¹General Practice and Primary Care Research Unit, Institute of Public Health, University of Cambridge, Cambridge

²School of Psychology, University of St Andrews, Scotland

³Department of Community Medicine, Institute of Public Health, University of Cambridge, Cambridge, UK

Correspondence to: W Hardeman, Department of Community Medicine, University of Cambridge, Institute of Public Health, Robinson Way, Cambridge CB2 2SR, UK. wh207@medschl.cam.ac.uk

OBJECTIVE: To identify and review published interventions aimed at the prevention of weight gain.

DESIGN: A systematic review of published interventions aimed at the prevention of weight gain.

METHODS: Search strategies¾we searched eight databases, manually checked reference lists and contacted authors. Inclusion and exclusion criteria¾studies of any design, in which participants were selected regardless of weight or age, were included. Interventions targeting a specific subgroup, multifactorial interventions, interventions aimed at weight loss, and those with an ambiguous aim were excluded. Data extraction¾data were extracted on behaviours targeted for change, psychological model, behaviour change methods and modes of delivery, methodological quality, characteristics of participants, and outcomes related to body weight and self-reported diet and physical activity. Classification and validation¾a taxonomy of behaviour change programmes was developed and used for classification of underlying model, behaviour change methods, and modes of delivery. The data extraction and subsequent classification were independently validated.

RESULTS: Eleven publications were included, describing five distinct interventions in schools and four in the wider community. Where diet and physical activity were described, positive effects were usually obtained, but all were measured by self-report. Effects on weight were mixed but follow-up was generally short. Smaller effects on weight gain were found among low-income participants, students and smokers. Many participants in the community-based studies were overweight or obese. Study dropout was higher among thinner and lower-income subjects.

CONCLUSION: Interventions to prevent weight gain exhibited various degrees of effectiveness. Definite statements about the elements of the interventions that were associated with increased effect size cannot be made as only one of the five studies that involved an RCT design reported a significant effect on weight. This intervention involved a correspondence programme and a mix of behaviour change methods including goal setting, self-monitoring and contingencies. Future interventions might be more effective if they were explicitly based on methods of behaviour change that have been shown to work in other contexts. Effective interventions would be more easily replicated if they were explicitly described. Effectiveness might be more precisely demonstrated if more objective measures of physical activity and diet were used, and if the follow-up was over a longer period.

International Journal of Obesity (2000) 24, 131-143

review; prevention; weight gain; obesity; intervention; behaviour change

Introduction

Obesity is rapidly increasing in prevalence in the UK¹ and many other developed countries,² and presents a major risk to health through its association with a wide spectrum of disease including type 2 diabetes, ischaemic heart disease, osteoporosis and some cancers.² Tackling this increasingly important public health issue requires action primarily at a societal level to counter the environmental influences on physical activity and dietary intake.^3,4 Action is also required to improve the effectiveness of weight loss programmes for individuals and groups who are already obese. A recent WHO consultation report described activities in these two areas.² A third strategy, which has received little rigorous evaluation,⁵ is the prevention of weight gain. This might be particularly effective for people at risk of the metabolic consequences of weight gain, such as people with a family history of type 2 diabetes.⁶ Although the interventions aimed at the prevention of obesity and its treatment may be superficially similar, the emphasis on physical activity rather than dietary change may differ between them,^7,8 and it is also likely that the content of the psychological intervention may be distinct.

This review was undertaken to inform an ongoing intervention aimed at the prevention of weight gain among people not selected by weight. Although previous reviews have focused on interventions aimed at preventing obesity and achieving weight loss,^{5,9,10,11,12,13,14,15} none of them were specific about the prevention of weight gain as a primary objective. Moreover, the reviews did not exclude studies in which participants were selected on the basis of their current weight.

The aims of this review were: (i) to describe the interventions aimed at the prevention of weight gain; and to characterise (ii) the target behaviours; (iii) the psychological models underlying the interventions, behaviour change methods and modes of delivery (for example newsletter, group); (iv) the methodological quality of the evaluation; (v) the characteristics of the participants; and (vi) the outcomes of the studies.

Methods

Inclusion and exclusion criteria

Published reports of interventions of any duration with the primary aim of preventing weight gain, in which participants were not selected by weight or age, were included in the review. Studies were included regardless of their design as our primary objective was to review the interventions, and there appeared to be few relevant randomised trials. Interventions in specific subgroups,^16,17,18 for example those stopping smoking, were excluded because approaches effective in these subgroups may not be generalisable to other populations. Secondly, studies were excluded if weight control was part of a multi-factorial intervention primarily aimed at a specific disease (for example diabetes, coronary heart disease, cancer).^{19,20,21,22,23} This was done because it would be difficult to disentangle conclusions about appropriate interventions specifically for the prevention of weight gain. Thirdly, interventions were excluded if they were aimed at weight loss,^{24,25,26,27,28,29} and finally, if the focus was ambiguous (for example weight control).^30,31,32

Identification and data extraction of papers

We searched the following databases: Medline (1966-present), EMBASE (1980-present), Psyclit (1974-present), Cochrane Library, Current Contents (Life Sciences), ERIC, HealthStar and Social Science Citation Index. The search strategy was limited to articles in English. The search strategies are shown in Table 1. Relevant papers were downloaded onto Reference Manager Version 7 (Seven WH),³³ and the reference lists of all papers were checked manually. Authors of included papers were contacted and asked to provide references of any published interventions that met the inclusion criteria. The following data were extracted from the studies onto standardised data collection forms by the first author: behaviours targeted for change, characteristics of participants, study design, underlying theoretical model, modes of delivery and study outcomes.

