Experimental design

The inclusion and exclusion criteria for the study are shown in Table 1.

Table 1.  Inclusion and exclusion criteria
Families eligible for inclusion should consist of at least one overweight (body mass index, BMI < 27 kg m−2) but otherwise healthy parent/adult aged less than 65 years, and at least one healthy child between 5 and 18 years of age, who was willing to participate in at least a minimum of the investigations (the so-called ‘light protocol’ for children) Families, in which no biological relationship existed between parents and children (e.g. adopted children) were allowedAdults with a BMI > 45 kg m−2 Adult offspring above the age of 18 years still living at home Adult members of the family who did not have a parental position (e.g. siblings to the parents and grandparents to the children) Subjects using prescription medication, or suffering from diseases or conditions that might influence the outcome of the study. Of special interest were diseases/medicine that influence body weight regulation (malabsorption, untreated hypo/hyperthyroidism, eating disorders, systemic use of steroids, etc.) and obesity-related cardiovascular risk factors (heart disease, systolic and diastolic blood pressures ≥ 160/100 mmHg, blood glucose > 6.1 mmol L−1, blood cholesterol > 7 mmol L−1, blood triglycerides > 3 mmol L−1) Marked alcohol habits > 21 alcoholic units week−1 (male), or >14 alcoholic units week−1 (female) Planned major changes in physical activity during the study to an extent that might interfere with the study outcome, as judged by the investigator Blood donation within the past 2 months prior to the study Adults with a weight change of >3 kg within 2 months prior to first clinical investigation day, CID1 Psychiatric disease (based on medical history only) Pregnant or lactating women, or women planning to become pregnant within the next 18 months Surgically treated obesity Participation in other clinical studies within the last 3 months Drug abuse (based on clinical judgment) Adults unable to give an informed consent Adults unable to engage in an 8-week low-calorie diet Individuals following a special diet (vegetarian, Atkins or other)

The weight status of the children was defined at screening according to the gender and age adjusted iso-BMI > 25 limits suggested by The International Obesity Task Force (9).

Recruitment of families was carried out by using a number of strategies, including a waiting list for weight-loss projects, referrals from local general practices or from other medical departments, flyers and posters in public places and advertising through radio, television, newspapers and internet. Families were interviewed by phone, whenever possible, before being invited to attend a screening examination. Some study centres also arranged information meetings before inviting the families to the screening visit.

The study was approved by the local ethical committees in the respective countries, confirming that the study protocol was in accordance with the Declaration of Helsinki (10) and the United Nations' Convention on the Rights of the Child (11). All study participants (including children where possible) signed an informed consent document after verbal and written instructions and according to local legislation.

Weight-loss phase and randomization

All adults eligible for inclusion initiated an 8-week weight-loss phase on a low-calorie diet (LCD) (Modifast®, Nutrition et Santé, Belgium), providing the subjects with 3.4–4.2 MJ d−1 with the additional use of about 200 g of raw vegetables. During their parents' LCD, the children received no instructions with respect to their food intake.


Families in which at least one of the overweight/obese parents achieved the target weight loss (≥8% of initial body weight) during the LCD period (from visits 3 to 8) were cluster-randomized to one of the five diets using a simple block randomization procedure with stratification. The randomization was stratified according to centre, the number of eligible parents within each family, and to the number of parents with a BMI > 34 kg m−2, resulting in the following four strata within each centre (i) Families with two eligible parents; no parents with BMI > 34 kg m−2; (ii) Families with two eligible parents; at least one parent with BMI >  34 kg m−2; (iii) Families with one eligible parent; parent with BMI ≤ 34 kg m−2; and (iv) Families with one eligible parent; parent with BMI > 34 kg m−2. The randomization was performed with a web-based randomization programme. The five intervention diets were:

  • Group 1: low protein (LP)/low glycaemic index (LGI)

  • Group 2: LP/high glycaemic index (HGI)

  • Group 3: high protein (HP)/LGI

  • Group 4: HP/HGI

  • Group 5: control (CTR) according to current national dietary guidelines in each of the countries, with a medium protein content and with no specific instructions on GI.

Study participants were instructed to maintain their weight loss during the intervention periods, but further weight reduction was also allowed. The five diets were all designed as low-fat (25–30 E%) ad libitum diets (i.e. no energy restrictions on total energy intake) so as to test the diets' ability to regulate appetite and body weight. In addition to addressing the efficacy of the five different diets in preventing weight (re)gain, the intervention study also generated a large amount of additional information, including responses to questionnaires, and blood, urine and fat biopsy samples, were collected.

The families were provided with recipes, together with cooking and behavioural advice, during the study. Parents with children below the age of 5 years were advised to feed these children according to local dietary guidelines. Specific details on the diets and the dietary counselling are given in a separate publication (C. S. Moore et al., unpublished data). Dietary counselling visits were scheduled every 2–4 weeks during the dietary intervention. The families were encouraged to bring the children to the dietary counselling visits whenever possible.

Dietary intervention study setup in the randomized phase

The Diogenes project involved two different dietary protocols. In Maastricht and Copenhagen (the ‘shop centres’) the intervention lasted 12 months, in the remaining six centres – Cambridge, Heraklion, Potsdam, Pamplona, Sofia and Prague (the ‘instruction centres’) 6 months. The experimental designs of the dietary intervention for the children and adults at the ‘shop centres’ and the ‘instruction centres’ are illustrated in Fig. 1.


Figure 1. Experimental design for adults and children at the shop centres and the instruction centres. *CID1 and 2 were combined in one visit for some of the children. CID, clinical investigation day; DI, dietary intervention; FU, follow-up; LCD, low-calorie diet.

Download figure to PowerPoint

The ‘shop centres’

During a 6-month period after the randomization, the families in the ‘shop centres’ received dietary instruction from dieticians and collected most of their foods from study shops, established at the research sites, free of charge, (Fig. 1, upper part, ‘shop centres’, DI1 [dietary intervention period 1]). After this 6-month period, the families went through a 6-month period buying of their foods as usual with continued dietary instruction by dieticians (Fig. 1, upper part, ‘shop centres’, DI2 [dietary intervention period 2]). The adults at the shop centres underwent four clinical investigation days (CIDs), before and at the end of the weight reduction phase, and 6 and 12 months after randomization, and a final follow-up visit 12 months after the end of the instruction period. The participating children had five planned CIDs and a follow-up visit (Fig. 1, lower part, ‘shop centres’).

The study shops were organized like commercial shops, with a variety of fresh, frozen and grocery products. Most of the products were sponsored by local food manufacturers. A Disc Operating System (DOS)-based computer programme was constructed for recording foods based on their existing bar code on the food (product database), and for the calculation of nutrient composition of each shopping session (shopping calculation software). The ability of this shop model to control dietary compliance to a targeted diet composition using the ad libitum principle has been validated in adults in a previous study at the Danish centre (12).

