Introduction

In the last 30 years, the prevalence of obesity in young children and teenagers has quadrupled. In 1980, 5 % of adolescents—defined as individuals aged between 12 and 19 years—were obese, but this figure has now risen to 21 % [1–4]. Obesity is associated with chronic diseases such as type 2 diabetes mellitus and hypertension and with abnormalities in lipid and glucose profiles. The joint presence of all these factors increases the risk of heart disease, the leading global cause of death, by a factor of four [5]. Cardiovascular risks are problems that can get aggravated or lead to premature death. There is a strong association between excessive weight in childhood and adolescence and health problems in early adulthood [6]. To prevent cardiovascular risk, interdisciplinary educational interventions designed to promote dietary changes and increase cardiorespiratory fitness are the first choice therapy for overweight and obesity [4, 7–12].

Several systematic meta-analytic reviews have evaluated the effectiveness of lifestyle change interventions in adolescents [13–16]; however, these reviews also included studies on younger children and did not discuss what components of interventions influence adolescents’ adherence to treatment. Furthermore, all of them included clinical trials of both preventive and treatment interventions for obesity, although these two intervention types use different strategies and techniques to achieve their objectives. Nevertheless, the literature shows that to ensure adherence to educational programs among adolescents, it is necessary to use strategies designed specifically for this stage of development [7, 8, 17]. Moreover, the extremely negative stereotypes of people who are overweight and obese may contribute to a low quality of life in overweight and obese adolescents as well as increase the probability that they will experience depression, suicide, academic difficulties, vocational limitations and social challenges [18].

Kelly and Melnyk [19] looked at the efficacy of various interventions in a review of 17 randomized controlled trials (RCTs) carried out between 1980 and 2007; they discussed the implications of these trials for clinical practice and research. Another review evaluated the quality of clinical trials in which children and adolescents were assigned to immersion treatments that removed them from their obesogenic environment [20]. The authors pointed out that the main limitations of their studies [19-20] were: 1) They were not able to comprehend the participants' change mechanism and 2) the short description of the protocols. The authors of these earlier reviews recognized the need to expand them to include the latest research and research reported in other languages so as to be able to investigate how all levels of the ecological model can improve the intervention effect on behavior change in adolescents.

There has been only one systematic review focused on adolescent interventions and it showed that having the opportunity to use technology was an important component of the intervention to be successful, because young people are the main users of these resources [21]. One systematic review concluded that delivering material via the Internet and apps can be a good alternative to traditional methods for improving young people’s knowledge of nutrition and another evaluated the use of video games as a tool for increasing the level of physical activity among adolescents [22, 23]. Neither of these reviews evaluated the effects of the interventions they considered on overweight and obesity and the samples did not consist exclusively of overweight or obese subjects [22, 23].

The objective of this systematic review is to determine what factors contribute to the efficacy of non-drug treatments designed to promote weight loss in overweight and obese adolescents. The effectiveness of interventions was evaluated via analysis of the effects on weight and cardiometabolic markers. This paper uses the World Health Organization’s [24] definition of adolescence, namely the period from 10 until 19 years old and only studies that investigated this population were considered for inclusion in our review.

Methods

This systematic review followed the PRISMA checklist of items for systematic reviews and meta-analyses [25] and the Cochrane Handbook for Systematic Reviews of Interventions version 5.1.0 [26].

Eligibility criteria

This review covers studies in which the main target of the intervention was the adolescent; studies which focused on the treatment of the family were excluded. The RCTs included in this review cover interventions based on lifestyle change, adherence to diet and regular physical activity aiming at the behavioral change.

Inclusion criteria

  • RCTs of overweight or obese adolescents.

  • Trials with adolescents aged 10–19 years.

  • Outcomes included body mass index (BMI) or percentile z-score for weight loss according to the World Health Organization criteria (1995).

  • Treatment lasted at least 2 months.

  • Published between 2004 and 2014.

  • Published in English.

  • Studies that tested lifestyle change interventions designed to promote weight loss in adolescents.

