Method: Data from individuals who received a brief alcohol intervention in a randomized trial in a Level-1 trauma center was used to conduct this analysis (Roudsari et al., 2010). The nine-item aggressive and illegal behaviors subscale from the Cognitive Appraisal of Risky Events measure (Fromme et al., 1997) was analyzed using latent profile analysis with Mplus 6 (Muthen & Muthen, 1998-2010). Items from this subscale capture risk-taking and violence behaviors commonly measured in brief alcohol intervention literature—such as drinking and driving (Schermer, et al, 2008) and fighting (Soderstrom et al, 2007). Latent profile analysis seeks to identify homogenous subpopulations amongst heterogeneous groups (Hagenaars & McCutcheon, 2002). Percent days heavy drinking and percent days abstinent were included in the model as distal outcomes to examine the relationship between the profiles and drinking across time.
Results: Model identification procedures (Lo, et al, 2001) identified a four profile solution (N=737; Lo-Mendell-Rubin Adjusted Likelihood Ratio Test=.01). Quality of classification for latent profiles was greater than .92, and entropy was .94. Profiles were categorized as: (1) high consequences perception profile (n=479, 64.9%), (2) moderate with high perception of consequences from damaging property/violence with a weapon profile (n=142, 19.3%), (3) moderate consequences perception profile (n=61, 8.3%), and (4) low consequences perception profile (n=55, 7.5%). Decreases in percent days heavy drinking and increases in percent days abstinent were observed for profiles across time. Most changes in drinking occurred within the first six-months following the intervention. At 12-months post-intervention, the largest decrease in drinking was for individuals in the moderate with high perception of consequences from damaging property/violence with a weapon profile who reduced percent days heavy drinking by 23%. The largest improvement at 12-months for drinking abstinence was for the moderate perception profile that increased percent days abstinent by 14%.
Implications: This study provides tentative evidence suggesting four latent profiles exist with respect to how injured individuals perceive the consequences of risk-taking and violence behaviors. The greatest changes in drinking were for those reporting relatively moderate levels of perceived consequences. Future research will examine additional outcome measures, covariate effects, and time points for when changes occur. Understanding which patients are changing and when changes happen will aid social workers and other behavioral health providers to better plan and deliver brief alcohol interventions to injured patients.
