(PS1-B36) In-session Language During a Substance-free Activities Session Predicts Alcohol Outcomes

Objective: Client language is hypothesized to be a key mechanism of behavior change in motivational interviewing (MI); however, measurement is limited to a specified target behavior (e.g., reductions in alcohol use) and primarily has been examined in interventions whose content focused almost exclusively on the target behavior. Behavioral economic theory suggests behaviors that are competing reinforcers of alcohol may also serve a critical role in reducing alcohol risk. This preliminary analysis examined whether in-session language during the substance-free activities session (SFAS) was associated with response to a brief intervention in college students.

Method: College students (N= 67, 56.7% female, M_age = 18.7 (1.4), 70.1% White) reporting 2 or more past-month binge drinking episodes participated in a randomized controlled trial of MI-based brief interventions. For the current analyses, only participants assigned to the SFAS condition were included, wherein all participants first received a traditional MI alcohol session followed roughly one week later by a SFAS session. Trained coders rated client language from SFAS sessions using a novel coding system integrating MI and behavioral economics. At the utterance level, client language receives a behavior code (target behavior of alcohol [TB] vs. alternative behaviors such as academics, physical activity [ALT]) and a motivation code (Approach vs. Avoid). Rates of each variable were computed as percentages of total client utterances. Regression analyses were conducted to examine the predictive validity of language variables on alcohol use at 1-month follow-up after controlling for baseline use.

Results: TB Avoid was not significantly correlated with either ALT Approach or ALT Avoid; TB Approach was associated with ALT Approach only, r= -.24, p= .047. Entered as simultaneous predictors, TB Avoid, b= -.58, SE= .29, p= .045, and TB Approach, b= 1.36, SE= .41, p= .002, were significant predictors of typical weekly drinking, while neither ALT variable was significant. Similar findings were observed for separate models examining frequency of binge drinking and alcohol-related problems, with the exception that only TB Approach, b= .79, SE= .29, p= .045 was a significant predictor of alcohol-related problems.

Conclusion: These preliminary findings of in-session mechanisms of the SFAS suggest alcohol-related language predicts intervention response, with greater amounts of TB Approach language associated with increased weekly drinking, binge drinking, and alcohol problems and higher rates of TB Avoid language associated with less drinking one month later. The more reliable association of TB Approach with outcomes is consistent with findings that sustain talk during alcohol MI sessions is more dependably associated with intervention response. Furthermore, ALT language did not add predictive power in terms of subsequent alcohol use. Thus, the SFAS may focus on behaviors acting as alternative reinforcers for alcohol, but differential evocation of alcohol-related language (i.e., decreasing Approach and increasing Avoid) may still be important in such contexts in order to improve client outcomes.