Exclusion criteria

  • Studies that used medication in any of the groups.

  • Studies where the sample included participants with normal BMI.

  • Studies that focused only on the family or the school rather than directly on adolescents.

  • Studies without a comparison group.

  • Studies of preventive interventions.

  • Studies which did not report treatment effectiveness.

  • Studies exclusively to treat eating disorders.

  • Studies that tested specific exercise protocols.

  • Studies that tested specific diets or restrained diets.

Information sources

Detailed individual search strategies were developed for each of the following bibliographic databases: Embase, Medline (via PubMed), Web of Science and PsychInfo. The Cochrane Central Register of Controlled Clinical Trials (Accessed by Cochrane Library) was also consulted to search registered trials and related publications. The end date for the search was November 23, 2014.

Search

The selection of truncations and word combinations was adapted for each database search. All references were managed by Thomson Reuters Endnote reference manager (Web version) and duplicate hits were removed.

Study selection

Two independent reviewers read the titles and abstracts of all studies to determine whether they met the eligibility criteria. The same two reviewers read all the articles independently and blindly to evaluate their methodology. A checklist based on the eligibility criteria was developed and used to classify studies as: positive (included), negative (not included) and neutral (doubtful). Studies were classified as positive if they met all the inclusion criteria, negative if they did not meet one or more of the criteria and neutral if the reviewer was uncertain about inclusion. One hundred and fifteen articles were assessed; overall inter-rater agreement was 86.9 % (n = 115). In the disagreement cases, a third reviewer, expert on the subject, evaluated the articles to decide whether to include them or not. The concordance index, evaluated by kappa test, was estimated to be 0.650 and considered substantial, resulting in 26 articles being included.

Data collection process

One author collected the required information from the selected articles. A second author cross-checked all the collected information and confirmed its accuracy. Disagreements were resolved by a third author who read the article concerned and made a final decision.

Items of data

The following information was recorded: author(s), program name (where applicable), year of publication, country, sample size, age range, gender, eligibility criteria, time of first assessment and any follow-up assessments, description of each treatment group, support therapy used and comparison between the treatment groups. Effect size (ES) was calculated as Cohen’s d for effects on weight (in kg) or BMI (kg/m2) in studies that reported means and standard deviations. Only the outcomes, which showed p values ≥0.05 for the difference between groups, were presented in the study.

Results

Study selection

The search strategies retrieved 12,948 citations of which 7421 were duplicates. The remaining 5527 articles were evaluated on the basis of the title and abstract. The full texts of 115 publications were examined by two reviewers and on this basis 89 studies were excluded: 29 focused only on food intake or exercise prescriptions and did not include an intervention with a health education component; 18 had samples which did not meet the age criteria; 16 had samples which were not limited to overweight or obese adolescents; 10 did not include a control group; 5 were not randomized clinical trials; 5 were not published in English; 4 were descriptions of study protocols; and 2 were studies of preventive interventions. Figure 1 displays the flowchart for the inclusion of the 26 remaining trials and Table 1 presents their characteristics [27–52].

Fig. 1
figure 1

Flowchart of the included studies

Full size image
Table 1 Study characteristics
Full size table

Study quality assessment

The 26 clinical trials included in the review were assessed using the Delphi list [53] that consists of 12 items which allows the comparison among studies in methodological aspects, such as design, participants, intervention and clinical outcomes. This procedure was used to assess the methodological quality of studies and was not used to determine eligibility. One point was awarded for meeting each Delphi quality criterion and thus the maximum score was 12 points. The mean score of the included studies was 5.58 (SD = 1.027). Twenty-four of the clinical trials reported eligibility criteria. All studies used random group allocation; in 6 the allocation was blinded and 22 studies matched groups with respect to baseline characteristics. Only 3 studies blinded the assessors to group allocation; 18 studies retained at least 85 % of the original sample at the end of the intervention and reported outcome variability statistics; 24 presented between-group comparisons.

