The Influence of a School Social Network Intervention on Adolescent's Health Behaviors: A Gender-Specific Agent-Based Model.

INTRODUCTION

Adolescence is a crucial stage for health behavior development, which is associated with health in adulthood. School closures caused by the coronavirus disease 2019 (COVID-19) pandemic have exposed adolescents to an increased risk of obesity due to a lack of physical activity. Although social network interventions provide an effective approach for promoting health-related behavior, current practices neglect gender differences in adolescent behavioral patterns and emotional preferences. The aim of this study was to examine the effectiveness of centrality-based methods integrated with of gender contexts in a social network intervention to improve adolescent's health behavior.

METHODS

We developed an agent-based model (ABM) that supports the small-world characteristics of adolescent social networks. Health-related data for junior middle school students ( = 234, 48% girls) were collected in November 2018, 2019 and 2020 in Tianjin, China. We simulated multiple network-based interventions with different criteria for influential agents (i.e., betweenness centrality, closeness centrality, eigenvector centrality, and PageRank) and a random condition. The rules for generating peer influence and accelerating behavioral changes were based on the diffusion of innovations theory, with gender specifications.

RESULTS

After the school closures, there was a significant increase in the prevalence of overweight and obesity among adolescents, with a greater increase in girls than in boys (+8.85% vs. +1.65%, < 0.001). Simulations showed that centrality-based network interventions were more effective than the random condition (average 6.17% per tick vs. 5.22% per tick, < 0.05), with a higher efficiency in girls than boys (average 3.68% vs. 2.99% per tick, < 0.05). PageRank outperformed other centrality conditions at the population level (6.37% per tick, p < 0.05). In girls, betweenness centrality was the best method (3.85% per tick, < 0.05), while in boys, PageRank still had the greatest efficiency (3.21% per tick, < 0.05).

CONCLUSIONS

We found evidence for gender differences in the negative impact of COVID-19-related school closures and the potential for centrality-based social network interventions to affect adolescent health behavior. Therefore, we emphasize the importance of gender-specific targeting strategies to further promote health-related school programs in the post-pandemic era.