Background

Adequate physical activity has a multitude of benefits for adolescents including a reduced risk of developing adverse psychological [1] and physical health conditions [2]. Moreover, physical activity is positively associated with academic outcomes, including cognitive skills (e.g., executive functioning, memory) [3], attitude (e.g., motivation, self-concept) [4], academic behaviour (e.g., on-task time, organization) [5], and academic achievement (e.g., standardized test scores) [6]. Conversely, excess sedentary time particularly recreational screen-time, during adolescence has negative implications for psychological [1] and physical health [7]. It is currently recommended that adolescents engage in at least an average of 60 min per day of moderate- to vigorous-intensity physical activity (MVPA) and limit the amount of time they spend sedentary [8].

Approximately 81% of adolescents globally do not meet physical activity guidelines [9]. Additionally, adolescents spend an average of nine hours per day sitting, which includes three hours per day engaged in sedentary screen-time [10]. Secondary (middle and high) school students can spend up to 75% of their class time sedentary; often accumulated in long, unbroken bouts of sitting [11]. Although secondary schools are typically required to provide regular physical education (PE) classes, just 36% of this time is spent in MVPA [12]. Moreover, scheduled PE time declines across secondary school years and is not always compulsory in the upper secondary or high school years [13]. There are many other opportunities for students to engage in movement behaviours (i.e., increase physical activity and/or reduce sedentary time) throughout the day, including recess and lunch breaks, and during, between and after lessons [14].

Several reviews have summarised school-based movement behaviour initiatives in the secondary school context, concluding that these have been largely ineffective (null to small positive effects) [15,16,17,18,19,20,21,22,23]. However, the majority have reported intervention effects on MVPA [15, 16, 18, 21], or sedentary time [23], included just PE interventions (already contributing to school-hours physical activity) [16, 19, 21], focused on older adolescents [17] or girls only [22], restricted inclusion criteria to specific intervention designs (only RCTs) [17, 18] and study lengths [15, 18], or focused on low-middle income countries [20]. Meta-analyses were performed in only five reviews [15, 16, 18, 21, 23] that described the impact of school-based interventions on physical activity (small or null effects) [16, 18, 21], and just two analyzed the intervention components used [15, 16]. None have reported the factors crucial for intervention implementation effectiveness.

To our knowledge, there is currently no synthesis of non-PE interventions delivered at or by schools to improve movement behaviours across the school day, and existing reviews have generally not reported outcomes beyond movement behaviours, such as cognitive and academic outcomes, physical health, and/or psychological outcomes. The three aims of our systematic review were to (a) identify intervention strategies used within secondary school settings to improve students’ movement behaviours throughout school-based initiatives, delivered at or by the school; (b) determine the overall effect of the interventions (meta-analysis) on physical activity (all intensities), sedentary time, cognitive/academic, physical health and/or psychological outcomes; and (c) describe factors related to intervention implementation.

Methods

Protocol and Registration

This review was registered with PROSPERO (CRD42020169988) and followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Statement guidelines (PRISMA). Following the PROSPERO registration, the eligibility criteria were expanded to include strategies targeting school-related work (i.e. homework) and a third aim was added: to describe factors related to intervention implementation.

Eligibility Criteria and Search Strategies

Eligibility criteria and search strategy were modelled on the Participants, Interventions, Comparisons, Outcomes, and Study design (PICOS) framework. Studies were included if they included: (a) Secondary/middle/high school-age adolescents (> 11 years and < 18 years); (b) Interventions delivered in the school setting (i.e., during class, recess and lunch time) and/or related to schoolwork (e.g., homework; c) Strategy/ies to increase physical activity of any intensity (i.e., light, moderate, vigorous or moderate to vigorous) and/or decrease sedentary behaviour); d) Any outcomes (for movement behaviours measures, have to report an effect on daily and/or school-based physical activity and sedentary behaviour); e) Any study design (e.g., randomized controlled trials [individual and cluster], controlled trials, pre-post studies design, quasi-experimental studies), and with any comparison (e.g., pre-post intervention comparison) or control groups (e.g., non-exposed control/comparison groups); f) written in English. Studies were excluded from this review if they included: (a) Target population was adolescents with special needs; primary (elementary), middle and secondary school studies were combined with results reported together; participants were in year 6 only (considered a primary school year in some countries); (b) Before- or after-school hours programs; (c) Intervention targeted PE lessons, before- or after-school hours, active travel, or educational programs (e.g., non-active lifestyle/health lessons) in isolation; (d) Non-experimental studies (e.g., cross-sectional and case studies); and (e) Findings were only reported in abstracts (including poster abstracts), conference proceedings, dissertations, commentaries, editorials, review articles, and letters. The search strategy is reported in Supplementary Table 1 (Search strategy used [i.e., EBSCO and EMBASE]).

