A1469
Title: Causal mediation analysis of engagement for randomized trials involving mobile health interventions
Authors: Andrew Spieker - Vanderbilt University Medical Center (United States) [presenting]
Abstract: Clinical studies of mobile health interventions have received attention in recent years. A key challenge that emerges in these studies is understanding the role of engagement in driving the effects of these interventions (e.g., response rate to text message-delivered interventions). Some of the key nuances associated with both instrumental variable and mediation-based analyses of engagement in clinical trials of mobile health interventions are discussed. In particular, key topics will include (1) the blessings and curses of strong access monotonicity, (2) the applicability of the exclusion restriction, and (3) the applicability of pure and total direct/indirect effects. These matters are discussed in the context of recent studies of text message-delivered interventions such as REACH, FAMS, and VERB.
