The transition from childhood to adolescence is often associated with less healthy dietary choices(Reference Lytle, Seifert and Greenstein1,Reference Kelder, Perry and Klepp2) , commonly a reduction in fruit and vegetables (FV) intake, and an increase in consumption of sugar-sweetened beverages (SSB)(Reference Lytle, Seifert and Greenstein1,Reference Lien, Klepp and Lytle3) . Data suggest that these and other unhealthful dietary behaviours track into adulthood and may lead to an increased risk of obesity and related disease in later life(Reference Craigie, Lake and Kelly4). It is, therefore, critical that both children and adolescents have the knowledge and ability to make positive food choices and develop good dietary habits that can be carried into adulthood.
Through food provision, nutrition education and healthy school policies, schools can create an environment promoting and enabling healthful dietary choices, with nutrition education embedded into a variety of subjects including science and health(Reference Carraway-Stage, Hovland and Showers5). However, school food provision can also be associated with large amounts of food waste(Reference Cohen, Richardson and Austin6–9), which has a negative impact not only on the environment(Reference Scherhaufer, Moates and Hartikainen10) but also on the nutritional benefit of the food provided to the pupils. Low food waste production in schools may be an indicator of a well-functioning system and positive food choices in the pupils. The school setting also provides a unique research opportunity to engage with children and adolescents across diverse socio-economic and ethnic backgrounds. It is not surprising, therefore, that there are large numbers of dietary intervention studies carried out in schools.
In line with the abundance of nutrition studies in schools, the publication of systematic reviews (SR) has risen in an attempt to summarise the evidence gathered from these interventions. Fundamental in translating the evidence into practical solutions that improve the diet is identification of the intervention components and characteristics associated with effectiveness. However, published SR differ in their scope and intervention focus, complicating the identification of characteristics that lead to improved dietary behaviours. Furthermore, primary and secondary school pupils differ in the freedom of food choice afforded to them, access to ‘competitive foods’ from vending machines or offsite outlets(Reference Capper, Brennan and Woodside11) and the cognitive development processes and social interactions associated with their age(Reference Story, Lytle and Birnbaum12), but are often grouped in reviews. The variation in the primary and secondary school systems means that effective components of dietary intervention studies may differ between educational levels. Therefore, studies in primary and secondary schools should be considered independently in order to reveal successful interventions for the appropriate age group. To date, a number of reviews of SR that have been published on obesity prevention or healthy eating interventions combine schools with other settings and focus on a wide age range(Reference Brand, Pischke and Steenbock13,Reference Khambalia, Dickinson and Hardy14) , making it difficult to elucidate successful components of school-based interventions for adolescents. A WHO report on food and nutrition policy for schools was published in 2006 and although it provides separate dietary recommendations and suggested food preparation skills for younger and older pupils, it does not consider which intervention components are most relevant to each age group(15).
This paper, therefore, systematically reviews published SR to summarise the evidence base on dietary interventions and food wastage in the secondary school environment. It adds to the literature by synthesising key findings from these reviews to consolidate successful components upon which secondary school food interventions can be based. This paper aims to identify intervention components targeting dietary behaviours specifically relevant to adolescents (aged 11–18). Ultimately, this could inform the development and implementation of policies and strategies aimed at improving food choices in secondary school pupils.
Methods
Inclusion/exclusion criteria
To be included in this review, SR had to meet the following criteria: (i) published in a peer-reviewed journal before May 2020; (ii) published in the English language; (iii) review school-based interventions; (iv) involve secondary school pupils (adolescents between the ages of 11 and 18) and (v) describe the effect of intervention or school policy on food choice, dietary behaviours or food waste. In addition, SR were excluded that: were conducted in clinical adolescent populations, i.e. overweight or obese; did not report results of school interventions independently if multiple settings were described; did not report results from adolescents aged 11–18 independently if interventions in younger age groups were also included; did not report dietary behaviours independently if other health behaviours were studied and were narrative reviews, reports or position statements. If studies on primary/elementary school pupils were included in the SR, they were included if there was a subgroup analysis for the secondary school pupils. Where possible, age limits were applied to database searches to reflect adolescents. SR were included if they were published after 2000 to ensure that included reviews reflected current or recent school policies and practices and the contemporary school environment.
Study selection
CINAHL, the Cochrane Database of Systematic Reviews, EMBASE, MEDLINE, PsychINFO and Web of Science were searched (see Additional File 1 for search strategy). An initial database search in May 2018 was updated in May 2020. Two authors independently reviewed all titles generated by the search and removed duplicates. These articles were then subjected to abstract review independently by two authors, and full texts of potentially relevant articles were obtained. Discrepancies regarding relevance of the full texts for inclusion were resolved by discussion with a third author. Reference lists of the remaining articles were searched to retrieve any additional relevant articles.
