Background: The aim of this study was to define optimal cut points for the Alcohol Use Disorders Identification Test (AUDIT) and its abbreviated versions (AUDIT-C, AUDIT-QF, and AUDIT-3), and to evaluate how effectively these questionnaires detect heavy drinking in the general population.
Methods: The study population consisted of a sub-sample of the National FINRISK Study. A stratified random sample of 3216 Finns, aged 25-64, was invited to a health check. Of these, 1851 (57.6%) completed the AUDIT and participated in person in the Timeline Followback (TLFB) interview regarding their alcohol consumption. The TLFB-based definition of heavy drinking was used as a primary gold standard (for males > or =16 standard drinks average in a week or > or =7 drinks at least once a month; for females, respectively, > or =10 and > or =5 drinks). Areas under receiving operating characteristics curves (AUROCs), sensitivities and specificities were used to compare the performance of the tests.
Results and conclusions: The AUDIT and its abbreviated versions are valid for detecting heavy drinking also in a general population sample. However, performance seems to vary between the different versions and accuracy of each test is achieved only by using tailored cut points according to gender. The AUDIT and AUDIT-C are effective for both males and females. The optimal cut points for males were found to be >/=7 or 8 for AUDIT and > or =6 for AUDIT-C. Among females the optimal cut points were found to be > or =5 for AUDIT and > or =4 for AUDIT-C. The study also indicates that AUDIT-QF among females and AUDIT-3 among males are relatively effective. The cut points for detecting all heavy drinkers (including binge drinkers without exceeding weekly thresholds) were lower than for detecting heavy drinkers excluding those who are only binge drinkers.