The ‘instruction centres’

At the six instruction centres the participating families received dietary instruction for a 6-month period after the randomization (Fig. 1, upper part, ‘instruction centres’, DI1). The adults completed three CIDs and a follow-up visit 6 months after the end of the instruction period. The children participating in the investigations completed four CIDs and a follow-up visit (Fig. 1, lower part, ‘instruction centres’).

Standardization of investigation procedures and data handling

Standard operation procedures were produced for most procedures or investigations undertaken during the study to ensure standardization across all the centres.

All data obtained during the intervention were entered electronically (single entry) into a data registration system EPIDATA (13) and transferred into a central database.

The data presented in this paper were extracted from the database in November 2008.

Clinical investigations on adults

The primary outcome measures for adults were body-weight-loss maintenance (kg), change in body composition as assessed by Dual energy X-ray Absorptiometry (DXA) or Bioelectrical Impedance Analysis (BIA), number of subjects maintaining more than 5% or more than 10% of the initial weight loss, and dropout rate during the dietary intervention. The secondary outcome measures were changes in abdominal fat mass (measured by DXA, waist circumference and sagittal diameter), risk factors for type 2 diabetes and cardiovascular disease, changes in appetite and satiety hormones, physical activity and identification and quantification of fat tissue messenger ribonucleic acid (mRNA), and certain peptide and protein biomarkers from blood. In addition, assessments of biological (genetic profiles, measurements of basal metabolic rate (BMR) and free-living energy expenditure, assessment of physical activity, etc.) and psychological features (questionnaires covering issues such as appetite and food preferences, health promoting behaviour, attitudes towards eating, social support, etc. [Table 2]) that determined the families' and the individuals' responses to the dietary intervention were performed.

Table 2.  Questionnaire plan for adults at screening, CID1–4 and the follow-up visit
(visit 1)(visit 2)(visit 9)(visit 17)(visit 24)(visit 25)
  • There was no CID4 at the six instruction centres.

  • ‘M’ indicates ‘main group + subgroup A and B’, ‘A’ indicates ‘subgroup A only’.

  • *

    The version at CID1 was different from that at CID2–4.

  • CID, clinical investigation day; FU, follow-up; Q., questionnaire; VAS, visual analogue scales.

General Q.M     
Baecke's Q.M  MM 
Eating attitudes test (EAT26)M     
Eysenck Personality Questionnaire Revised (EPQ-R)M     
Marlowe CrowneM     
Previous dieting experience (Health Seeking Behaviour)M     
Eating Self-Efficacy Scale (ESES) MMM  
Quality of Life – IWQOL-lite MMM  
Theory of planned behaviour MMM  
Three Factor Eating Q. MMM  
Flexible and rigid control of eating behaviour MMM  
Profile of Mood States MMM  
Preoccupying Cognitions MMM  
Psychosocial/attitudinal questionnaire (Surrey) MMM  
Food Choice Q. AA   
Food Preference Checklist AA   
VAS-scales AA   
End of Day Q.* MMMM 
Q. on weight-loss attempts and social support     M
Study protocols for adults

At the screening visit the adults chose one of four different protocol types they would participate in (i) ‘Main protocol’; (ii) ‘Subgroup A’; (iii) ‘Subgroup B’; or (iv) ‘Subgroup A plus B’. Adults who were not eligible for inclusion in the study but who had an eligible partner could participate in a so-called ‘light protocol’, which only required a minimum amount of participation in the study. Table 3 lists the investigations for each of these study protocols.

Table 3.  Investigations on adults Thumbnail image of
Main protocol

A total of 17 visits (‘instructions centres’) and 25 visits (‘shop centres’) were planned for each adult in the course of the study. Details of the procedures and investigations are given in Table 3. A number of examinations were carried out, including blood pressure, measurement of waist and hip circumference, sagittal abdominal height, body composition and blood and urine samples on the CIDs (visits 2, 9, 17 and 24), and a number of questionnaires were provided. Subjects taking part in the main protocol or in any of the subgroups fasted (no intake of foods or liquids for at least 10 h prior to the investigations except for a water intake of 350–500 mL).


Body weight was measured on the CIDs and at all dietary counselling sessions. Other measures were midway made at the same time. Waist circumference was measured between the bottom of the ribs and the top of the hip bone. Hip circumference was measured at the widest point between the hips and buttocks, when observing the subject from the front. Sagittal diameter was measured at the highest point of the abdomen during expiration, with the subject lying on their back. At CIDs body composition was assessed by DXA at two centres, by BIA at five centres and by both DXA and BIA at one centre.

Registration of food intake using food diaries

The adults were asked to complete a 3-d weighed food record for three consecutive days, including two week days and one weekend day, three or four times during the study (prior to CID1, within 2–4 weeks after the randomization and prior to CID3 and 4) (Table 3). The participants were instructed to weigh all their foods whenever possible and to supply information on brand names, cooking and processing. When weighing was not possible (e.g. when dining out) they were instructed to record the food in household measures (cups, glasses, tablespoons, etc.). Before and after each meal or snack the subjects recorded their motivation to eat, mood, prior activity levels, and selected social and environmental factors (such as number and relationship of other diners) in the food diaries, in accordance with the method of de Castro et al.(14). All foods noted in these diaries were coded to foods listed in country specific food databases. By use of a standardized and centralized procedure combining the weight, the coding and the nutrient information for each food item, the nutrient intake was thereafter calculated for each food diary.

Blood sampling and collection of urine samples

Fasting blood samples were drawn from a venflon at each of the CIDs and were followed by an oral glucose tolerance test lasting for 120 min. The subjects consumed 75 g of glucose (82.5 g of glucose monohydrate) diluted with 250 mL of tepid water at time = 0 min of the oral glucose tolerance test. An overview of the blood samples drawn at each CID is given in Table 3. Buffy coat was collected for later DNA extraction and genetic single nucleotide polymorphism analysis.

The subjects were asked to complete a 24-h urine collection four to five times during the study period. The completeness of the urinary collection was checked by recovery rate of Para amino benzoic acid taken as tablets three times during the collection period. In addition, the subjects were asked to deliver a spot urine sample at some of the CIDs. The components analysed in the 24 h and the spot urine samples are listed in Table 3.

Fat biopsies

Fat biopsies were drawn from the abdominal subcutaneous fat under local anaesthesia. A maximum of 2 g of adipose tissue was drawn at each of the relevant visits. The fat samples were stored at −80°C for future analyses of mRNA expression.

Central analyses of clinical samples

Blood samples, urine samples and fat biopsies were sent from each centre to a central laboratory, depending on the type of analysis. Labelling, storage, packing and shipment of samples were performed according to identical standard operation procedures, and the centres were provided with most of the equipment necessary for obtaining the samples.

Total cholesterol, high-density lipoprotein cholesterol and triglycerides (TG), for which results are presented in this manuscript, were analysed at the Research Laboratory, Department of Clinical Biochemistry, Gentofte University Hospital, Denmark. Low-density lipoprotein cholesterol was calculated from the measured values of total cholesterol, high-density lipoprotein cholesterol and TG, according to Friedewald's equation (15).


