The Alcohol Use Disorders Identification Test (AUDIT): Exploring the Factor Structure and Cutoff Thresholds in a Representative Post-Conflict Population in Northern Uganda

• Published in: Alcohol and alcoholism (Oxford) 2017-05, Vol.52 (3), p.318
• Main Authors: Blair, Alden Hooper, Pearce, Margo Ellen, Katamba, Achilles, Malamba, Samuel S, Muyinda, Herbert, Schechter, Martin T, Spittal, Patricia M
• Format: Article
• Language: English
• Subjects: Adolescent; Adult; Alcohol Drinking - epidemiology; Alcohol Drinking - psychology; Alcohol-Related Disorders - diagnosis; Alcohol-Related Disorders - epidemiology; Alcohol-Related Disorders - psychology; Armed Conflicts - psychology; Armed Conflicts - trends; Cohort Studies; Factor Analysis, Statistical; Female; Humans; Male; Middle Aged; Population Surveillance - methods; Prospective Studies; Self Report; Surveys and Questionnaires; Uganda - epidemiology; Young Adult
• ISSN: 0735-0414; 1464-3502
• DOI: 10.1093/alcalc/agw090

Despite increased use of the Alcohol Use Disorders Identification Test (AUDIT) in sub-Saharan Africa, few studies have assessed its underlying conceptual framework, and none have done so in post-conflict settings. Further, significant inconsistencies exist between definitions used for problematic consumption. Such is the case in Uganda, facing one of the highest per-capita alcohol consumption levels regionally, which is thought to be hindering rebuilding in the North after two decades of civil war. This study explores the impact of varying designation cutoff thresholds in the AUDIT as well as its conceptual factor structure in a representative sample of the population. In all, 1720 Cango Lyec Project participants completed socio-economic and mental health questionnaires, provided blood samples and took the AUDIT. Participant characteristics and consumption designations were compared at AUDIT summary score thresholds of ≥3, ≥5 and ≥8. Confirmatory factor analyses (CFA) explored one-, two- and three-factor level models overall and by sex with relative and absolute fit indicators. There were no significant differences in participant demographic characteristics between thresholds. At higher cutoffs, the test increased in specificity to identify those with hazardous drinking, disordered drinking and suffering from alcohol-related harms. All conceptual models indicated good fit, with three-factor models superior overall and within both sexes. In Northern Uganda, a three-factor AUDIT model best explores alcohol use in the population and is appropriate for use in both sexes. Lower cutoff thresholds are recommended to identify those with potentially disordered drinking to best plan effective interventions and treatments. A CFA of the AUDIT showed good fit for one-, two, and three-factor models overall and by sex in a representative sample in post-conflict Northern Uganda. A three-plus total AUDIT cutoff score is suggested to screen for hazardous drinking in this or similar populations.