Information Sources

A systematic search was conducted using six online databases: MEDLINE complete, Embase, CINAHL, SPORTDiscus, APA PsycINFO, and ERIC. Further articles were identified via forwards and backward citation tracking of included articles and relevant systematic reviews. Peer-reviewed articles between January 2000 and January 2023 were considered for inclusion, as most school-based movement behaviour interventions have been conducted in the last ~ 20 years. Reference lists of the included articles were also screened to identify additional eligible interventions.

Study Selection

All search results were exported into a reference manager (Endnote x9, Clarivate analytics) and duplicates were removed. Titles and abstracts were exported to Covidence (Melbourne, Australia). Two authors (KP, AMC) screened all titles and abstracts; discussed discrepancies and came to a consensus for inclusion. Both authors reviewed the full text and discussed full-text discrepancies. Any disagreements were solved in a meeting involving four authors (AMC, KP, AT, JS).

Fig. 1
figure 1

Study selection process (PRISMA − 2020 version). Legend: IV, intervention; PA, physical activity; SB, sedentary behaviour

Data Collection Process

After identifying published articles, the authors identified whether they constituted individual studies or multiple articles from the same study. Hereafter, “articles” refers to the count of papers identified, while “studies” pertains to the number of distinct research studies/interventions represented by these articles. Data extraction was performed by two authors (AMC and KP), and all articles were cross-checked for accuracy by an additional author (EM). Data extracted and quantified included study and participant characteristics, intervention strategies and intervention effects. Intervention outcomes were classified into five categories: movement behaviours (e.g., physical activity, sedentary time, energy expenditure), cognitive and academic outcomes (e.g., working memory, on task behaviour), physical health (e.g., fitness, obesity, and musculoskeletal health), and psychological outcomes. Data extraction of factors relevant to the implementation of the programs (performed by AMC and KP; checked for accuracy by NL) was guided by the United Kingdom Medical Research Council Process Evaluation of Complex Interventions [24], which includes the following three components: Implementation (implementation process, fidelity, dose, adaptations, reach), Mechanism of impact (participants’ responses, mediators, unanticipated pathways/consequences, and Context (contextual factors which affect/shape intervention). Descriptive analysis was used to describe studies, participants’ characteristics and intervention components and effects on physical activity (all intensities), sedentary time, cognitive/academic, physical health, and/or psychological outcomes.

Methodological Quality

Quality assessment screening was completed by three authors (AMC, KP, NL), in duplicate, using an adapted version of the Effective Public Health Practice Project tool for quantitative studies [25]. The three authors discussed discrepancies and came to a consensus for the scores. The components assessed were selection bias (sample representation), study design (e.g., RCT), confounders (e.g., control for baseline differences), blinding (researcher and participant), data collection methods (e.g., validity and reliability of the assessment tool used), withdrawals and dropouts (e.g., report of number of withdrawals and reason), intervention integrity (e.g., percentage of participants that received the intervention), and statistical analysis (e.g., use of intention to treat approach) [26]. Individual criteria for methodological quality scores and an overall quality score were created and rated as strong (no weak rating), moderate (one weak rating) or weak (> 1 weak rating) [25]. Inter-rater reliability for assessment of study quality between authors was moderate (Cohen’s Kappa = 0.6, p < 0.001).

Statistical Analysis

Descriptive analysis (number and/or percentage) was used to summarise studies and participant characteristics, intervention strategies and the effect of these strategies on physical activity (all intensities), sedentary time, cognitive/academic, physical health and/or psychological outcomes, and factors related to intervention implementation. In addition, meta-analyses were conducted using Stata v18 when at least three articles reported intervention effects on outcomes with a comparable device (e.g., actigraphy), same outcome (e.g. daily MPA, and/or measurement scales/unit (kg/m2, cm), provided complete data for pre- and post- intervention measurements, and were randomised controlled trials. Self-reported movement behaviour outcomes were excluded from the meta-analyses. Consequently, meta-analyses included articles reporting on comparable: (a) device-measured movement behaviours (i.e. sedentary time, LPA, MPA, VPA, MVPA assessed during the whole day); and (b) body mass index and waist circumference (measured with standardized protocols). The sample size, mean difference, and standard error/confidence interval between intervention and control groups were entered into Microsoft® Excel and effect sizes were calculated (due to the varying units for a given outcome, standardized mean differences were calculated and used in meta-analysis). Reported immediate intervention effects were used for these meta-analyses and, in case of incomplete reporting, the corresponding authors were contacted to request additional information. For cluster randomized controlled trials, the point estimate and the 95% confidence interval intervention-reported data from their intervention effect that accounted for clustering were extracted. For cluster-randomised controlled trials, we ensured that all studies included in the meta-analyses accounted for clustered data in their statistical analyses. In cases where a study included multiple groups receiving the same or different intervention, a group combination to create a single pair-wise comparison was used [28]. A random-effects restricted maximum-likelihood model was utilized for the meta-analyses.