The questionnaires completed by the adult participants at the screening, CIDs and the follow-up visits are listed in Table 2.


All adults were asked if they would like to record the number of steps they took in periods of 7 d each, at three time points during the study, by means of a pedometer (Tanita CalorieWalk [5666]).

Investigations on adults – subgroup A

The subjects attending subgroup A underwent a number of extra investigations on each of the CIDs 1, 2 and 3. BMR was determined by indirect calorimetry (ventilated hood system); free-living energy expenditure was determined by the double-labelled water (DLW, 2H218O) technique for a period of 14 d. For this, urine samples were collected at days 1, 7 and 14, after ingestion of the DLW dose. Physical activity patterns were monitored for a period of three consecutive days with an Intelligent Device for Energy Expenditure and Physical Activity (IDEEA, MiniSun LLC, Fresno, CA, USA), which is designed to register limb movement, posture changes, body transition, steps taken and heart rate (16). At the end of CID1 and CID2 the subjects consumed a test meal consisting of pasta with tomato sauce. Before and after the meal the subjects filled in a Food Choice Questionnaire and a Food Preference Checklist (17,18). Subjective appetite ratings were recorded before and up to 180 min after administration of the meal by using visual analogue scales (19). Details on the sub-study procedures will be described in subsequent publications. Thirty adult participants were enrolled in subgroup A at each of the centres in Maastricht, Copenhagen, Cambridge, Potsdam, Pamplona and Prague, making a total of 180 subjects.

Investigations on adults – subgroup B

The subjects attending subgroup B had an extra fat biopsy performed 3 d into the LCD period. Fifty adult participants were enrolled in subgroup B at the centres in Maastricht, Copenhagen and Cambridge, making a total of 150 subjects.

Clinical investigations in children

Primary outcome measures for children were changes in BMI Z-score, body composition and dropout rate during the dietary intervention. Secondary end points were changes in the proportion of overweight and obese children, changes in waist/hip circumference ratio, blood parameters and physical activity. In addition to the outcome measures related to body-weight regulation, the safety of a high-protein diet in relation to bone health and kidney health was assessed in those of the children who were willing to deliver blood and urine samples. Safety end points included changes in urinary and blood proteins, such as albumin, creatinine, plasma osteocalcin and urinary N-telopeptide (U-NTx).

Sub-study protocols

At the screening visit the children (and their parents on behalf of their children) were asked to choose between taking part in all planned investigations (‘full’ protocol), taking part in some of the investigations (‘light’ protocol) or not to take part in the investigations at all (‘minus’ protocol). The investigations carried out on the full protocol and the light protocols are illustrated in Table 4.

Table 4.  Investigations on children Thumbnail image of
All children taking part in the investigations

Both children on ‘light’ and ‘full’ protocols had their weight, height, waist and hip circumference measured at each CID (visits 2, 9, 11, 17 and 24). The children were asked if they would like to record the number of steps they took during a period of seven consecutive days at around visits 2, 17 and 24 by means of a pedometer. In addition, the children at two of the centres (Potsdam and Copenhagen) were asked if they would like to record their physical activity by means of translated versions of the Habitual Activity Estimation Scale questionnaire (20) during a period of three consecutive days, including two week days and one weekend day, prior to their visits 2, 11, 17 and 24. All children were encouraged to fill in a 3-d food diary (same principles as for the adults) prior to their parents CID1, about 4 weeks after the randomization and prior to CID4 and CID5.

When planning the study, CIDs were scheduled for the children both before (CID1) and after their parents' LCD (CID2) to allow for analyses on the influence of the parental dieting on the children's weights. However, during the start-up at the first centres, the investigators realized that it might put too much work load on the families and the participating centres with all these CIDs for the children, and it was therefore decided to allow that the CID1 and CID2 for the children were combined in one single visit as close to the randomization as possible. These combined visits (CID1 + 2) are addressed in the same way as CID2 visits (children having CID1 and CID2 as separate visits) in this paper, and together they are chosen to represent ‘baseline’ for the children.

Full protocol

The children following the full protocol had their blood pressures and body composition (DXA or BIA) measured and delivered blood and urine samples at the CIDs. Children who participated in blood sampling and/or determination of body composition by DXA or BIA were fasting (no intake of foods or liquids for at least 4 h prior to the investigations except for a water intake of 350–500 mL).

Statistical considerations

Sample size estimations

It was anticipated that there would be less difference between the LGI vs. HGI diets in weight loss after the 6-month dietary intervention than between the LP and HP diets. A conservative approach to the calculations on power was therefore chosen. Calculations were based on findings of a study by Sloth et al.(21), which found no significant difference between the effect of LGI or HGI diets on body weight during 10 weeks of controlled dietary intervention. A group size of 125 (all five groups of equal size), and thus a total of 625 adults, was estimated to be necessary in order to obtain a significant (P < 0.05) difference in body-weight change of 1.0 kg (SD 2.01 kg) with a power of 97%. Any effect size above 0.72 kg would give a power of more than 0.80.

Given an estimated dropout rate of 20% during the LCD and a subsequent dropout rate of 15% during the 6-month intervention equally distributed among the five dietary groups, 918 adult subjects had to pass the screening successfully to end up with 125 subjects per intervention group at CID3.

Presentation of baseline characteristics

Baseline characteristics are presented centre-wise. The data for the adults originate from CID1 (anthropometrics, blood pressures and lipid profile), and data related to screening (demographics and height) are only presented and analysed for those adults also present at CID1. The data for the children originate from CID2 (anthropometrics and blood pressures), and data from screening (demographics) are only shown for children present at CID2. For both adults and children the maximum number of subjects (the number of subjects registered as present at the CID1 and CID2, respectively) for each centre is stated below the centre names in Tables 5–7. For each variable the number of subjects for whom a value was available is shown in each of the table cells.

Table 5.  Characteristics of the female adults attending CID1 at the eight centres*
 Total max. n = 620Maastricht (the Netherlands) max. n = 89Copenhagen (Denmark) max. n = 80Cambridge (UK) max. n = 84Heraklion (Greece) max. n = 80Potsdam (Germany) max. n = 66Pamplona (Spain) max. n = 59Sofia (Bulgaria) max. n = 72Prague (the Czech Republic) max. n = 90 P
  • Age, family type and height were recorded at screening; the other variables are from CID1. The variable ‘Family type’ states whether the female subject participated alone (mother) or if she had a spouse, who was included in the study too (also a spouse on light protocols counts as ‘both’). The number of mother only vs. both parents participating are compared centre-wise by using a Pearson Chi-squared test and the percentage of women that participated alone are given in brackets. For the remaining variables centres are compared using Kruskal-Wallis rank test, and values are expressed as median and inter-quartile range in brackets.

  • *

    Numbers are excluding ‘Superlight adults’.