Twenty-seven papers, comprising all papers included by the first author and a random sample of 16 excluded papers, were checked against inclusion and exclusion criteria by two reviewers (SG, PL). Any disagreements were resolved by discussion. As a result, no further papers were included. Three reviewers (SG, MJ, NW) independently validated one-third of the data extraction, and any disagreements were resolved by discussion. The resulting descriptions of the intervention, including the underlying model and behavioural change methods as mentioned in the paper, were sent to the original authors for final validation. Six out of nine authors responded to the request.

Classification of models, behaviour change methods and modes of delivery

A clinical and health psychologist (MJ) developed a preliminary taxonomy of behaviour change programmes, using textbooks^34,35 and consulting experts in the field (see Appendix). Behaviour change methods were then classified (MJ), using the taxonomy. The classification was validated by a second clinical and health psychologist (DJ) and any disagreements were resolved by discussion (MJ, DJ, WH). If the authors did not explicitly mention an underlying model, the classification was based on the behaviour change methods used (MJ, DJ, WH) and the model is mentioned in parentheses. Modes of delivery were also classified independently in the same way (MJ, DJ) and validated by a third researcher (WH). Any disagreements were resolved by discussion (WH, MJ).

Results

Search results

The combined search strategy identified 11 publications that met the inclusion criteria, describing nine distinct interventions.^{36,37,38,39,40,41,42,43,44,45,46} Seven interventions were undertaken in the USA, and two in Italy. Two papers referred to an intervention with African-American mothers and daughters,^40,46 two papers to the Pound of Prevention study,^43,44 and two papers to the Pathways study.^36,42 Four interventions took place in the community ^{40,41,43,44,46} and five were school-based.^{36,37,38,39,42,45}

Description of the interventions as checked with authors

Forster et al ⁴¹ reported the community-based Pound of Prevention programme, a distinct activity within the Minnesota Heart Health Program aimed at the prevention of weight gain by diet and exercise. Monthly newsletters were distributed for a period of one year. Each newsletter had a postage-paid return postcard, on which participants filled in their current weight, date of weighing, and anything they had done to control their weight. The participants had $10 withdrawn from their bank account, and $20 was paid back if their weight at 12 months was not more than at baseline. An optional educational course was held with four sessions about midway through the year.

The Pound of Prevention study, reported by Jeffery and French,⁴³ had similar aims and target behaviours, but involved a larger population for longer (36 months). The study comprised an education-only group (25% of participants), an education plus lottery incentive group (25%), and a control group (50%). Participants in the intervention groups received a monthly two to four page newsletter with a stamped return postcard. They filled in the frequency of recommended behaviours (for example weighing) and their current body weight. In the education plus lottery incentive group the participants had their names entered into a $100 monthly lottery, contingent on returning the postcard. Participants could participate in a face-to-face weight control short course conducted by a health educator, aerobics classes, group walks, exercise seminars, a one-month membership of exercise facilities with free childcare, a corres-pondence weight control course, a marathon, a recipe contest, and the '5&3' Challenge, in which participants were challenged to exercise at least 20 min and to eat five servings of fruit and/or vegetables on a minimum of 3 days per week.⁴⁴

Fitzgibbon, Stolley et al ^40,46 reported two interventions among African-American mothers and daughters aimed at obesity prevention by dietary change. During the 6-week pilot-study,⁴⁰ weekly one-hour meetings led by a clinical psychologist or a graduate clinical psychology student took place. Measures were taken in the first and last sessions from mother and daughter pairings, or dyads. In the remaining sessions intervention dyads learnt to read food labels, calculate fat content, reduce fat intake while being able to eat in fast food restaurants, and use low-fat foods in meal planning. Group discussions focused on pros and cons of food choice, and obesity, hypertension, diabetes and analysing restaurant menus. Dyads also made a 'Rap against Fat' with the information learned in the previous weeks. Each set of lyrics was put to music and tape-recorded.

The full intervention⁴⁶ had a similar structure, but targeted physical activity alongside diet, and lasted 12 weeks. A clinical psychology student and a dietician led the sessions. The sessions were similar to the pilot study, but also included tasting low-fat and high-fat foods, using food labels to buy more healthy foods, analysis of caloric content, a presentation about medical complications of obesity, discussions about exercise and coping with making difficult dietary changes, a low-impact aerobics class, and altering favourite recipes. Mothers and daughters in the control arm met in groups and received a programme focusing on general health issues such as communicable disease control, effective communication skills, relaxation techniques, and stress reduction.

Davis et al ³⁸ evaluated a one-semester school intervention among Navajo and Pueblo Indians, aimed at the prevention of obesity by changing diet and exercise. Attention was also paid to tobacco use as a risk behaviour. Three intervention schools received a cardiovascular curriculum alone, and four schools received the same curriculum with an additional social influences component. This comprised 18 h of lessons aimed at increasing knowledge and changing health behaviours. The lessons involved culturally relevant prevention activities, and five teaching training units on the cardiovascular system, exercise, nutrition, tobacco and social influence. Members of the community were brought in to discuss the cultural importance of healthy traditions. In a home component children interviewed their family members about their eating and exercise habits. Children were sent home with nutrition recommendations, and community health fairs and a fitness competition for school staff were organised.