Data extraction
Data were extracted against a template by one author and checked by the other authors. Relevant data were extracted from identified reviews using the following elements: aim, inclusion criteria, search period, geographical region of included studies, number and type of study, intervention approaches used in the included studies, main results of the SR, as well as any results specifically related to intervention approach, e.g. environmental restructuring v. education only, peer or parental involvement, intervention intensity, intervention provider and theoretical basis of includes studies. All results were reported as extracted from the original research paper; authors did not refer back to the primary studies.
Quality assessment
To determine the quality of the included SR, the AMSTAR 2 (A Measurement Tool to Assess Systematic Reviews) was applied(Reference Shea, Reeves and Wells16) (Additional File 2). Through discussion amongst the authors prior to quality assessment, three critical domains of the AMSTAR 2 were agreed. These were (1) explanation of selection of the study designs; (2) use of a comprehensive literature search strategy and (3) account for risk of bias in individual studies when interpreting/discussing the results. These three domains were thought to be most relevant to quality assessment of studies in this field. The included SR were then assessed based on adherence to these critical domains, as well as the presence of non-critical weaknesses determined by their relevance to the current topic and SR included. SR were marked ‘low’ or ‘critically low’ if they failed to address one or more than one of the critical domains respectively, as guided by AMSTAR 2. Two questions (7 and 10) were deemed irrelevant to the topic through initial discussions amongst the authors and, therefore, SR were not penalised if they failed to address these questions. Subsequently, SR with no or one relevant, non-critical flaw were deemed to be of ‘high’ quality and SR with more than one relevant, non-critical flaw were deemed to be of ‘moderate’ quality. Two authors independently conducted the quality assessment and any disagreements were discussed with the other authors until consensus was reached.
Data synthesis
Statistical analyses or meta-analyses were not conducted due to the heterogeneity in outcomes among SR. Instead, the authors extracted the results of existing analyses in the SR and reported them in a systematic format. In accordance with reporting of SR, PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines were followed(Reference Moher, Liberati and Tetzlaff17) (Additional File 3).
Results
The study selection process is outlined in Fig. 1. Thirteen SR on food choice and dietary behaviours met the inclusion criteria. No SR reporting outcomes related to food wastage in the secondary school environment were found.
Fig. 1 PRISMA flow chart of the systematic review process
Systematic review characteristics
Table 1 presents the main characteristics of the SR. They varied considerably in their focus and inclusion criteria. Three SR examined the effect of general healthy eating promotion in schools on dietary behaviours(Reference Van Cauwenberghe, Maes and Spittaels18–Reference Robinson, Webster and Whitt-Glover20), three examined the impact of nutrition education interventions(Reference Murimi, Moyeda-Carabaza and Nguyen21–Reference Ajie and Chapman-Novakofski23), three focused on FV interventions(Reference Knai, Pomerleau and Lock24–Reference Nørnberg, Houlby and Skov26), two focused on beverage intake (SSB or water)(Reference Vézina-Im, Beaulieu and Bélanger-Gravel27,Reference Vézina-Im and Beaulieu28) , one examined interventions using the WHO Health Promoting Schools framework(Reference McHugh, Hurst and Bethel29) and one explored barriers and facilitators for healthy eating(Reference Shepherd, Harden and Rees30).
Table 1 Characteristics of included systematic reviews
FV, fruit and vegetable; HPS, health-promoting schools; RCT, randomised-controlled trial; SR, systematic review; SSB, sugar-sweetened beverages.
Two SR included only randomised-controlled trials(Reference Meiklejohn, Ryan and Palermo22,Reference McHugh, Hurst and Bethel29) , while one stipulated no restrictions on intervention design(Reference Van Cauwenberghe, Maes and Spittaels18), resulting in a wide range in the number of included studies (Table 2). In general, the reviews did not evaluate which primary studies held greater weight based on study design and methodological rigour. Results from 168 primary studies in secondary schools were included in the SR. There was little overlap in the primary studies included Haerens et al. (2006)(Reference Haerens, Deforche and Maes31) appeared in five SR(Reference Calvert, Dempsey and Povey19,Reference Meiklejohn, Ryan and Palermo22,Reference Ajie and Chapman-Novakofski23,Reference McHugh, Hurst and Bethel29) , and Lytle et al. (Reference Lytle, Murray and Perry32) and Haerens et al. (2007)(Reference Haerens, Deforche and Maes33) were both included in four SR(Reference Meiklejohn, Ryan and Palermo22,Reference Knai, Pomerleau and Lock24,Reference De Sa and Lock25,Reference McHugh, Hurst and Bethel29) and(Reference Van Cauwenberghe, Maes and Spittaels18,Reference Meiklejohn, Ryan and Palermo22,Reference Ajie and Chapman-Novakofski23,Reference De Sa and Lock25) . All SR included interventions carried out in both males and females. Eight SR examined interventions in secondary schools only, while five also included interventions in pre- and primary schools but reported results for each educational level independently, thus allowing inclusion in this SR. Most SR that included a range of age groups from pre-school to secondary school(Reference Van Cauwenberghe, Maes and Spittaels18,Reference Robinson, Webster and Whitt-Glover20,Reference Murimi, Moyeda-Carabaza and Nguyen21,Reference Knai, Pomerleau and Lock24,Reference De Sa and Lock25) had a higher number of studies in primary schools compared with secondary schools. Eighty-nine percentage of primary studies were conducted in North America and Europe, with Australia, the Middle East, South and East Asia and South America represented in a small number of studies (Table 1).