Characteristics of the studies

Table 1 shows the descriptive information of the studies. There was variability in the BMI classification criteria used: 19 studies used those of the International Obesity Task Force (IOTF), while 4 studies used those of the Centers for Disease Control and Prevention (CDC). Most of the studies included both male and female participants, but six included only female subjects [29, 30, 37, 44, 46, 52]. Ethnicity was an important component in four studies, where the aim was to develop an intervention targeted specifically at Koreans, African-Americans, Hispanic Americans or Mexican-Americans [27, 29, 35, 44]. One study had a sample including normal weight adolescents [35], but we only used outcome data from overweight and obese adolescents in our analyses. All studies involved the adolescents’ family and eight also involved the school in some way [35–37, 41–43, 48, 49]. Half the studies were carried out in the USA and the majority had been published in the last 5 years.

Intervention components

The studies covered many types of interventions of varying lengths and meeting frequencies (see Table 1). Data were collected from clinics, homes and schools or through the Internet. Changing the lifestyle of overweight and obese adolescents was an objective of all the interventions and they included elements designed to equip participants with the behavioral skills that would help them to follow a healthy diet and practice physical activity. Interventions were provided by various types of health professional: psychologists, nurses, nutritionists, physicians, fitness instructors and physiotherapists; the important point is that in all cases the intervention was multidisciplinary, taking into account all the various environments adolescents encountered, as required to treat obesity effectively. All studies lasted at least 10 weeks and the longest lasted 24 months.

In ten interventions, the parents were given advice on healthy lifestyle and how to motivate their children, but did not participate directly in the intervention [32, 35, 36, 38, 44, 46–48, 50, 51]. In six interventions, parents and adolescents both participated in intervention meetings [27, 31, 37, 40, 43, 45, 47], and in one further study obese and overweight parents, but not normal weight parents, were invited to join the intervention with their children [52]. In six of the studies where both parents and adolescents participated, they had separate meetings [29, 34, 39, 41, 42, 49] and only in the Loozit study did the parents attend the same number of meetings as their children [41, 42].

None of the programs used low-calorie diets as their nutritional component; all used nutrition education based on international and national obesity treatment guidelines. The main recommendations and components were: healthy food choices, maintain an appropriate energy balance, reduce portion sizes, reduce energy and fat intake, increase fruit and vegetable consumption, make healthy beverage and snack choices, increase water consumption, learn how to categorize foods according to their health benefits and learn how to read labels on processed food. One of the studies gave to the adolescents a DVD with healthy cooking recipes aiming to modify the family eating [38]. Another study provided lessons in cooking healthy food for the adolescents.

All the programs promoted physical activities such as walking, running, team sports, strength exercises and guidance to play video games with motion sensor [29, 30, 32, 35, 36, 39, 44, 49]. In two studies, adolescents practiced yoga [30, 46]. In another study, participants were given free access to a gym [49]. All interventions recommended exercising for at least 90 min per week and in two interventions the teenagers were given pedometers to enable them to monitor their activity [30, 44].

The majority of studies used cognitive behavioral therapy (CBT) as a method of promoting behavioral change [27, 28, 30, 31, 34–36, 39, 41, 42, 44, 46–48]. The main techniques used to modify eating habits and increase rates of physical exercise were setting goals to increase healthy behaviors, self-monitoring, cognitive re-programming, control of stimuli, problem-solving, self-efficacy enhancement, social skills, assertive communication, stress management, contingency management, environmental support, token economy, positive reinforcement and relapse prevention. Motivational Interviewing (MI) is another therapy widely used to promote behavioral change in adolescents and this was the basis of treatment in seven studies; these studies also mentioned the importance of training health professionals in the use of MI [29, 33, 38, 49–52]. Family therapy was used in some interventions [37, 40, 43, 45]. Only one study did not report the form of therapy used; however, it was reported that therapy sessions focused on self-esteem, the relationship between food and feelings and anxiety and depression [32].