  • BMI, body mass index; BW, body weight; CID, clinical investigation day; dBP, diastolic blood pressure; HC, hip circumference; HDL, high-density lipoprotein; LDL, low-density lipoprotein; SagD, sagittal diameter; sBP, systolic blood pressure; WC, waist circumference.

Age (years)40 (36–45) n = 62040 (37–45) n = 8940 (37–45) n = 8940 (37–45) n = 8940 (37–45) n = 8940 (37–45) n = 8940 (37–45) n = 8940 (37–45) n = 8940 (37–45) n = 890.0001
Family type mother/both (mother%)417/203 (67) n = 62056/33 (63) n = 8942/38 (53) n = 8066/18 (79) n = 8455/25 (69) n = 8045/21 (68) n = 6623/36 (39) n = 5961/11 (85) n = 7269/21 (77) n = 90<0.0001
BW (kg)92.5 (83.3–103.8) n = 62091.8 (84.7–102.6) n = 8996.0 (85.8–105.2) n = 8091.2 (81.7–102.9) n = 8494.3 (81.0–108.9) n = 8093.0 (84.5–102.9) n = 6688.8 (81.2–96.4) n = 5994.0 (87.0–110.0) n = 7292.6 (83.2–103.1) n = 900.0844
Height (cm)166.0(161.0–170.0) n = 620168.0 (164.0–170.0) n = 89168.0 (163.0–172.0) n = 80165.0 (161.0–167.0) n = 84162.0 (156.5–166.0) n = 80167.0 (163.0–171.0) n = 66160.0 (158.0–165.0) n = 59167.0 (161.5–170.0) n = 72168.0 (164.0–172.0) n = 900.0001
BMI (kg m−2)33.8 (30.7–37.6) n = 62032.6 (30.4–35.9) n = 8934.0 (30.8–37.2) n = 8033.1 (30.1–38.4) n = 8436.1 (31.5–41.2) n = 8033.0 (30.5–35.7) n = 6634.3 (31.0–37.2) n = 5934.6 (31.5–40.2) n = 7232.5 (30.3–36.8) n = 900.0050
WC (cm)103.0 (95.4–111.1) n = 615103.3 (96.0–111.6) n = 89104.7 (96.1–111.3) n = 80104.8 (97.6–115.5) n = 84107.1 (99.1–114.9) n = 80102.3 (93.6–110.3) n = 66102.8 (95.6–110.0) n = 59103.0 (94.0–110.0) n = 6799.8 (91.5–109.0) n = 900.0059
HC (cm)117.7 (110.8–126.3) n = 615103.3 (96.0–111.6) n = 89104.7 (96.1–111.3) n = 80104.8 (97.6–115.5) n = 84107.1 (99.1–114.9) n = 80102.3 (93.6–110.3) n = 66102.8 (95.6–110.0) n = 59103.0 (94.0–110.0) n = 6799.8 (91.5–109.0) n = 900.0003
SagD (cm)24.1 (22.0–26.5) n = 59822.8 (20.4–25.1) n = 8723.8 (22.1–26.3) n = 7923.6 (22.1–25.8) n = 7724.5 (22.6–27.3) n = 7926.4 (24.0–28.5) n = 6522.6 (21.0–24.2) n = 5725.9 (23.9–28.7) n = 6724.0 (21.0–26.0) n = 870.0001
sBP (mmHg)120 (111–130) n = 603122 (116–133) n = 86119 (110–127) n = 79117 (109–126) n = 84116 (108–124) n = 79120 (109–130) n = 63120 (113–126) n = 59140 (135–145) n = 65118 (111–125) n = 880.0001
dBP (mmHg)76 (68–82) n = 60378 (71–84) n = 8675 (68–82) n = 7968 (62–74) n = 8477 (70–82) n = 7981 (73–87) n = 6364 (60–71) n = 5990 (80–95) n = 6575 (70–80) n = 880.0001
Fasting glucose (mmol L−1)4.90 (4.60–5.30) n = 5584.80 (4.50–5.10) n = 854.90 (4.50–5.20) n = 795.10 (4.90–5.60) n = 614.80 (4.50–5.20) n = 745.00 (4.90–5.30) n = 585.20 (4.90–5.60) n = 414.90 (4.50–5.20) n = 724.80 (4.50–5.20) n = 880.0001
Fasting insulin (microIU mL−1)8.70 (5.83–12.70) n = 53811.60 (8.50–15.30) n = 869.18 (6.49–12.40) n = 787.86 (6.04–12.70) n = 577.56 (5.45–11.35) n = 727.27 (5.01–11.70) n = 6011.20 (9.30–15.20) n = 417.70 (4.85–10.50) n = 636.79 (4.63–9.61) n = 810.0001
Cholesterol (mmol L−1)4.80 (4.22–5.49) n = 5674.08 (3.46–4.67) n = 854.37 (3.84–5.00) n = 805.40 (4.67–5.81) n = 615.19 (4.45–5.61) n = 745.24 (4.70–5.72) n = 645.18 (4.46–5.55) n = 414.77 (4.32–5.44) n = 724.93 (4.24–5.49) n = 900.0001
HDL (mmol L−1)1.27 (1.04–1.51) n = 5671.03 (0.90–1.22) n = 851.16 (0.97–1.36) n = 801.31 (1.11–1.47) n = 611.41 (1.11–1.59) n = 741.35 (1.11–1.60) n = 641.30 (1.18–1.58) n = 411.25 (1.10–1.59) n = 721.32 (1.14–1.57) n = 900.0001
LDL (mmol L−1)2.96 (2.39–3.54) n = 5642.39 (1.91–3.14) n = 852.62 (2.09–3.14) n = 783.38 (2.71–3.97) n = 613.23 (2.67–3.75) n = 743.21 (2.78–3.75) n = 643.09 (2.64–3.66) n = 412.91 (2.37–3.38) n = 723.03 (2.41–3.48) n = 890.0001
Triglycerides (mmol L−1)1.13 (0.82–1.52) n = 5671.01 (0.74–1.55) n = 851.10 (0.81–1.53) n = 801.13 (0.82–1.53) n = 611.09 (0.79–1.43) n = 741.27 (0.92–1.50) n = 641.07 (0.76–1.25) n = 411.16 (0.81–1.58) n = 721.20 (0.90–1.58) n = 900.2621
Table 6.  Characteristics of the male adults attending CID1 at the eight centres*
 Total max. n = 312Maastricht (the Netherlands) max. n = 57Copenhagen (Denmark) max. n = 55Cambridge (UK) max. n = 26Heraklion (Greece) max. n = 35Potsdam (Germany) max. n = 40Pamplona (Spain) max. n = 40Sofia (Bulgaria) max. n = 29Prague (the Czech Republic) max. n = 30 P
  • Age, family type and height was recorded at screening, the rest of the variables are from CID1. The variable ‘Family type’ states whether the male subject participated alone or if he had a spouse, who was included in the study too (spouse on light protocols counts as ‘both’). The number of father only vs. both parents participating are compared centre-wise by using a Pearson Chi-squared test and the percentage of men that participated alone are given in brackets. For the remaining variables the centres are compared by using Kruskal-Wallis rank test, and values are expressed as median and inter-quartile range in brackets.

