Randomised comparison of diets for maintaining obese subjects' weight after major weight loss: ad lib, low fat, high carbohydrate dietv fixed energy intakeBMJ 1997; 314 doi: http://dx.doi.org/10.1136/bmj.314.7073.29 (Published 04 January 1997) Cite this as: BMJ 1997;314:29
- a Research Department of Human Nutrition, Royal Veterinary and Agricultural University, Rolighedsvej 30, 1958 Frederiksberg, Copenhagen, Denmark
- Correspondence and reprint requests to: Professor Toubro
- Accepted 16 October 1996
Objectives: To compare importance of rate of initial weight loss for long term outcome in obese patients and to compare efficacy of two different weight maintenance programmes.
Design: Subjects were randomised to either rapid or slow initial weight loss. Completing patients were re-randomised to one year weight maintenance programme of ad lib diet or fixed energy intake diet. Patients were followed up one year later.
Setting: University research department in Copenhagen, Denmark.
Subjects: 43 (41 women) obese adults (body mass index 27-40) who were otherwise healthy living in or around Copenhagen.
Interventions: 8 weeks of low energy diet (2 MJ/day) or 17 weeks of conventional diet (5 MJ/day), both supported by an anorectic compound (ephedrine 20 mg and caffeine 200 mg thrice daily); one year weight maintenance programme of ad lib, low fat, high carbohydrate diet or fixed energy intake diet (≤7.8 MJ/day), both with reinforcement sessions 2-3 times monthly.
Main outcome measures: Mean initial weight loss and proportion of patients maintaining a weight loss of >5 kg at follow up.
Results: Mean initial weight loss was 12.6 kg (95% confidence interval 10.9 to 14.3 kg) in rapid weight loss group and 12.6 (9.9 to 15.3) kg in conventional diet group. Rate of initial weight loss had no effect on weight maintenance after 6 or 12 months of weight maintenance or at follow up. After weight maintenance programme, the ad lib group had maintained 13.2 (8.1 to 18.3) kg of the initial weight loss of 13.5 (11.4 to 15.5) kg, and the fixed energy intake group had maintained 9.7 (6.1 to 13.3) kg of the initial 13.8 (11.8 to 15.7) kg weight loss (group difference 3.5 (-2.4 to 9.3) kg). Regained weight at follow up was greater in fixed energy intake group than in ad lib group (11.3 (7.1 to 15.5) kg v 5.4 (2.3 to 8.6) kg, group difference 5.9 (0.7 to 11. 1) kg, P<0.03). At follow up, 65% of ad lib group and 40% of fixed energy intake group had maintained a weight loss of >5 kg (P<0.07).
Conclusion: Ad lib, low fat, high carbohydrate diet was superior to fixed energy intake for maintaining weight after a major weight loss. The rate of the initial weight loss did not influence long term outcome.
Obesity has reached epidemic proportions in the Western world, but weight loss reverses almost all the health hazards of obesity
Obese patients lose weight when they keep strictly to an energy restricted diet, but weight losses tend not to be maintained in the long term
We conducted an intensive one year weight maintenance programme (comparing an ad lib, low fat, high carbohydrate diet with a fixed energy intake diet) after a major weight loss (eight weeks of low energy diet or 17 weeks of conventional diet)
The rate of initial weight loss did not influence the long term outcome
The ad lib, low fat diet was superior in maintaining weight loss during weight maintenance programme and at one year follow up
Obesity has reached epidemic proportions in the Western world; in Britain the prevalence of obesity doubled between 1980 and 1991, and it is still increasing.1 Obesity is responsible for a considerable proportion of the mortality attributable to circulatory disease, non-insulin dependent diabetes, and certain cancers, and it is one of the most important avoidable risk factors. In the United States obesity has been estimated to contribute 8% of all health costs.2 There is strong evidence to suggest that weight loss reverses almost all the health hazards of obesity3 and normalises mortality.4 Obese patients inevitably lose weight when they keep strictly to an energy restricted diet,5 but long term results of energy restriction (the so called calorie counting method) are modest: Typically, after a six month hypocaloric diet (5 MJ/day) has caused 96% of subjects to lose >5 kg weight, only 52% will have maintained a weight loss of ≥5 kg at one year's follow up and only 11% will have done so at five years.6
Uncertainty over the aetiology of obesity remains a major barrier to developing effective strategies for prevention and treatment. Recent evidence indicates that obesity develops when individuals from the relatively large proportion of the population with a genetic predisposition to obesity are exposed to certain environmental and behavioural conditions, such as an inactive lifestyle and an energy dense diet. A high fat diet seems to promote energy intake by overriding normal signals of satiety,7 8 and studies from various research disciplines indicate that a high fat diet plays a crucial role in the development and maintenance of obesity.9 A high fat diet is a risk factor for weight gain among overweight subjects with a family history of obesity,10 and obese subjects report a habitual diet with a higher fat content than do subjects of normal weight.1112
The apparent role of dietary fat in obesity has lead to the evaluation of ad lib, low fat, high carbohydrate diets for weight loss, but in a short intervention study the weight loss was found to be smaller than from calorie counting.13 However, as it may take years to develop obesity on a high fat diet, so it may also take years to reverse it by dietary changes. In the present study we therefore decided to induce weight loss by traditional energy restriction and subsequently to compare the long term effects of either an ad lib low fat, high carbohydrate diet or fixed energy intake on weight maintenance. We included two different rates of weight loss in the study in order to examine any effect of this on the long term outcome.
