Bridging Disciplinary Boundaries (January 11 - 14, 2007)

Problem: Traditionally, outcome from alcoholism treatment has been considered a binary event. Either the person was drinking or the person was abstinent. More recently outcome data from clinical trials with longer follow-up periods suggest a different perspective is warranted. For many individuals, post-treatment outcome is not marked by an abrupt switch from drinking to sustained abstinence. Instead post-treatment behavior involves multiple fluctuations between drinking and abstinence prior to the emergence of either drinking or abstinence as the dominant behavior pattern. Recognition of drinking as a dynamic behavior allows the question of the relationship between the pre-treatment drinking dynamic and the post-treatment drinking dynamic. This study examines pre and post drinking as a dynamic process.

Methods: A secondary analysis of data collected in a large (n=1726) RCT, Project MATCH was conducted. Baseline drinking was obtained with the Form 90 interview for each day in the 90-days prior to randomization. Finite mixture modeling techniques (Nagin, 2005) were used to identify drinking trajectory groups at baseline. Monthly averages of percent days abstinent represented post-treatment drinking over 12 months. Person characteristics previously reported as prognostic of outcome were considered as covariates.

Results: The analysis grouped drinking trajectories over the 90-day baseline period in 5 categories (low use, declining use, increasing use, high use, and chronic use). The 5-group model performed quite well with average posterior probabilities of group membership ranging from .937 to .986, where 1.0 indicates certainty. The odds of correctly classifying into a particular group ranged from 89:1 to 495:1. Multinomial analysis resulted in the identification of three person characteristics (sociopathy, Involvement in Alcoholics Anonymous, and alcohol dependence) that statistically influenced group membership. A repeated measures ANOVA of percent days abstinent over 12 months post-treatment and including factors for treatment condition and the 5 trajectory groups, resulted in a significant between subjects main effects for trajectory group (F=30.27, df=4/769, p=.0001) and alcohol dependence (F=9.88, df=1/769, p=.002).

Discussion: The findings, if corroborated by additional independent research, have important implications for baseline assessment and treatment assignment. The results show a strong relationship between baseline trajectory group and post-treatment drinking, stronger than treatment condition in this study. The results underscore the dynamic character of drinking behavior. Collecting daily drinking over a 90-day baseline period provides social workers with information on the probable outcome from treatment. The issue is particularly salient for clients whose assessment and treatment assignment often involve interactions with a multi-disciplinary team whose members and agencies have differing perspectives on the nature of drinking among alcoholics. The binary state of abstinence/drinking is a static view and does not consider the dynamic nature of drinking behavior, as is evidenced by the emphasis placed on consideration of trait-like personal characteristics as prognostic of treatment outcome. These results show that assessing the trajectories of pre-treatment drinking behavior has prognostic value for post-treatment drinking.