Main outcomes

The mean sample size at baseline was 87.76 (SD = 60.91, range 38–288). All studies assessed BMI through Z-scores and BMI was one of the main outcome variables. Many studies also measured at least one of the variables used to assess the metabolic syndrome, namely waist circumference (WC), diastolic and systolic blood pressure (SBP and DBP, respectively), total cholesterol (TC), triglyceride (TGC), low-density lipoprotein cholesterol (LDLC), high-density lipoprotein cholesterol (HDLC) and insulin resistance (HOMA-IR). Other biological variables related to exogenous obesity that were reported are as follows: glycated hemoglobin (HbA1c) levels, C-reactive protein (CRP) levels, leptin concentration, adiponectin, 30-min insulin secretion and 0- to 120-min insulinemia. One study used increase in cardiorespiratory capacity as the primary outcome and assessed this in terms of peak oxygen uptake (VO2 max), heart rate (HR), heat rate variability (HRV) and heart frequency (HF). The quality of nutrition was measured with self-report instruments in the majority of the studies; these instruments were designed to assess the intake of fat, vegetables, sugar, fruits and the eating habits that constitute a healthy diet for this population. In most studies, the primary outcomes were related to nutrition or biological fitness; however, nine studies also reported psychological and behavioral variables such as quality of life, body satisfaction, self-esteem, family support, self-concept, self-efficacy, attitude to appearance, frequency of family meals, emotional eating, restrained eating, use of cognitive skills, concerns about body shape, motivation to change behavior and processes of change [30, 31, 33, 38, 39, 41, 42, 46, 50].

All clinical trials that presented variability statistics reported changes in BMI over time (p < 0.005). Six studies found that the intervention had no effect on any of the variables assessed based on group comparisons at various time points [21, 27, 33–35, 44]; one reported group differences in BMI z-score (p = 0.001), length of PA (p < 0.05), energy expenditure in PA (p < 0.05), motivation regulation (p < 0.01) and motivation over time (p < 0.01), but only on the second follow-up assessment [27]. All the studies reported BMI z-score, BMI (kg/m2), BMI percentile, weight and waist circumference (WC). More than half of the studies (n = 14) reported improvements in BMI; nine reported reductions in weight and six reported reductions in waist circumference variables in the intervention groups when compared with the control groups; these results are presented in detail in Table 1.

Five studies reported group differences in a variable related to metabolic syndrome at follow-up assessments [33, 36, 42, 45, 51]. Six studies reported improvements in psychological outcomes and other variables, such as the learning of new behavioral skills, motivation, hours spent watching TV, academic achievement, body satisfaction, attitude to appearance, emotional eating, restrained eating and concerns about body shape (see Table 1). In all cases except one [31] where there were group differences between the intervention group and a control group, the intervention group had improved relative to the control group.

The reported effectiveness of interventions was greatest in studies that used a waiting list control or standard health care as a control group [28, 37, 43, 44, 48], whereas clinical trials that found group differences with respect to only a few variables also found low significance in the comparison [27, 39–41, 50]. Clinical trials that included regular exercise (at least 40 min of exercise 3 times per week) as a component of the intervention also found a bigger effect on their outcomes [32, 35, 36, 44, 46, 48].

One program based on motivational techniques was not effective when the comparison was between baseline data and data collected at the end of the intervention; however, 3 months later there were differences relative to baseline in the intervention group with respect to BMI (p = 0.001), length of PA (p < 0.05), energy expenditure in PA (p < 0.05), motivation regulations (p < 0.01) and motivation over the time (p < 0.01) [33].

All the studies we analyzed were RCTs and this is the best design for establishing a cause and effect relationship between treatment and outcome variables. However, ten studies did not present means or standard deviations for pre- or post-intervention data, which made it impossible for us to evaluate ESs based on Cohen’s d [27, 29, 33, 35–37, 46, 49, 50, 52]. Only one intervention produced a large reduction in adolescents’ weight (ES:0.99) [48]; one other resulted in weight loss with a medium ES (ES:0.71) [32]. In seven interventions, the ES for weight was negative; in other words the adolescents weighed more after the intervention [30, 34, 38, 41–43, 51].