Table 2 Intervention effectiveness as reported in the systematic reviews
AA, African American; B-A, before–after; CO, cohort study; CT, crossover trial; EVAL, evaluation using post-test survey; EXP, experimental intervention; FV, fruit and vegetable; LS, longitudinal study; MVPA, moderate-vigorous physical activity; PA, physical activity; PC, prospective cohort; PP, pre-post; QE, quasi-experimental; RCT, randomised-controlled trial; SSB, sugar-sweetened beverage.
Study quality
Two SR were rated ‘high’(Reference McHugh, Hurst and Bethel29,Reference Shepherd, Harden and Rees30) ; one was rated ‘moderate’(Reference Van Cauwenberghe, Maes and Spittaels18), seven were rated ‘low’(Reference Calvert, Dempsey and Povey19,Reference Murimi, Moyeda-Carabaza and Nguyen21,Reference Meiklejohn, Ryan and Palermo22,Reference Knai, Pomerleau and Lock24–Reference Vézina-Im, Beaulieu and Bélanger-Gravel27) and three were rated ‘critically low’(Reference Robinson, Webster and Whitt-Glover20,Reference Ajie and Chapman-Novakofski23,Reference Vézina-Im and Beaulieu28) . The critical weaknesses identified in the low and critically low-quality SR were a lack of explanation for the study designs included in the SR and not accounting for risk of bias in interpretation/discussion of results. For moderate quality SR, the authors did not register or report a pre-designed protocol or perform data extraction in duplicate.
Evidence synthesis
Reported effectiveness at improving dietary behaviours
Of the three SR that examined the effect of general healthy eating promotion, two low-quality SR(Reference Calvert, Dempsey and Povey19,Reference Robinson, Webster and Whitt-Glover20) concluded that there was evidence for improvement, and one moderate quality SR(Reference Van Cauwenberghe, Maes and Spittaels18) concluded that there was moderate and limited evidence for the effectiveness of educational and multicomponent interventions, respectively. For nutrition education interventions(Reference Murimi, Moyeda-Carabaza and Nguyen21–Reference Ajie and Chapman-Novakofski23), authors concluded that there was evidence for effectiveness based on positive results in the majority of primary studies included, provided that they involved certain intervention components; two SR were of low quality and one was of critically low quality. Of the SR that focused on FV intake(Reference Knai, Pomerleau and Lock24–Reference Nørnberg, Houlby and Skov26), all of which were of low quality, one concluded that there was inconclusive evidence for change in vegetable intake and two concluded that FV interventions had a positive effect on FV intake but these both included studies in primary/elementary school children. The two SR focusing on beverage intake(Reference Vézina-Im, Beaulieu and Bélanger-Gravel27,Reference Vézina-Im and Beaulieu28) both concluded that there was evidence for improvements in SSB and water intake; one SR was rated low quality and one was rated critically low. The high-quality SR by McHugh et al. concluded that there was limited evidence of improved dietary behaviours for nutrition interventions following the WHO Health Promoting Schools framework. Finally, the high-quality SR examining barriers and facilitators of dietary improvement(Reference Shepherd, Harden and Rees30) concluded that there was mixed evidence for improvements in knowledge and dietary behaviours, with differences according to gender. Overall effectiveness as described by the SR is summarised in Table 2.
As adolescence is associated with a low intake of FV and high intake of SSB, evidence for effective interventions targeting these dietary behaviours, alongside knowledge of and attitudes towards nutrition, was synthesised.
