Bio-psycho-social changes may lead to unhealthy lifestyles in adolescence, which tend to co-occur and have similar determinants. This has a major impact in terms of disease, disability, death, and high costs to families and community.1,2

Adolescence has been associated to unhealthy nutrition and physical activity is reduced in this period.3 Obesity is linked to body image disorders that affect mental health and well-being.4 In 2010 the Health Behavior in School-Aged Children (HBSC) international study found that 14.3% of participants were overweight. The cited study also found that 19% of 15 year-old's reported smoking at least once a week, 26% a prevalence of weekly alcohol consumption, 33% having been drunk twice or more, and 18% using cannabis at least once in life. Regarding sexuality, 27% of 15 year-old's reported having had sexual intercourse.1 In addition, traffic injury is the leading cause of death among 15-19 year-old's and among 10–14 year old's.5

In order to prevent unhealthy lifestyles health promotion programmes are performed at schools. These are appropriate contexts to improve public health that allow access to higher numbers of adolescents.6,7 Systematic reviews have been performed to evaluate the effect of school-based interventions regarding nutrition and physical activity,6,8 substance use,9,10 or violence,11 which found some positive effects of variable magnitude. Peters and cols performed a review on three domains, i. e., nutrition, substance use and sexuality.7

This is a systematic review across different behavioural domains since they have similar determinants, and school-based interventions usually deal with all risks that affect adolescence. We questioned ¿What are the main characteristics of recent school-based health promotion and risk prevention interventions targeted to adolescents? ¿What are their main effects? The objective was to summarize characteristics and effects of school-based interventions leading to behaviours such as nutrition, physical activity, mental and holistic health, and risk prevention (eating disorders, substance use, sexuality, violence and road safety).

In February 2012 a search was performed in PubMed, PsycINFO, ERIC and Scopus, for both review and original articles. Revised databases include relevant information on health, psychology, education and multidisciplinary sciences. Since research information is updated every five years and previous works become obsolete,12,13 we considered articles published from 2007.01.01 to 2011.12.31.

The following search terms: adolescent, program evaluation, health promotion, and whenever it was possible the thesaurus tool were used (complementary material: table I). We considered getting in touch with authors if there were doubts or difficulty in retrieving full-text.

Reference lists of both original and meta-analytic articles selected were verified to identify additional studies which reached inclusion criteria. These were English, Spanish, and Portuguese languages, with abstract, quantitative experimental or observational studies in which an intervention group was compared at least twice (pretest and posttest), interventions aimed at healthy adolescents (11-17 years) without specific illness or risks (such as substance consumers, mental disorders or obese), the primary interventions were conducted at school, and the article had a moderate to strong score in quality assessment.

In the final sample, exclusion criteria were different to original and had weak quality scores. The standardized quality assessment tool of Effective Public Health Practice Project was used.14 Eight components were scored (weak/moderate/strong): (1) Selection bias, (2) study design, (3) confounders, (4) blinding, (5) data collection, (6) withdrawals and drop-outs, (7) intervention integrity, and (8) analysis.

When there were multiple publications for the same study a project account was created. The most important information was extracted and summarized in tables attending the following categories: health nutrition and physical activity, mental and holistic health, and risk prevention interventions. Regarding the sustainability, we considered population and setting, whether outcomes were measured over multiple time points and at short or long-term. Regarding the applicability we also reviewed if effect varied across different populations and settings.15 Whenever it was possible, the following subcategories were identified: reference (author, year), study characteristics (design, randomized level, country, population, follow-ups, number of variables, participants’ age), intervention characteristics (setting, objectives, theoretical framework, implementation characteristics), main effects (only those that p<.05). When it was possible, Cohen's d was calculated.16 We considered 0.2 low, 0.5 moderate, and 0.8 large effect. When offered, we used odds ratio (1.50 low, 3.50 moderate, 9.00 large effect), and R2 (0.01 low, 0.06 moderate, 0.14 large effect).17 However, given the heterogeneity of studies regarding interventions, participants, outcome measures, no pooled effect sizes were calculated but a narrative systematic review was performed.