  • *

    Numbers are excluding ‘Superlight adults’.

  • BMI, body mass index; BW, body weight; CID, clinical investigation day; dBP, diastolic blood pressure; HC, hip circumference; HDL, high-density lipoprotein; LDL, low-density lipoprotein; SagD, sagittal diameter; sBP, systolic blood pressure; WC, waist circumference.

Age (years)42 (38–45) n = 31243 (40–45) n = 5742 (38–45) n = 5540 (35–45) n = 2640 (37–43) n = 3542 (38–46) n = 4043 (40–47) n = 4044 (41–46) n = 2942 (38–46) n = 300.1163
Family type father/both (father%)118/194 (38) n = 31224/33 (42) n = 5716/38 (30) n = 546/20 (23) n = 269/26 (26) n = 3519/21 (48) n = 4020/20 (50) n = 4014/15 (48) n = 299/21 (30) n = 300.0762
BW (kg)106.0 (96.7–120.7) n = 31299.1 (93.1–111.5) n = 57110.0 (102.8–122.8) n = 55104.0 (100.3–116.7) n = 26110.2 (102.6–121.7) n = 35104.1 (94.7–117.0) n = 4098.2 (87.3–111.6) n = 40118.0 (106.0–132.0) n = 29115.8 (99.8–128.7) n = 300.0001
Height (cm)178.0 (173.0–183.0) n = 312179.0 (175.0–183.0) n = 57180.0 (176.0–184.0) n = 55178.0 (174.0–180.0) n = 26174.0 (170.0–180.0) n = 35178.0 (175.0–185.0) n = 40172.0 (169.0–176.5) n = 40176.0 (173.0–186.0) n = 29180.0 (176.0–185.0) n = 300.0001
BMI (kg m−2)33.8 (30.7–37.4) n = 31231.5 (29.1–34.0) n = 5734.0 (30.7–37.1) n = 5533.8 (31.8–36.2) n = 2636.3 (33.6–40.5) n = 3532.3 (30.5–34.8) n = 4032.9 (30.1–36.9) n = 4038.2 (33.1–40.0) n = 2934.3 (31.0–38.0) n = 300.0001
WC (cm)112.4 (105.8–122.5) n = 306106.3 (102.3–113.8) n = 57112.5 (104.5–122.6) n = 55114.6 (107.8–125.5) n = 26119.0 (110.0–127.0) n = 35109.9 (102.6–119.6) n = 40111.0 (103.1–120.6) n = 40119.0 (109.0–133.0) n = 23117.5 (108.0–128.0) n = 300.0001
HC (cm)111.8 (106.7–118.6) n = 306105.6 (100.3–111.1) n = 57112.1 (107.8–116.7) n = 55114.0 (111.8–122.0) n = 26116.3 (113.0–121.0) n = 35109.8 (106.3–114.9) n = 40109.8 (104.5–117.8) n = 40121.0 (113.0–130.0) n = 23114.7 (110.0–123.0) n = 300.0001
SagD (cm)26.5 (24.1–29.4) n = 29924.1 (22.5–25.6) n = 5526.5 (24.1–28.8) n = 5526.4 (24.5–29.1) n = 2527.8 (25.2–30.6) n = 3528.1 (26.4–30.8) n = 4023.5 (21.9–26.6) n = 3830.1 (27.2–32.9) n = 2328.0 (26.0–32.0) n = 280.0001
sBP (mmHg)131 (123–141) n = 301140 (129–148) n = 54127 (122–137) n = 55125 (118–133) n = 26127 (122–140) n = 35135 (125–143) n = 38128 (120–134) n = 40140 (140–145) n = 23126 (120–134) n = 300.0001
dBP (mmHg)81 (74–88) n = 30181 (79–90) n = 5479 (73–86) n = 5573 (64–83) n = 2681 (76–89) n = 3587 (82–93) n = 3869 (63–74) n = 4095 (85–100) n = 2380 (75–87) n = 300.0001
Fasting glucose (mmol L−1)5.20 (4.90–5.50) n = 2905.10 (4.80–5.40) n = 545.10 (4.80–5.40) n = 555.10 (4.95–5.65) n = 205.15 (4.80–5.50) n = 345.20 (5.20–5.80) n = 355.60 (5.30–6.20) n = 335.00 (4.70–5.60) n = 295.10 (4.60–5.60) n = 300.0002
Fasting insulin (microIU mL−1)11.60 (7.98–16.50) n = 28611.45 (9.33–18.40) n = 5411.90 (9.39–16.20) n = 559.73 (6.90–15.10) n = 2012.80 (9.16–19.10) n = 3211.30 (7.71–15.20) n = 3713.20 (10.50–20.10) n = 3211.65 (6.76–15.80) n = 269.30 (5.58–12.20) n = 300.0997
Cholesterol (mmol L−1)4.96 (4.28–5.81) n = 2954.43 (3.64–5.41) n = 554.44 (3.91–4.93) n = 555.07 (4.68–5.93) n = 205.57 (4.66–6.29) n = 345.28 (4.73–6.19) n = 395.31 (4.98–6.09) n = 335.04 (4.49–5.91) n = 294.96 (4.46–5.72) n = 300.0001
HDL (mmol L−1)1.04 (0.90–1.21) n = 2950.88 (0.79–0.97) n = 550.97 (0.86–1.10) n = 551.06 (0.94–1.25) n = 201.13 (0.99–1.29) n = 341.07 (0.91–1.27) n = 391.17 (1.04–1.37) n = 331.06 (0.96–1.28) n = 291.11 (0.92–1.30) n = 300.0001
LDL (mmol L−1)3.11 (2.56–3.89) n = 2922.81 (2.29–3.42) n = 542.67 (2.26–3.22) n = 553.26 (3.07–4.15) n = 203.56 (3.03–4.42) n = 343.34 (2.79–4.06) n = 393.46 (2.98–4.28) n = 322.96 (2.57–3.95) n = 283.13 (2.86–3.54) n = 300.0001
Triglycerides (mmol L−1)1.49 (1.08–2.09) n = 2951.49 (1.03–2.23) n = 551.48 (1.09–1.91) n = 551.31 (1.05–1.92) n = 201.27 (0.97–2.13) n = 341.62 (1.28–2.40) n = 391.38 (0.96–1.80) n = 331.65 (1.01–2.32) n = 291.49 (1.16–2.02) n = 300.3647
Table 7.  Characteristics of the children attending CID2 at the eight centres
 Total max. n = 841Maastricht (the Netherlands) max. n = 149Copenhagen (Denmark) max. n = 165Cambridge (UK) max. n = 93Heraklion (Greece) max. n = 79Potsdam (Germany) max. n = 83Pamplona (Spain) max. n = 98Sofia (Bulgaria) max. n = 82Prague (the Czech Republic) max. n = 92 P
  1. Age and gender were recorded at screening; the other variables are from CID2. The distributions of gender and weight status at the different centres are compared by using Pearson Chi-squared test, and the values are expressed as the number in each category and as percentage in brackets. The other variables are compared by using Kruskal-Wallis rank test, and values are expressed as median with inter-quartile range in brackets.

