The purpose of this study was to evaluate current primary prevention interventions implemented in the after school setting for child and adolescent obesity. Based on this review it is evident that the after school time frame is increasing in popularity for intervention and research. The experimental rigor of the studies reviewed in this article greatly varied as approximately one third (n = 7 studies) were RCT’s, a third were quasi-experimental studies (n = 9 studies) and a third were pilot studies (n = 9 studies). It is clear more RCT’s are needed in this area, since they are generally considered the gold standard for program evaluation. A greater number of RCT’s will also be useful for conducting more in-depth reviews in the future, such as a meta-analysis to yield a common effect size for measures such as BMI-percentile, and behaviors such as physical activity.
Obesity prevention programs were also incorporated into many extracurricular activities that attracted children to participate. For example, one intervention utilized Girl Scout troops, which is nationally known as an enrichment program for young girls. During the program troop leaders served as positive role models and merit badges were given to incentivize the young girls to adopt healthy behaviors [38
]. Sports that some children may not have experience with were also used to promote physical activity, including Pilates [23
], soccer (for inner city youth) [31
], and culturally tailored dance routines [37
]. Communications was utilized in an intervention to help children learned aspects of media campaigning, in which they developed refrigerator magnets, a web site, a commercial, and a rap song to promote healthy behaviors for among their family members [22
]. Other innovative programs included teaching various aspects of theater production, which culminated with a play performances at the school [32
], and teaching children agriculture through developing and maintaining a community gardening [40
]. From these examples it is clear that the opportunities in the after school environment are vast. Researchers should use this opportunity to incorporate obesity prevention strategies in fun and exciting activities that are available to them, and that also peak the interest of their children.
The age and/or school grade range of the children in the studies in this review were from kindergarten through middle school, however the average age range was from 9 to 10 years. This indicates that children were generally in the fourth or fifth grade. Targeting this age group is useful since dietary and physical activity behaviors start to develop in these years and interventions designed to influence and build healthy behaviors at this juncture have the potential for long-term impact. This might also be indicative of the age-range researchers and practitioners should expect to find in this setting. As children grow older parents may be more likely to allow their children to stay home unsupervised, and when they enter middle school (the sixth or seventh grade) after school programs are likely replaced by sports or academic teams. Therefore, this may be a limitation of this setting; while accessibility is high, the availability of older children including preteens and teens, is likely much lower. More research is needed to address this issue.
A little over half of the interventions in this review targeted both nutrition and physical activity behaviors (n
= 12), while four aimed to modify physical activity alone, and three aimed to modify nutrition behaviors alone. Among the intervention that targeted nutrition behaviors either alone or with physical activity, fruit and vegetable consumption and snacking were the two most common behaviors targeted. The pattern that a majority of interventions targeted both physical activity and nutrition behaviors is similar to that of school-based obesity prevention interventions [43
]. While multifaceted, comprehensive programs are beneficial and ultimately needed for obesity prevention there is however some value in testing single-component programs to better test their efficacy. Therefore, we recommend more studies are needed for testing both types of interventions: the effectiveness of multi-component interventions and the efficacy of single-component interventions. Results from efficacy trials should also ultimately inform researchers of efficacious practices that can be used in multi-component interventions.
