Abstract
Text messaging (SMS) interventions incorporating a combination of behavior change techniques can assist reductions in alcohol consumption among young adult hazardous drinkers, but mechanisms of action remain unknown. In this secondary analysis, we test the hypothesis that desire to get drunk (DD) recorded prior to drinking episodes would mediate SMS intervention effects on the likelihood of event-level heavy drinking (4+/5+ drinks for women/men). We recruited young adult hazardous drinkers to a trial where they were randomized to 1 of 5 SMS interventions: TRACK (self-monitoring of alcohol use), PLAN (feedback on drinking plans and DD), USE (postdrinking feedback on alcohol consumed), GOAL (goal prompts/support), and COMBO (i.e., 4 interventions combined). Up to 3 days per week for 14 weeks, when participants reported a plan to drink, they were asked to report DD on a scale from 0 (none) to 8 (completely) and next day asked to recall drink quantity. Multilevel structural equation models showed that DD mediated the treatment effect of GOAL on heavy drinking. This work illustrates the importance of goal support features in digital alcohol interventions and the utility of measuring desire to get drunk as a key mediator in alcohol studies.