The Q statistic and I2 [27], and visual inspection of the forest plots, were used to examine statistical heterogeneity. I2 indicated high heterogeneity when > 75% and moderate when > 50%. Random effect meta-regression models were used to explore heterogeneity induced by the relationship between moderators (i.e., participants’ age [years], duration of the intervention [weeks], type of intervention [single-or multi-component]) and study effect sizes, when there were more than ten studies in the meta-analysis [28]. Where the meta-regression suggested the presence of a potentially important covariate, subgroup analyses were used to further investigate the data. For all meta-analysis and meta-regression models, statistical significance was set at p < 0.05 and effect sizes were interpreted as 0.2 small, 0.5 medium and 0.6 as large [29]. To assess potential small-study effects and publication bias for meta-analysis with at least 10 studies (i.e. MVPA, BMI, and WC) [30], funnel plots were produced and the Egger regression asymmetry test [31] was conducted. The trim-and-fill [32] computation was also used to assess the effect of publication bias on the interpretation of results.

Results

Characteristics of the Included Studies

In total, 85 articles (from 63 studies) fulfilled the inclusion criteria (Table 1). Of the included studies, 17 (27%) were conducted in Australia, followed by the United Kingdom (n = 9, 14.3%), the USA (n = 6, 9.5%) and other countries (n = 31, 49%). Study sample sizes ranged from 33 [33] to 6,476 [34] (Table 1). In total, the studies included 45,733 participants, with 50 articles (58.9%) reporting on physical activity, 31 (36.4%) on sedentary time, 35 (41.2%) on physical health, and 12 (14.1%) on psychological outcomes. Participant ages ranged from 11 to 17 years and the average age was 13.7 years (reported in 44 articles, 51.8%). The percentage of girls varied between 36% [35] and 100% [36,37,38,39,40,41]. The participants’ sex was not reported in five studies [34, 42,43,44,45]. Most of the studies (n = 46, 73%) used a randomised controlled trial design [33,34,35,36,37,38,39, 42, 44, 46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78], 15 studies (23.8%) used a quasi-experimental design [40, 79,80,81,82,83,84,85,86,87,88,89,90,91], one study (1.6%) used a cross-over trial design [43] and one (1.6%) a hybrid effectiveness-implementation design [92]. Intervention length ranged from a single 20-min session [52, 62] to three years [81].

Table 1 Characteristics of the included studies and study populations description

Strategies Used within Secondary-School Interventions and their Effect on Movement Behaviours, Cognitive/Academic, Physical Health and/or Psychological Outcomes

Various unique intervention strategies were reported (n = 23) across 11 categories (i.e., active lessons, community involvement, educational resources, environmental, incentives/rewards, peer support, physical activity session/s, research support, school policy, teacher training, and technology strategies). A definition, detailed description, and examples of these categories can be found in Table 2 and Supplementary Table 2, respectively. Single component interventions using only one strategy (n = 19 studies, 28.6%) involved physical activity sessions (n = 10), peer-led support (n = 6), or an environmental modification strategy (n = 3). Most interventions (n = 45 studies, 71.4%) involved the use of a combination of two or more strategies (‘multicomponent interventions’). The strategies most frequently used were: physical activity sessions (n = 31); environmental modifications (n = 29); educational resources (n = 25 interventions); peer support (n = 20); teacher training (n = 18); supporting technology (n = 11); active lessons (n = 8); community involvement (n = 7); research support (n = 6); incentives/rewards (n = 4); and school policies (n = 3) (Table 3). Although some interventions targeted several elements of the school day (e.g., class, recess/lunch, homework), 29 studies (46%) included a class-time component [37, 43, 44, 46, 50,51,52,53,54,55, 57,58,59, 63, 64, 66,67,68,69,70,71,72, 74, 79, 80, 84, 87, 89, 90], 22 studies (34.9%) a recess/lunch component [34, 35, 37, 38, 46, 49, 56, 57, 63, 65, 68,69,70, 72, 76, 77, 83, 86, 88, 91,92,93,94], 15 (23.8%) studies a whole-school day approach [36, 39, 40, 50, 56, 57, 61, 63, 68, 69, 72, 73, 75, 81, 83, 95], and one intervention (1.6%) [96] included an active homework component. Table 4 summarizes studies that report a positive impact on movement behaviours, cognitive/academic performance, physical wellbeing, and psychological outcomes, sorted by their methodological quality. These interventions encompassed a range of strategies, including physical activity sessions, environmental enhancements, teacher training, peer support, and educational resources.

