Discussion

PA promotion in young people is currently a key public health strategy. Schools are often the preferred setting for these initiatives, not least because young people spend many hours at school, and it provides researchers with a captive population. Nonetheless, considering the variety of influences on young people's behaviours, interventions in other settings should also be considered. The aim of this article was to present an overview of the effectiveness of family- and community-based PA interventions using a mixed-methods review strategy. Previous reviews of the literature highlighted the paucity of evidence in this area, especially from community-based interventions, and the limited methodological quality of the studies. The updated review of the evidence shows important progress, with an additional 10 controlled studies published since August 2007, which were generally found to be of higher methodological quality than previous work. However there remains a lack of (good quality) evaluations of community-based interventions and identifying the components of family-based interventions associated with effectiveness proved challenging. This was also highlighted by a previous review studying how to engage parents to increase youth PA,46 where they concluded that no obvious pattern could be identified. However they suggested that a medium-to-low level engagement, such as parent training, family counselling or telephone-based interventions, as opposed to actually engaging parents in a family PA programme with their children, may be effective.

In the updated review, three of the six family-based studies reported significant positive effects, such as 2327 more steps per day38 or 3.9 h of additional PA per week.40 However comparisons of the interventions, target populations or evaluation methods did not identify common components across the effective and ineffective interventions. As in the previous reviews, most studies targeted children, which may be a reflection of the perception that families become less of an influence throughout adolescence. Neither of the two studies including adolescents, a community-based mentoring scheme42 and family-based educational intervention,39 showed a positive effect. Only one community-based intervention43 was shown to be effective, which was the only environmental intervention and the only intervention solely targeting PA. On the other hand, the evaluation of this intervention, where the school playground was made available during out-of-school hours, had a low methodological quality score, raising doubts about the validity of the conclusions.

The earlier reviews tentatively showed that family-based interventions may be more likely to be effective if they are set in the home and include self-monitoring (using pedometers) and goal setting for specific lifestyle changes (such as increasing baseline steps by 2000 steps per day). These strategies, however, were not replicated in more recent research. Instead most interventions focused on provision of information in group settings and activity sessions for the children. Although a few studies included elements of self-monitoring, these were only minor intervention components and unlikely to have made substantial contributions to a potential intervention effect. Self-monitoring with pedometers can be effective in increasing PA levels in both adults and young people, especially when combined with goal setting.47 48 However the few pedometer-based studies conducted in young people to date have tended to be set in schools and were of medium-to-low methodological quality.48 A recent review of behaviour change strategies in family-based childhood obesity prevention interventions also highlighted ‘prompting specific goal setting’, ‘prompting self-monitoring’, ‘setting graded tasks’ and ‘providing performance feedback’ as common components of effective interventions.49 Based on the previously identified potential of these strategies, it is important that their applicability in the family or community setting is further explored and tested, using high-quality research with longer-term follow-up.

Intervention implementation is known to affect outcomes,50 and eight different aspects of implementation can be identified: fidelity, dosage, quality, participant responsiveness, program differentiation, monitoring of control condition, programme reach and adaptation. Few of the studies included in this review reported on attempts to assess aspects of implementation beyond attendance rates. Where reported in the family-based interventions,37,–,40 attendance was generally found to be high, despite none of them being delivered in the home. The required attendance of both children and parents at locations outside the home has previously been suggested to affect compliance,10 but this does not seem to have been the case in the more recent studies. Intervention delivery in general was poorly described, and only one study45 reported on the quality of implementation of the intervention, as assessed by the intervention deliverers. One area of concern in this literature is the reach of the intervention programme. Studies often reported using wide-ranging recruitment methods, with the potential to reach a large population, such as media campaigns, brochures and flyers. However no study reported on the recruitment rates compared with those eligible or invited to participate or on representativeness of the study sample to the target study population, raising questions regarding the generalisibility of the results to wider populations and the potential for wider implementation.

Previous reviews have highlighted the potential of environmental interventions both inside and outside the school setting.8 9 Most of these interventions aim to create safe opportunities for active play or travel. The only environmental intervention included in our updated review also elicited significant positive effects.43 Two recent reviews also showed that after-school programmes can be effective in promoting children's PA.51 52 These programmes generally also include the provision of safe and accessible opportunities for PA, next to health education. A small but growing body of evidence therefore seems to indicate that increasing safe opportunities for children to engage in play-based activities both at school and out of school may be a worthwhile strategy to pursue. Nonetheless the reported results must be interpreted cautiously because most studies used small sample sizes and were of poor quality, affecting the internal validity of the studies and the reliability of the conclusions drawn. The evaluation of environmental interventions or other natural experiments potentially affecting PA or other health behaviours is known to be challenging and fraught with difficulties.53 54 It is however crucial for future decision making that these collective interventions are evaluated with appropriate study designs and robust measures of the outcome to better estimate their effectiveness and allow for comparisons with other approaches.53

Several studies reported positive effects on anthropometry outcomes. Those studies that targeted overweight or obese participants were more likely to report significant positive effects, irrespective of the reported effect on PA. This is likely to be predominantly explained by the greater potential to obtain an effect on body composition among these participants than in the general population. Nevertheless several other issues may have contributed to this finding. Body composition can be measured more precisely and with less error than PA.55 Indeed, for many studies, body composition was the main outcome measure, and the sample size will have been determined accordingly. In addition, the interventions targeted both dietary and PA behaviour, and two of the studies targeting overweight or obese participants reported significant positive outcomes for dietary behaviour, but not for PA39 41 (the remaining study did not report on a dietary outcome40). It may be that for overweight or obese populations, improvements in dietary behaviour are – at least initially – more important to obtain improvements in body composition, although no mediation analyses were presented for the studies included in this review. However it is likely that for weight maintenance, improvements in both PA and dietary habits are necessary. Moreover, it is important to stress that PA has important health benefits independent from weight.1 56

Although the methodological quality of the more recently conducted evaluations seems to have improved, several challenges still remain. Only three of the 10 studies in the updated review used an objective measure of PA, and only four reported on the validity of their PA measure. We found that the type of PA measure used was not related to intervention effectiveness. However increasing the precision of the measurement of PA and studying how changes occur across the week and in different domains will allow for a more thorough investigation of the effects of interventions. Only three studies reported a follow-up of 6 months or longer, with one reporting a significant positive intervention effect. The long-term effect of family- and community-based interventions therefore also remains unclear. Indeed previous literature seems to suggest that repeated intervention efforts may be necessary to obtain established changes in PA habits.57 Few studies investigating PA promotion in young people have investigated mediation of the intervention effect,58 and to our knowledge, none of the studies in the updated review reported on mediation effects. Studying mediation is an important strategy to develop an understanding of effective intervention elements and the mechanisms of change, and it is recommended that future evaluations include measures along the causal pathway to study this.

As with the primary studies, this review has its strengths and limitations. We conducted a systematic literature search, assessed final study inclusion and methodological quality by two reviewers independently, assessed methodological quality and systematically investigated potential common components of effective interventions. Limitations include the restriction of the literature search to three databases, only one reviewer undertaking the initial reviewing stages and the potential for publication bias. However we aimed to minimise the effect of the former two issues by selecting the three key databases based on previous review experience9 and by using broad initial review criteria. We based our quality assessment on the published evidence rather than contacting all the primary authors to allow us to both study the quality of the studies and highlight continuing issues with the reporting of trials.