Article Text

Download PDFPDF

Effectiveness of after-school interventions at increasing moderate-to-vigorous physical activity levels in 5- to 18-year olds: a systematic review and meta-analysis
  1. Ruth Mears1,2,
  2. Russell Jago2
  1. 1Centre for Academic Primary Care, School of Social and Community Medicine, University of Bristol, Bristol, UK
  2. 2Centre for Exercise, Nutrition and Health Sciences, School for Policy Studies, University of Bristol, Bristol, UK
  1. Correspondence to Dr Ruth Mears, Centre for Academic Primary Care, School of Social and Community Medicine, University of Bristol, 39 Whatley Road, Bristol BS8 2PS, UK; rm14101{at}bristol.ac.uk

Abstract

Aim Physical activity in children improves cardiovascular, mental, metabolic and skeletal health. Many children fail to meet the national recommendation of at least 60 min per day of moderate-to-vigorous physical activity (MVPA). After-school programmes provide an opportunity to engage children in physical activity. This systematic review and meta-analysis examine the effectiveness of after-school interventions at increasing MVPA levels in children and adolescents.

Design Systematic review and meta-analyses.

Data sources A literature search was conducted using MEDLINE, EMBASE and PsychINFO databases from January 1950 to April 2015.

Eligibility criteria for selecting studies Inclusion criteria—Population: participants aged 5–18 years. Intervention: an after-school programme in a school-based setting as the main component of an intervention to increase physical activity levels. Outcomes: individual-level measure of time spent in MVPA. Study design: quasi-experimental, pilot, non-randomised or randomised trials. Exclusion criteria: conference abstracts, unpublished articles, dissertations and non-English language papers.

Results 1387 records were identified through database searching. After removal of duplicates, there were 748 records. 15 articles met the inclusion criteria for the systematic review. 6 studies were eligible for meta-analysis and the pooled intervention effect at end point follow-up was 4.84 min/day of MVPA (95% CI −0.94 to 10.61). The effectiveness of after-school interventions varied considerably and comparisons between studies limited by different methodological study designs. Subgroup analyses within a small minority of studies revealed significant benefits in overweight/obese children and boys. There was a lack of convincing evidence that interventions based on theories of behaviour change were more effective than those with no underlying theory.

Conclusions After-school physical activity interventions to date have had mixed effectiveness on increasing MVPA levels. More robust evaluations of extracurricular physical activity interventions are required, particularly studies that use objective assessment of physical activity.

  • Intervention
  • Child
  • Adolescent
  • Physical activity
  • School
View Full Text

Statistics from Altmetric.com

Introduction

The health benefits of physical activity (PA) for school-aged children include reduced adiposity, improved cardiovascular fitness, academic performance, mental health, skeletal health, lipid levels and blood pressure.1–4 PA levels in childhood predict adult PA levels.5 ,6 Regular adult PA has been shown to reduce the risk of coronary heart disease, stroke, diabetes, hypertension, breast and colon cancer, depression and osteoporosis.7 ,8 Many children fail to meet the national recommendation of at least 60 min/day of moderate-to-vigorous PA (MVPA).9 Developing strategies to increase PA levels is critical to reduce the comorbidity and mortality associated with inactivity.10

After-school (extracurricular) programmes provide an opportunity to engage children in PA11 but evidence supporting their effectiveness has been mixed.12–14 Strategies to increase PA levels in this time period include the adaptation of existing after-school programmes, single sport-specific interventions and multicomponent interventions.12–14 The school-based setting has potential to provide a cost-effective site for a PA intervention15 ,16 as transport of children to the site is not required, resources are readily available and staff may be willing to be trained for involvement in the programme providing a potentially sustainable option. Two systematic reviews in 2011 provide further support for focusing on the school-based setting. Atkin et al12 report that effective studies were mainly based in schools rather than the community though this review did not include a meta-analysis. A separate review reported on the positive efficacy of school-based interventions,17 though this review was not limited to the after-school time period.

