Assessing the Impact of De Novo Social Ties within Health Intervention Settings: New Questions for Health Behavior Intervention Research

Abstract

Recent developments in the study of health and social networks have focused on linkages between health outcomes and naturally-occurring social relations, such as friendship or kinship. Based on findings in this area, a new generation of health behavior intervention programs have been implemented that rely on the formation of new social relations among program participants. However, little is known about the qualities of these de novo social relations. We examined the social networks of 59 participants within a randomized controlled trial of an intervention designed to prevent excessive gestational weight gain. We employed exponential random graph modeling techniques to analyze supportive relationships formed between participants in the intervention arm, to detect unique effects of program participation on the likelihood of forming ties. Program participation had a positive effect on the likelihood of forming supportive social relations, however, in this particular timeframe we did not detect any additional effect of such relations on the health behaviors or outcomes of interest. Our findings raise two critical questions: do short-term group-level programs reliably lead to the formation of new social relations among participants; and do these relations have a unique effect on health outcomes relative to standard methods of health behavior intervention?

The "Madre Sana" Data Set

“…The intervention showed efficacy in preventing excessive weight gain during pregnancy in normal weight women (47.1 % usual care vs. 6.7 % intervention; absolute difference 40.4 %; p = .036) (Gesell, Katula, Strickland, & Vitolins, 2015). We found that the intervention activities had a significant and positive effect on the likelihood of tie formation, however, in this particular timeframe we did not detect any additional effect of such relations on gestational weight gain (Tesdahl & Gesell, In Press). As the sum of intervention attendances increased among dyads, the likelihood of their forming a tie increased proportionally with the number of sessions attended (β=0.09, p<.001).…”

Individual- and group-level network-building interventions to address social isolation and loneliness: A scoping review with implications for COVID19

“…Group-level interventions involved delivering social skills training [ 50 , 55 , 56 ], or providing opportunities for co-participation in activities of common interest [ 56 ]. Two studies involved peer support and training through pairing isolates with highly connected actors [ 57 , 58 ], while another study encouraged peer support and training through co-participation of individuals and their support workers [ 55 ].…”

“…Kasari et al (2016) [ 50 ] reported social network salience scores (see above). Tesdahl (2015) [ 56 ] developed an exponential random graph model to assess how personal, interpersonal, and structural features of the network were associated with the existence of ties between pairs of actors. The model showed that the total number of sessions that pairs of participants attended, as well as similarity in physical activity and pregnancy due date would significantly increase the chance that they are connected in the conversation network.…”

“…This is particularly useful for individuals who need assistance in building and maintaining social relations. In some studies, the skillset training have been enhanced through peer learning and reflection [ 55 , 56 ].…”

Social Support and Strain Across Close Relationships: A Twin Study

“…This has significant implications, particularly because social relationships have substantial effects on health, and having close ties is increasingly seen as a health behavior. If we are able to identify the factors that simultaneously and inversely influence social support and strain, and the factors that operate independently, then this could help the inform prevention and intervention strategies for behavior change and healthier aging (Lachman and Agrigoroaei 2010 ; Tesdahl and Gesell 2015 ).…”