Subjects and methods
We consecutively recruited 43 obese adults (2 men, 41 women) who were otherwise healthy from the outpatient waiting list of the Research Department of Human Nutrition, Royal Veterinary and Agricultural University, Copenhagen. All had stable weights, with a body mass index (weight (kg)/ (height (m))2) between 27 and 40. There was one man in each of the two treatment arms, during both weight reduction and weight maintenance. None of the subjects had clinical or biochemical evidence of diabetes or other endocrine disorders or hepatic or renal disease, and none was taking prescribed drugs. They were screened by analysis of blood samples–including haematology (haemoglobin concentration, white cell count, and differential count) and biochemistry (plasma glucose concentration and serum concentrations of sodium, potassium, urate, aspartate aminotransferase, lactate dehydrogenase, alkali phosphatase, and creatinine)–as well as by measurement of blood pressure and electrocardiography. The subjects gave their informed consent to the study according to the declaration of Helsinki II. The protocol was approved by the municipal ethics committee of Copenhagen and Frederiksberg.
Weight reduction programmes
The first phase of the study consisted of weight reduction programmes. All the subjects were randomly assigned to either eight weeks of low energy diet (2 MJ/day) (n=21) or to 17 weeks of conventional hypocaloric, high protein diet (5 MJ/day) (n=22). Both diets were supported by an anorectic compound (ephedrine 20 mg and caffeine 200 mg thrice daily,14 Let-igen, Nycomed DAK A/S, Roskilde, Denmark). The duration of the diets was chosen to achieve similar weight losses in both groups. During this phase, the subjects were split into four groups of 7-13 subjects each (two of low energy diet and two of conventional diet).
The low energy dietconsisted of nutrition powder (Bli-Let, Nycomed DAK A/S) dissolved in water and was taken as five daily meals (six for men). This met all recommendations for daily intake of high quality protein (women 60 g, men 72 g), essential amino acids, carbohydrate (30.5 g), vegetable fat (6 g), and fibre (17.5 g). The subjects took a daily supplement of a tablet containing vitamins, minerals, and trace elements and a 1 g fish oil capsule containing at least 350 mg essential omega-3 fatty acids to ensure that their daily intake met recommended amounts.15
The conventional dietconsisted of ordinary foods and a daily vitamin and mineral tablet (Apovit, Nycomed DAK A/S). The diet plan included recommended amounts of listed food items, together with recipes for low energy marinades and dressings.
The subjects attended the department weekly as outpatients, and the two low energy diet groups were kept apart from the two conventional diet groups. All the groups received nutritional instruction and behaviour therapy16: they were instructed by the same staff in dietary guidelines, basic nutritional education, and behaviour therapy sessions of 1-2 hours. Every fortnight the patients were interviewed about adverse effects, and their body weight was measured on a decimal scale (Seca model 707, Copenhagen). Body composition was estimated by the bioimpedance method with an Animeter (HTS-Engineering, Odense, Denmark), and fat free mass and fat mass were calculated with Danish standard equations.17
Weight maintenance programmes
After the weight reduction phase the use of low energy diet and anorectic compounds was stopped and subsequently forbidden. The 37 subjects who completed the weight reduction phase were re-randomised to two different one year, non-pharmacological, weight maintenance diet programmes (the low energy diet groups and conventional diet groups being randomised separately). Two groups of patients were assigned to a low fat, high carbohydrate diet consumed ad lib and two groups to a diet of fixed energy intake.
