Many paper based screening instruments have been developed for identifying different levels of alcohol misuse. Individual studies reporting the screening performance of these instruments for identifying hazardous drinking or anyone with an alcohol use disorder have produced variable results. Research has also suggested that discrimination between a positive and negative screen may vary by demographic factors such as age, gender and ethnicity. The aims of our study were to estimate the accuracy of short alcohol screening instruments (< 13 items) for the identification of hazardous drinking or an alcohol use disorder in primary care, to compare the accuracy of one instrument with another, and determine if accuracy values vary according to sex, age and ethnicity.
We conducted comprehensive database searches and reference screening of included articles to identify relevant studies. We selected studies that administered any alcohol screening instrument with 13 items or fewer in a primary care population, used a valid reference standard defined as a structured interview using recognized criteria, avoided spectrum bias and reported relevant accuracy data or sufficient data to allow generation of summary statistics. Each study was assessed for the likelihood of bias using a standardized checklist (Quality Assessment of Studies of Diagnostic Accuracy included in Systematic Reviews – QUADAS). We conducted meta-analyses using hierarchical models using SAS. Results for summary measures of accuracy, threshold and gender effects will be presented for the identification of hazardous drinking and alcohol use disorders. Initial results suggest that AUDIT with a threshold score of five, identified a greater proportion of true positives and true negatives for hazardous drinking compared with an AUDIT score of eight. This was true for both men and women. The evidence also suggests that AUDIT-C with a threshold score of four may be better for identifying hazardous drinking than AUDIT threshold score of eight.