Alcohol Reduction and the TTM: A Secondary SEM Analysis

Background

Alcohol exposed pregnancies (AEP) are the leading preventable cause of birth defects in the United States (CDC, 2009). Project CHOICES, a multi-site two group randomized controlled trial, was conducted to test the effect of an intervention to prevent AEP in at-risk women (Floyd et al., 2007). Participants assigned to the intervention condition received a four session motivational intervention aimed at decreasing alcohol use and one a session consultation designed to increase effective contraception. In light of the dual focus of the project, successful participants could either reduce alcohol use or increase effective contraception use (Floyd et al., 2007; Velasquez et al., 2009). Outcomes for Project CHOICES showed a twofold reduction in AEP risk for women in the intervention condition (Floyd et al., 2007).

By conducting a secondary analysis of data from the Project CHOICES efficacy trial, the present study examined Transtheoretical Model (TTM) constructs (DiClemente & Velasquez, 2002) to explicate the reduction of alcohol use by participants. Specifically, post intervention alcohol reduction and an increased use of behavioral processes of change were posited to follow participants' increased readiness to change and use of experiential processes of change. In addition, session attendance by participants was hypothesized to contribute to participants' use of behavioral processes of change and alcohol reduction. Alcohol severity was included within the model to describe participants' baseline level of alcohol misuse. Depression was included as a control variable to alcohol severity as depression is the most prevalent co-morbid psychiatric disorder for women who exceed daily drinking levels or who are heavy drinkers (NIAAA, 2006).

Methods

Structural equation modeling was used to test participants' alcohol reduction. Analyses were conducted using Amos 17.0. Baseline and post intervention data from Project CHOICES were used to test hypotheses. Indicator variables used within the model were all items from the Alcohol Use Disorders Identification Test (AUDIT), average drinks per drinking day calculated from the Timeline-Follow-Back, number of treatment sessions attended, and subscale items from the Processes of Change Questionnaire and Brief Symptom Inventory 53 (BSI-53). Latent variables used within the model were alcohol severity and depression, modeled from the AUDIT and BSI-53 items respectively.

Results

Data from 830 women were examined. Within the measurement model, Cronbach's Alpha for alcohol severity was .88 and for depression was .90. After co-varying error terms in the depression subscale, the full adjusted model showed a significant change in fit (Δχ2= 235.8, Δdf=4, <.000) and indicated excellent model fit (CFI=.97; RMSEA=.04; CMIN/DF=2.5; and χ2=464.2 (df=186), p<.000) according to SEM model fit conventions (Byrne, 2001; Hair, Black, Babin, & Anderson, 2009). Furthermore, all hypothesized relationships were significant and showed stability after adding the additional co-variances.

Conclusion/Implications

Results from this analysis indicate that health behavior change from interventions aimed at increasing motivation for alcohol reduction in AEP at-risk women may occur as posited by the TTM. Therefore, additional research is necessary in order to replicate structural models predicting similar relationships and outcomes to further test TTM behavior change hypotheses.