obesity reviews
doi: 10.1111/obr.12047
Childhood Health Promotion
Effects of school-based interventions for direct delivery of physical activity on fitness and cardiometabolic markers in children and adolescents: a systematic review of randomized controlled trials C. Sun1,2, A. Pezic1, G. Tikellis1, A-L. Ponsonby1,2, M. Wake2,3, J. B. Carlin2,4, V. Cleland5 and T. Dwyer1
1
Environmental and Genetic Epidemiology
Research Group, Murdoch Childrens Research Institute (MCRI), Royal Children’s Hospital, Melbourne, Victoria, Australia; 2
Department of Paediatrics, The University of
Melbourne, Melbourne, Victoria, Australia; 3
Centre for Community Child Health, MCRI,
Royal Children’s Hospital, Melbourne, Victoria, Australia; 4Clinical Epidemiology and Biostatistics Unit, MCRI, Royal Children’s Hospital, Melbourne, Victoria, Australia; 5
Menzies Research Institute Tasmania,
University of Tasmania, Hobart, Tasmania, Australia
Received 19 December 2012; revised 2 April 2013; accepted 20 April 2013
Address for correspondence: Dr C Sun, Murdoch Childrens Research Institute, Royal
Summary To evaluate the effectiveness of school-based physical activity interventions on fitness, adiposity and cardiometabolic outcomes among schoolchildren. Medline, Embase, EBSCOhost CINAHL and ERIC databases were searched up to October 2012. Inclusion criteria: intervention delivered at school with controls having no intervention or usual physical education classes; participants aged 5–18 years; outcomes spanning some or all of the above. We assessed levels of evidence for identified trials based on methodological quality and sample size. Dose of the interventions (a total summary measure of intensity, frequency and duration) were considered. Eighteen randomized controlled trials (RCTs, total participants = 6,207) were included, of which six were large, higher quality trials with high dose of the intervention. The intervention was consistent in increasing fitness with large, higher quality studies and high dose of intervention providing strong evidence. Dose of school-based physical activity is an important determinant of trial efficiency. Some large, higher quality RCTs provided strong evidence for interventions to decrease skin-fold thickness, increase fitness and high-density lipoprotein cholesterol. Evidence for body mass index, body fat and waist circumference, blood pressure and triglycerides, low-density lipoprotein cholesterol and total cholesterol remain inconclusive and require additional higher quality studies with high dose of interventions to provide conclusive evidence.
Children’s Hospital, Flemington Road, Parkville, Melbourne, Vic. 3052, Australia. E-mail:
[email protected];
Keywords: Obesity, physical activity delivery, randomized controlled trials, school-based.
[email protected]
obesity reviews (2013) 14, 818–838
Introduction Childhood obesity is an escalating public health problem worldwide (1,2). Of particular concern is the persistence of obesity into adult years that is accompanied by long-term obesity-related cardiometabolic morbidity and mortality. The search for successful population-based strategies to prevent childhood obesity has become increasingly urgent. 818 14, 818–838, October 2013
Much effort has been devoted to early intervention targeting physical inactivity given the significant health benefits of physical activity and the recognition that its pattern tracks from childhood to adulthood (3–5). Schoolchildren spend a significant amount of their time in school settings. Although formal physical education classes are not the only opportunity for students to engage in physical activity, such classes offer a regular schedule and © 2013 The Authors obesity reviews © 2013 International Association for the Study of Obesity
obesity reviews
structured format both for engaging in physical activity and for establishing habitual activity early in life. Such a setting provides an ideal and easily accessible environment and has the potential to offer benefits to all children (6), particularly those with no or limited access to play areas outside school (7). Such interventions thus, have the potential to reduce population-wide obesity and its related chronic diseases. Yet, despite this appeal, school-based interventions in which children engage in structured physical activity sessions delivered within the school curriculum have not been widely implemented. For instance, they did not feature among the 13 different approaches for obesity prevention in children and adolescents in a recent large-scale assessment of cost-effectiveness and likely population health benefit (8). The majority of studies included in recent systematic reviews of school-based physical activity interventions focused primarily on promoting (rather than directly delivering) an increase in physical activity (duration of physical activity during school and leisure time) and fitness levels in children and adolescents (9,10). The term ‘delivering physical activity’ is defined as meaning that the exercise prescription was actually delivered in a school in a setting where ‘instruction under direct supervision’ was provided to the children in the study group. Generally, these programmes were compulsory (e.g. the interventions are part of the school curriculum). Overall, there is good evidence that school-based physical activity interventions built primarily around promotion of physical activity are effective in promoting an increase in physical activity levels and fitness but the evidence for a beneficial effect on obesity levels (e.g. body mass index [BMI]) and cardiometabolic outcomes (e.g. blood pressure) remains inconsistent (9). These inconsistent findings may be attributed to a high level of heterogeneity among studies with regard to (i) the extent to which the intervention actually delivered as opposed to promoted the physical activity programmes (9–11), (ii) the type of outcome used to assess the effect of the intervention and the methodology used to collect such measures and more importantly (iii) failure to document the intensity and ‘dose’ of the physical activity intervention programme. As a result, the intensity, frequency and duration of the actual physical activity undertaken by the children is frequently unclear and, in many trials, may have been insufficient to produce significant health changes during the varied duration of the interventions (11). Few trials, in fact, have, as their primary intervention, directly manipulated children’s physical activity level (intensity, frequency and duration). In a recent Swiss randomized controlled trial (RCT) that focused purely on delivering exercise to children attending primary school, the multi-component physical activity programme included structuring the three existing physical education lessons © 2013 The Authors obesity reviews © 2013 International Association for the Study of Obesity
School exercise interventions and adiposity C. Sun et al.