Table 2 Definitions of the intervention strategies reported
Table 3 Frequency of intervention strategies used across studies
Table 4 Studies that reported a positive significant effect on movement behaviours, academic outcomes, physical health and/or psychological outcomes and methodological quality of the studies

Intervention Effects on Movement Behaviours

Sixty-three articles (74%) described intervention effects on movement behaviours (self-reported and/or device measured) (Table 1, outcomes). Of these, 40 articles (47%) reported on device-measured physical activity (i.e., light- [LPA], moderate- [MPA], vigorous- [VPA], moderate- to vigorous-intensity [MVPA] physical activity). Sedentary time was reported in 31 articles (36.4%); of these, 20 (23.5%) used device-measured sedentary/sitting time. A descriptive summary of the number of intervention strategies and their impact on different outcomes is presented in Supplementary Table 3.

Sufficient data for meta-analysis were available for sedentary time, LPA, MPA, VPA, MVPA assessed during the whole day (Summary statistics used for calculation of standardized mean difference across the studies are presented in Supplementary Table 7). All meta-analyses showed non-significant effects: Sedentary time, SMD = -0.02 (95% CI: -0.14, 0.11), Fig. 2.A; LPA, SMD = -0.01 (95% CI: -0.08, 0.05), Fig. 2.B; MPA, SMD = 0.06 (95% CI: -0.09,0.22), Fig. 2.C; VPA, SMD = 0.08 (95% CI: -0.02, 0.18), Fig. 2.D; MVPA, SMD = 0.05 (95% CI: -0.01, 0.12), Fig. 2.E).

Fig. 2
figure 2

Forest plots for the effect of school-based initiatives on device measured sedentary time (A), light intensity physical activity (B), moderate physical activity (C), vigorous physical activity (D), moderate-to-vigorous physical activity (E). Legend: Risk of bias: (SB) selection bias; (SD) study design; (C) confounder; (B) blinding; (DC) data collection; (W) withdrawal and (OR) overall risk

Intervention Effects on Cognitive and Academic Outcomes

Cognitive and academic outcomes were measured in 12 articles (14.1%) which did not fulfill the inclusion criteria for meta-analyses. The most frequently explored outcomes were executive functions. Evidence of a positive impact was reported on working memory [52, 57, 64], inhibitory control [62], cognitive flexibility [47], verbal fluency [47], attention and concentration [47, 110], and on-task behaviour [66]. No effects were found on inhibitory control [72, 99], and working memory [99]. A positive intervention effect was found on health course grades [67], numeracy and reading performance [78] and academically-relevant social skills ratings (cooperation, assertion, academic competence) [67]. No effect was found on mathematical skills [72].

Intervention Effects on Physical Health Outcomes

Physical health related outcomes were assessed in 29 articles (34.1%). Meta analysis of 16 articles (18.8%) showed a statistically significant small effect on BMI (SMD= -0.09 [95% CI: -0.16, -0.02]) with a moderate level of heterogeneity and 56% of the articles were assessed with high-risk of bias; Fig. 3.A). For waist circumference, 5 articles (5.9%) provided necessary data for meta-analysis. A non-significant intervention effect was found (SMD = 0.09 [95% CI: -0.03, 0.21]; Fig. 3, B).