Systematic reviews enable the results of multiple studies to be integrated to synthesise a higher level of evidence and provide objective critical appraisal of the literature to date. There have been several previous reviews of after-school interventions but these were conducted in 2009 and 2011, and therefore require updating in order to ensure that they remain relevant.12–14 Previous reviews of after-school PA interventions have included studies with group-level outcome measures (eg, System for Observing Fitness Instruction Time—SOFIT18) as well as individual-level outcome measures which make comparison across studies difficult and they have been limited to narrative reviews. Reviews have also been limited by a focus on all forms of activity with studies that included measures of volume of activity but not time spent in a moderate-to-vigorous intensity of PA. This limitation is important as current public health guidance in the UK, the USA, Europe and many other countries is based on minutes of MVPA. None of the previous reviews have reported on the extent to which theories of behaviour change have underpinned the intervention design, which is an important omission as theory-based interventions and theory-based derived mediators of behaviour change are now considered to be best practice for intervention design.19–21 As such, understanding how interventions were intended to function is important for assessing the factors that may have affected intervention effectiveness.

The primary aim of this systematic review was to examine the effectiveness of after-school interventions at increasing MVPA levels in children and adolescents using a meta-analysis approach where possible. The secondary aim was to report on intervention design, based on theories of behaviour change.

Methods

Search strategy

A literature search was conducted using MEDLINE, EMBASE and PsychINFO databases from January 1950 to April 2015. The search strategy included the following search terms relating to children (MeSH terms ‘Child’ and ‘Adolescent’, free-text words ‘child*’ or ‘teenager*’ or ‘adolescent*’), afterschool (free-text terms ‘after school’ and ‘extra-curricular’) and PA (MeSH terms ‘sports’ and ‘exercise’, free-text word ‘sport*’, ‘exercise*’ and ‘physical activity’). Retrieved titles and abstracts were screened and the full text obtained for potentially eligible articles. References cited within the included studies and relevant review articles were also examined using the inclusion and exclusion criteria to assess for eligibility. Results were reported in accordance with PRISMA guidelines.22

Inclusion criteria

Population: Participants aged 5–18 years.

Intervention: An after-school programme in a school-based setting as the main component of an intervention to increase PA levels.

Outcomes: Individual-level measure of time spent in MVPA.

Study design: Quasi-experimental, pilot, non-randomised or randomised trials.

Exclusion criteria

Conference abstracts, unpublished articles, dissertations and non-English language papers were excluded.

Data extraction

A reviewer (RM) extracted data from included papers which was checked by a second reviewer (RJ). Discrepancies between the data were resolved through discussions. The data extracted have been summarised in table 1.

Table 1

Data extraction

Assessment of study quality

Critical appraisal of study quality was conducted by a reviewer (RM) using an adapted version of the ‘Quality Assessment Tool for Quantitative Studies’.23 ,24 This tool was selected due to the nature of the review including a range of different quantitative study designs. A second reviewer (RJ) also appraised the included studies and discrepancies were resolved through discussion. The reliability and validity of this tool has been documented by the ‘National Collaborating Centre for Methods and Tools’ and it is deemed of strong methodological rating.25 Selection bias, study design, blinding, data collection, withdrawals and drop outs, intervention integrity and statistical analyses were appraised and each given a rating of weak, moderate or strong.

Qualitative synthesis

For the qualitative section of the systematic review, synthesis was discussed among the authors until consensus was reached. A reviewer (RM) wrote the initial qualitative synthesis, and this was checked and amended by the second reviewer (RJ).

As the included studies provided some evidence of differences by gender and body mass index (BMI) at baseline, we conducted an additional qualitative synthesis of differences by these subgroups.

Meta-analysis

To minimise heterogeneity within the meta-analysis, studies were only included if they measured the same outcome measure (adjusted difference in means of MVPA in the intervention group compared with the control group at follow-up). Random-effects meta-analyses were performed in STATA V.11 (Statacorp, College Station, Texas, USA) for these studies. The χ2 test was used to assess statistical heterogeneity. Statistical heterogeneity (I2) provides a quantitative estimation of the clinical heterogeneity and/or methodological heterogeneity within included studies in the meta-analysis. Heterogeneity was further minimised by conducting additional separate meta-analyses for accelerometer-only studies.

The first meta-analysis examined baseline to end point data for accelerometer studies (where end point data are defined as data collected at a time point closest to the end of the intervention). The second meta-analyses included self-report studies and examined baseline to end point data. The third meta-analysis focused on data from baseline to initial follow-up data collection point for accelerometer studies (where initial follow-up data collection point is defined as the first follow-up data collection point after baseline). A further analysis included self-report studies and examined baseline to initial follow-up data collection point.

Results

Literature search

In total, 1387 records were identified through database searching. Fifteen papers met the inclusion criteria. One paper was identified via the references cited from an included study. Figure 1 provides an overview of how papers were identified, included and excluded in accordance with the PRISMA guidelines.22

Figure 1

Summary of study selection process.

