Psychometric properties of the Drug Use Disorders Identification Test (DUDIT) and prevalence of drug use among SA site-based construction workers

Construction workers in South Africa are an at-risk group for HIV/AIDS. Substance misuse, other than alcohol, is associated with poor decision-making, risky lifestyle behaviour, and non-use of condoms. Substance misuse is also associated with harmful and hazardous alcohol consumption and depression. The Drug Use Disorders Identification Test (DUDIT) is a widely used screening tool to detect drug abuse. The present study examines both the psychometric properties of the DUDIT and the prevalence of drug abuse among construction workers. A field study involving 496 male workers at 18 construction sites of seven construction companies in the Western Cape province of South Africa, was undertaken. Descriptive statistics, reliability, and exploratory and confirmatory factor analyses were used to determine the prevalence of substance use, and the dimensionality, reliability and construct validity of the DUDIT. Approximately 10% of participants reported some form of substance use. Four hundred and sixty-seven employees (94.2%) were determined not to have drug-related problems (DUDIT score < 6). Of the at-risk employees (score >5), 5.0% (n = 25) were classified as 'possibly having a drug-related problem' (score 6-24) and 0.8% (n = 4) as 'possibly highly dependent on drugs' (score 25 or more). Of the 47 employees who reported drug use, 29 (61.7%) were classified as having a medium to high risk in relation to drug use. The internal consistency of the DUDIT was very good. The DUDIT reported a one-factor factorial structure, and the model fit indices indicated good model fit to the data. All factor loadings were significant, but there was some evidence of collinearity. Concurrent validity was demonstrated. DUDIT items 1 and 10 require further investigation due to their poor performance compared to the other items. The contribution of these two items needs to be investigated using item response theory (IRT).