Reported effectiveness in key dietary behaviours
Sugar-sweetened beverage/water intake
Two SR(Reference Vézina-Im, Beaulieu and Bélanger-Gravel27,Reference Vézina-Im and Beaulieu28) reported on the effects of interventions specifically to reduce SSB or increase water intake, and four SR included change in SSB consumption as one of a variety of dietary behaviours(Reference Calvert, Dempsey and Povey19,Reference Meiklejohn, Ryan and Palermo22,Reference Ajie and Chapman-Novakofski23,Reference Vézina-Im and Beaulieu28) . Overall, there is limited evidence that environmental restructuring involving reduction in the availability of SSB or increased availability of water may be beneficial. The SR by Vézina-Im et al. reported that 72 % of studies resulted in significant reductions in SSB, with legislative or environmental interventions being the most effective at prompting this change. However, authors describe that over 60 % of studies received a weak quality rating. The SR did not report on how SSB intake was measured, i.e. ml or servings per day. Calvert et al. reported eight studies that included SSB intake as an outcome and described that 75 % of studies resulted in significant improvements in dietary behaviours. However, in this SR, as well as the others that included SSB intake as one outcome amongst several dietary behaviours, it was not possible to identify intervention effectiveness specifically for SSB consumption.
Fruit and vegetable consumption
Three SR reported solely on FV interventions(Reference Knai, Pomerleau and Lock24–Reference Nørnberg, Houlby and Skov26) and eleven SR included FV consumption as a measurement outcome. Results from the SR with a focus on FV were mixed; however, provision of free FV was the most promising component. The SR by Nørnberg et al. reported on choice architectural nudge interventions involving the distribution of free vegetables and modifications to serving style and found limited effects on vegetable intake. The distribution of free vegetables was found to have a positive impact on attitudes towards vegetables and willingness to try them but no effect on vegetable consumption. Conversely, De Sa and Lock, who reported on school FV schemes, concluded that the provision of free or subsidised FV increased consumption, although this was not always sustained at follow-up. Knai et al. reported positive outcomes on FV intake in only one out of four studies on adolescents. The lack of success was suggested to be due to short intervention duration.
The remaining SR(Reference Van Cauwenberghe, Maes and Spittaels18–Reference Robinson, Webster and Whitt-Glover20,Reference Meiklejohn, Ryan and Palermo22,Reference Ajie and Chapman-Novakofski23,Reference Vézina-Im and Beaulieu28–Reference Shepherd, Harden and Rees30) reported change in FV consumption as one of a variety of other dietary behaviours, including fat and sugar intake. Studies in these SR changed the school environment and increased the availability of FV through provision, and results suggested overall positive change in intake.
The majority of dietary interventions include FV intake as a single behaviour. However, Nørnberg et al. focused on interventions to increase only vegetable consumption in schools. This separation was deemed important given the higher intakes of fruit in adolescents compared with vegetables and the type of intervention that might be used to promote consumption.
Nutritional knowledge and attitudes
Seven SR(Reference Robinson, Webster and Whitt-Glover20,Reference Murimi, Moyeda-Carabaza and Nguyen21,Reference Ajie and Chapman-Novakofski23–Reference Nørnberg, Houlby and Skov26,Reference Shepherd, Harden and Rees30) reported on changes in nutrition knowledge and/or attitudes to food behaviours. Shepherd et al. concluded that 86 % of studies were effective at improving dietary behaviours and/or knowledge but increased knowledge was not consistent. Interventions involved classroom education, parental involvement, peer-taught lessons and changes to school meals, but authors did not identify specific intervention components that contributed to success. However, Shepherd et al. did identify gender- and age-specific results, reporting that interventions were more successful in females and 15- to 16-year-olds compared with 12- to 13-year-olds. The three studies that included nutritional knowledge as an outcome in the nutrition education-focused SR by Murimi et al. all reported improvements. De Sa and Lock reported one study in secondary schools with nutritional knowledge as a primary outcome, but no differences were found at follow-up. Robinson et al. reported four studies that included measurement of nutritional knowledge, all of which led to significant improvements in knowledge, but studies had multiple components and authors did not report on components contributing to improved nutritional knowledge.
As evidence is mixed for successful interventions specific to certain dietary behaviours, characteristics that contributed to successful interventions for multiple dietary behaviours were identified from the SR.
Intervention approaches identified in included systematic reviews as contributing to positive outcomes
SR identified a number of characteristics that contributed to the success of dietary interventions. Common components were increased availability of or exposure to healthy foods (six SR; (Reference Brand, Pischke and Steenbock13,Reference Khambalia, Dickinson and Hardy14,Reference Van Cauwenberghe, Maes and Spittaels18–Reference Robinson, Webster and Whitt-Glover20,Reference Meiklejohn, Ryan and Palermo22) ); multicomponent interventions, i.e. education plus environmental restructuring (five SR;(Reference Brand, Pischke and Steenbock13,Reference Calvert, Dempsey and Povey19–Reference Meiklejohn, Ryan and Palermo22) ); the use of online content/media and messaging (four SR;(Reference Brand, Pischke and Steenbock13,Reference Khambalia, Dickinson and Hardy14,Reference Robinson, Webster and Whitt-Glover20,Reference Ajie and Chapman-Novakofski23) ); peer or parent involvement (four SR;(Reference Brand, Pischke and Steenbock13,Reference Khambalia, Dickinson and Hardy14,Reference Robinson, Webster and Whitt-Glover20,Reference Meiklejohn, Ryan and Palermo22) ) and the use of behavioural theories (four SR;(Reference Shea, Reeves and Wells16,Reference Robinson, Webster and Whitt-Glover20,Reference Ajie and Chapman-Novakofski23,Reference Knai, Pomerleau and Lock24) ).