Participant characteristics

Online supplementary table S1 summarises participant and study characteristics. The majority of studies were conducted in the USA with only two UK-based trials.26 ,27 The total number of participants enrolled in each trial varied from 13 to 1422.28 ,29 Only one study involved children younger than 8 years old.30 The oldest child enrolled in a study was 15 years old.31 Some studies targeted specific populations such as African-American or black children in three studies,31–33 BMI>85th centile in one study34 and females in three studies.26 ,33 ,35

Supplementary table 1

Study Characteristics

Study characteristics

There were nine randomised controlled trials (RCTs)26 ,27 ,29 ,31–34 ,36 ,37 including five pilot RCTs26 ,27 ,31 ,32 ,34 and one cross-sectional RCT.36 The remaining six studies were quasi-experimental,30 ,35 ,38 longitudinal28 ,39 and cross-sectional.40 The nature of the after-school PA component of the intervention included structured or unstructured play, planned MVPA, multisport PAs, single sport PA programme (eg, soccer or dance offered alone) or adhering to specific principles such as the SPARK or CATCH Kids Club curriculum or the Young Men's Christian Association (YMCA) environmental change principles.

Study quality

The methodological critical appraisal of studies included in the systematic review is summarised in figure 2 and those included in the meta-analysis is summarised in figure 3.

Figure 2

Assessment of study quality for systematic review.

Figure 3

Assessment of study quality for meta-analysis. N/A, not available.

No studies demonstrated that participants who consented and participated in the trial were similar in baseline demographics and activity levels to those who did not. Although Iversen et al39 attempted to minimise bias by randomly selecting six schools from 115 schools for participation, they failed to report the percentage of eligible students within these schools who agreed to engage in the study. Jago et al26 was one of the few studies to attempt to compare trial participants to that of the general population. There was no difference in BMI but trial participants engaged in 19.5 fewer minutes of MVPA per day (33.2 vs 52.3) at baseline than those of a similar demographic.26

Withdrawal and drop-out rates were <20% in eight of the studies;26 ,29 ,32 ,34 ,36 ,37 ,38 ,40 however, mean programme attendance was <50% or not reported in five of these studies.34 ,36 ,37 ,38 ,40 Most studies failed to measure the consistency of the intervention delivered with only one study describing an ‘independent evaluator systematically observing after-school programme activities to assess the fidelity of intervention implementation, delivery and reach’.29 Some studies reviewed written documentation by intervention staff regarding on-site activities.30 ,32 Sample size calculations were absent in six studies.28 ,29 ,30 ,33 ,38 ,40 Where the unit of allocation and unit of analyses differed, almost all studies took account of clustering in their analysis.

Outcome measures

MVPA was measured by accelerometers in 12 studies,26 ,27 ,29 ,30–32 ,34–38 ,40 heart rate (HR) monitor in one study28 and self-report in two studies33 ,39 (table 2). There was little consistency in the unit of measurement utilised for MVPA with studies reporting hours33 or minutes per weekday26 ,38 or day,29–31 ,34 ,36 minutes per after-school time period,37 minutes per hour,35 minutes per intervention session,40 minutes per week39 and percentage lesson time in MVPA.28 Subgroup analyses according to BMI or sex were reported in a minority of studies.27 ,36–40 The majority of studies reported MVPA at baseline and at the end of the intervention time period. Only two studies provided an indication of the long-term impact of the intervention, measured by MVPA data recorded over 4 weeks after the intervention had ceased (table 3).26 ,27

Table 2

Physical activity assessment

Table 3

Moderate-to-vigorous physical activity measurement points

Theories of behaviour change

Eight studies (53%) reported that the design was based on an underpinning theory of behaviour change. The most commonly reported theory of behaviour change was social cognitive theory, which was used in four studies29 ,31 ,32 ,36 (table 4). Three studies involved self-determination theory,26 ,27 ,29 two studies an ecological approach,30 ,36 one study the health promotion model35 and one study strategic self-presentation.31 Studies based on a theory of behaviour change were effective at significantly increasing overall PA levels across all participant subgroups at all time points in one study,30 at mid-intervention only in one study,29 at 3 months after the intervention had ended in one study,26 for boys only in one study27 and ineffective in achieving any significant difference in MVPA in two studies.32 ,35 One intervention with no underlying theory of behaviour change specified was effective at increasing overall PA levels across all participant subgroups at all time points,33 one at mid-point only34 and two within certain subgroups.37 ,39

