Implementation processes of social network interventions for physical activity and sedentary behavior among children and adolescents: a scoping review | BMC Public Health

This scoping review aimed to evaluate and synthesize the implementation process of social network interventions to modify physical activity and sedentary behaviors in childhood and adolescence. Several findings were identified in the scoping review and are presented in the following sections. The analyzed SNI for PA and SB have been reported with high replicability (TIDieR), were conducted pragmatically, i.e., with very similar conditions to real settings (PRECIS-2), and the report of some RE-AIM domains support the generalizability of SNI.

Theoretical support for interventions

The most frequent theoretical framework reported in the interventions was the self-determination theory (SDT), used in four out of the seven analyzed studies along with other theories (27, 28, 33, 34). The SDT is based on intrinsic or autonomous motivation and psychological needs as relation, autonomy, and competence (38). Recently, a systematic review found that SDT-based interventions improved different health domains (39). In children and adolescents, the evidence suggests that SDT constructs help to explain physical activity behavior (40).

Diffusion of innovation theory (DOI) was used as a theoretical framework in other social network interventions (32,33,34). DOI theory is considered relevant for social network interventions (16) since it is helpful to explain how new ideas and practices spread within the social network (41, 42). Likewise, integrating the theory of social networks with DOI theory has been suggested to improve the implementation of interventions, achieving a comprehensive approach (43). In the present scoping review, two interventions applied a combination of SDT and DOI theory (33, 34). However, only the study of Sebire et al. (2018) positively affected PA and SB.

Other theories, such as social cognitive theory (SCT), theory of planned behavior (TPB), and self-persuasion theory (SPT), were used to inform the interventions. However, only one study reported positive effects on PA (35). However, SCT has been previously used in interventions for PA and obesity with no significant results, and low efficacy (44, 45). The theoretical support for SNI is required not only for identifying those constructs to intervene during the implementation process and how they interact with each other to influence the outcome of interest but also, more importantly, for helping to explain how patterns of the social system may be modified by fostering or intensifying some social interaction mechanisms to optimize the system behavior. This perspective implies the adoption of the systems lens for using theoretical frameworks for complex interventions (46).

Interpersonal relationships promoted

The social mechanism most promoted in the studies included in this review was social influence, analyzed in all the studies except for the study of Jago et al. (33). This finding is consistent with what has been pointed out in the literature about the role of social influence in the relationship between health behavior and social networks (47). The evidence shows that social influence is positively associated with health behaviors in school adolescents (9) and that individual PA levels may result from the influence of friends and peer (12, 14, 15). The other social mechanisms reported in the studies included social norms, social pressure, social support, and social modeling. The identification of these mechanisms is a key step to define the implementation of strategies (8). However, it is difficult to determine the actual social mechanisms responsible for behavioral changes due to social networks’ inherent, complex dynamics and their natural evolution (47).

Types of social network interventions

The identification of the social network intervention approach used is an important aspect to understand the implementation process, since each typology of SNI describes different strategies to be implemented (16). Six out of seven studies were based on individual approaches of SNI. Generally, researchers used friend’s nomination to identify the agents by a network parameter (centrality) and then trained them to implement strategies within the network for behavior change; two of these studies had positive effects (34, 35). This type of SNI is supported by the most substantial evidence of effectiveness in the adult population (8). The study that used a social approach based on the specific type of induction and alteration based on reciprocity positively affected the PA level (29). A review of SNI for health behaviors in adults identified that the social approach, especially induction, was the most frequent SNI typology used (8).

In the present review, two interventions used “closeness centrality”: influence agents closely connected to all other network members (27, 28). Four interventions used “indegree centrality”: influence agents who received the most nominations or were popular in the network (32,33,34,35). This strategy is supported by the fact that popularity in social networks has been associated with engagement in health behaviors and significant associations for predicting health behavior (9). However, the SNI type’s effectiveness for behavior change still requires more research attention.

It is important to note that previous evidence syntheses have also explored PA interventions based on peer-to-peer approaches (48, 49). Studies analyzed in these reviews applied diverse peer-leadership approaches, where leaders were trained to foster PA among peers. In these studies, the peer leader was identified using different criteria than those applied in SNI, i.e., applying teacher criteria or based on the student’s leadership skills. The difference with these reviews is that our scoping review focuses on those interventions in which sociometric data from networks is used to learn from the community, a particular characteristic of SNI (16). In other words, the difference lies in the methods used to identify who delivers the strategies. For this review, we strictly adhere to the definition of SNI, in which network information is required to inform the design of the intervention (16).

It is worth noting that four of the seven studies were identified using manual search and previous knowledge of the researchers. This may be explained by the fact that these studies do not include the term of social network intervention in the title or abstract. However, in the methods section, authors describe the procedures that fit with the operational definition of SNI (32,33,34,35).

External validity: replicability

In this scoping review, the replicability of interventions was analyzed by the TIDieR tool, and most of the items were appropriately reported. In general, all the SNI had good replicability scores. However, “Tailoring”, “Modification”, and “How well actual” were the items with the lowest frequency of reporting, crucial aspects for assessing intervention fidelity, adherence, and adverse events. These items are useful to complete the description of the implementation process and should be addressed in future studies. Similar findings about frequencies of complete reports have been documented in overviews of systematic reviews (50, 51). Even though the analyzed studies had acceptable replicability, more complete and accurate reporting could reduce research waste, improve evidence synthesis, and implementation in other contexts (52).

