Testing the Efficacy of Interactive Alcohol Decision-Making Program

The National Survey on Drug Use and Health estimates that 19.3 million individuals ages 12 and over met criteria for an alcohol use disorder in the past year, fully 87.4% of whom neither received any treatment in the past 12 months, nor wanted it. Recognition of this tremendous gap has led to efforts, both nationally and internationally, to implement proactive screening and brief intervention. Brief interventions have proven efficacious in reducing unhealthy drinking with effect sizes in meta-analyses ranging from small to moderate. Brief interventions are also uniquely applicable to non-treatment-seeking populations, who may refuse extended treatment but accept a minimal, opportunistic intervention. However, the public health impact of brief interventions has been limited by substantial difficulty with implementation. Further, positive overall findings in meta-analyses obscure the results of multiple rigorous efficacy trials showing no brief intervention effect on alcohol use. Both of these issues-the implementation challenges and the inconsistency in outcomes-suggest that changes are needed before brief alcohol interventions can meet their full potential.

Technology offers exciting potential in both respects. First, computer-delivered brief interventions (CDBIs) can be presented inexpensively, with perfect fidelity, and without the need for training or provider time. Second, their replicability, flexibility, and modularity makes them the perfect platform for (a) isolating the active ingredients that are associated with positive outcomes; and thereby (b) continually optimizing CDBIs to achieve cumulative increases in efficacy.

The overall goal of this research is to develop a maximally effective and replicable CDBI for reducing heavy alcohol use. To accomplish this, the investigators will use the Multiphase Optimization Strategy (MOST), an efficient method for optimizing intervention content that uses factorial designs to evaluate main and interaction effects of specific intervention components. The selection of components will be guided by: (a) Common Factors Theory, which highlights the tremendous contribution of relational factors, such as empathy, alliance and positive regard to therapy outcomes, but which is of unknown relevance to CDBIs; and (b) the Media Equation Theory, which suggests that people automatically respond to computers in social ways, particularly when those computers replicate human characteristics. Guided by this literature, the investigators will examine outcomes of a CDBI in which empathic statements, the gender of the narrator, the presence/absence of a narrator backstory, and the use/non-use of motivational interviewing techniques are systematically manipulated using a factorial design.

Specific aims are as follows:

1. Develop 16 distinct, time-consistent, computer-delivered sessions representing all possible combinations of the following four factors (each of which will be binary; i.e. either present vs. absent or male vs. female): (1) narrator gender, (2) empathic statements and reflections, (3) narrator backstory, and (4) use of motivational techniques.
2. Evaluate main and interaction effects of the five factors by randomly assigning 352 heavy drinkers to the 16 sessions (each of the four factors being presented to half of the participants) and evaluating (1) subjective reactions to the CDBI at baseline and (2) alcohol use at 1-month follow-up.