Table 4

Theory of behaviour change

Impact of intervention on MVPA

Table 5 summarises the effect of the intervention on MVPA. The greatest difference in mean MVPA (22.2 min/day, 95% CI 9.6 to 34.2, p=0.0006) was reported by Barbeau et al,33 though these data were collected through self-report. From accelerometer-based data, the largest significant mean difference in MVPA from baseline to end of intervention in the intervention versus the control group was 10.5 min/day (95% CI 1.5 to 18.6, p=0.017).30

Table 5

MVPA outcome

Difference in means of MVPA for subgroups

MVPA levels did not significantly change among all participants in each study. Subgroup analyses within certain studies, however, revealed significant differences.27 ,37 ,39 ,40 Table 6 provides a summary of included studies that have conducted a separate analysis examining the impact of the intervention on MVPA according to gender and/or weight status.

Table 6

Subgroup analysis by gender and weight status

Difference in means of MVPA at specific time points only

Wilson et al29 reported 4.87 min/day (95% CI 1.18 to 8.57, p<0.05) more MVPA in the intervention group at mid-intervention though this significant effect was lost at follow-up 2 weeks after the intervention had ceased. Weintraub et al34 similarly noted a dwindling effect of the intervention from a significant difference of 10.57 min/day of MPA (95% CI 1.42 to 19.73, p=0.03) at 3 months to a non-significant difference of 3.02 mins/day of MPA (CI −3.68 to 9.72, p=0.36) at 6 months (end point of the intervention). Jago et al26 reported a postintervention effect (3 months after intervention ceased) of 8.7 min/weekday more MVPA (95% CI 5.5 to 11.9) in the intervention group.

No difference in MVPA

Three studies reported no difference in means of MVPA in any of the participants.31 ,32 ,38 One of these was a pilot study.32 One study reported a non-significant positive trend towards a greater change in MVPA levels in the intervention group at follow-up.35

Meta-analysis

Six studies were eligible for inclusion in the meta-analysis.26 ,27 ,29 ,30 ,33 ,34 Five of these studies used an objective measure of MVPA (accelerometry)26 ,27 ,29 ,30 ,34 and one study used a self-report measure of PA.33

The first meta-analysis (figure 4) focused on accelerometer-based studies and examines the adjusted mean difference in min/day of MVPA in the intervention versus the control group from baseline to end point follow-up (where end point follow-up is defined as data collected at a time point closest to the end of the intervention). There was an effect size of 2.57 min/day of MVPA (95% CI −1.74 to 6.87) and I2 value of 44.8%.

Figure 4

Adjusted mean difference in min/day of MVPA in I versus C at end point follow-up: accelerometer-only studies. C, control group; ES, effect size; I, intervention group; MVPA, moderate-to-vigorous physical activity.

The second meta-analysis (figure 5) includes all six studies (five accelerometer, one self-report) studies and again examines the adjusted mean difference in min/day of MVPA in the intervention versus the control group from baseline to end point follow-up. There was an effect size of 4.84 min/day of MVPA (95% CI −0.94 to 10.61) and I2 value of 70.0% (figure 5).

Figure 5

Adjusted mean difference in min/day of MVPA in I versus C at end point follow-up: self-report and accelerometer studies. C, control group; ES, effect size; I, intervention group; MVPA, moderate-to-vigorous physical activity.

The third meta-analysis (figure 6) focuses on accelerometer-based studies and examines the adjusted mean difference in min/day of MVPA in the intervention versus the control group from baseline to initial follow-up point (where initial follow-up data collection point is defined as the first follow-up data collection point after baseline). The initial follow-up data collection point was at mid-intervention for two studies and at the end of the intervention for three studies. There was an effect size of 5.18 min/day of MVPA (95% CI 0.75 to 9.62) and I2 value of 46.4%. A further analysis including the self-report study resulted in an effect size of 7.04 min/day of MVPA (95% CI 1.59 to 12.5) and I2 value of 65.5.%.

Figure 6

Adjusted mean difference in min/day of MVPA in I versus C at initial follow-up: accelerometer-only studies. C, control group; ES, effect size; I, intervention group; MVPA, moderate-to-vigorous physical activity.