Fig. 3
figure 3

Forest plots for the effect of school-based initiatives on body mass index (A) and waist circumference (B). Legend: Risk of bias: (SB) selection bias; (SD) study design; (C) confounder; (B) blinding; (DC) data collection; (W) withdrawal and (OR) overall risk

Evidence of a positive effect was reported for blood pressure [48, 83, 84, 94], high-density lipoprotein (HDL) to total cholesterol ratio [84], and glucose [84]. No effects on body fat [33, 106], blood pressure [76] or musculoskeletal health [80] were found. Fitness-related outcomes were assessed in 19 articles (22.4%). Positive effects on cardiorespiratory fitness (e.g. increase distance run) [44, 48, 65, 72, 87, 92, 99], and muscular fitness (e.g. muscular endurance strength) were found [69, 70, 83, 87, 99]. No effects on cardiorespiratory fitness [33, 41, 61, 76, 82] and muscular fitness [41, 50, 65, 82, 112] were reported. Hippocampal metabolism was assessed [113], showing a significant intervention effect on N-acetylaspartate and glutamate + glutamine in the left hippocampus.

Intervention Effects on Psychological Outcomes

Psychological outcomes were assessed in 12 articles (14.1%), which did not fulfill the inclusion criteria for meta-analyses. Positive effects were found in: anxiety and self-esteem [87], social support [51, 76], wellbeing [41, 51, 54, 69], self-perception [114], and lower depressive symptoms in participants with elevated depressive symptoms at baseline [111]. No effects on wellbeing [73, 99] or on stress and internalization/externalization problems [99] were reported, and one study indicated that the relationship with classmates deteriorated in the intervention group [74]. Also, there were no effects on social cognitive variables for physical activity (i.e., self-efficacy, perceived environment, social support, behavioural strategies, outcome expectations and outcome expectancies related to physical activity) [103].

Methodological Quality

Of the included articles, 65% received an overall weak quality rating, 25% were rated as moderate, and 11% were considered strong quality (Supplementary Table 4). A breakdown of the methodological quality per article is shown in Table 5. Study design, data collection methods, confounders, withdrawals and dropout were assessed as strong in > 50% of the articles. Selection bias was considered weak in 45% of the articles. For articles that had a positive effect on an outcome reported in this review (n = 58) [68%], 16 were classified as strong, 17 as moderate, and 25 as weak (Table 4).

Table 5 Methodological quality assessment for each article

Factors Relevant to the Implementation

At least one factor relevant to the implementation of the programs (implementation, mechanism of impact and/or context) was reported in 43 articles (50.5%). Specific components reported included fidelity (n = 26, 30.5%) [33, 34, 38, 45, 53, 57, 58, 60, 63, 66, 70, 73, 77, 82, 92, 99, 101, 102, 105, 107, 111, 115,116,117]. Fidelity was assessed based on adherence to the intervention protocol and consistency in delivery, of which 14 articles (16.4%) indicated that interventions were delivered as planned (> 60% of consistency with the original plan); participant responses (n = 25, 29.4%)) [33, 38, 46, 51, 53, 55,56,57, 67, 70, 73, 74, 77, 79, 88, 92, 101, 106, 107, 111, 115,116,117], where 20 articles (23.5%), indicated the majority of the participants enjoyed or participated in the intervention; dose (n = 18, 21.2%) [34, 45, 48, 57, 60, 63, 66, 70, 73, 77, 85, 87, 99, 101, 102, 105, 115, 117], of which 10 indicated that the intervention was delivered and received as planned (> 50% of the frequency and duration of the intervention was delivered); reach (n = 18, 21.2%) [33, 34, 45, 55, 60, 63, 64, 70, 88, 92, 93, 106, 107, 115,116,117], where 13 indicated that the intervention targeted the expected audience; context (n = 15, 17.6%) [33, 38, 42, 47, 48, 51, 74, 87, 99, 107, 115, 118], of which 4 indicated no adverse effects of the intervention, and 5 articles showed that the intervention needed some modifications (e.g. different devices, time for planning, resources); and adaptation was required in 6 studies (i.e. content and frequency adaptations) [34, 45, 63, 74, 77, 117].

Publication Bias

Funnel plots for MVPA, BMI and WC are presented in Supplementary material 6, and show little sign of asymmetry. In addition, the Egger’s tests for all three outcomes were not statistically significant, indicating absence of small-sample effects. There was little evidence of publication bias with pooled effect size estimates using the trim-and-fill method similar to the main findings for MVPA (SMD = 0.06 vs. 0.05), with overall interpretations of effects unchanged for BMI (SMD=-0.08 vs. -0.08) and WC (SMD = 1.46 vs.1.46).