The ES for BMI was large in Tsiros et al.’s study (ES:1.69) [48] and in Farah et al.’s study (ES:0.93) [32]. In three studies, the intervention resulted in a medium ES with respect to BMI: Park et al. (ES:0.61) [44], Reinehr et al. (ES:0.46) [45] and Lloyd-Richardson et al. (ES:0.43) [39]. Small ESs were obtained in studies by Brennan et al. (ES:0.15) [28], Naar-King et al. (ES:0.11) [40, and Hofsteenge et al. (ES:0.10) [34]. In four studies, the intervention had an adverse effect on BMI, i.e., produced an increase in BMI [33, 38, 41, 42]. None of the clinical trials which reported means and standard deviations found a beneficial change (positive ES) in weight or BMI in the control group.

Discussion

There was a low mean Delphi score (5.58; SD = 1.027) for elements of design intended to minimize bias in clinical trials. Some elements of good design, such as blinding the people delivering the treatment to group assignment, are impossible to apply in this type of study, but only three studies blinded the assessors pre- and post-intervention and the allocation of the subjects was blinded in just six of them. It was noticed that the interventions that had a larger number of variables associated with obesity were statistically significant when comparing groups. These interventions were the ones that compared the treatment with an unequal care group. This makes it difficult to assess whether good results are associated with the number of meetings or with multicomponent interventions.

Studies which offered a similar type of treatment in the experimental intervention and the control group found no differences between groups [27, 39–41, 50], thus confirming the importance of controlling confounding variables to evaluate the main variables of interest accurately [13]. Even so, in all clinical trials it was possible to find differences regarding BMI during the intervention. The differences regarding BMI have been discussed in one meta-analysis that concluded that treatment interventions were more effective than preventive on this variable [16].

The number of meetings with the participants influenced the success of interventions. Interventions involving more meetings produced better results in terms of effects on BMI and biological markers associated with overweight and obesity. Previous meta-analytic reviews have noted that interventions lasting at least 6 months tend to result in greater differences between groups of children and attribute this to the difference in ‘dose’ [13, 17, 39]. However, the treatment dose effect can be weakened by the time and due to this some clinical trials found differences over time in intervention effectiveness [35, 36, 39]. Other treatments can have its effect enhanced, for example one of the clinical trials that compared conventional therapy with therapy plus motivational interviewing found that at the first follow-up assessment (third month after the intervention end) there was no difference between groups, but in the second assessment follow-up assessment (sixth month after the intervention end) there were improvements, in the intervention group, in BMI (p = 0.001), length of PA (p < 0.05), energy expenditure (p < 0.05), motivation regulation (p < 0.01) and motivation over the time (p < 0.01), which complies with the proposed approach [33].

Only one of the 16 studies for which it was possible to calculate ESs presented an intervention which had a large effect on weight; in this study, the experimental intervention focused on cognitive behavioral therapy (CBT), so this finding highlights the importance of CBT-based approaches to promoting weight loss through behavioral change [48]. Earlier meta-analyses have also shown that the interventions failed in weight reductions [16, 54]. In research on overweight and obesity in adolescents, it is problematic to rely solely on weight as an outcome variable, because there are many variables that affect weight and these can be particularly important in longitudinal studies since in this population changes in body composition or growth can mask the effects of weight loss programs on weight. In the studies we analyzed, the interventions appeared more effective when assessed in terms of the ES for change in BMI: there was an intervention effect in eight studies [28, 32, 34, 39, 40, 45, 48]. Therefore, BMI is an outcome that is less influenced by other variables that cannot be controlled in the trial. Moreover, as well as considering effects on variables that are directly associated with overweight and obesity, it is important to consider the effects on eating behavior, physical exercise habits and metabolic indicators.