Discussion

Main findings

This systematic review found considerable variation in the effectiveness of after-school PA interventions, with comparisons between studies limited by different study designs. Studies reporting a beneficial effect on MVPA across all subgroups of participants were often limited in terms of precision by wide CIs. The only meta-analyses demonstrating some evidence of a difference in MVPA were those which included mid-intervention data instead of end-intervention data for two studies. However, as mid-intervention data were only available for two of the studies in the meta-analyses, caution is needed regarding the strength of evidence supporting a beneficial change in MVPA from baseline to mid-point versus baseline to end point of the intervention.

Subgroup analyses within a small minority of studies revealed specific benefits in overweight/obese children36 ,37 ,39 and boys in two studies.27 ,40 These findings suggest that adaptations to content to suit the needs of particular groups may be needed. However, due to the small number of studies undertaking subgroup analyses and the lack of consistent methodology for these analyses, the significance of subgroup differences should be interpreted cautiously until further evidence is available.

There was a lack of convincing evidence that interventions based on theories of behaviour change were more effective than those with no underlying theory.

Possible explanations

The potential of after-school programmes to influence MVPA levels may be understood more fully by studying potential effective components within an intervention strategy. Barbeau et al33 reported a mean difference in MVPA between the control and the intervention groups of 22.2 min/day (95% CI 9.6 to 34.2), though this was through self-report measures. One strategy employed by Barbeau et al was to provide immediate feedback to participants on whether they were achieving sufficient intensity of exercise during a session using HR monitors and teaching participants on maintaining a HR of above 150 bpm. Ignico and Ethridge28 also utilised the concept of children ‘self-regulating activity intensity to stay within a target HR zone’ and reported this to be the motivating force behind the 38 min (95% CI NR) of MVPA reported per intervention session. A separate study specifically examined the use of HR feedback to increase PA in children and demonstrated a significant increase in vigorous PA levels.42

It is important to note that Barbeau et al randomised students within schools to the intervention or the control group at the individual level. This potentially minimised the effect of any concurrent school physical or educational factors that may influence MVPA levels and contaminate the intervention or the control group. Weintraub et al34 also randomised at the individual level and the study reported the intervention group to obtain 10.57 min/day of MPA (95% CI 1.42 to 19.73) more than the control group mid-intervention, though this significant difference was lost at 6 months. The majority of other trials randomised at the school level.26 ,27 ,29 ,30 ,36–38 Although measures were taken to try and control the potential confounders through adjustment for cluster-level effects, different schools with different characteristics have already been selected and the objective of randomisation potentially diminished by unknown confounding variables.43

Some studies identified positive changes in overweight/obese children36 ,37 or the ‘at-risk’ population defined by Iversen et al39 as those with a BMI>85th centile, PA<300 min per week or <5 fruit and vegetable servings per day. A previous study has reported that obese children tend to be less active than non-obese children particularly outside of school time.44 Given this research, it is possible that an after-school PA programme may replace a normally sedentary time for obese children and active time for non-obese children, thus explaining the potential discrepancy in effect between these subgroups. Of note, the only intervention within this review specifically targeting obese children was found to be effective at the mid-intervention point, though the sample size was small, the CI large and the effect was lost by the end of the intervention time period.34 Madsen et al37 recorded overweight and obese students attending more sessions than normal weight students (60% vs 39%, 95% CI for difference 2 to 38) potentially indicating that it may be feasible to target this weight group.

The analysis also showed that there may be some evidence of a gender difference with greater effect on the MVPA of boys. Jago et al27 found that boys in the intervention group obtained 8.6 min more of weekday MVPA than the control group (95% CI 2.8 to 14.5), with no evidence of an effect for girls. Similarly, Schuna et al40 found that boys achieved greater MVPA levels than girls in the Keep It Moving (KIM) after-school programme. This difference between boys and girls has been reported elsewhere in the literature,45 ,46 though the reasons underlying this remain unclear. This finding suggests that there is a particular need to find ways to increase girls MVPA during extracurricular interventions.

Interventions in the context of daily PA levels

Data from the Avon Longitudinal Study of Parents and Children (ALSPAC) showed that the highest peaks of PA within a day occurred during the after-school time period.47 After-school programmes may therefore occur within an already active time period where children are already engaging in PA, resulting in minimal change in overall daily MVPA levels with the intervention. This may explain why the study by Gortmaker et al30 was successful (MD 10.5 min of MVPA per day, 95% CI 1.5 to 18.6) as it targeted children already enrolled in an after-school programme and aimed to optimise PA through modification of this programme through a set of environmental standards. This meant that the setting the children were in had not changed, but optimisation of this setting had taken place potentially leading to higher rates of sustainability of the intervention.