Increased availability of or exposure to healthy foods
Six SR described improvements in outcomes following an increased availability of or exposure to healthy foods. One SR(Reference De Sa and Lock25) discussed that increasing availability of or exposure to healthy foods, specifically FV in this SR, could be achieved in a number of ways, including provision as snacks, a change in school meals, a school garden, healthy breakfast provision or via cooking or tasting sessions at school. FV intake was the most commonly explored dietary behaviour in relation to increased availability.
Linked to this are findings from three SR(Reference Murimi, Moyeda-Carabaza and Nguyen21,Reference Vézina-Im, Beaulieu and Bélanger-Gravel27,Reference McHugh, Hurst and Bethel29) that approached the same issue from a different perspective, suggesting pursuing efforts to restrict access to unhealthier foods such as SSB and high energy or sugary snacks. This is based on evidence for reduced consumption following restriction of these foods in schools.
Multicomponent interventions
Interventions included in the SR can be characterised into three types: educational only (or behavioural), environmental only (or legislative/policy) and educational and environmental combined (multicomponent). The type of interventions included in each SR are summarised in Table 3.
Table 3 Summary of key variables
FV, fruit and vegetable; SCT, social cognitive theory; SSB, sugar-sweetened beverage.
Five SR conclude that multicomponent interventions, i.e. those combining education and changes to the environment, are more successful than environmental or educational alone. However, Van Cauwenberghe et al. describe moderate evidence suggesting a benefit of educational-only interventions. Conversely, two SR by Vézina-Im et al. in 2017 and 2019 concluded that environmental or legislative interventions alone were more effective than educational or multicomponent interventions. In both SR, authors focused on a single dietary behaviour, i.e. SSB and water consumption, respectively. Therefore, although evidence is mixed and success may be dependent on the dietary behaviour targeted, multicomponent interventions were most commonly reported to be successful.
Use of online content/media and messaging
Four SR reported successful dietary improvements and increased knowledge following interventions incorporating media, computer-based feedback and messaging and/or online content. These components were deemed more ‘age appropriate’ by some SR and, therefore, relevant to a secondary school environment. The SR by Murimi et al. discussed internet use and multimedia CD, for example, as contributors to successful interventions. Ajie et al. reviewed computer-based interventions only, and recommended the use of tailored feedback. Similarly, almost all of the studies that employed personalised computer-based dietary feedback in the SR by Calvert et al. were successful in reducing intake of SSB and increasing FV, dairy product and protein intake. This personalised feedback involved comparison with standard behaviour towards foods such as dairy products. Additionally, seven studies in this SR included media content in the form of, for example, radio or television shows promoting healthy eating behaviours, and these all reported positive change in dietary behaviours such as FV intake.
Peer or parent involvement
Four SR discussed peer and parent involvement as contributors to successful interventions. Interventions involving peers included peer-led education sessions, role models, group projects and discussions and consistent peer support. De Sa and Lock concluded that motivation from peers or fictional role models were features in three out of seven studies in adolescents that led to an increased intake of FV. The SR by Calvert et al. identified peer involvement as a main contributing factor to successful interventions, reporting that all nine studies that included peer involvement, the majority of which were rated moderate to high quality, were successful in promoting positive behavioural change. One SR (30) concluded that peers could effectively deliver nutrition education in schools.
Parents were involved in a number of ways across studies, including measurement of their own FV intake, health camps, invites to school meetings, homework assignments or by receiving written material. One SR(Reference Murimi, Moyeda-Carabaza and Nguyen21) suggested that face-to-face engagement with parents is necessary, as interventions that involved parents through, for example, nutrition classes or tasting sessions with pupils were more effective than passive methods such as receiving written material. However, SR and primary studies did not directly compare interventions with and without parental support.
Theoretical basis of the intervention
Four SR reported the use of behavioural theories as a feature of successful interventions. The most frequently used theories were Social Cognitive Theory, Transtheoretical Model and the Theory of Planned Behaviour. However, the most effective of these upon which to base interventions could not be distinguished. SR often stated that one theory could not be described as more effective than others could, but overall it was recommended that the use of theory was better than no theory. One SR(Reference Murimi, Moyeda-Carabaza and Nguyen21), however, reported that the use of theories was not associated with success but discussed that this may be because the interventions were ‘informed by theory’ rather than ‘theory-driven’. This limitation is also discussed in one other SR(15). Two SR(Reference Vézina-Im, Beaulieu and Bélanger-Gravel27,Reference Vézina-Im and Beaulieu28) attempted to code behaviour change techniques in interventions to identify those most relevant to effectiveness but neither were able to because of different combinations of behaviour change techniques used. However, the most frequently reported behaviour change techniques in the SR by Vézina-Im et al. (2017) were information about the health consequences of the behaviour, restructuring the physical environment, behavioural goal setting, self-monitoring of behaviour, threat to health and social support.
