(DCP060) SOCIAL SUPPORT IN A VIRTUAL WORLD: NETWORK ANALYSIS AND DIRECTED CONTENT ANALYSIS OF A PEER NETWORK FOR YOUTH WITH TYPE 1 DIABETES

Background: Virtual peer communities in diabetes may be effective sources of social support, but their evaluation is largely limited to enumeration and thematic analyses of posts, comments, and reactions. In this evaluation, we aimed to quantify social network metrics such as the ‘centrality’ of individual members; ‘views’ as a measure of passive participation alongside more active communication; and types and subtypes of social support. We examined the Virtual Peer Network private Facebook page for young persons 14 to 24 years old with type 1 diabetes (VPN-T1D) in Canada, which we built in partnership with peer leaders.

METHODS AND RESULTS: We selected peer leaders and recruited regular members from prior study and clinic populations and through social media (regular members only). We offered leaders a one-day workshop and small monthly stipend. We quantified posts, comments, reactions, and views (June 21, 2017, to March 20, 2020). Using social network analysis methods, we computed each member’s centrality ([80% number of posts, comments, and reactions + 20% number of members with whom they connected through these]). We divided each centrality value by the highest centrality across members, to derive ‘relative centrality.’ We compared relative centrality between leaders and regular members (linear regression). Applying directed content analysis, we classified and counted communications as social support-related (informational, emotional, esteem, network, tangible assistance) or unrelated, adapting a rubric developed for HIV communities. VPN-T1D had ten leaders and gained 212 members over 33 months, with 26 exiting (11·7%) over this period. There were 5,109 posts/comments (6·8%), 6,233 reactions (8·3%), and 63,709 views (84·9%). Two hundred and three members (91%) connected at least once through a high to medium engagement interaction (post, comment, reaction). Relative centrality averaged 0·53 (SD 0·26) for peer leaders and 0·04 (SD 0·05) for regular members (0·49 difference, 95%CI: 0·44, 0·53). Fifty-five percent of posts/comments were social support-related, over half informational (e.g., insurance challenges, travel preparation) and one fifth esteem-related (e.g., relieving blame by acknowledging challenges).

Conclusion: Informational support and esteem-building were key messages shared amongst members. Even reading such exchanges was likely valued, given that most network activity was passive in the form of views, and rateof exit from the group was low. The 10-fold higher relative centrality of leaders captures their role as the anchor of the network, suggesting that training and stipends foster a critical core membership that generates ‘content.’