Pathways is a continuing school-based intervention to prevent obesity in Native American children grades 3-5 from seven communities/nations: White Mountain Apache, San Carlo Apache, Tohono O'odham, Gila River, Oglala Lakota, Sicangu Lakota and Navajo. The aim is to provide an environment that supports and encourages healthy eating choices and increased physical activity.³⁶ The formative assessment phase of the work⁴² was used to develop the intervention during an initial 3-year feasibility phase in eight schools, with full collaboration and participation of American Indians from each of the communities. It included a school-focused assessment, a community/home-based assessment, and an assessment of obesity risk behaviours. In 1996 Pathways moved to a full-scale randomised trial involving 40 schools. The programme has four components: a school curriculum, food service, physical activity and family.³⁶

Donnelly et al ³⁹ evaluated a school-based intervention to improve physical and metabolic fitness by diet and physical activity over two school years. The school lunch had a reduced energy, fat and sodium content. Cooking techniques, commodities and vendor products were altered to reduce fat and sodium, and students received 18 grade-specific modules. Three times a week teachers delivered 30-40 min sessions of physical activity designed to use large muscle groups. The emphasis was on activities that could be easily incorporated into the students' lifestyle.

Simonetti et al ⁴⁵ described a 12-month intervention in three groups of schools in Italy aimed at the prevention of obesity by changes in diet and nutrition. In multi-media action schools (MA), printed and audio-visual materials were used, and discussion meetings organised with parents and teachers. Age-specific rules were used for correct diet and nutrition. In written action schools (WA) only printed material was distributed, and rules for correct diet and nutrition disseminated. The third group of schools functioned as control schools.

A second school-based intervention in Italy was aimed at the prevention of childhood obesity by improving nutritional knowledge and promoting an active lifestyle.³⁷ A second aim was to validate a manual about dietary education. The manual was used to teach primary school teachers and involve parents in the prevention of obesity. One school received dietary education and a second school functioned as control (G Cairella & G Tarsitani, personal communication).

Models, behaviour change methods and modes of delivery

Table 2 summarises the underlying psychological model, behaviour change methods and modes of delivery. Only two studies^36,40,42 explicitly mentioned using a theoretical model, in each case Bandura's Social Learning⁴⁷ or Social Cognitive Theory⁴⁸ was mentioned, putting emphasis on increasing self-efficacy. Four studies^37,38,39,45 used a health education model with the main emphasis on providing information. One study³⁹ additionally made environmental changes. While these might be classified as behavioural, the study did not use other behavioural methods, so we classified it separately. The Pound of Prevention programme and study^41,43,44 and the study by Stolley and Fitzgibbon⁴⁶ appeared to use a behavioural model with behaviour change methods commonly used in clinical psychology, such as self-monitoring, use of incentives, increasing skills and goal setting.

The descriptions of the interventions were quite limited and, as a result, some behaviour change methods used may not have been described, or were not classifiable as described. All interventions used information, of which two^37,45 exclusively relied on this method, and one used environmental changes alongside information.³⁹ Otherwise, the most commonly used methods were rehearsal of relevant skills,^36,40,44,46 use of incentives or rewards,^41,44,46 setting goals,^36,41,44 self-monitoring,^36,41,44 homework,^38,44,46 environmental changes^36,44,46 and the use of social encouragement or support.^36,44,46 A wider range of behaviour change methods was typically used by those interventions characterised as using a behavioural or Social Learning/Cognitive Model.

Interventions most commonly took place in classrooms^{36,37,38,39,42,45} and involved the family.^{36,38,40,42,43,44,45,46} Group interventions^{40,41,43,44,46} and involvement of friends^43,44 were less common. Six interventions^{36,38,41,42,43,44,45,46} used printed or written materials, and video³⁸ and audio-visual materials⁴⁵ were each used in one study. School teachers,^36,39,42,45 nutritionists/dieticians^43,44,46 and food service staff ^36,39,42 most commonly delivered the intervention, and to a lesser degree clinical psychologists,^40,46 physical educators^36,42 and health educators.^43,44

Methodological quality

Table 3 describes participants and target behaviours and Table 4 summarises study design and duration of intervention and follow-up. Five interventions were randomised controlled trials,^{38,41,42,43,46} in which schools,^38,42 dyads of mothers and daughters,⁴⁶ and individuals^41,43 were the units of randomisation. Of these, two^38,46 did not report on appropriately corrected analyses, and in one^36,42 analyses are in progress. No study reported on intention to treat analyses and allocation concealment at randomisation or outcome assessment. Four interventions^37,39,40,45 were non-randomised trials. Recruitment rates were reported in one study,⁴¹ in which 7.3% of those who received a recruitment letter enrolled in the programme. When reported, dropout from the intervention ranged from 6%⁴¹ to 17%.⁴⁰ Dropout in the trials varied between 0%⁴³ and 23% ⁴⁶ at 1 year follow-up, and was higher in low-income⁴³ and thinner⁴⁶ individuals. The intervention and follow-up tended to be of short duration. Interventions lasted from 6 weeks⁴⁰ to 36 months,⁴³ and no completed study reported a follow-up beyond immediate post-intervention measurements.