  2. BMI, body mass index; BW, body weight; CID, clinical investigation day; dBP, diastolic blood pressure; HC, hip circumference; NW, normal weight; OB, obese; OW, overweight; sBP, systolic blood pressure; WC, waist circumference.

Age (years)12.0 (8.5–14.5) n = 84111.0 (9.0–14.0) n = 14911.5 (8.5–14.0) n = 16512.0 (8.5–13.5) n = 9311.0 (8.0–14.0) n = 7912.5 (8.5–15.5) n = 8311.5 (8.0–15.0) n = 9814.5 (11.5–15.5) n = 8212.3 (10.5–14.5) n = 920.0007
Gender female/male (female%)450/391 (54) n = 84184/65 (56) n = 14990/75 (55) n = 16548/45 (52) n = 9337/42 (47) n = 7939/44 (47) n = 8340/58 (41) n = 9855/27 (67) n = 8257/35 (62) n = 920.0096
BW (kg)52.7 (37.6–67.9) n = 82949.3 (33.7–63.8) n = 14850.8 (34.7–66.6) n = 16548.8 (34.4–58.5) n = 8660.9 (39.5–75.5) n = 7955.0 (38.6–70.2) n = 8047.3 (34.2–59.8) n = 9865.0 (50.0–75.0) n = 8155.6 (43.9–68.7) n = 920.0001
Height (cm)155.0 (139.0–167.0) n = 828152.0 (136.0–167.0) n = 148153.0 (136.0–168.0) n = 165151.5 (139.0–166.0) n = 86154.0 (135.0–164.0) n = 79159.0 (138.0–171.0) n = 79153.5 (133.0–164.0) n = 98163.0 (150.0–170.0) n = 81160.0 (143.5–169.0) n = 920.0012
BMI (kg m−2)21.5 (18.2–25.1) n = 82820.5 (17.4–23.6) n = 14820.8 (17.9–24.9) n = 16519.8 (17.5–22.5) n = 8624.6 (21.1–29.2) n = 7920.8 (17.6–24.6) n = 7920.6 (17.9–23.7) n = 9824.8 (21.9–27.0) n = 8122.4 (18.5–24.9) n = 920.0001
Weight status (OB/OW/NW in numbers and as % in brackets)161/256/402 (20/31/49) n = 81922/38/79 (16/27/57) n = 13931/56/78 (19/34/47) n = 1658/19/59 (9/22/69) n = 8638/25/16 (48/32/20) n = 7912/18/49 (15/23/62) n = 7914/30/54 (14/31/55) n = 9821/37/23 (26/46/28) n = 8115/33/44 (16/36/48) n = 92<0.0001
WC (cm)72.0 (62.9–81.3) n = 75571.3 (61.3–79.4) n = 14471.2 (61.4–80.3) n = 16570.0 (63.0–77.3) n = 8680.0 (71.0–98.0) n = 7969.9 (60.4–80.8) n = 7469.5 (62.5–79.3) n = 9778.0 (68.0–85.0) n = 1873.2 (65.7–81.6) n = 920.0001
HC (cm)86.5 (74.5–96.5) n = 75378.0 (65.0–88.6) n = 14485.5 (72.8–96.9) n = 16587.3 (77.0–96.3) n = 8595.7 (84.0–106.4) n = 7986.2 (75.5–96.8) n = 7484.3 (76.5–94.3) n = 9792.0 (90.0–97.0) n = 1790.0 (81.8–98.0) n = 920.0001
sBP (mmHg)109 (101–118) n = 650109 (99–120) n = 123110 (105–115) n = 89104 (98–111) n = 86114 (103–120) n = 78108 (102–118) n = 77109 (102–118) n = 94100 (90–105) n = 15115 (103–122) n = 880.0001
dBP (mmHg)61 (55–68) n = 65063 (59–71) n = 12363 (58–67) n = 8954 (50–58) n = 8667 (61–76) n = 7867 (62–72) n = 7751 (47–57) n = 9470 (60–75) n = 1562 (57–68) n = 880.0001

Weight status of the children was determined at screening according to the gender- and age-adjusted iso-BMI ≥ 25 limits as suggested by The International Obesity Task Force (9).

Categorical data are presented as actual numbers (e.g. the number of normal weight, overweight or obese children), with this number as percentage of the total given in brackets. The centres were compared by using a Pearson Chi-squared test. Numeric data are presented as median with inter-quartile range (IQR) in brackets. The centres have been compared by using Kruskal-Wallis rank test because the distribution of data for the majority of the variables deviated substantially from a normal distribution judged by histograms and quantile-quantile (q-q) plots.

The median weight loss during the LCD (from visits 3 to 8) was calculated for all adults weighed at both visits (in kilograms and as percentage of the initial weight). This median weight loss was divided into a weight loss for those that reached the targeted 8% weight loss (‘succeeders’), one for those that did not (the ‘non-succeeders’), and finally into a median weight loss for female and male succeeders respectively. Results are presented as median with IQR in brackets. The weight losses (in kilograms and as percentage of the initial weight) are compared for female and male succeeders by using the Kruskal-Wallis rank test. The analyses were performed by using Stata 10.0 (StataCorp. 2007. Stata Statistical Software: Release 10. College Station, TX: StataCorp LP).

The study is registered with, number NCT00390637.


  1. Top of page
  2. Summary
  3. Introduction
  4. Subjects and methods
  5. Results
  6. Discussion
  7. Conflict of Interest Statement
  8. Acknowledgements
  9. References

A total of approximately 891 families attended screening visits in the period from January 2006 to August 2007 (1209 parents with 1164 children were registered. It was not mandatory for the children to attend the screening visit). Figure 2 is a flow diagram showing the number of families, adults and children from screening to randomization. One hundred and seventy-eight adults took part in at least one of the subgroup A procedures at CID1, and 90 adults had the subgroup B procedure performed. Five hundred and thirty children delivered blood samples at CID2.


Figure 2. Progress of study participants until the end of the first dietary intervention period (DI1). CID, clinical investigation day; HGI, high glycaemic index; HP, high protein; LCD, low-calorie diet; LGI, low glycaemic index; LP, low protein.

Download figure to PowerPoint

The characteristics of the adults attending CID1 at the eight different centres are shown for women (Table 5) and for men (Table 6). The adults attending CID1 differ in various variables between the different centres (Tables 5 and 6). The women differed regarding all variables except for TG and body weight, and the men differed for all variables except for TG, age, fasting insulin and the number of men that participated alone or as a part of a couple (family type).

