Participant baseline characteristics.

In 2021, nearly 66 of every 1,000 adolescent girls ages 15–19 in the Dominican Republic gave birth. Adolescent mothers face health disparities including increased risk for rapid repeat pregnancy and lower breastfeeding rates. Mobile health (mHealth) is a growing approach for reaching adolescents. FAMA (Fortaleciendo la Autodeterminación de Madres Adolescentes) was a 12-week moderated digital education support group with adolescent mothers in the Dominican Republic, associated with improvements in health knowledge and contraceptive uptake. This study explores the FAMA intervention’s mechanism of action through a mixed-methods secondary data analysis using WhatsApp messages and post-intervention interviews to characterize experiences of social support and patterns of intervention engagement. We assessed associations between multiple measures of engagement and intervention outcomes. Linear or Poisson regression was used to evaluate association with health knowledge, and social support. or contraceptive uptake, respectively, selected based on type and distribution of each outcome variable. Models adjusted for key confounders. Findings indicate FAMA was largely used to exchange companionship and informational support. We found a significant positive association between engagement as measured by acknowledging intervention messages and improved health knowledge (adjusted coefficient: 2.84, CI: 0.83-4.84, p= 0.01). In contrast, we found a negative association between engagement as measured by social support exchange and improved health knowledge (adjusted coefficient: -5.78, CI: -10.42- -1.00, p= 0.02), suggesting that interactions focused on support may not reinforce informational content as directly as other forms of engagement. Our findings suggest that engagement with FAMA was associated with increases in knowledge and a close reading of message content is most beneficial for knowledge gain. This analysis enhances understanding of user engagement with group mHealth interventions and contributes new approaches to measure engagement, accounting for different engagement styles participants may have. Future digital interventions may leverage our findings to design interventions that encourage beneficial engagement types.