The instrument consists of 3 questions
(1) number of days consuming alcohol over past year
(2) number of drinks on an average day of drinking
(3) number of times 6 or more drinks consumed on a single occasion
Responses range from never or none (0 points) to high (4 points).
where
• The time frame is during the past year rather than "ever".
• A drink was considered a standard drink: can or bottle of beer, bottle of wine cooler, glass of wine, one cocktail, shot of hard liquor.
total score =
= SUM(points for all 3 questions)
Interpretation:
• minimum score: 0
• maximum score: 12
• The higher the score, the greater the risk for alcohol misuse.
• A variety of cutoffs (from >= 3 to >= 8) can be used depending on the sensitivity and specificity for the different endpoints (heavy drinking, dependence, at risk drinking).
Performance:
• A cutoff of >= 4 had a sensitivity of 86% and specificity of 72% for heaving drinking and/or alcohol dependence.
• A cutoff of >= 5 had a sensitivity of 68% and specificity of 90% for heaving drinking and/or alcohol dependence. This had the same Youden index as a cutoff >= 4.
• The overall performance depends on what your criteria are for light, moderate and heavy alcohol use. Most people who use alcohol do not abuse it, but many people who drink should drink less.
• Asking the third question (consuming 6 or more drinks) alone was almost as effective as all 3 questions.
Limitations:
• There does not appear to be an adjustment for lower intake by women. For example, the third question might be 4 or 5 drinks for a woman. This appears to be because few women were respondents.
• As with any screening test, the prevalence of problem drinking in the population and the selected cut-off will affect how well the test will perform in a clinical setting.
• I would think that drinking 6 or more times per week should be weighted more than 4 or 5 times per week.
• I wonder if you could get better discrimination by making it a product of the 3 questions rather than a sum (range of scores from 0 to 64).