819
each week and adding two additional lessons a week, daily short activity breaks and physical activity homework. This trial increased total daily moderate-to-vigorous physical activity (objectively measured by accelerometry) and led to lower body fatness and greater aerobic fitness among the intervention compared to control groups of primary schoolchildren over the course of a year (12). An earlier trial conducted in 1978 showed similar outcomes although as accelerometry was not available at that time it could not demonstrate the important intermediary effects of increased physical activity (13,14). In this trial, the number of hours per week of physical activity (5.75 h week-1) was greater than in the Swiss trial, but the duration was shorter – one school term of 14 weeks, compared to a year in the Swiss trial. Therefore, to better understand the specific role of increasing school-delivered physical activity on children and adolescence’s health, we undertook a systematic review of RCTs. In this report, we examined whether such delivered interventions have an effect on the various measures of adiposity, fitness and cardiometabolic health among school-aged children and adolescence taking into account the potential role of the ‘dose’ (a quantitative summary measure derived by multiplying intensity of the average session by frequency by the session duration per week and multiplying this in turn by the total duration of the intervention) of the intervention on the outcomes.
Methods We conducted and reported this systematic review in accordance with the Preferred Items for Systematic Reviews and Meta-analyses statement (15).
Inclusion and exclusion criteria We restricted studies to RCTs (cluster or individual) conducted among schoolchildren in primary (elementary) and/or secondary schools (aged 5–18 years). There was no restriction on sex, ethnicity or duration of follow-up. We also included trials with selective groups (e.g. overweight or low socioeconomic status children or adolescents). We excluded studies where the intervention was not delivered in a school setting and studies with other intervention components (e.g. an education session). We also excluded studies which reported only stratified analysis, which we were not able to combine. There was no restriction on the variation in terms of the components of interventions such as type, intensity, duration, frequency and timing of delivery (during and/or after school hours) of physical activity. Control groups received either the ‘standard’ or no physical education programme.
14, 818–838, October 2013
obesity reviews
820 School exercise interventions and adiposity C. Sun et al.
1,360 records identified from five databases*
5 additional records identified from hand search of included studies, reviews and key journals
1,189 records after duplicates removed
1,189 records screened
68 full-text articles assessed for eligibility
18 studies included for qualitative synthesis
1,121 records excluded based on abstract
50 papers excluded with reasons: 10 not in a school setting 5 outcomes not of interest 21 multicomponent interventions 7 not RCT 5 studies provided stratified analyses- unable to combine results 2 multiple reports
Included outcomes were adiposity measures (BMI, skinfold thickness, body fat and/or lean mass), fitness measures and other cardiometabolic health outcomes including blood pressure and fasting lipids.