Discussion

Our systematic review provided a narrative synthesis of the intervention strategies used within the secondary school settings to improve students’ movement behaviours throughout school-based initiatives and the factors related to intervention implementation, and a meta-analysis (where possible) of the effectiveness of school-based initiatives to increase physical activity and reduce sedentary time on adolescents’ movement behaviours, energy expenditure, cognitive/academic, behavioural and physical and psychological health outcomes. Intervention strategies that reported favourable effects on movement behaviours, cognitive/academic, physical and psychological outcomes tended to include physical activity sessions, environmental modifications, teacher training, peer support and/or educational resources. Despite some promising findings for BMI, the meta-analysis showed no significant effects of interventions on the total accumulated daily movement behaviours, assessed using accelerometers, and waist circumference.

Interventions incorporated a wide variety of strategies, with the majority being multicomponent. Previous reviews have concluded multicomponent interventions elicit greater effects on youth health and wellbeing compared to single component interventions [119], particularly those that use a whole-school approach [9]. However, it can be difficult to establish which strategies embedded within multicomponent interventions are most effective and may be context specific. Homework and recess time were seldom targeted in interventions. This is consistent with a recent review and meta-analysis of school-based recess interventions, which reported that just one of 43 interventions targeted secondary school students [120]. Physical activity sessions, educational resources, environmental modifications, peer-support, and teacher training were the most commonly used strategies, consistent with a previous review of school-based interventions to increase adolescent physical activity [16]. Although active lessons are commonly used and can have positive effects on lesson-time physical activity in primary/elementary school interventions [121], only eight of the 63 interventions in this review incorporated active lessons. There is clearly a need for evidence of the effectiveness of interventions involving active lessons, recess and lunch breaks and active homework.

Our meta-analyses revealed small positive effects of overall interventions on BMI; however, most articles were assessed with a high risk of bias. Results related to BMI are consistent with findings from a previous review indicating that school-based physical activity interventions result in a very small decrease in BMI z-score in children and adolescents [23]. We did not find significant effects for accelerometer-determined physical activity of any intensity. This is consistent with previous meta-analyses, which found that school-based physical activity interventions had small or no benefits on whole-day moderate-intensity physical activity levels of adolescents [15,16,17]. This review found few studies reported the impact on sedentary time and LPA, especially during school hours. It found a statistically significant effect for sitting and stepping time (data not shown, as two out the four studies reporting these data did not meet the inclusion criteria for analysis [i.e. randomised controlled trials]). The effect of school-based initiatives on movement behaviours may have been under-estimated as these initiatives could be more effective at reducing sedentary time and increasing LPA rather than higher intensity activity during school hours. Future research should consider assessing multiple movement behaviour intensities (i.e., sedentary time, LPA) with the appropriate devices [122].

The quality assessment of the studies showed different distributions of effectiveness between weak, moderate and strong quality. Further analyses showed no apparent differences in quality between studies that reported significant effects and those that did not for each outcome; however, only few studies were classified with low risk of bias. Only 43 articles (50.5%) reported factors relevant to the implementation of the programs. Therefore, with the lack of and inconsistencies in reporting implementation-relevant information, it is difficult to determine whether the interventions were sufficiently or appropriately implemented as intended. Insufficient equipment or supportive environments (e.g., standing desks), high dropout rates due to academic commitments, cancelled sessions due to unplanned activities, lack of time to plan and deliver, and programs not viewed as a priority were reported as factors that could influence intervention effectiveness. Evaluating factors related to implementation before and during the intervention, that allows researchers and schools to adapt and improve the implementation strategies to the school’s and students’ needs, could achieve better results [123].

Limitations of included studies were the high level of heterogeneity between outcome measures, and that only a small number of studies reported movement behaviours during school hours. It could be possible that interventions were effective during school hours, but effects were compensated for outside the school [124]. Future school-based interventions should consider (a) incorporating whole of school approaches (b) including additional outcomes such as academic performance and reporting movement behaviour outcomes for school hours and the whole day and multiple intensities, (c) reporting factors relevant to the implementation to assist in interpreting effectiveness.

Schools should be encouraged to develop and implement policies that support whole-of-school physical activity strategies. These policies could include providing teaching relief so teachers can receive professional development (e.g., to deliver active lessons in the classroom), and allocating space in the curriculum for additional physical activity sessions and delivery of educational resources.

Conclusions

While some intervention approaches for increasing adolescents’ physical activity and reducing sedentary time in secondary schools demonstrated promise (e.g. physical activity sessions, environmental modifications, teacher training, peer support, educational resources and/or active lesson strategies), small or non-significant effects were found in the meta-analyses. Future movement behaviour interventions in secondary schools should utilize a whole-school approach to beneficially change adolescents’ activity levels. Consistent reporting of implementation will increase understanding of how interventions effect outcomes.