The children attending CID2 at the different centres differed for all variables included in Table 7. A remarkable difference in the percentage of children classified as obese, overweight or normal weight existed between the centres. Greece and Bulgaria had the highest percentages of overweight and obese children.

A total of 634 families (773 adults and 784 children) were randomized to the five different diets.

Data on start and completion weights (visits 3 and 8) for the LCD were available for 775 subjects, of whom 750 achieved a weight loss of ≥8% of their initial body weight during the LCD. The median weight loss for subjects attending both visits 3 and 8 was 10.3 kg (IQR: 8.7–12.8 kg; n = 775) corresponding to 10.4% (IQR: 9.0–12.5%; n = 775) of the initial (visit 3) body weight. Subjects achieving a weight loss of at least 8% of their initial body weight had a median weight loss of 10.4 kg (IQR: 8.8–12.9 kg; n = 750), corresponding to 10.5% (IQR: 9.1–12.6%; n = 750) of their initial body weight. The group of subjects that did complete the LCD period and attended visit 8, but who did not loose as much weight as required, had an average weight loss of 6.6 kg (IQR: 5.9–7.4 kg; n = 25), corresponding to 7.4% (IQR: 6.6–7.7%; n = 25) of their initial body weight. The male succeeders lost more weight during LCD than the female succeeders, both when expressed in kilograms (12.1 kg [IQR: 10.0–14.9; n = 268] for male succeeders and 9.8 kg [IQR: 8.2–11.6 kg; n = 482] for female succeeders, P = 0.0001) and as a percentage of their initial body weight (11.4% [9.4–13.3%; n = 268] for male succeeders and 10.2% [9.0–12.1% for female succeeders; n = 482]; P = 0.0001).

Five hundred and forty-six adults and 442 children completed the first 6-month dietary intervention period (Fig. 2), corresponding to 71% and 56% of the adults and children being randomized respectively.


  1. Top of page
  2. Summary
  3. Introduction
  4. Subjects and methods
  5. Results
  6. Discussion
  7. Conflict of Interest Statement
  8. Acknowledgements
  9. References

The Diogenes dietary intervention study is the largest dietary intervention study, specifically focusing on long-term weight loss (children) and weight-loss maintenance (adults) in obese families, conducted to date. The study design is unique in that it integrates a number of scientific disciplines, including a classical prospective long-term, randomized, controlled study design, rigorously examining issues influencing energy balance via energy intake (food diaries, compliance questionnaires and the supermarket system), physical activity (questionnaires and pedometers), physical activity patterns (IDEEA), energy expenditure measurements (BMR and DLW) and various socio-psychological factors (questionnaires). Furthermore, a large number of blood, urine and fat biopsy samples were collected for a detailed analysis of risk factors related to type 2 diabetes and cardiovascular diseases and for biomarker analysis. The diets were controlled through two different approaches: the shop model, providing the highest possible control of food intake among free-living subjects and intended to test the efficacy of the study regimes (the ability of a treatment to produce benefit if applied ideally); and the instruction only model, intended to show the effectiveness of the study regimes (the benefit that actually occurs when a treatment is used in practice) (22). The use of these two different study designs may mean that not all results from all the centres are universally comparable.

The study design includes an initial weight-loss phase for the adults only. This design was chosen as it will take years in a large population to study the effects of a certain diet on weight gain. In such situation the compliance becomes a real problem. Therefore a design was chosen that focused on weight regain after a period of weight loss. From previous studies we know that in the first 6 months subjects are quite vulnerable to regain weight. This design makes it possible to undertake such a large scale dietary intervention within a reasonable time period without running into compliance and financial problems.

One study by McMillan-Price et al. is very close in design to the dietary intervention in Diogenes (7). However, McMillan-Price et al. tested the ability of different diets varying in protein and GI to induce weight loss, not their ability to maintain an initial weight loss. They compared four diets (corresponding to the diets 1–4 in Diogenes) and not five diets as in Diogenes, the number of subjects was lower (129 subjects randomized) than in the Diogenes intervention, the intervention lasted for a shorter period (12 weeks), the study population consisted of single individuals (adults) from one country, the subjects were guided towards an appropriate energy intake (not ad libitum) and provision of free food was not a part of the study design. The conclusions that will be drawn on the results from Diogenes might therefore not necessarily be consistent with the conclusion by McMillan-Price et al. that both HP and LGI regimens increase body fat loss, but cardiovascular risk reduction is optimized by a high-carbohydrate (LP), LGI diet.

Philippou et al. found no effect of dietary GI on the ability to maintain a weight loss throughout a period of 4 months. However, the number of subjects in their study was low (HGI n = 19 and LGI n = 23) and the dietary instructions on GI seemed less restrictive than those offered in Diogenes – subjects were just asked to include at least one HGI or LGI carbohydrate containing food item with each of their meals and snacks (23).

Baseline characteristics of adults and children show some heterogeneity across the study centres. When comparing the percentages of overweight children from each of the centres with those numbers presented by the International Association for the Study of Obesity (24), the percentages are higher among the Diogenes children in all countries except for Greece and UK – the countries that according to IASO has the highest prevalence of childhood overweight and obesity. When including families with at least one parent with a BMI above 27 kg m−2, one might of course expect a high number of overweight or obese children. However, these numbers might also be influenced by other factors like the location of the centres, the strategy and media used for recruitment and the type of centres (free foods or not). The numbers of overweight and obese children at the centre in Cambridge, UK were surprisingly low. According to the investigators from that centre, the education level in Cambridge is higher and health generally better than in other parts of the UK. Besides, the investigators at Cambridge had the feeling that the study attracted the more health conscious parents, and together these factors might explain the low prevalence of obesity/overweight among the children at the centre in Cambridge.

The randomization procedure, with stratification according to centre, aimed to minimize the risk of finding effects of the different diets due exclusively to the heterogeneity of the study populations across the different centres and/or the country specific dietary habits (e.g. a HP/LGI diet in Spain is made up of different foods than in Germany). This does not exclude the possibility of diet × centre interactions, but although such interactions might complicate the analyses and interpretations of the results, the heterogeneity should be viewed as a strength of the study design. Thus, the inclusion of a heterogeneous population of European overweight and obese families with diets varying according to country specific dietary habits and food supplies is expected to promote the generalizability of the conclusions drawn on the impact of dietary protein and GI on weight-loss maintenance and on metabolic and cardiovascular risk factors, and at the same time enables analyses on the impact of country and/or centre type (shop vs. instruction) on these results.

It proved very challenging to work with whole families and to engage all the family members. The dieticians had to instruct both children and parents in how to get through the initial 8-week LCD period for the adults, where the parents had to cook acceptable food for their children without becoming too tempted themselves. In the subsequent dietary counselling sessions the dieticians had to communicate with all the family members, and they also had to deal with the interrelationships between the different family members. The large number of investigations at the centres, the homework for the families (food diaries, questionnaires, pedometers, etc.), the dietary counselling sessions, the dietary changes and the shopping sessions (at the shop centres) comprised a heavy work load for the families and investigators.