Further observational studies examining what activities active children do and where they engage in these activities in the after-school time period may be useful to consider when developing strategies to engage less active children in PA. The focus on whether to develop pre-existing after-school programmes or create new research programmes may depend on country-level factors. In the USA, for example, after-school programmes (eg, YMCA) are more widespread than the UK and are used as a form of aftercare for working parents. This suggests that interventions targeted at this population may reach a different population to that of newly created PA after-school programmes. However, given the contextual difference, where there is a lack of current provision, there is a clear need to create, optimise and evaluate new programmes.

It is important for studies to clearly identify the target population for their intervention and the clinically relevant outcome they are trying to achieve within this target population. For example, Herrick et al38 noted that participants in their study were already achieving an average of 21 min of MVPA in the after-school period and nearly 60 min of total daily MVPA at baseline questioning the clinical relevance of an intervention within this population.

Long-term effect

There is a lack of data regarding the long-term impact of an intervention on MVPA with only two pilot studies measuring MVPA 3–4 months after the intervention had ceased.26 ,27 Jago et al reported 8.7 min more MVPA per weekday (95% CI 5.5 to 11.9) in the intervention group 3 months after the intervention had ceased compared with the control group. Interestingly, they did not find a difference between the same groups in the last couple weeks of the programme.26 This conflicting effect may be explained due to the nature of the study as a pilot feasibility trial, not powered to detect group differences.

The longest follow-up time period of studies reporting favourable intervention effects on MVPA was 2 weeks after the intervention ceased.29 In this study, the beneficial effect of the intervention at mid-point was lost 2 weeks postintervention.

Weintraub et al34 also measured MVPA at two time points (mid-intervention and end of intervention). They found a reduction in intervention efficacy as time progressed from 3 to 6 months. This may be associated with a decline in mean attendance at the intervention soccer group from 53% in the first 3 months to 35% for the second 3 months. Considering there was only a total of nine students, this implies that some sessions in the latter part of the programme involved very few participants only.

Limitations

Several after-school school-based PA intervention studies did not meet the inclusion criteria for the review as they did not specifically measure MVPA levels but instead used other measures of PA. Despite included studies all measuring MVPA levels, the lack of consensus in reporting units of MVPA made direct comparison of studies difficult. Some studies only reported MVPA achieved during the intervention session alone.28 ,40 As such, it is difficult to then comment on the overall effect of the intervention on a child's PA levels as the intervention may be replacing a more active or less active time period. The meta-analysis was limited to a small number of studies which measured the adjusted mean difference in minutes per day of MVPA in the control versus the intervention group. This limitation highlights the importance of consistency in reporting measures of MVPA to allow for future meaningful comparisons on the efficacy of interventions to be made and progress the literature forward.

It is also important to highlight that there was variance in the accelerometer cut-point used for MVPA. This may lead to differing interpretations of an interventions' effectiveness. A previous study evaluating the accuracy of the various accelerometer cut-points recommended that Evensons cut-points should be used.48 There seems to be little conclusive evidence regarding the number of days an accelerometer should be worn in order to accurately calculate daily MVPA levels, with included study protocols ranging from 3 to 7 days. Additionally, the definition of non-wear time and criteria for inclusion in analysis varied across studies.

In this review, ‘postintervention’ MVPA was defined as an outcome measure of MVPA taken >4 weeks after the intervention had ceased. This outcome measure is less relevant for studies which aim to provide and maintain PA through a structured ongoing after-school programme than for those studies aiming to promote PA seeking behaviour change which persists after an intervention has ceased. There is a need to identify programmes that children will attend and which can increase MVPA in a sustainable way. This could either be via improving current after-school provision or where no provision exists adding new programmes that are shown to be effective.

Critical appraisal of studies using the ‘Quality Assessment Tool for Quantitative Studies’ revealed certain aspects of study design which were poorly conducted by many of the studies. The majority of studies did not assess the fidelity of intervention implementation which may mean that the intervention was not delivered in the format it was designed or in accordance with the theory of behaviour change. Studies also often failed to report or demonstrate good levels of attendance to their intervention. This may explain the small or negligible effect of most interventions on MVPA but also the lack of difference in effectiveness of those interventions based on a theory of behaviour change compared with those with no theoretical basis.