Search strategy Studies were identified from five electronic databases: Medline (Ovid), 1948 to October, week 2, 2012; Embase (Ovid), 1980 to October, week 2, 2012; Cochrane Central Register of Controlled Trials (up to October 2012); EBSCOhost CINAHL (1937 to October 2012); and ERIC (1910 to October 2012) using the keywords ‘physical fitness’, ‘physical exertion’, ‘physical activity’, ‘physical education’, ‘exercise therapy’, and ‘exercise’ combined with ‘students’, ‘schools’ and ‘randomized controlled trials’ (Fig. 1). There was no restriction on language. We also searched databases of ongoing trials: Current Controlled trials (http://www.controlled-trials.com – with links to other databases of ongoing trials). Additional articles were retrieved from hand searching reference lists of the included articles, relevant reviews and meta-analyses identified from the above five databases and other sources. The table of contents for journals that commonly publish articles in this area (e.g. Preventive Medicine, Journal of Pediatrics, Pediatrics, Archives of Pediatrics and Adolescent Medicine, European Journal of Pediatrics, Adolescent Medicine and Health Education Quarterly) were also had searched for the years 1980–2012.
14, 818–838, October 2013
Figure 1 Flow chart of the systematic review. *Databases were Medline, Embase, CINAHL, ERIC and Cochrane Clinical Trials.
Study selection process Two reviewers (CS and AP) independently screened titles and abstracts for relevance. Duplicates were identified by using Endnote software (Thomas Reuters). After removing irrelevant titles and abstracts, two reviewers (CS and AP) independently identified potential eligible abstracts and retrieved full-text articles for further assessment. Full papers were assessed for inclusion or exclusion by five reviewers (CS, AP, GT, ALP and TD) and disagreement was solved through discussion. For studies that have published multiple reports, we included the report with less follow-up time to minimize the potential that loss to follow-up could have on the results.
Data extraction Data extraction was conducted by two reviewers (CS and AP) and checked by a third reviewer (GT) using a standardized coding sheet (available from the authors on request). Data were extracted without blinding to authorship, the details of the physical activity programme for both the intervention and control groups (e.g. duration, intensity, frequency of exercise intervention), and the trial design. We extracted data for the mean, standard deviations (SD) and standard errors for the post-intervention measures, change-from-baseline and estimated mean difference at follow-up after adjusting for baseline outcome measures. © 2013 The Authors obesity reviews © 2013 International Association for the Study of Obesity
obesity reviews
Data analysis and synthesis Due to the heterogeneity of studies in terms of the study design, target population, intervention and outcome measures, a formal meta-analysis was not appropriate for studies included in this review. The primary focus of our data synthesis was descriptive and included detailed tabular summaries. To overcome the heterogeneous nature of the physical activity interventions, we used metabolic equivalents (METs) to estimate the intensity of different types of physical activities that were reported for each included trial (16). METs were based on values reported in the study or from a recent reference for METs values based on activities reported in the study (17). METs were classified as low (1.8 to 2.9, average 2.35 was used), moderate (3 to 5.9, average 4.45) or of vigorous intensity (7.5 was used) (17). The METs value reflects the different intensities of a specific physical activity intervention session, e.g. low intensity was applied to a warm-up and a cool-down session unless otherwise specified in the original trial. Duration of the intervention was summarized in weeks with 1 month being equal to 4 weeks. An academic year was assumed to be of 9 months duration for studies reporting duration as year(s) unless otherwise specified in the original trial. Subsequently, we generated two quantitative summary measures of physical activity for each intervention: (i) a weekly quantitative summary measure derived by multiplying the METs, session duration (minutes) in each week, by the frequency; (ii) a total quantitative summary measure derived by multiplying the weekly quantitative summary measure by the duration of the intervention (weeks). This latter measure represents the ‘dose’ of the intervention, which we considered high if the value was greater than the median value of 13,333 units derived from the studies examined. This is equivalent to about two 60-min sessions of moderate-to-vigorous physical exercise per week. The common element of any intervention shared with the control group was subtracted from the estimated summary measures. That is, overall dose = (intervention arm dose) minus (control arm dose); where dose = intensity by frequency by session duration per week by total weeks duration of the intervention.
School exercise interventions and adiposity C. Sun et al.
821
‘unclear’. We did not consider allocation concealment to be a critical element in the risk of bias assessment given the majority of studies were cluster RCTs randomized by school or class. A higher quality study was defined as a study with total score of 4 (out of a possible 6) or higher (19).
Grading body of evidence We used a rating system of levels of evidence, described by van Sluijs et al. to provide evidence for the effectiveness of the intervention (19,20). Size of the study was scored as ‘1’ if there were more than 200 participants (>100 in each group; large) and as ‘0’ if there were 200 participants or less (small; Table 1).