Another finding was that a majority (n
= 13) of the interventions reviewed were based on some behavioral theory, a trend that is similar to school-based obesity prevention interventions [43
]. Theories are beneficial for promoting healthy behaviors for several reasons; for example they discern measurable intervention objectives, and provide guidance for intervention strategies. Social cognitive theory (SCT) was the most commonly used theory among the interventions in this review, which posits that human behavior can be explained by reciprocal determinism, or a continuous interaction between behavioral, personal and environmental factors [44
]. This was not surprising, given the popularity of this theory in obesity prevention research. In a meta-analysis spanning from 1985 to 2003 authors reviewed randomized controlled trials (RCT’s) designed to favorably impact nutrition and physical activity among children and interventions that were most successful were implicitly or explicitly based on SCT [45
]. When using theory it is particularly helpful to measure and document changes in behavioral constructs or antecedents of behavior the theory has reified. Among the thirteen studies based on some theory, four did not measure any antecedent of behavior change. For studies that did, self-efficacy was the most commonly measured antecedent. This again was not surprising, since self-efficacy is the principle construct of SCT. From this review it can be concluded that there is an apparent need in this area. More research is needed in the advancements of operationalizing theoretical constructs into programmatic activities, and research is needed in evaluating what programmatic activities are ultimately most beneficial for behavior change. For example self-efficacy has been found to be significantly associated with exercising daily for 30 minutes and consuming five servings of fruits and vegetables among fifth grade children [46
]. While future interventions should target self-efficacy for both behaviors, program activities may not be the same, given the inherent differences in the two behaviors. Along side this recommendation, the need to validate instruments measuring behavioral antecedents is greatly needed. Smith [47
] found that among all articles published from 2006−2007 in four of the top journals in Public Health Education (Health Education and Behavior, Health Education Journal, Health Education Research, and International Electronic Journal of Health Education), less than half (41.6%) reported any psychometric property when needed, and the most commonly reported coefficient was Cronbach’s alpha. For step-by-step guidance on the proper methodologies for validating surveys measuring theoretical constructs, please refer to Barry and colleagues [48
With regards to the duration of the interventions in this review, they greatly varied from 3 weeks to 3 years. Since there is no universally accepted criterion for what is considered a ‘brief’ or ‘long term’ intervention, it was difficult to fully describe this feature in this review. However, by using the criteria Cook-Cottone and collegues [10
] used in their meta-analysis of school-based obesity prevention interventions (programs ranging from 0 to 12 weeks were considered short, 13 to 27 weeks as low-moderate, 28 to 32 weeks as moderate, and those lasting more than 32 weeks long) it was found that a majority (10 interventions) could be considered short
, 5 were low-moderate
, and 5 were long
. From these findings it appears that greater efforts have been given to shorter interventions, which may have contributed to the low amount of impact variables found to be significantly mediated for the studies in this review. In the future longer interventions (greater than 12 weeks) should be developed and evaluated to contribute to the existing evidence.
presents various methodological issues for the studies in this review. The first issue is in regards to the impact measures. The most commonly reported measure was some type of weight status, body composition or other functional assessment (n
= 19 or 76% of studies) following behavioral measures (n
= 17 or 68% of studies) and the least common measure used were behavioral antecedents (n
= 13 or 52% of studies). Very few studies (n
= 4 or 16% of studies) included all three types of measures, and most studies used at least two types (n
= 16 or 64% of studies). There were five studies (or 20% of studies) that only included one type of measure. To evaluate physical activity and diet a variety of methods were used. Both behaviors can be measured using either subjective (or self-report) or objective (or independently measured) means. Physical activity measurements mainly relied on self-report, as four studies utilized brief surveys [18
], three utilized physical activity recalls [23
], and three [35
] used accelerometry. One study also used parents to recall the amount of physical activity their child(ren) participated in over a period of time [40
]. Diet was similar as six studies relied on self-report [21
], and three relied on parent recalls [29
]. Self-report methods did vary however, with some studies utilizing surveys and others using 24-hour recalls. Planning models such as the Precede-Proceed model call for a comprehensive evaluation of interventions, and often stress the importance of evaluating all three types of measures. By including all three, researchers can also better understanding whether or not program activities are robust enough to impact behavioral antecedents, whether the impact on the behavioral antecedents are sufficient for mediating behavior change, and finally whether behavioral changes are strong enough to impact other variables such as weight status or body composition. Future studies would benefit from including all three types of measures described in this review.