In several studies,29 ,34 the control group was provided with a programme which may have been more sedentary than the activities which they would normally have engaged in during the after-school time period. This may result in the effectiveness of the intervention at increasing MVPA levels to be overestimated.

The only meta-analyses providing some evidence of a difference (p<0.05) in MVPA examined the adjusted mean difference in min/day of MVPA in the intervention versus the control group from baseline to initial follow-up point. However, the I2 value for accelerometer-based studies indicated moderate statistical heterogeneity (46.4%) and when self-report studies were also included, the I2 value was 65.5% signifying substantial statistical heterogeneity. The heterogeneity observed in the meta-analyses is likely due to the variation in methodological approaches between the different studies.

Future directions

Given that the desired clinically relevant outcome end point is an increase in daily MVPA levels, it would seem sensible to report change in MVPA in terms of min/day. In order to fully understand the potential benefits of translation of this research into clinical practice, in terms of children meeting national and WHO recommendations for MVPA, another end point that future studies may wish to consider reporting, would be the percentage of participants achieving 60 min of MVPA per day at baseline and follow-up.

This review has focused on school-based interventions, but there is clearly a role for interventions within other settings. Further reviews exploring the influence of the intervention setting on efficacy, sustainability and cost-effectiveness are needed. Specifically, interventions targeted at enhancing existing programmes may require alternative theories of behaviour change that focus on increasing capacity among staff as opposed to individual behaviour change.

Subgroup analyses needs to be interpreted with caution as studies may conduct these analyses after no intervention effect was found, and therefore the study may not be statistically powered to determine the effects on subgroups. Future studies need to be clear in their intervention design of the statistical analyses they intend to perform and ensure that they are adequately powered to answer the research question posed.

Contextualisation

A systematic review from 2009 by Beets et al13 concluded that after-school programmes can be effective at improving PA levels with a meta-analysis indicating an effect size of 0.44 (95% CI 0.28 to 0.60). A separate review by Pate and O'Neill14 reported mixed findings with regard to the effectiveness of after-school interventions at increasing PA. Atkin et al12 found only three of the nine studies within their review to be effective. There is some evidence for PA interventions49 not confined to the after-school time period, though these only had a small effect on overall PA.50 No other reviews have commented on subgroup analysis of overweight/obese children or theories of behaviour change.

Conclusion

This meta-analysis and systematic review has found that after-school PA interventions to date have had mixed effectiveness on increasing MVPA levels. The meta-analyses provided some evidence supporting a difference (p<0.05) in MVPA when mid-intervention data from two studies were used instead of end-intervention data. However, due to only two studies included in the meta-analyses measuring mid-intervention data, this finding warrants further investigation.

The qualitative synthesis has identified that overweight/obese children may show significant changes in MVPA levels compared with controls even when there is no significant change in MVPA in the intervention group as a whole. Owing to the lack of subgroup analyses in many of the included studies, this finding needs to be interpreted with considerable caution and further exploration of subgroups with adequately powered studies is warranted. Similarly, the few studies indicating that boys may benefit to a greater extent than girls from an intervention also need to be investigated more robustly and extensively. It is also important to understand the reasons for this potential gender difference in order to develop strategies to adapt content to increase impact on girls and non-overweight children.

The secondary aim of the review was to report on intervention design based on theories of behaviour change. In this review, the presence or absence of a stated theory of behaviour change underlying an after-school intervention had no convincing effect on the effectiveness of the intervention. This may be due to the large number of factors contributing to whether an intervention yields a change in PA.

What are the findings?

  • There is mixed evidence supporting the effectiveness of after-school physical activity interventions at increasing MVPA but current evidence is of variable study design making comparisons difficult.

  • A small number of previous programmes have had more effect on overweight/obese children and boys.

  • There is limited evidence supporting the use of any single theory of behavioural change as the basis for an intervention.

How might it impact on clinical practice in the future?

  • After-school programmes may provide an opportunity for children to increase their MVPA levels and help tackle the fourth leading risk factor (physical inactivity) for global mortality as identified by the WHO.8

References

View Abstract

Footnotes

  • Contributors The systematic review was conducted by RM and RJ. The first draft of the paper was written by RM with sections written by RJ. Both authors critically reviewed and edited the final manuscript.

  • Funding RM is an Academic Clinical Fellow in Primary Care, funded by the NIHR.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles