The Alcohol Use Disorders Identification Test for Consumption (Audit-C) for Predicting Hazardous Drinking

Summary

A cross-sectional study 1 investigated whether the AUDIT score is useful for predicting hazardous drinking and whether the AUDIT score was more useful than pre-existing laboratory tests. 334 outpatients who consulted internal medicine department in Japan completed self-reported questionnaires and underwent a diagnostic interview, physical examination, and laboratory testing. 40 (23 %) male patients reported daily alcohol consumption ≥ 40 g, and 16 (10 %) female patients reported consumption ≥ 20 g. The optimal cutoff values of hazardous drinking were calculated using a 10-fold cross validation, resulting in an optimal AUDIT score cutoff of 8.2, with a sensitivity of 95.5 %, specificity of 87.0 %, false positive rate of 13.0 %, false negative rate of 4.5 %, and area under the receiver operating characteristic curve of 0.97. Multivariate analysis revealed that the most popular short version of the AUDIT consisting solely of its 3 consumption items (AUDIT-C) and patient sex were significantly associated with hazardous drinking. The aspartate transaminase (AST)/alanine transaminase (ALT) ratio and mean corpuscular volume (MCV) were weakly significant.

Another cross-sectional study 2 evaluated the accuracy of alcohol use disorders identification test (AUDIT) and an abbreviated version of this test, in the detection of hazardous drinking at a single Australian major trauma centre. 523 trauma admissions were identified and of these 146 (28%) were screened. The optimum cut off scores for AUDIT and AUDIT-C were 8 and 5 respectively corresponding to sensitivities of 88% and 91% and both tests had excellent overall accuracy for the detection of hazardous alcohol consumption. There was no significant difference between AUDIT-C and AUDIT performance (p=0.395) (AUDIT-C AUROC 0.96 95%CI 0.93, 0.99).

A cross-sectional validation study 3 compared screening questionnaires with standardized interviews in 392 male and 927 female adult outpatients at an academic family practice clinic in USA. The AUDIT-C, full AUDIT, self-reported risky drinking, AUDIT question #3, and an augmented CAGE questionnaire were compared with an interview primary reference standard of alcohol misuse, defined as a Diagnostic and Statistical Manual, 4th ed. alcohol use disorder and/or drinking above recommended limits in the past year. Based on interviews with 92% of eligible patients, 128 (33%) men and 177 (19%) women met the criteria for alcohol misuse. Areas under the receiver operating characteristic curves (AUROCs) for the AUDIT-C were 0.94 (0.91, 0.96) and 0.90 (0.87, 0.93) in men and women, respectively (p=0.04). Based on AUROC curves, the AUDIT-C performed as well as the full AUDIT and significantly better than self-reported risky drinking, AUDIT question #3, or the augmented CAGE questionnaire (p-values <0.001). The AUDIT-C screening thresholds that simultaneously maximized sensitivity and specificity were > or =4 in men (sensitivity 0.86, specificity 0.89) and > or =3 in women (sensitivity 0.73, specificity 0.91).

A meta-analysis 4 included 135 discrete validation studies. Summary estimates indicated that the screening instruments performed well : area under the curve (AUC) 0.91 (95% CI 0.88 to 0.93); sensitivity 0.98 (0.95 to 0.99); specificity 0.78 (0.74 to 0.82). Noting a paucity of validation evidence for existing assessment instruments, aggregated reliability estimates suggest a reliability of 0.81 (0.78 to 0.83) adjusted for 10 items. AUDIT or AUDIT-C (the first three questions of AUDIT) were recommended.

A cross-sectional data 5 of health surveys from 5 401 university students in the Netherlands were used. 20 % of students were hazardous and harmful drinkers. The area under the ROC (receiver operating characteristic) curve was 0.922 (95% CI 0.914 to 0.930). At an AUDIT-C cutoff score of ≥7, sensitivity and specificity were both >80%, while other cutoffs showed less balanced results. A cutoff of ≥8 performed better among males, but for other subgroups ≥7 was most suitable.