(PS2-16) Moderating theoretical mediators of the FAMS 2.0 RCT by support person cohabitation: Application of PROCESS modeling

Abstract Text:

Introduction: Observational evidence has long underscored the importance of family/friend involvement for diabetes management, but meta-analyses of interventions seeking to improve family involvement demonstrate mixed results. More work is needed to understand for whom (moderation) and how (mediation) family interventions for diabetes management work. The FAMS 2.0 randomized controlled trial (RCT) tested a remotely delivered family-focused diabetes self-management support intervention versus enhanced usual care over 9 months. The intervention included monthly coaching sessions to set self-management goals and build interpersonal skills to elicit social support; daily text messages to provide education and encourage medication adherence and goal completion; and the option to co-enroll a support person in the trial. The FAMS 2.0 RCT demonstrated improved glycemic management (HbA1c) for adults with type 2 diabetes who elected to co-enroll a non-cohabitating support person in the trial, but overall main effects of the intervention on HbA1c were not significant. Post-intervention improvements in HbA1c overall were mediated by intervention targets including self-efficacy, dietary behavior, and family/friend involvement during the intervention period. We employed a moderated mediation approach to understand if the moderating effect of support person cohabitation on HbA1c worked through these identified mediators

Methods: These secondary analyses used data from the FAMS 2.0 RCT. Adults with type 2 diabetes (N=329) were randomized to intervention or enhanced usual care. They were 52% male, on average 57.0 ± 10.8 years old, 39% reported minoritized race or ethnicity, with mean HbA1c 8.6 ± 1.7%. Most participants (N=208; 63%) enrolled a cohabitating support person. We ran a single imputation to address data missing from the outcome (20%) and the mediators (15%). We employed the PROCESS macro in R with 5,000 bootstraps of all effects to test the moderating effect of cohabitation with support persons versus not on four previously identified mediators of HbA1c change: diabetes self-efficacy, family/friend involvement, dietary behaviors, and physical activity.

Results: There were significant indirect effects (mediation) on HbA1c both for adults with diabetes who did and did not have cohabitating support persons through diabetes self-efficacy and dietary behaviors. However, the index of moderated mediation, which directly tests differences in mediation by the cohabitation moderator, was not significant for any of the mediators tested.

Discussion: Self-efficacy and dietary behaviors mediated the effect of the intervention on HbA1c, regardless of cohabitation status. Hypothesized mediators did not explain the difference in the intervention’s effect on HbA1c by cohabitation status. Future work should investigate other mediators such as social network expansion or harmful family involvement, specifically, as possible drivers for this prior finding.