With regards to sample size, seven studies reported an a priori
sample size calculation, 5 of which were RCT’s, and two had a quasi-experimental design. As Eng [49
] reported, it is important for studies to have an adequate sample size, since it directly impacts the statistical power of the study. Studies with inadequate power run the risk of reporting false-negative findings, which are commonly known as a type II error. This is positive finding, that most researchers evaluating RCT’s are recruiting an adequate number of research participants. Most of the quasi-experimental studies did not have sample size calculations, however this could strengthen their results. Sample size calculations are not generally warranted for pilot studies, since their true purpose is to test the feasibility of the intervention, and gather information to justify future implementation. Future studies should continue reporting their a priori
sample size calculations, especially for RCT’s.
The next issue reviewed in this article deals with the utilization of some type of process evaluation. Monitoring the implementation of obesity prevention interventions, or any type of health promoting program, is extremely important. This is especially true when multiple facilitators implement interventions across multiple sites for the same study. By failing to monitor program activities, researchers run the risk of making what is known as a type III error, where weak or null results can be attributed to poorly executed or incorrectly implemented interventions [50
]. Most process evaluations focus on two dimensions; dose, or the amount of time research participants spend engaged in program activities, and fidelity, or to what extent an intervention was delivered according to the intended delivery [50
]. While few frameworks exist for process evaluations, Saunders and colleagues [51
] outline a useful six-step framework for developing and using six types of process evaluations for health promotion programs. The steps include: fidelity (whether the intervention was implemented as planned), dose delivered (assurance that program lessons were implemented in order and for the amount of time planned), dose received (whether the intervention was well received by the participants), reach (attendance), recruitment (an assessment of what tasks were implemented to approach and invite participants to be involved with the study), and context (aspects of the environment that could have influenced the implementation of an intervention or study variables or contamination the comparison group might have by being exposed to the experimental program). From the studies in this review, 19 (or 76% of studies) reported using at least one type of process evaluation. In a further evaluation of these studies, it was found that attendance (or reach)
was the most commonly used process evaluation method. More attention should be given to process evaluations in future studies, and researchers should consider using the Sauders model [51
], or other models such as the Process Evaluation Model (PEM) [52
], or the RE-AIM Framework, which stands for Reach, Effectiveness, Adoption, Implementation, and Maintenance [53
Another common limitation in the design of the studies in this review was that only three studies evaluated any measure past the time of post intervention. Follow-up evaluations are greatly needed with obesity prevention research, to show whether effects are sustained after a set amount of non-intervention time. This is especially true for measures of weight status, such as BMI-percentile or z
-scores; while weight status may not change in the short-term, there is a great deal of interest in showing longer-term weight maintenance of children participating in experimental interventions. Drawing upon Prochaska’s Transtheoretical Model, six months appears to be an appropriate amount of time to implement a follow-up, since the theory purports that individuals typically need at least this amount of time to maintain a behavior change [54
]. Nonetheless, while a six month follow-up would be beneficial, practically any follow-up assessment would be beneficial for evaluating a program’s ability to make long-lasting behavior change.
A final issue not appearing on is with regards to reporting the use of intra-class correlation (ICC) in data analysis, when appropriate. While RCT’s do appear to be the strongest design for evaluating obesity prevention programs, researchers can rarely assign children to intervention conditions and often must assign groups of children to conditions, such as children attending the same school or after school program. Stevens and colleagues [55
] explain that RCT’s carry the unique challenge of having correlations among study variables within these assigned groups. The magnitude of this association is known as the ICC. It is important to be aware that ICC can impact study outcomes and should be properly controlled for, however is not always properly used or recognized in the literature. In a review of 59 grouped RCT’s authors concluded that only 54% used “appropriate analyses” accounting for ICC, while 25% used a mixture of ‘appropriate and inappropriate analyses’, and 20% used ‘all inappropriate analyses’ not accounting for ICC [56
]. The magnitude of this correlation has the potential to impact study results, which could lead to misleading or erroneous conclusions. In the articles reviewed for this study five of the seven RCT’s mentioned using the ICC as part of their data analysis. As more rigorous studies are employed in this area, future researchers should be sure to take the ICC into account in data analysis.