Outcomes measured

All studies measured body weight, body mass index (BMI) or skinfold thickness, and all but two^39,41 measured self-reported diet and physical activity. Measurements of blood pressure,^39,43 fitness levels,³⁹ physical activity level,³⁶ peak aerobic capacity,³⁹ smoking,⁴³ and blood chemistry³⁹ were less common.

Characteristics of participants

Five interventions^{37,38,39,42,45} recruited school children (see Tables 3 and 4). All studies were undertaken in primary schools with students ranging from 6 to 13 y old. Davis et al ³⁸ reported that one-third were obese, Cairella et al ³⁷ that 20.5% of males and 29% of females were obese, and Simonetti et al ⁴⁵ reported that one-third were overweight or obese. In the study by Donnelly et al ³⁹ mean BMI was close to the 85th percentile for NHANES,⁴⁹ a national cross-sectional survey of the US population.

The Pound of Prevention programme,⁴¹ and the study⁴³ and intervention with African-American mothers and daughters^40,46 recruited in the community. Mean age in the Forster et al programme⁴¹ was 45.9 y and that in the Jeffery and French study⁴³ 31.0-36.6 y. Both recruited subjects who turned out to be overweight. The African-American mothers^40,46 were on average 31-34 y old, their daughters 9-11 y, and both groups were overweight compared with NHANES I.⁴⁹ Comparison of BMI across studies is difficult as it was inconsistently reported and criteria for being overweight and obese differed. Half of the studies targeted individuals with a low income or a socially disadvantaged background.

Study findings

The effectiveness of the interventions regarding weight is reported in Table 4. The Pound of Prevention programme by Forster et al ⁴¹ resulted in a net weight loss of 2.1 lb (s.e. 0.6; equivalent to 1.0 kg, s.e. 0.3) in the intervention adults at 1 y post-intervention, compared with 0.3 lb (s.e. 0.6; equivalent to 0.1 kg, s.e. 0.3) in the control adults (P=0.03). Weight maintenance or loss was more frequent in the intervention group (82% of participants) than in the control group (56%; P<0.001). Male and older participants benefited most, whereas smokers showed no positive effects. The subsequent Pound of Prevention study by Jeffery and French⁴³ resulted in more frequent weight monitoring in the intervention groups at 12 months (P=0.01 in men, P=0.004 in high-income women, P=0.06 in low-income women), but no differences in diet and physical activity. Weight change did not differ between the groups. A trend of less weight gain was observed in intervention men and high-income women, but among low-income intervention women there was more weight gain. Evaluation of activities⁴⁴ showed that the correspondence course on weight control had the highest participation rate, and participation in the exercise options offered was very low.

The pilot-study by Fitzgibbon et al ⁴⁰ with African-American mother and daughter dyads reported a decrease in daily fat intake in grams (P<0.05) and in daily percentage of total calories (P<0.05) consumed by the intervention group, compared to controls at 6 weeks follow-up, but no significant differences in nutrition knowledge or attitude (P=0.13). The mothers' self-reported eating patterns (absolute and percentage fat intake) improved significantly (P<0.05), more so than the daughters. Differences in BMI were not reported. The full trial⁴⁶ demonstrated differences at 12 months between intervention and control mothers in saturated fat intake (P<0.05) and percentage of calories from fat (P<0.001), and smaller differences for daughters in percentage of calories from fat (P<0.05). No differences in BMI were found between intervention and control mothers.

After 3 years of intervention with Native American students, Davis et al ³⁸ found a significant increase in overall knowledge (P<0.001) and self-reported exercise habits (P<0.001), and a decrease in self-reported use of butter on bread and tortillas (P<0.01), in comparison with the control students.³⁸

Donnelly et al ³⁹ found after 2 years that lunches at intervention schools contained significantly less energy (9%), fat (25%), sodium (21%), and more fibre (17%) compared to control schools (P<0.05). Differences in 24-h dietary intake between intervention and control schools were significant only for sodium (P<0.05). At the intervention schools, observed physical activity was 6% greater in the classroom than in the control schools (P<0.05), but was 16% lower outside school (P<0.05). Body weight and body fat did not differ between schools for normal weight or obese children, and no differences were found for serum levels of cholesterol, insulin and glucose. However, HDL cholesterol was significantly greater and the cholesterol/HDL ratio was less for intervention compared to control schools (P<0.05). The authors concluded that compensatory changes in energy intake and physical activity outside school may have been responsible for the lack of differences between groups.

After 1 year of school-based intervention, Simonetti et al ⁴⁵ found a decrease in the number of obese (-12.2%) and overweight (-12.1%) students in the multi-media action schools, and no differences in the written action and control schools (P-values not reported). Overall, smaller weight changes were found in obese students compared to overweight students (P-value not reported). Cairella et al ³⁷ report only baseline measures and found that obese and normal weight students did not differ in physical activity, leisure time and nutritional knowledge.

The Pathways study⁴² has yet to publish its outcome data. Currently only information about process measures has been reported.³⁶

Overall, half ^38,40,46 of the interventions that measured diet and physical activity by selfreport^{38,39,40,43,46} found positive changes. Effects on observed weight were mixed, with two studies^39,46 finding no significant differences between intervention and control group, two studies^41,45 reporting less weight gain in the intervention group and one study⁴³ finding less weight gain only in subgroups.

Discussion

The nine distinct interventions primarily aimed at the prevention of weight gain that we identified had mixed effects. Effectiveness appeared to be greater among older, male and high-income participants, and lower among low-income participants, school students and smokers. However, robust conclusions about relative effectiveness require randomised controlled designs, longer follow-up periods, bigger sample sizes, intention to treat analysis, and precise measurement.

Due to the variability of study designs, samples and outcome variables, it is difficult to identify effective types of interventions. Only one of the five studies that involved an RCT design reported a significant effect on weight. This intervention involved a correspondence programme and a mix of behaviour change methods including goal setting, self-monitoring, and contingencies. Other reviews^50,51 show that, among other things, the use of behaviour change theories, feedback on behaviour change, and individual exercise programmes are associated with effectiveness. It remains important to establish whether more complex techniques add to the effectiveness of provision of information alone, and whether face-to-face interventions are more effective than those delivered by post, as this will have implications for the likelihood of implementation, the costs and the potential geographical scope of programmes.

The underlying theoretical models and methods of behavioural change were rarely fully reported, perhaps due to perceived or actual limitations in journal space. In our attempt to classify components of interventions, we are more confident that identified components were included than that non-identified components were actually absent from the intervention. Unless underlying models and methods of behaviour change are identified and described accurately, it will be difficult to replicate effective interventions in different situations. Use of our taxonomy (see the appendix) may assist individuals developing interventions to consider the full range of possible approaches. Moreover, if underlying models and methods of behaviour change are articulated, systematic reviews in the future may be able to identify more clearly which are most effective. At present, reviews suggest that the use of psychological methods in developing interventions might render them more effective.^50,52 However, often more energy is devoted to the evaluation of poorly defined interventions of limited generalisability than to appropriate prior development.⁵³

Diverse models put differing degrees of emphasis on the key psychological processes which underlie behaviour change and which, therefore, need to be targeted in an intervention study. The model which we have characterised as the Health Education Model, focuses on the provision of information about risk and the behaviour changes required. Other interventions also include information provision, but are based on models which propose that other factors are necessary to influence behaviour change. Social Learning⁴⁷/Cognitive Theory⁴⁸ proposes that the key cognition which needs to be enhanced is self-efficacy, while the behavioural models use methods directed at changing behaviour directly rather than through changing cognitions. The models differ in their emphasis on early or late stages in the process of behavioural change, with Health Education methods and the provision of information focusing on the early, motivational changes and behavioural and environmental models being directed at the action phase.⁵⁴ Future research might explore the comparative effectiveness of individualised approaches derived from Social Learning Theory and behavioural theory on one hand, with Health Education approaches and persuasive communication, designed for larger target groups, on the other hand.

Most studies were aimed at changes in both diet and physical activity, although the majority of strategies targeted diet. When interventions targeted physical activity, they focused on exercise, and any distinction between the two concepts was not always made clear. Physical activity includes all usual activities such as walking, cycling and climbing stairs, whereas exercise usually refers to structured programmes.^55,56 Future interventions aimed at prevention of weight gain might focus on small changes in habitual physical activity, as organised exercise seems to attract relatively few participants⁴⁴ and may evoke behavioural compensation.³⁹ We noted that the technique of starting with easy changes that were then gradually increased was only reported in one study,³⁸ even though this behavioural technique has been shown to be effective.³⁴ Physical activity that is repeated and practised in a given setting will become habitual,⁵⁷ whereas exercise is more likely to require conscious decision-making and the development and implementation of specific plans to perform the behaviour.⁵⁸ As physical activity has been shown to be predictive of future weight gain in prospective epidemiological studies,^{59,60,61,62,63} it has considerable potential as a means of prevention of weight gain in non-overweight individuals. More research is needed on effective strategies to achieve this.

When diet and physical activity were measured, most studies showed positive changes. However, the great majority relied on self-report measures that may lack validity.⁵⁵ More objective measures are needed, such as heart rate monitoring with individual calibration to measure energy expenditure.⁵⁵ This technique is a feasible way of assessing the pattern and total level of energy expenditure in medium-sized epidemiological studies,⁶⁴ and has been used in high and low intensity exercise interventions.⁶⁵ Similarly, more objective assessment of food intake might be achieved by measuring relevant biomarkers.⁶⁶

Although most included studies had a randomised controlled design, they had short follow-up. As the effects of interventions on weight gain can only be measured in the long term, longer follow-up periods of about 5 years would be informative. The high dropout rates among low-income and thin individuals indicate that specific strategies are needed to retain these participants in relevant programmes. It would be useful if short-term effects on immediate processes such as uptake of the programme and changes in knowledge were reported, and if medium to long term effects on diet, physical activity and weight were described.

The methods used in this review have several limitations. We did not search the grey literature or include any unpublished interventions. However, we reduced the chance of not including relevant interventions by requesting information from authors about any other interventions that they were aware of. It sometimes proved difficult to identify the exact aim of the intervention. We may have excluded interventions that reported weight control as the aim, but which in reality, were aimed at weight gain prevention. The taxonomy that we used as a first attempt to classify underlying theoretical models, methods of behaviour change and modes of delivery, and may require refinement. However, we feel that it should be helpful to anyone developing future interventions, as it indicates the range of possible approaches to behavioural change. Finally, we might not have classified all of the components of the intervention, as the authors may not have reported everything that was actually being done, or they may have reported it in a way that made it difficult to classify. However, the process of checking the description of the components of the interventions with authors should have minimised such omissions. In addition, the use of a second abstractor to classify each component of the intervention will have diminished the likelihood of misclassification.

In conclusion, interventions aimed at the prevention of obesity or weight gain in which participants are not selected by weight are scarce, and studies provided little detail on the intervention, underlying theoretical model, and study methods. Where diet and physical activity were described, positive effects were usually obtained but all were measured by self-report, and effects on weight were mixed. Future interventions might be more effective, if they were explicitly based on methods of behaviour change that have been shown to work in other situations. Effective interventions might be more easily replicated if they were explicitly described. Effectiveness might be more precisely demonstrated if more objective measures of physical activity and diet were used, and if the follow-up was longer.

Acknowledgements

This study was funded by the DoH/MRC Nutrition Initiative. The views expressed in this publication are those of the researchers, and not necessarily the Department of Health or the MRC. We wish to thank Paul Little for his contribution to the validation of inclusion and exclusion criteria, and Derek Johnston for the validation of components of behaviour change programmes. We also thank the authors for checking the description of their intervention and for their help with the identification of other papers. NJW is an MRC Clinician Scientist Fellow.

Appendix

Table 5.

1 Prentice AM, Jebb SA. Obesity in Britain: Gluttony or sloth&quest. Br Med J 1995; 311: 437-439.

2 World Health Organisation. Obesity. Preventing and managing the global epidemic. Report of a WHO Consultation on Obesity. World Health Organisation: Geneva, 1997,

3 Obesity: The report of the British Nutrition Foundation Task Force. Blackwell Science: Oxford, 1999,

4 James WP. A public health approach to the problem of obesity. Int J Obes Relat Metab Disord 1995; 19: (Suppl 3) S37-S45. MEDLINE

5 Glenny A-M, O'Meara S, Melville A, Sheldon TA, Wilson C. The treatment and prevention of obesity: A systematic review of the literature. Int J Obes Relat Metab Disord 1997; 21: 715-737. MEDLINE

6 Knowler WC, Pettitt DJ, Savage PJ, Bennett PH. Diabetes incidence in Pima indians: Contributions of obesity and parental diabetes. Am J Epidemiol 1981; 113: 144-156. MEDLINE

7 Lissner L, Barker DJP, Blundell JE, Dietz WH, Epstein LH, Jeffery RW et al. What are the bio-behavioural determinants of body weight regulation? In: Bouchard C, Bray GA (eds). Regulation of bodyweight. Biological and behavioural mechanisms. John Wiley & Sons: Chichester, 1996, 159-177.

8 Blundell JE, King NA. Effects of exercise on appetite control: Loose coupling between energy expenditure and energy intake. Int J Obes Relat Metab Disord 1998; 22: (Suppl 2) S22-S29. MEDLINE

9 Jackson MY, Proulx JM, Pelican S. Obesity prevention. Am J Clin Nutr 1991; 53: (Suppl) S1625-S1630.

10 Battle EK, Brownell KD. Confronting a rising tide of eating disorders and obesity: Treatment vs. prevention and policy. Addict Behav 1996; 21: 755-765. MEDLINE

11 Foreyt JP, Cousins JH. Primary prevention of obesity in Mexican-American children. Ann N Y Acad Sci 1993; 699: 137-146. MEDLINE

12 Melnyk MG, Weinstein E. Preventing obesity in black women by targeting adolescents: A literature review. J Am Diet Assoc 1994; 94: 536-540. MEDLINE

13 National Institutes of Health. Clinical guidelines on the identification, evaluation, and treatment of overweight and obesity in adults, 1998;

14 Douketis JD, Feightner JW, Attia J, Feldman WF, with the Canadian Task Force on Preventive Health Care. Periodic health examination, 1999 update: 1. Detection, prevention and treatment of obesity. Can Med Assoc J 1999; 160: 513-525.

15 Scottish Intercollegiate Guidelines Network. Obesity in Scotland. Integrating prevention with weight management 1996; 8: 1-57.

16 Spring B, Pingitore R, Kessler K. Strategies to minimize weight gain after smoking cessation: Psychological and pharmacological intervention with specific reference to dexfenflu-ramine. Int J Obes Relat Metab Disord 1992; 16: (Suppl 3) S19-S23. MEDLINE

17 Hall SM, Tunstall CD, Vila KL, Duffy J. Weight gain prevention and smoking cessation: Cautionary findings. Am J Public Health 1992; 82: 799-803. MEDLINE

18 Spring B, Wurtman J, Gleason R, Wurtman R, Kessler K. Weight gain and withdrawal symptoms after smoking cessation: A preventive intervention using d-fenfluramine. Health Psychol 1991; 10: 216-223. MEDLINE

19 Taylor CB, Fortmann SP, Flora J, Kayman S, Barrett DC, Jatulis D, Farquhar JW. Effect of long-term community health education on body mass index. The Stanford Five-City Project. Am J Epidemiol 1991; 134: 235-249. MEDLINE

20 Jeffery RW. Community programs for obesity prevention: The Minnesota Heart Health Program. Obes Res 1995; 3: (Suppl 2) S283-S288.

