Psychometric Properties of the Alcohol Use Disorders Identification Test (AUDIT) and Prevalence of Alcohol use Among SA Site-based Construction Workers

Construction workers in SA are regarded a high-risk group in the context of HIV/AIDS. Excessive alcohol use is associated with risky lifestyles and lack of condom use, decreased uptake of HIV testing, and poor adherence to ARV treatment. Excessive alcohol consumption is also associated with depression and illicit drug use. Screening is widely employed in the detection of problematic alcohol consumption; the Alcohol Use Disorders Identification Test (AUDIT) being extensively used for this purpose. This study examines both the psychometric properties of the AUDIT (one-, two-, and three-factor models) and the prevalence of alcohol use among construction workers. A field-administered survey was used to gather data from 496 male workers drawn from 18 construction sites of 7 construction firms. Descriptive statistics, internal consistency, and confirmatory factor analyses were used to analyze the prevalence of alcohol use, as well as the dimensionality, reliability, and construct validity of the AUDIT. Nearly 50% of participants reported never consuming alcohol. Including abstainers, three quarters of participants were classed as low risk (score <8). The at-risk workers were categorized as follows: 17.3% at medium risk (score 8-15); 3.6% at high risk (score 16-19); and 3.8% at very high risk (score 20+). Notably, of the 250 workers who reported using alcohol, 14.8% may be categorized as being at high-to-very high risk. In essence, 24.8% of construction participants were classed as engaging in hazardous or harmful drinking. Internal consistency of the AUDIT was very good. A 1-factor measurement model was indicated, the output indices presenting satisfactory model fit to the data. All factor loadings were significant. Concurrent validity was demonstrated. Further work is indicated in relation to items 9 and 10 of the AUDIT, as these particular items do not perform as well as the remaining items. The contribution of these two items needs to be examined using item response theory (IRT).