Ad lib, low fat, high carbohydrate diet–The subjects were given a 24 page (A5 format) dietary leaflet specifying details to be included in their daily habits: (a) use a thin layer of butter or margarine on bread or none at all, (b) use the frying pan less often and throw the dripping away, (c) use cooking methods that require less fat, (d) select lean meat and meat products (<10 g fat/100 g of food item), and (e) eat more carbohydrates, especially complex carbohydrates. Alcoholic beverages were allowed only on special occasions. The aim of the diet was to achieve a macronutrient composition that produced 20-25% of energy intake from fat, at least 55% of energy intake from carbohydrate, and the rest from protein.
Fixed energy intake diet–In order to achieve quantitative as well as qualitative self control, the subjects were introduced to an educational system, which consisted of isoenergetic interchangeable units, represented by 144 counters, each with a small picture of the food it symbolises.18 The counters were also colour coded, with each colour representing a food group: blue counters for foods rich in protein, green counters for foods rich in fibres and low in energy content, and red counters for foods rich in fat and sugar. The energy content of the food represented by each counter was 260 kJ (62.5 kcal). The subjects were encouraged to restrict the number of red counters they used and to use at least seven blue counters a day. At the start of the programme, the subjects had a daily “ration” of 30 counters (7.8 MJ/day), which could be subsequently adjusted: if a subject's weight increased the daily ration of counters was reduced stepwise by two to a minimum of 20. At a ration of 20 counters, subjects were ordered to complete a seven day record of the weight of their food intake. Once a month, the subjects' body weight, body composition, and any side effects were recorded.
During this phase the subjects were allowed, but not encouraged, to loose weight. They were seen in groups two or three times a month for the first six months, and once a month for the following six months. The sessions consisted of dietary instruction, reinforcement, support, and nutritional education and practical instruction in food preparation in the department kitchen.
One year follow up
At the end of the weight maintenance programme, the 34 subjects who completed it were not contacted again for a year. When they were invited to a follow up visit 28 of these subjects accepted.
Venous blood samples were drawn from an antecubital vein of each subject after an overnight fast at entry to the study; at the end of the weight reduction phase (after 8 or 17 weeks); after three, six, and 12 months of the weight maintenance phase; and at the one year follow up. Concentrations of plasma glucose, total cholesterol, high density lipoprotein cholesterol, and triglycerides were determined by enzymatic methods (Hoffmann la Roche, Basel, Switzerland; Boehringer Mannheim, Mannheim, Germany), while insulin concentrations were determined by radioimmunoassay kits (Novo, Copenhagen, Denmark).
Unless stated otherwise, all results are expressed as means (95% confidence interval). Differences between visits and comparisons between different groups were done by either one way or two way analysis of variance (SigmaStat Version 1.02, Jandel Scientific, Germany). All analyses were based on the principle of intention to treat–that is, last body weight was carried forward for the few patients who dropped out. Only the 37 subjects who entered the weight maintenance phase were included in the comparison of the effect of the weight reduction programmes on weight maintenance.
Weight reduction phase
Table 1 shows the subjects' anthropometric data at the start of this phase and the weight loss achieved during this phase. The mean weight loss in the low energy diet groups (12.6 kg (95% confidence interval 10.9 to 14.3 kg)) was similar to that in the conventional diet groups (12.6 (9.9 to 15.3) kg). The rate of weight loss in the low energy diet groups was about twice that in the conventional diet groups (1.6 (1.4 to 1.8) kg/week v 0.8 (0.7 to 1.0) kg/week) (see fig 1).
Five subjects withdrew during this phase, two from the low energy diet groups (one refused to drink the nutrition powder and one had an alcohol problem) and three from the conventional diet groups (two due to lack of weight loss and one due to side effects from the anorectic compound). One subject from the low energy diet groups dropped out between the weight reduction and maintenance phases because of psychological problems after involvement in a car accident. The mean weight loss of the 37 subjects who entered the weight maintenance phase was 13.0 (11.4 to 14.6) kg in the low energy diet groups (n=18) and 14.2 (11.9 to 16.5) kg in the conventional diet groups (n=19) (group difference 1.2 (-1.5 to 3.9) kg).