Other intervention characteristics
Population
None of the included SR reported if effectiveness was related to nationality or socio-economic status, as these were not examined in most SR. However, a number of SR reported gender differences in results(Reference Calvert, Dempsey and Povey19,Reference Ajie and Chapman-Novakofski23,Reference Knai, Pomerleau and Lock24,Reference Shepherd, Harden and Rees30) , concluding that many dietary interventions were solely effective in females(Reference Ajie and Chapman-Novakofski23,Reference Knai, Pomerleau and Lock24,Reference Shepherd, Harden and Rees30) . In one SR(Reference Calvert, Dempsey and Povey19), it was reported that four studies targeted females only but none targeted males. Calvert et al. described that different genders responded to different intervention components, for example, girls increased fruit consumption while boys reduced snacking.
Intervention duration and exposure
Intervention duration ranged from a one-off session to three years. Three SR concluded that longer duration interventions were more successful at improving dietary behaviours, suggesting at least six months(Reference Murimi, Moyeda-Carabaza and Nguyen21) and 12 months(Reference Knai, Pomerleau and Lock24) duration and an average of 6·5 computer-based education sessions(Reference Ajie and Chapman-Novakofski23). However, one SR(Reference Robinson, Webster and Whitt-Glover20) concluded that duration did not impact study findings, although authors described that few interventions were longer than 12 weeks. One SR(Reference Calvert, Dempsey and Povey19) concluded that exposure to interventions was more important for effectiveness than duration, which is supported by conclusions drawn by Murimi et al. who further suggest that interventions with contact time in intervals longer than two weeks were less likely to be successful.
Intervention provider
SR did not provide evidence for comparison between specific intervention providers. However, there is evidence for effectiveness with various implementers, including teachers, researchers, nutrition professionals and peers, suggesting that a range of individuals can deliver successful interventions. One SR(Reference Murimi, Moyeda-Carabaza and Nguyen21) reported that providing training for the implementers contributed to the success of interventions, suggesting that the provision of resources was not enough to ensure fidelity and success.
Limitations and recommendations
SR identified a number of limitations in the included primary studies, which have been summarised in Table 4. The most commonly reported were inconsistent measurement tools(Reference Van Cauwenberghe, Maes and Spittaels18–Reference Robinson, Webster and Whitt-Glover20,Reference Meiklejohn, Ryan and Palermo22,Reference Ajie and Chapman-Novakofski23,Reference De Sa and Lock25–Reference Vézina-Im and Beaulieu28) , short duration or intensity of intervention(Reference Van Cauwenberghe, Maes and Spittaels18–Reference Murimi, Moyeda-Carabaza and Nguyen21,Reference Ajie and Chapman-Novakofski23–Reference De Sa and Lock25,Reference Vézina-Im and Beaulieu28) and a lack of generalisability to other countries(Reference Van Cauwenberghe, Maes and Spittaels18,Reference Calvert, Dempsey and Povey19,Reference Knai, Pomerleau and Lock24–Reference Nørnberg, Houlby and Skov26,Reference Shepherd, Harden and Rees30) . Table 4 also contains recommendations for future interventions as described by the SR. Recommendations from high- and moderate-quality studies are highlighted. The most common recommendations from SR were for consistency on measurement tools for dietary intake, addressing the environment outside as well as inside schools and taking demographics such as gender into account during interventions.
Table 4 Limitations and research recommendations
* Limitation or recommendation from high- or moderate-quality SR.
Discussion
Overall findings
This SR has systematically identified and synthesised evidence on components of successful interventions for improving dietary behaviours specific to secondary school environments. Results indicate that while interventions are heterogeneous, there are common characteristics that contribute to improved dietary behaviours in secondary school pupils. These are multicomponent interventions combining environmental and educational strategies, increased availability of healthy foods, multimedia and computer-based education and feedback, peer and/or parental involvement and the use of behavioural theory as a basis to the intervention. However, of the thirteen SR, only two received a high confidence rating and reported mixed evidence for overall improvements in dietary behaviours. It is important to highlight that this SR of SR aimed to review studies on food waste in the school environment, but no relevant studies were found for inclusion and few studies in this area overall were found.