However, the family approach of Diogenes offered many benefits, not only from a scientific point of view, but also for the families. More specifically, the family approach is highly relevant for any subsequent implementation in real life, as most overweight subjects living in a family (children in particular) will have major difficulties in changing and sustaining their lifestyle if they cannot get the rest of their family to support the initiative.


  1. Top of page
  2. Summary
  3. Introduction
  4. Subjects and methods
  5. Results
  6. Discussion
  7. Conflict of Interest Statement
  8. Acknowledgements
  9. References

The Diogenes project, coordinated by Professor Wim Saris, Maastricht University, The Netherlands, is funded by a grant of 14.5 million Euros of EU funding under the Food Quality and Safety Priority of the Sixth Framework Programme for Research and Technological Development of the European Union (2005–2009), through the Directorate-General for Research of the European Commission. Financial contributions from local sponsors were also provided to the shop centres, which also received a number of foods free of charge from food manufacturers. A full list of these sponsors can be seen at The study protocol of the Diogenes dietary intervention study (RTD1) was prepared in close collaboration with the other RTD lines within the Diogenes project, specifically RTD2 (including a number of blood samples and fat biopsies) coordinated by Dominique Langin, and RTD4 (including a number of questionnaires, DLW investigations, BMR measurements, meal test procedures) coordinated by Monique Raats, and with the Datahub (collection and cleaning of study data) coordinated by Claus Holst.


  1. Top of page
  2. Summary
  3. Introduction
  4. Subjects and methods
  5. Results
  6. Discussion
  7. Conflict of Interest Statement
  8. Acknowledgements
  9. References
  • 1
    International Association for the Study of Obesity. Adult overweight and obesity in the European Union (EU27) & Switzerland. 2008 [WWW document]. URL (accessed September 2008).
  • 2
    Prentice A, Branca F, Decsi T, Michaelsen KF, Fletcher RJ, Guesry P, Manz F, Vidailhet M, Pannemans D, Samartín S. Energy and nutrient dietary reference values for children in Europe: methodological approaches and current nutritional recommendations. Br J Nutr 2004; 92(Suppl. 2): S83S146.
  • 3
    Saris WH, Tarnopolsky MA. Controlling food intake and energy balance: which macronutrient should we select? Curr Opin Clin Nutr Metab Care 2003; 6: 609613.
  • 4
    Paddon-Jones D, Westman E, Mattes RD, Wolfe RR, Astrup A, Westerterp-Plantenga M. Protein, weight management, and satiety. Am J Clin Nutr 2008; 87: 1558S1561S.
  • 5
    Brand-Miller JC, Holt SH, Pawlak DB, McMillan J. Glycemic index and obesity. Am J Clin Nutr 2002; 76: 281S285S.
  • 6
    Ludwig DS. Dietary glycemic index and obesity. J Nutr 2000; 130: 280S283S.
  • 7
    McMillan-Price J, Petocz P, Atkinson F, O'neill K, Samman S, Steinbeck K, Caterson I, Brand-Miller J. Comparison of 4 diets of varying glycemic load on weight loss and cardiovascular risk reduction in overweight and obese young adults: a randomized controlled trial. Arch Intern Med 2006; 166: 14661475.
  • 8
    Thomas DE, Elliott EJ, Baur L. Low glycaemic index or low glycaemic load diets for overweight and obesity. Cochrane Database Syst Rev 2007; (3): Art No: CD005105 DOI: 10 1002/14651858 CD005105 pub2 2007.
  • 9
    Cole TJ, Bellizzi MC, Flegal KM, Dietz WH. Establishing a standard definition for child overweight and obesity worldwide: international survey. BMJ 2000; 320: 12401243.
  • 10
    World Medical Association. Declaration of Helsinki – ethical principles for medical research involving human subjects. 2004. [WWW document]. URL (accessed July 2008).
  • 11
    United Nations. Convention on the rights of the child. 1989. [WWW document]. URL (accessed July 2008).
  • 12
    Skov AR, Toubro S, Raben A, Astrup A. A method to achieve control of dietary macronutrient composition in ad libitum diets consumed by free-living subjects. Eur J Clin Nutr 1997; 51: 667672.
  • 13
    Lauritsen JM, Bruus M. Epidata (version 3) – a comprehensive tool for validated entry and documentation of data. The Epidata association, Odense, Denmark. 2004. Ref Type: Computer Program.
  • 14
    De Castro JM, McCormick J, Pedersen M, Kreitzman SN. Spontaneous human meal patterns are related to preprandial factors regardless of natural environmental constraints. Physiol Behav 1986; 38: 2529.
  • 15
    Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem 1972; 18: 499502.
  • 16
    Zhang K, Werner P, Sun M, Pi-Sunyer FX, Boozer CN. Measurement of human daily physical activity. Obes Res 2003; 11: 3340.
  • 17
    Blundell JE, Jebb S, Stubbs RJ, Wilding JR, Lawton CL, Browning L, Whybrow S, Halford JCG. Effect of rimonabant on energy intake, motivation to eat and body weight with or without Hypocaloric Diet: the REBA Study. Obes Rev 2006; 7: 104 (abstract).
  • 18
    Blundell JE, Stubbs RJ, Golding C, Croden F, Alam R, Whybrow S, Le Noury J, Lawton CL. Resistance and susceptibility to weight gain: individual variability in response to a high-fat diet. Physiol Behav 2005; 86: 614622.
  • 19
    Flint A, Raben A, Blundell JE, Astrup A. Reproducibility, power and validity of visual analogue scales in assessment of appetite sensations in single test meal studies. Int J Obes Relat Metab Disord 2000; 24: 3848.
  • 20
    Hay J, Cairney J. Development of the Habitual Activity Estimation Scale for clinical research: a systematic approach. Pediatr Exerc Sci 2006; 18: 193202.
  • 21
    Sloth B, Krog-Mikkelsen I, Flint A, Tetens I, Björck I, Vinoy S, Elmståhl H, Astrup A, Lang V, Raben A. No difference in body weight decrease between a low-glycemic-index and a high-glycemic-index diet but reduced LDL cholesterol after 10-wk ad libitum intake of the low-glycemic-index diet. Am J Clin Nutr 2004; 80: 337347.
  • 22
    Turnock BJ. Public Health: What is It and How It Works. Aspen: Bethesda, MD, 1997.
  • 23
    Philippou E, Neary NM, Chaudhri O, Brynes AE, Dornhorst A, Leeds AR, Hickson M, Frost GS. The effect of dietary glycemic index on weight maintenance in overweight subjects: a pilot study. Obesity 2009; 17: 396401.
  • 24
    International Association for the Study of Obesity. Childhood overweight in the European Union (EU27) & Switzerland. [WWW document]. URL (assessed February 2009).