Weight maintenance phase
The 37 patients were re-randomised to the weight maintenance programmes, producing two similar groups with respect to anthropometry and weight loss. The two groups randomised to the ad lib, low fat diet had lost 13.5 (11.4 to 15.5) kg, and the two groups with fixed energy intake groups had lost 13.8 (11.8 to 15.7) kg. Table 2 shows the subjects' anthropometric data at the start of this phase and the weight regained after two years.
After six months of the weight maintenance phase the groups with the ad lib diet showed an additional weight loss of 2.3 (0.0 4.6) kg (P<0.05) while the groups with fixed energy intake had gained 0.5 (-2.6 to 3.6) kg (group difference 2.8 (-1.0 to 6.6) kg). During the second six months of this phase three patients dropped out–two from the fixed energy intake groups (both due to non-compliance and unavailability to follow up) and one from the ad lib groups (due to development of familial muscle dystrophy). In accordance with analyses being based on the principle of intention to treat, these subjects' last recorded measurements were carried forward. After one year of weight maintenance the ad lib groups showed a non-significant weight gain of 0.3 (-3.0 to 3.6) kg, while the fixed energy intake groups gained 4.1 (1.2 to 6.9) kg (P<0.01) (group difference 3.8 (-0.4 to 8.0) kg, P=0.08).
The dietary records carried out three and six months after the start of the weight maintenance programmes did not show any difference between groups in energy intake. The reported percentage of energy intake from dietary fat was lower in the ad lib than in the fixed energy intake groups after three months, but their recorded food intake for seven days covered only about 70% of their estimated energy intake.
Thirteen (76%) of the subjects from the ad lib groups and 15 (75%) from the fixed energy intake groups were assessed at follow up one year after the end of the weight maintenance phase. When last recorded body weight was carried forward for all nine subjects who had dropped out, the ad lib groups had regained 5.4 (2.3 to 8.6) kg of the 13.5 kg initially lost while the fixed energy intake groups had regained 11.3 (7.1 to 15.5) kg of the initial 13.8 kg weight loss (group difference 5.9 (0.7 to 11.1) kg, P<0.03). The maintained weight loss was also greater in the ad lib groups than in the fixed energy intake groups (8.0 (3.5 to 12.6) kg v 2.5 (-1.7 to 6.6) kg, group difference 5.6 (-0.3 to 11.5) kg, P=0.06).
Whole study period
Figure 2 shows the subjects' absolute weight changes. Among the subjects who completed the study, fat mass contributed 67% (3.9 kg) of the total weight regained by those in the ad lib groups compared with 75% (8.5 kg) of the total weight regained by those in the fixed energy intake groups (group difference 3.4 (-1.3 to 8.1) kg). For all subjects who entered the weight maintenance phase, 65% of those in the ad lib groups and 40% of those in the fixed energy intake groups maintained a weight loss >5 kg (P<0.07); for the subjects who completed the study, these percentages were 58% and 25% respectively (P<0.07).
Figure 3 shows the proportion of subjects for a given weight loss obtained at follow up. The relative risk of subjects in the fixed energy intake groups of maintaining a weight loss <5 kg was 1.7 (0.8 to 3.7) compared with the ad lib groups (P<0.05). Analysis of covariance showed that maintenance of weight loss was unaffected by the preceding weight reduction programme (low energy diet or conventional diet) (P>0.30); this was true after six months' weight maintenance (group difference 2.4 (-3.2 to 7.9) kg), 12 months' weight maintenance (group difference 3.0 (-2.8 to 8.9) kg), and after one year follow up (group difference 3.0 (-3.0 to 9.1) kg).
We found no group differences in any of the plasma concentrations (glucose, insulin, cholesterol, and triglycerides) measured for blood analyses (see table 3).
Our study shows that a low fat, high carbohydrate diet consumed ad lib was superior to a more traditional fixed energy intake diet in maintaining weight and preventing relapse two years after a major weight loss. The mean weight loss after two years was three times higher in the ad lib group than in the fixed energy intake group (8.0 kg v 2.5 kg), and more patients in the former group maintained a substantial proportion of their initial weight loss (maintained weight loss of >5 kg: 65% v 40%).
We found that the rate of initial weight loss had no effect on subsequent weight maintenance, which suggests that different procedures to induce weight loss may be equally suitable providing they are followed by an effective, long term dietary programme of weight maintenance. However, with a less intensive weight maintenance programme than the one in this study, we would anticipate the long term outcome after an initial weight loss to be unsatisfactory.