21 Jeffery RW, Gray CW, French SA, Hellerstedt WL, Murray D, Luepker RV, Blackburn H. Evaluation of weight reduction in a community intervention for cardiovascular disease risk: Changes in body mass index in the Minnesota Heart Health Program. Int J Obes Relat Metab Disord 1995; 19: 30-39. MEDLINE

22 Angelico F, Del Ben M, Fabiani L, Lentini P, Pannozzo F, Urbinati GC, Ricci G. Management of childhoodobesity through a school-based programme of generalhealth and nutrition education. Public Health 1991; 105: 393-398. MEDLINE

23 Macaulay AC, Paradis G, Potvin L, Cross EJ, Saad Haddad C, McComber A, Desrosiers S, Kirby R, Montour LT, Lamping, DL, Leduc J, Rivard M. The Kahnawake Schools Diabetes Prevention Project: Intervention, evaluation, and baseline results of a diabetes primary prevention program with a native community in Canada. Prevent Med 1997; 26: 779-790.

24 Epstein LH. Family-based behavioural intervention for obese children. Int J Obes Relat Metab Disord 1996; 20: (Suppl 1) S14-S21. MEDLINE

25 Wing RR, Marcus MD, Epstein LH, Jawad A. A "family-based" approach to the treatment of obese type II diabetic patients. J Consult Clin Psychol 1991; 59: 156-162. MEDLINE

26 Mellin LM, Slinkard LA, Irwin CE Jr. Adolescent obesity intervention: Validation of the SHAPEDOWN program. J Am Diet Assoc 1987; 87: 333-338. MEDLINE

27 Klem ML, Wing RR, Simkin Silverman L, Kuller LH. The psychological consequences of weight gain prevention in healthy premenopausal women. Int J Eat Disord 1997; 21: 167-174. MEDLINE

28 Epstein LH, Coleman KJ, Myers MD. Exercise in treating obesity in children and adolescent. Med Sci Sports Exerc 1996; 28: 428-435. MEDLINE

29 Foster GD, Wadden TA, Brownell KD. Peer-led program for the treatment and prevention of obesity in the schools. J Consult Clin Psychol 1985; 53: 538-540. MEDLINE

30 Jeffery RW, Forster JL, French SA, Kelder SH. The Healthy Worker Project: A work-site intervention for weight control and smoking cessation. Am J Public Health 1993; 83: 395-401. MEDLINE

31 Wadden TA, Stunkard AJ, Rich L, Rubin CJ, Sweidel G, McKinney S. Obesity in black adolescent girls: A controlled clinical trial of treatment by diet, behavior modification, and parental support. Pediatrics 1990; 85: 345-352. MEDLINE

32 Sallis JF, McKenzie TL, Alcaraz JE, Kolody B, Hovell MF, Nader PR. Project SPARK. Effects of physical education on adiposity in children. Ann NY Acad Sci 1993; 699: 127-136. MEDLINE

33 Beutler EB, Cambell EP, Frincke LA, Hannon DP, Heim S, Kenney NB, Schneider KE, Smith BS, Vaucher B. Reference Manager Version 7.00N. Carlsbad: CA, USA,, 1995,

34 Kanfer FHE, Goldstein APE. Helping people change: A textbook of methods. Pergamon: New York, 1986,

35 Bootzin R. Behaviour modification and therapy: An introduction. Winthrop: Cambridge, Mas, 1975,

36 Caballero B, Davis S, Davis CE, Ethelbah B, Evans M, Lohman T, Stephenson, L, Story M, White, J. Pathways: A school-based program for the primary prevention of obesity in American Indian children. J Nutr Biochem 1998; 9: 535-543.

37 Cairella G, Romagnoli F, Cantarelli P, Valentini P, Tarsitani G. School oriented intervention on dietary education: Results of phase 1. Int J Obes Relat Metab Disord 1998; 22: (Suppl) S254 (abstract).

38 Davis S, Gomez Y, Lambert L, Skipper B. Primary prevention of obesity in American Indian children. In: Williams CL, Kimm SYS (eds). Pevention and treatment of childhood obesity. New York Academy of Sciences: New York, 1993, 167-180.

39 Donnelly JE, Jacobsen DJ, Whatley JE, Hill JO, Swift LL, Cherrington A, Polk B, Tran ZV, Reed G. Nutrition and physical activity program to attenuate obesity and promote physical and metabolic fitness in elementary school children. Obes Res 1996; 4: 229-243. MEDLINE

40 Fitzgibbon ML, Stolley MR, Kirschenbaum DS. An obesity prevention pilot program in African-American mothers and daughters. J Nutr Educ 1995; 27: 93-97.