Strength of the evidence We followed a systematic approach to ascertain the level of evidence provided by each study based on a previously used rating system (Fig. 2) (19). We used seven levels of evidence based on methodological quality and sample size: strong positive, strong no effect, moderate positive, moderate no effect, limited positive, limited no effect and inconclusive. Conclusions were based on the consistency of results among studies with the highest level of quality. That is, if at least two-thirds (66.6%) of the relevant studies reported significant results in the same direction, then this was considered an overall consistent finding. We contacted the authors of studies that presented stratified results without a block randomization, but only the trial by Weeks et al. was able to provide combined results for all outcomes of interest (21), enabling us to use a two-sample combined mean t-test to estimate postintervention effect size between intervention and control. For the trial by Dwyer et al. (13), we used the two sample mean t-test to estimate the SDs assuming equal SDs in the intervention and control groups. Using the method described by Higgins and Deeks (20), we were able to generate combined means and SDs for four studies (Tables 4 and 5) (22–25). All analyses were performed using STATA 11 (StataCorp, College Station, TX, USA).
Results Risk of bias assessment We used the Cochrane Collaboration recommended tool (18) to assess the risk of bias (and therefore study quality) for each included trial. The tool consists of six domains that included assessment for randomization procedure, blinding of outcome assessors, level of compliance (>70%), loss to follow-up (100 in each group) or if a power calculation was provided to justify the sample size are considered ‘large’ and studies included 200 or less participants are considered ‘small’.
one RCT was conducted among overweight and obese adolescents (26).
Characteristics of included studies We identified a total of 18 RCTs (Table 1) involving 6,207 participants. Eleven trials used a cluster design, seven of which were randomized by schools (12,22,24,25,27–29) and four randomized by class (13,30–32). Four studies were conducted in the USA (23,28–30), one in Canada (27), eight in European countries (one each in Denmark (33), Romania (34), Germany (31), UK (35), Switzerland (12) France (25) and two in Spain, (24,32)), one in Egypt (36), one in China (26) and three in Australia (13,21,22). Duration of interventions varied considerably from 6 weeks to 3 years, demonstrating the need to construct a ‘dose’ measure. All evaluated effects of physical activity immediately post intervention. Studies are ranked in decreasing order based on ‘dose’ (defined as total quantitative summary measure derived by multiplying the weekly quantitative summary measure by the duration of the intervention in weeks) of the physical activity intervention. Detailed summaries of study-specific information on type, intensity, duration and ‘dose’ of exercise interventions among the included studies can be found in Table 2 (Ardoy et al. (32) was not included in the tables as there was no description of the intensity of control and one of the intervention groups in the trial). The control group in most trials participated in regular physical education lessons. The studies varied in age range (Table 1). However, no marked differences were quantitatively observed by age.
14, 818–838, October 2013
Risk of bias assessment The results of the detailed risk of bias assessment for the 18 included studies can be found in Table 3. All trials performed randomization sequence generation and reported similar baseline characteristics between the intervention and control groups, except that one study showed a trend towards lower baseline BMI in the control group after randomization (35). Overall, details regarding allocation concealment were not adequately described in most trials. Only two trials (12,34) performed allocation concealment and an additional three studies which reported concealment did not implement it (29,33,35). Although blinding of participants and study personnel delivering the interventions was not feasible due to the nature of the trials, three studies did report blinding of outcome assessors for BMI as a primary outcome (12,13,29). Compliance Generally, studies reported good compliance due to the structured setting of the intervention (detailed in Table 3). Three studies also selected a random subsample of participants to wear accelerometers, thus providing an objective measure of physical activity to assess compliance (12,24, 29). One study reported heart rate as an objective indicator of physical activity delivered (22). Loss of follow-up Most studies provided detailed data on the number of participants randomized but then lost to follow-up. The extent of loss to follow-up was low, ranging from 2.5% © 2013 The Authors obesity reviews © 2013 International Association for the Study of Obesity
obesity reviews
School exercise interventions and adiposity C. Sun et al.