We found no persistent changes in the plasma variables after weight loss, except for plasma insulin concentration, which was significantly reduced in both groups. Hypertriglyceridaemia has been reported in patients with non-insulin dependent diabetes when they were given a high carbohydrate diet19; we did not find this in our study, perhaps because of the simultaneous weight loss in our obese subjects who were otherwise healthy.
Role of fat intake in weight maintenance
The mechanism by which a low fat, high carbohydrate diet prevents relapse to a positive energy balance and weight regain may be both by reducing energy intake and by increasing energy expenditure,12 although the effect on appetite seems to be predominant.20 In observational studies levels of body fat and obesity are positively associated with dietary fat content,9 and case reference studies indicate that obese subjects generally consume a diet with a higher fat content than do lean counterparts.11 21 Although obese subjects tend to underreport their energy intake22 and possibly also fat intake,23 when fat oxidation is used as a biological marker of fat intake a larger proportion of dietary energy seems to be derived from fat in obese subjects than in subjects with normal weight.24
Meal test studies and short term experimental studies indicate that the satiating effect of fat is weaker than that of carbohydrate25 and that passive overconsumption of energy can easily be provoked by covertly increasing the fat content of the diet.7 8 While a high fat diet is not necessarily a prerequisite to the development of obesity, it may promote obesity in subjects with a familial background of obesity26 or with low levels of physical activity.1 26 The principle of fixed energy intake is based on voluntary restriction of energy intake, which may tend to keep the dietary composition unchanged–that is, high in fat–whereas a low fat, high carbohydrate diet probably provides greater satiety for less energy intake.
Comparison with other studies
Previous studies have compared the efficacy of ad lib low fat diets with fixed energy intake or calorie counting to induce weight loss, and results for ad lib low fat diets have not been encouraging. Jeffery et al found that dietary counselling focusing on fat reduction (20 g/day) and unrestricted carbohydrate intake tended to be more effective than energy restriction to 4.2-5.0 MJ/day27: after 12 months' treatment, the low fat group had lost 2.1 kg and the calorie counting group only 0.5 kg. These small weight losses indicate that compliance, at least to the calorie counting, was poor. Though the low fat diet was inefficient in inducing weight loss, the authors found that patients in this group were more compliant with treatment directions, found the diet more palatable, and reported greater reduction in binge eating. In contrast, Schlundt et al found that calorie counting produced a larger weight loss in obese subjects than did a low fat ad lib diet during a 16-20 week programme13: weight loss in the low fat group was 8.0 kg in men and 3.9 kg in women compared with 11.8 kg in men and 8.2 kg in women in the calorie counting group.
Several points must be considered when comparing the effectiveness of different diets. Firstly, energy restriction as a tool for inducing weight loss is highly effective in obese subjects, and trials have reported mean weight losses of 10-15 kg over six months of treatment.5 14 16 The relapse rate, however, is remarkably high–50% of all patients regain or exceed their pretreatment weight at 12 months' follow up.6 Because a high fat content of the diet plays a role for the development of obesity, the low fat principle seems more appropriate as a tool for weight maintenance. We therefore chose to test the two diet principles after an initial weight loss of 12-13 kg.
Secondly, compliance is a crucial factor for weight loss. During their trial of ad lib, low fat diets, Lyon et al gave subjects an evening meal to be consumed at home. The meal was enriched with radiolabelled glucose, and the subjects were asked to collect expired air in a test tube after the meal. The recovery of radiolabelled carbon was used as an index of adherence to the diet, and the authors found a strong correlation between adherence and loss of body fat (r =0.74).28 Consequently, a small or no weight loss may be attributed to lack of compliance to the diet.
Finally, for nutritional public health policy, our results support the theory that a low fat diet could contribute to preventing obesity. Even a mean weight loss of 2-3 kg may produce a substantial reduction in the prevalence of obesity on a population basis. The recent observation that the prevalence of obesity is increasing concomitantly with a decrease in dietary fat content does not in conflict with this, because the level of daily physical activity is also decreasing.1 There is an important interaction between exercise level and fat oxidation,29 indicating that the amount of fat in the diet has to be reduced for a given reduction in total energy expenditure in order to achieve energy balance.
We thank all subjects who participated and the staff of the metabolic unit of the Research Department of Human Nutrition–Lis Kristoffersen, Tina Cuthbertson, Charlotte Kostecki, Inge Timmermann, Bente Knap.
Funding: Drugs and part of the study's funding were provided by Nycomed DAK A/S, Roskilde, Denmark.
Conflicts of interest: None.