41 Forster JL, Jeffery RW, Schmid TL, Kramer FM. Preventing weight gain in adults: A pound of prevention. Health Psychol 1988; 7: 515-525. MEDLINE

42 Gittelsohn J, Evans M, Helitzer D, Anliker J, Story M, Metcalfe L, Davis S, Cloud PI. Formative research in a school-based obesity prevention program for Native American school children (Pathways). Health Educ Res 1998; 13: 251-265. MEDLINE

43 Jeffery RW, French SA. Preventing weight gain in adults: Design, methods and one year results from the Pound of Prevention study. Int J Obes Relat Metab Disord 1997; 21: 457-464. MEDLINE

44 Sherwood NA, Morton N, Jeffery RW, French SA, Neumark-Sztainer D, Falkner NH. Consumer preferences in format and type of community-based weight control programs. Am J Health Prom 1998; 13: 12-18.

45 Simonetti D-A, Tarsitani G, Cairella M, Siani V, De Filippis S, Mancinelli S, Marazzi MC, Palombi L. Prevention of obesity in elementary and nursery school children. Public Health 1986; 100: 166-173. MEDLINE

46 Stolley MR, Fitzgibbon ML. Effects of an obesity prevention program on the eating behavior of African American mothers and daughters. Health Educ Behav 1997; 24: 152-164. MEDLINE

47 Bandura A. Social learning theory. Prentice-Hall: Englewood Cliffs, NJ, 1977,

48 Bandura A. Social foundations of thought and action: A social cognitive theory. Prentice Hall: Englewood Cliffs, NJ, 1986,

49 Must A, Dallal GE, Dietz WH. Reference data for obesity: 85th and 95th percentiles of body mass index (wt/ht²) and triceps skinfold thickness. Am J Clin Nutr 1991; 53: 839-846. MEDLINE

50 Health Promotion Effectiveness Reviews. Health promotion interventions to promote healthy eating in the general population: A review. Health Education Authority: London, 1997,

51 Health Promotion Effectiveness Reviews. Effectiveness of physical activity promotion in primary care. Health Education Authority: London, 1998,

52 Kalichman SC, Carey MP, Johnson BT. Prevention of sexually transmitted HIV-infection: A meta-analytic review of the behavioral outcome literature. Ann Behav Med 1996; 18: 6-15.

53 Medical Research Council. Primary Health Care 1997. MRC Topic Review. Medical Research Council: London, 1997,

54 Abraham C, Sheeran P, Johnston M. From health beliefs to self-regulation: Theoretical advances in the psychology of action control. Psychol Health 1998; 13: 569-592.

55 Wareham NJ, Rennie KL. The assessment of physical activity in individuals and populations: Why try to be more precise about how physical activity is assessed&quest. Int J Obes Relat Metab Disord 1998; 22: (Suppl) S30-S38. MEDLINE

56 US Department of Health and Human Services. Physical activity and health: A report of the Surgeon General. Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion: Atlanta, GA, 1996,

57 Quellette JA, Wood W. Habit and intention in everyday life: The multiple processes by which past behavior predicts future behavior. Psychol Bull 1998; 124: 54-74.

58 Gollwitzer PM, Brandstätter V. Implementation intentions and effective goal pursuit. J Pers Soc Psych 1997; 73: 186-199.

59 Rissanen A, Hellovaara M, Krekt P, Revnanen A, Aromaa A. Determinants of weight gain and overweight in adult Finns. Eur J Clin Nutr 1991; 45: 419-430. MEDLINE

60 Haapanen N, Miilunpalo S, Pasanen M, Oja P, Vuori I. Association between leisure time physical activity and 10-year body mass change among working-aged men and women. Int J Obes Relat Metab Disord 1997; 21: 288-296. MEDLINE

61 Coakley EH, Rimm EB, Colditz G, Kawachi I, Willett W. Predictors of weight change in men: Results from the Health Professionals Follow-up Study. Int J Obes Relat Metab Disord 1998; 22: 89-96. MEDLINE

62 Williamson DF, Madans J, Anda RF, Kleinman JC, Kahn HS, Byers T. Recreational physical activity and ten-year weight change in a US national cohort. Int J Obes Relat Metab Disord 1993; 17: 279-286. MEDLINE

63 Rennie KL, Wareham NJ. The validation of physical activity instruments for measuring energy expenditure: Problems and pitfalls. Public Health Nut 1998; 1: 265-271.

64 Wareham NJ, Hennings SJ, Prentice AM, Day NE. Feasibility of heart-rate monitoring to estimate total level and pattern of energy expenditure in a population-based epidemiological study: The Ely Young Cohort Feasibility Study 1994-5. Br J Nutr 1997; 78: 889-900. MEDLINE

65 Gossard D, Haskell WL, Taylor CB, Mueller JK, Rogers F, Chandler M, Ahn DK, Miller NH, DeBusk RF. Effects of low- and high-intensity home-based exercise training on functional capacity in healthy middle-aged men. Am J Cardiol 1986; 57: 446-449. MEDLINE

66 Bates CJ, Thurnham DI, Bingham SA, Margetts BM, Nelson M. Biochemical markers of nutrient intake. In: Margetts BM, Nelson N (eds). Design concepts in nutritional epidemiology. Oxford University Press: Oxford, 1991, 192-265.

Table 1 Search strategies

Table 2 Main underlying model, behaviour change methods, and modes of delivery

Table 3 Target behaviours and target group

Table 4 Baseline BMI/weight, sample size, intervention, and effect on weight

Table 5 Taxonomy used in classifying programmes^34,35