Two independent reviewers were involved in every step, i. e., screening the results from database searches, carrying out quality assessment, extracting data, reviewing drafts and the final report. Disagreements were solved by discussion. Standardized tools were adapted and used independently.6

We have found good-quality projects in physical activity, health nutrition, mental and holistic health, and risk behaviour prevention. Implementation activities were performed in six areas, which were classified in interactive lessons, peer mediation, improving accessibility by environment changes, activities with parents, with community, and tailored message; and they mainly addressed facilitors and barriers, social influences (peers, family, school, media and community), cognitive, emotional and behavioural skills, and including multiple components. Similar activities and behavioural determinants were approached in interventions that led to different behaviours, this point that integrative programs addressing multiple behaviours could be implemented in schools by using those effective activities and elements.7 In addition, the majority of projects were based on theoretical models; previous authors emphasized that theory-based programs produce better effects.6,7,15

Some projects found moderate to large short and long-term effect and some of them measured outcomes over multiple time points. Most projects were carried out in “real conditions”, in a wide number of schools with large samples, and in some cases included family and community contexts, these situations contribute to sustainability of interventions.15

The majority of the projects conducted to promote physical activity and health nutrition found moderate to large effects on behavioural variables. Most of them were based on health promotion models, but effects were found in short to medium-term such as Haerens and cols that performed environment changes plus tailored-computer training and activities with parents,20,21 Dunton and cols19 and Covelli22 that carried out interactive lessons plus physical activity training, or Chatzisarantis& Hagger26 that changed the teaching style of physical education teachers to provided autonomy (feedback, rationale, sense of choice and acknowledge difficulties) for physical activity. However, Dzewaltowski and cols’ intervention found large long-term effects (two years). This was based on social cognitive theory and action teams improved accessibility to physical activity by school-environment changes.27

Subsequently, a few projects aimed at mental and holistic health found large effects. Some of them in a short-term, such as Hampel and cols, in stress prevention, that performed interactive lessons and activities with parents,31 and Jemmott III and cols, which based on social-cognitive and planned behaviour theories, carried out lessons taught by community co-facilitators.40,41 Likewise, Shek and cols that performed interactive lessons found effect on positive development after two years.34–36 Finally, some projects intended for risk prevention found large effects, we emphasize those in long-term. For instance, in smoking prevention, De Vries and cols, who based on social influences theory, performed active lessons combined with activities with peers, parents, and community,51–54 Perry and cols, who based on social cognitive theory, carried out interactive lessons, peer mediation, school events, and parent information.59 In violence prevention, Swaim& Kelly performed activities at school and community led by peers, based on social influence theory and ecological model.46

Regarding applicability, although most interventions were performed in a wide number of schools with large populations, only two studies compared an intervention in different contexts. De Vries and cols51–54, compared their program in different countries across Europe and found some disparity effects that were explained by administrative difficulties, design limitations (regarding response rates or random assignment), and although it was based on core objectives and methods many differences appeared amongst countries.51 Germeni y cols50 also found some disparity effect across school types that were explained by the characteristics of the population, and other school-related factors.

Our review points to well-implemented programs that may promote healthy lifestyles in adolescence. Results are consistent with those reported in recent reviews;6–11 and since risk behaviours have similar determinants, the strength of this article is the review has been conducted across different behavioural domains that affect adolescence and are usually addressed in school-based interventions. In this respect, practitioners and researchers in health and education can use those elements that were effective, especially those similar across domains, in delivering health promotion interventions.7

Regarding limitations, although databases included were relevant with the interventions analyzed, we limited the search to four databases, and also limited the time period (over last five years) and languages, so some articles might be missed. The heterogeneity concerning differences in interventions, study designs, methodologies and variables used to outcome measures precludes giving a synthesis of results and performing a meta-analysis. In the future, it could be interesting to generate sub-groups or intervention types to solve this limitation. Furthermore, although we have included some aspects related to sustainability and applicability, first it was not easy identifying aspects regarding these criteria in all the projects, second our study was mainly oriented to summarize characteristics and effects including interventions. However, in the future it may be pertinent to analyze these criteria thoroughly.15

This review has some implications for practice and research. Since school-based interventions are associated with some positive effects, should be continued and encouraged by public health institutes; and health promotion contents would be included more frequently in student curriculums.6There is enough evidence to encourage delivering school-based interventions based on theories about behavioural change o health promotion. There is some evidence that encourages performing interventions that promote environmental changes, dealing with social influence and improving cognitive, emotional and behavioural skills. Since the fact that similar domains were found in the areas reviewed, it would be interesting performing integrative interventions aimed at whole risks that affect adolescence. On the other hand, assisted computer and on-line intervention must be studied more.

There are many gaps in the literature that may be researched. It is necessary to find out which the main variables are that influence behavioural change in order to develop health promotion programs to intervene in said variables. It is required both to implement and evaluate (process and effect) interventions that include named evidence in a comprehensive way, and participatory action research may be useful in this kind of research. Thus, since it is very important to promote health and prevent risk in adolescence, it is necessary to take notice of research related to school-based intervention as a major priority and recognize the need to fund projects in this research field.

Mª Felicitas Domínguez-Berjón.

We declare that both authorshave made substantial contributions to all of the following: (1) the conception and design of the study, or acquisition of data, or analysis and interpretation of data, (2) drafting the article or revising it critically for important intellectual content, (3) the final approval of the version to be submitted.

There is not funding.

We have not conflict of interest