External validity: applicability

Applicability was assessed with Precis-2; in general, SNI tended to be pragmatic. There were no studies considered rather explanatory or very explanatory. Social network interventions have the particularity of using established networks, in this case, the classrooms. This allows many elements to be pragmatic. In this review, the most pragmatic items were “Eligibility”, “Recruitment”, and “Primary outcome”. At the same time, “Follow-up” tends to be equally pragmatic and explanatory, and “Organisation” and “Primary analysis” tend to be rather explanatory. These results differ from other reviews where “Primary analysis” tends to be more pragmatic (53, 54), and “Eligibility”, “Organisation”, and “Follow-up” tend to be more explanatory in primary health care (53, 54).

In this review, two studies considered pragmatic positively affected PA (34, 35). However, the most explanatory study also positively improved PA in adolescents in SNI (29). A systematic review concluded that pragmatic trials were ineffective in improving PA in children (55), and others concluded that more pragmatic studies were associated with smaller increases in PA (56). It is worth mentioning that in most cases, SNI has aspects that are inevitably considered explanatory in social networks. For instance, although the interventions were implemented in usual school settings, characteristics of the social network of students are not commonly analyzed to inform pedagogical interventions. Therefore, this exploration could be considered explanatory.

External validity: generalizability

Generalizability was assessed with RE-AIM. Overall, three RE-AIM domains (reach, adoption (staff), and implementation) support an acceptable level of the generalizability of findings. The report of the rest of the dimensions needs to be improved, particularly maintenance (organizational). The percentage mean for studies varied from 26.6% (35) to 61.2% (34). The most reported dimension was reach with 55%, adoption (staff) and implementation both with 57%, and the least reported was maintenance (organizational) with only 2%. Previous systematic reviews have also found that Reach is the dimension where more items are reported in PA school interventions among adolescents (57). In specific populations, such as indigenous youth, the result was different. A review concluded that the dimensions of reach and implementation were the most poorly reported in PA interventions (58). However, items could have been better described with more information (57), while the dimension of maintenance is the most poorly reported (59). In this regard, it has been pointed out that maintenance is one of the most challenging elements for physical activity interventions in schools since it requires the integration of multi-stakeholder perspectives from classroom, school, and policy levels (60).

External validity: integration

All the analyzed studies were conducted in schools, reinforcing the premise that the school setting offers a good opportunity to promote health and prevent disease (61). In the analyzed SNI studies, students enacted the strategies deployed in their social networks in real-world education settings. Social interactions are the core of SNI. However, students should sometimes receive training to improve their social skills and unfold social mechanisms to promote the intended behaviors. Since social learning skills are part of the learning objective in school settings, SNI fits the requirements for implementation within schools. Although the analyzed studies specify the intervention time, the exposure dose, and other relevant aspects for external validity, social interactions occur naturally during school hours and are not restricted to educational intentions. Therefore, it is a challenge to effectively register the quantity and quality of social interactions that may occur during school days that have yet to be planned within the study. In addition, there is evidence of the effectiveness of SNI in the short and long term (more or less than six months) in some health-related behaviors (8). All the studies analyzed in the present scoping review had a duration of no more than ten weeks of intervention, and three of them reported positive effects on PA in interventions between 6 and 10 weeks (29, 34, 35). The SNI for PA and SB tend to be conducted with high external validity in the domains of actor, action targets, temporality, dose, and theoretical justification. The domains of action and implementation outcomes need to be improved in the reports for SNI replicability, applicability, and generalizability.

It is worth indicating that when integrating the three dimensions of external validity according to the domains proposed to describe the implementation of strategies (20), there is a risk of omitting relevant aspects of each intervention. This, added to the need to attend to contextual conditions, as suggested from the systems perspective (18), implies that a reading of the context must be carried out to adapt and contextualize the implementation of the strategies.

The findings of this scoping review contribute to the initiative of whole systems approaches for physical activity promotion (62) because tools such as social network analysis are used to understand systems as a central issue within implementation science (17). Its application by public health and educational practitioners requires adopting a system approach (46), e.i., embracing the uncertainty and unpredictable nature of the relationship between actions and their consequences. From this perspective, SNI needs to be understood as events within the system (63). Since schools are complex social systems (64), SNI requires identifying particular social dynamics within the network and being attentive enough to recognize changes produced by the system during the intervention (46). For instance, the identification of central actors within the social network may be used to activate “leverage points” (65) that generates meaningful changes in the social systems of students (66). In this regard, integrating context- and practice-based evidence is encouraged. Regarding implications for research, the present review’s findings help identify aspects to be explored in future research. Among these aspects, examining social network mechanisms and studies about the effectiveness and implementation processes of the SNI typologies, for both PA and SB, are elements to study in greater depth. Also, future research should be focused on SNI as a complementary component along with other participatory strategies and system mapping methods to better align with the whole system approach.

This review presents some limitations. First, the researchers used tools to evaluate replicability, applicability, and generalizability. However, this could have been more accurate if the program implementers had participated in the task. Second, all the SNI analyzed were conducted in school settings. The present review did not identify other spheres of socialization, such as neighborhoods, sports teams, churches, and the like. A third limitation is that in this scoping review, and as part of the narrative description of findings about the implementation process, the direction of the effect of the interventions has been mentioned, but not extensively analyzed. This procedure is usually done in systematic reviews of the effectiveness of interventions, as previously conducted in the field of SNI (8). Among the strengths, integrating the three tools used as dimensions of external validity gave a novel approach to evaluate intervention studies from the D&I research perspective. Also, considering that diverse local circumstances define social network processes, structures, and dynamics, it is challenging to generalize these interventions. In this regard, the present scoping review highlights essential aspects of SNI. Endeavors to scale up SNI must pay special attention in the implementation process to identify these local circumstances responsible for the social dynamics unfolding in the networks.