The treatment components can configure an important element for success of the treatment. Multidisciplinary weight loss interventions that focus on changing eating habits and encourage exercise practice to achieve emotional and behavioral change are most effective [15]. In addition, all interventions were unanimous in considering the family support as an important component in the weight loss of the adolescents. The results of the seven clinical trials that provided programs in schools and social interactions reinforce the importance of covering the different levels of the ecological model of obesity treatment, as a similar review carried out in 2008 already noted [19]. Although another study has referred the importance of dietary restriction for weight loss, in this review no restrictive diet was found in the interventions. All studies agreed that the nutritional education to healthy eating is more effective for eating maintenance than restrictive diets in the long term [4, 7]. Providing physical activity as one of the treatment elements directly influences the outcomes, but participants in such programs revert to a sedentary lifestyle after treatment [54].

Support therapies are important components in adolescent weight loss interventions, but they need to be tailored to the needs of this particular age group; the emotional causes and consequences of overweight and obesity are specific to each stage of development. The use of technology in educational work can have highly beneficial effects on nutrition and physical exercise behavior in adolescents [22, 23]. Studies which used computer programs, the Internet, short message service, CDs and video games in the delivery of the intervention reported good adherence. The success of these delivery methods is presumably due to the fact that the adolescent participants were born in the digital era [29–31, 38, 41, 42, 52].

Twenty-five studies used CBT, motivational interviewing or family therapy to help participants change their lifestyle [20, 55]. The setting where the clinical trials were conducted reinforces the importance of joining interdisciplinary forces in health care, considering the complexity in the treatment of obesity, especially in adolescence. All studies were interdisciplinary or involved multidisciplinary teams.

Obesity treatment techniques in adolescence are far from being conclusive. However, this review has identified some potentially effective strategies and indicates that some non-drug interventions have promise as methods of promoting weight loss and improving psychological and biological markers that are consequences of overweight. Therefore, one of the main limitations of this review was to find a complete description of the clinical trials’ methodology. In addition, the best evidence in this area depends on the outcome of homogeneous variables, which are no longer possible, considering that there is no consensus on the markers’ overweight and obesity in adolescence.

The methodological weakness of the RCTs, in the data analysis description of the studies, made us describe in the results only the outcomes, which presented p value ≤0.05 in the difference between groups. It was not considered type II errors; in other words, the null hypothesis was rejected. Thus, factors such as underpower test or sample size’s power of the RCTs were not considered. We reiterate that the clinical relevance of a finding does not depend exclusively on a p value [56], whereas a study may have had a significant difference between groups based on p value, but not present a significant effect for the participant, or both groups were effective (or ineffective) not presenting significant difference. In our study, we emphasized the discussion of the components that are common to all the interventions tested by the RCTs that help adolescents to lose weight. Therefore, we focus on the main aspects that can compose a specific treatment for this group age.

The effect of non-drug treatments on weight loss in adolescents depends on a multiplicity of variables; nonetheless, this review indicates the importance of taking an interdisciplinary approach that encompasses all the various levels of the ecological model. Weight loss interventions must be appropriate to the developmental stage of users, and it is important to take into account that today’s adolescents were born in the digital era. We found that there were benefits to using technology involving the family and the school and using motivational and behavioral techniques to enhance adherence to programs based on healthy, but not restrained eating and regular physical exercise.

Limitations

There are some limitations to this review: the lack of methodological rigor and the heterogeneity of the clinical trials that were analyzed. The lack of methodological rigor can be considered a bias to the effectiveness. Language limitation: we only considered English language publications.

Another important limitation in this study is that type II errors were not considered. The null hypothesis was rejected, when we considered only statistical significance between groups based on p value to describe some results. Therefore, we suggest that future studies consider type II errors and discuss the interventions’ components that can have a robust effect on the weight loss and the cardiometabolic markers in obese adolescents.

Conclusion

This study is not conclusive about the most effective treatment for obesity in adolescents. However, changes in eating habits and inclusion of exercise in the routine are important elements of effective weight loss interventions targeted at adolescents. To achieve changes in eating and exercise habits, it is normally necessary to provide psychological therapy. The most commonly used types of therapy are CBT and motivational approaches. The family support in the treatment and using technology also seem to enhance adherence to a healthy lifestyle designed to ensure weight reduction in overweight and obese adolescents.