Targeting problematic dietary behaviours – fruit and vegetables and sugar-sweetened beverages
Although adolescents have been shown to have low intakes of FV and higher than recommended intakes of SSB(34), the majority of SR included here report the results of interventions on multiple behaviours, making it difficult to discern how FV and SSB intakes, explicitly, are influenced. Vézina-Im et al. (2017) suggested that interventions aimed at multiple behaviours should clearly report which BCT was used for each behaviour, as the selection of BCT based on the behavioural theory behind the intervention was not always clear in the primary studies. Furthermore, finding interventions on established behavioural theories may improve effectiveness. Approaches that increase the availability of FV, for example, by free provision, may improve uptake, or at least attitudes towards FV, but it is likely that interventions such as this would need to be implemented long term. Cost-effectiveness and adherence would then need to be assessed. Considering fruit and vegetable intakes as separate dietary behaviours, which was suggested by two SR and supported by other literature(Reference Appleton, Hemingway and Saulais35), is likely to have an impact on how environmental interventions that involve restructuring of the physical environment or provision of FV are implemented. This may be an important consideration for future interventions.
Similarly, environmental changes including an increase or decrease in availability of a single dietary component, i.e. SSB or water intake were reported by two SR to be more effective than education in influencing intake. However, it was cautioned that implementing only environmental change may increase the risk of unintended consequences associated with the intervention, described by Von Philipsborn et al. as adverse compensatory behaviour(Reference von Philipsborn, Stratil and Burns36). Interventions that restrict certain foods or beverages at school may lead to increased consumption before and after school. Nutrition education alongside environmental restructuring may mitigate the risk of these unintended consequences. A larger number of SR included here concluded that multicomponent interventions combining environmental changes with education were more successful. Furthermore, a review on SSB interventions by Avery et al. (Reference Avery, Bostock and Mccullough37), which was not included in this SR, highlights nutrition education delivered by peer, teachers or nutritionists can be effective at reducing SSB consumption. However, in general positive results were not maintained at follow-up. This may be due to short-term effects of education-only interventions and highlights a need for maintenance sessions as well as environmental restructuring alongside education.
Targeting adolescents
The development from child to adolescent is associated with more independence, fuller schedules, eating away from home, growing concern over appearance and weight and peer acceptance(Reference Story, Lytle and Birnbaum12), all of which influence dietary behaviours(Reference Neumark-Sztainer, Story and Perry38). Several SR and a SR of SR on physical activity in children and adolescents(Reference Wassenaar, Williamson and Johansen-Berg39) highlight the small number of studies conducted in secondary schools compared with primary schools, weakening the evidence from SR on adolescents. For example, Murimi et al. reported that only 20 % of their studies took place in secondary schools. Addressing this imbalance with more interventions targeting secondary school settings will strengthen the evidence base for interventions that positively influence the dietary behaviours of a range of age groups. Considering interventions in secondary schools separate to those in primary schools will also allow the development of age-appropriate strategies, such as those that require more developed cognitive skills like receiving personalised dietary feedback. Furthermore, results from this SR have highlighted that peers can successfully lead interventions, strengthened by process evaluations included in the high-quality SR by Shepherd et al. that suggest that interventions are acceptable to both the peer leaders and the receivers. This reiterates the process evaluation results of a peer-led education study, which concluded that peer-led nutrition education is feasible and well accepted by pupils and teachers(Reference Story, Lytle and Birnbaum12). This educational strategy may be a useful tool in the secondary school environment.
Gender differences in body image and body composition also become more apparent in adolescence(Reference Perry and Pauletti40). Khambalia et al. (Reference Khambalia, Dickinson and Hardy14) discuss that male and female motivations and responses to intervention components differ, which was apparent in results from three SR included here(Reference Calvert, Dempsey and Povey19,Reference Ajie and Chapman-Novakofski23,Reference Shepherd, Harden and Rees30) , as well as a SR not included(Reference Kropski, Keckley and Jensen41). Significant effects were often seen in females only, suggesting that gender-specific intervention components should be further explored.
Limitations and research gaps
The majority of primary studies lacked process or cost-effectiveness evaluations. This echoes the findings of other SR of SR on the control and prevention of obesity and interventions to reduce free sugar intake(Reference Khambalia, Dickinson and Hardy14,Reference Kirkpatrick, Raffoul and Maynard42) . A lack of information on acceptability and feasibility for in-school interventions limits their ability to be effectively implemented. Although it has been suggested that interventions delivered by non-school individuals are too expensive and unsustainable(Reference Brown and Summerbell43), McHugh et al. caution unnecessarily burdening school staff and curricula with interventions that do not align with the school’s priorities. These together suggest that there is a balance to strike between the cost of non-school staff as implementers of the intervention, and time and resource constraints of school staff if expected to deliver the intervention themselves. In addition, retention of knowledge and maintenance of dietary patterns in the secondary school environment is currently unknown, as interventions tended to lack long-term follow-up. Future research should consider maintenance sessions and adequate follow-up when designing interventions.
