A Confirmatory Factor Analysis of the Alcohol Use Disorders Identification Test in a National Sample of Caregivers Involved With Child Protective Services

Background/Purpose:  Within the child protective services (CPS) system, caregiver substance abuse is a prevalent concern which has been associated with maltreatment and poor child outcomes.  Developed by the World Health Organization to assess for harmful patterns of alcohol consumption, the Alcohol Use Disorders Identification Test (AUDIT) has been used in numerous populations.  Different factor structures that fit the data, ranging from one to three factors, have been found in different samples indicating that variations exist in how populations interpret or view alcohol consumption.  Two factor models include consumption and alcohol-related problem subscales.  Three factor models include consumption, dependence, and problem subscales.  However, no studies have examined the type or number of factors present in a sample of CPS-involved adults.  Considering the relationship between caregiver substance abuse and child maltreatment, it is important to determine if this measure works with this population and what the factor structure is to guide research and interventions.  Using confirmatory factor analysis (CFA), this study compares the one, two, and three factor structures of the AUDIT with a sample of CPS-involved adults.

Methods:  This analysis utilized data from the National Survey of Child and Adolescent Well-Being II (NSCAW II).  In this national sample of CPS-involved caregivers, the AUDIT was administered to caregivers whose children remained in the home following the initial maltreatment report.  Consistent with best practices for model development, a randomly draw half sample was used for the CFA model resulting in 2004 cases in which the AUDIT was administered.  One, two, and three factor CFA models were conducted.  The robust Weighted Least Squares (WLSMV) estimator was used due to the categorical nature of the response items.  All analyses were conducted in Mplus 7.0.  Model fit was assessed using fit indices, factor loadings, and item-level R².

Results:  The two and three factor models provided a better fit to the population than a single factor model.  For the single factor model, fits statistics were acceptable but not strong for the RMSEA (RMSEA=.048, 90% CI: 0.042–0.055; CFI=0.987; TLI=0.984).  Both the two factor (RMSEA=.028, 90% CI: 0.020–0.035; CFI=0.996; TLI=0.995) and three factor models provided a better fit for CPS-involved caregivers (RMSEA=.028, 90% CI: 0.021–0.036; CFI=0.996; TLI=0.994).  All standardized estimates loaded highly and significantly (p<.0001) for both the two and three factor models. Item-level R² were highest in the three factor model for all 10 items.

Conclusions/Implications:  This first CFA of the AUDIT in a sample of CPS-involved caregivers indicates the measure is stable within this population and that latent constructs are present within the AUDIT.  Although empirical support is available for both the two and three factor models, a three factor model is recommended by the authors because the domain of problems associated with drinking are central to the issues with this population.  These results indicate the AUDIT is a valid measure, in research and practice, for assessing alcohol use in the CPS population that captures key components of their use including consumption, dependence and problems due to alcohol use.