825
Table 2 Characteristics of intervention for included studies Author, year
Type
Intensity/METs*
Session duration (mins)
Dwyer, 1979 (13)
Endurance exercise¶
Low/2.35 Moderate/4.45 Vigorous/7.5
12 13 50
Hansen, 1991 (33)§
Warm-up and cool-down Organize games, gymnastics, exercises
Low/2.35 Moderate/4.45
Hergenroeder, 1993 (30)
Aerobic exercise
Vandongen, 1995 (22)
Frequency/ week
Duration (weeks)
INT_total‡
Duration/ week (mins)
Total PA/ week†
5
375
2,305.25
14
32,273.5
10 40
3
150
604.5
32
19,344
Moderate/4.45 Vigorous/7.5
Varied, average 21 min
3
Varied, average 63 min
Varied, average 343.7
18
6,186
Running, relays, skipping, ‘health hustles’ (moving to music)
Vigorous/7.5
15
5
75
562.5
36
20,250
Sadowsky, 1999 (23)
Walking Aerobic activity (aerobic dance, walk/jog) Cool-down
Low/2.35 Moderate/4.45 Low/2.35
5 30 5
4
160
8
5,024
McKay, 2000 (27)
Jumping and other activities in PE lessons and classroom
Low/2.35 Moderate/4.45
20 10
3
90
337.5
32
10,800
McMurray, 2002 (28)
Warm-up and cool-down Aerobic activities
Low/2.35 Vigorous/7.5
10 20
3
90
520.5
8
4,164
Serbescu, 2006 (34)§
Warm-up: light running, strengthening, callisthenic exercises Cool-down Jumping, running, impact-loading and weight-bearing exercises exercise¶
Moderate/4.45
10
2
100
512.8
26
13,332.8
Low/2.35 Moderate/4.45 Vigorous/7.5
10 12 18
Weeks, 2008 (21)§
Jumping
Vigorous/7.5
10
2
20
150
32
4,800
Donnelly, 2009 (29)§
Running and jumping
Moderate/3.4
10
9
90
306
108
33,048
Martinez-Vizcaino, 2008 (24)
Non-competitive PA
Moderate/4.45
15 60 15
3
270
1,201.5
24
28,836
Duncan, 2009 (35)§
Warm-up and cool-down Medicine ball cheat pass, overhead throws Skipping, lateral shuffles, bunny hops
Low/2.35 Moderate/4.45 Vigorous/7.5
10 12 18
2
80
423.8
6
Walther, 2009 (31)
Not specifically reported Not specifically reported Endurance training
Low/2.35 Moderate/4.45 Vigorous/7.5
10 20 15
3
135
675
36
24,300
Kriemler, 2010 (12)§
Warm-up and cool-down Motor skill: impact loading Not specified Not specified Motor skills (jumping) Aerobic strength, motor skills, hopping, rope jumping
Low/2.35 Moderate/4.45 Moderate/4.45 Vigorous/7.5 Moderate/3.4 Moderate/4.45
10 15 10 10 4 10
2
220
914
36
32,904
628
2,542.8
20 5
El Ansari, 2010 (36)
Self-chosen outdoor and indoor activities
Moderate/4.45
60
3
180
801
12
9,612
Thivel, 2011 (25)§
Supervised physical exercise
Low/2.35 Moderate/4.45 Vigorous/7.5
10 35 15
2
120
583.5
24
14,004
Sun, 2011 (26)
Warm-up and cool-down Physical activities/exercise games (jogging, running, jumping rope, basketball)
Low/2.35 Moderate/4.45
20 40
4
240
900
10
9,000
For studies that mention activity as low, we used average of 1.8–2.9 METs as (1.8 + 2.9)/2 = 2.35 METs. For studies that mention activity as moderate, we used average of 4.45 (range 3–5.9) METs. For studies that mention activity as vigorous: we used 7.5 METs based on the specific activities described in individual studies. *Reference: Ridley et al.(16) †Converted METs multiply cumulative duration per week. ‡Summary physical activity per week multiply duration. §Shared common elements of control group and this part was not included in the estimation of summary intervention physical activity measure ¶Running, skipping, vigorous dancing.
to 19.0%, with three studies reporting no losses (26,34, 36). Only two studies used an intention-to-treat analysis approach without exclusion of randomized students who refused to participate or were absent for the baseline © 2013 The Authors obesity reviews © 2013 International Association for the Study of Obesity
measurements, however, the details of such analysis were not adequately described (21,33). Based on the 6-item score described above, we judged six studies to have a high overall risk of bias and to thus be of
14, 818–838, October 2013
obesity reviews
826 School exercise interventions and adiposity C. Sun et al.
Table 3 Quality assessment of included studies, based on risk of bias Author, year
Randomization sequence generation performed
Blinding of outcome (BMI or other) assessors
Compliance
Loss to follow-up (%)
Intention-to-treat analysis
Total score of 5 Dwyer, 1979 (13) Hansen, 1991 (33) Weeks, 2008 (21) Walther, 2009 (31) Kriemler, 2010 (12)
Yes Yes Yes Yes Yes
Yes No Unclear Yes Yes
Compulsory 100% 76% 80% 100% Guaranteed compliance 100%