Some other limitations of the primary studies included heterogeneity in measurement tools, self-report data and a lack of description or selective reporting. However, as described by Meiklejohn et al., self-report data collection may be the only time- and cost-effective method for use in school-based interventions. An additional criticism of individual studies was that implementation, fidelity and participation rates were frequently not reported. This limits the development of recommendations for future interventions and the ability of decision makers to stimulate changes in practice. Reporting of participation and fidelity may help to explain why some interventions were ineffective at promoting healthful dietary behaviours. Furthermore, reviewed interventions were mostly conducted in Europe and the USA, and, therefore, may not be generalisable. This is particularly important considering that dietary guidelines differ between countries and, thus, studies in less developed countries are needed.
Finally, this SR aimed to synthesise evidence on reducing food wastage in the secondary school environment, but no relevant SR were found in this area. One review(Reference Byker Shanks, Banna and Serrano44) discussed inconsistent measurement techniques for food wastage in the National School Lunch Program in the USA and similar to SR included in this SR, report on few studies in post-primary/elementary school. A review by Reynolds et al. reported on three positive school-based interventions on food waste reduction involving changing dietary guidelines and education on food waste(Reference Reynolds, Goucher and Quested45), but these were either in primary school or the age group was not reported. In conclusion, analysis of food wastage in secondary school food environments is lacking and future research should include an aspect of food waste analysis, given the negative impact it can have on nutrient intake in pupils(Reference Byker, Farris and Marcenelle46,Reference Boschini, Falasconi and Giordano47) .
Quality of the systematic reviews
The majority of included SR received a low or critically low-quality rating. However, it must be noted that some SR were conducted before the quality assessment tool, AMSTAR, was published and updated in 2007 and 2017, and so less guidance was available for conducting SR at the time. Furthermore, the domains of AMSTAR-2 are open to interpretation and, thus, ratings do not represent a universal quality score. Based on the research setting of the present SR, not all critical domains advised by Shea et al. (Reference Shea, Reeves and Wells16) were applied. Despite this, many SR had fundamental flaws such as not accounting for the quality of primary studies when interpreting results. One critical domain applied in assessment was that there was explanation of study designs included in the SR. For public health interventions such as those targeting dietary behaviours, randomised controlled trials may not always be achievable and appropriate(Reference Maes, Van Cauwenberghe and Van Lippevelde48,Reference Harris, McGregor and Perencevich49) , particularly in a school setting, and do not necessarily increase the quality of evidence(Reference Rychetnik, Frommer and Hawe50), so it is important that authors explain their inclusion of one or multiple study designs in SR.
Strengths and weaknesses
Results from this SR have generated a comprehensive overview of intervention components specific to secondary school environments that have been reported to contribute to successful interventions. Nevertheless, this SR is not without limitations. The SR included in this SR were generally of low quality, lacking explanation of included study designs and not taking risk of bias into account when synthesising evidence from primary studies. Therefore, results should be interpreted with caution. Furthermore, although the aim was to investigate interventions in secondary schools only, a number of SR included interventions in primary schools and, thus, some of the conclusions drawn about intervention effectiveness in secondary schools may overlap with those that apply more generally to all educational levels or to primary schools. However, due to strict inclusion criteria, this risk was minimised. There may also be sample bias driven by the higher numbers of studies, and therefore SR, in primary schools than secondary schools.
Conclusion
There is currently limited evidence on school-based dietary interventions that can positively influence the dietary behaviours of adolescents. No single intervention type appears more effective than others in this setting and age group. Interventions should consider a design that incorporates a number of key characteristics that have been repeatedly reported to improve chance of success. These include the combination of education and environmental change, a theoretical basis, the use of ‘age-appropriate’ formats such as computer-based feedback, media and messaging, an increase in the availability of healthy foods and the involvement of peers and/or parents in education or support roles. Future research studies and interventions based on them would benefit from process evaluations and cost-effectiveness analyses for a variety of intervention durations and intensities, as well as implementers. This would guide the development of interventions on dietary interventions that have a positive effect on food choices in secondary school pupils without placing excessive burden on an already demanding school system.
Acknowledgements
Acknowledgements:
Not applicable.
Financial support:
This work is supported by the EIT Food Horizon 2020 funding. EIT Food had no role in the design, analysis or writing of this article.
Conflict of interest:
‘There are no conflicts of interest’.
Authorship:
S.F.B. and T.E.C. conducted the database search, abstract, full-text screens and quality assessment independently. T.E.C. extracted the data into predefined templates, which were checked by S.F.B., J.V.W. and M.C.M. T.E.C. wrote the manuscript guided by S.F.B., J.V.W. and M.C.M.
Ethics of human subject participation:
Not applicable.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S1368980022000829
Open access
