AUDIT-C Alcohol Screening & Unit Calculator
Score the AUDIT-C with the WHO manual’s own wording, then see the result judged against all three published threshold sets — US, Korean population and Korean emergency-department — instead of one. The converter alongside shows what a drink actually weighs under the UK, US, Korean and WHO definitions, which differ by up to 75 percent.
| Threshold set, population and reference standard | Cut-off applied | Result | Sens / Spec |
|---|
| Definition | Pure alcohol | Your week in these units | What the source actually says |
|---|
| Serving | Per week | Pure alcohol |
|---|
| Authority, document and date | Applies to this respondent | In pure alcohol | Your intake |
|---|
The instrument itself is not this site’s invention and neither is any number on the page. The three questions, their response options and their point values are transcribed verbatim from Box 4 of AUDIT: The Alcohol Use Disorders Identification Test — Guidelines for Use in Primary Care, second edition, document WHO/MSD/MSB/01.6a, published by the World Health Organization in 2001. The AUDIT-C is simply questions 1 to 3 of that instrument, and it was first derived and evaluated by Bush and colleagues in 1998 in 243 male Veterans Affairs general medical clinic patients — female patients were excluded from that derivation sample, which is worth knowing before applying a sex-specific cut-off drawn from it.
Three separate validation studies supply the cut-offs, and they do not agree. Bradley 2007 studied 1,319 adult outpatients at a United States academic family practice and validated against a standardised diagnostic interview, reporting thresholds of 4 or more in men (sensitivity 0.86, specificity 0.89) and 3 or more in women (sensitivity 0.73, specificity 0.91). Lee JH 2018 used the Korean National Health and Nutrition Examination Survey and validated against the full 10-item AUDIT score, arriving at 6 for men, 5 for women and 3 for adults over 65. Lee KW 2018 studied 640 injured patients in a Korean emergency department and arrived at 5, 4 and 4. The tool shows all three with their populations and reference standards and never collapses them into one verdict.
The intake limits come from four different bodies: the UK Chief Medical Officers’ 2016 low-risk guidelines, the US Dietary Guidelines for Americans as published by NIAAA, NIAAA’s own current heavy-drinking figures, and the NIAAA risky-use limits as the USPSTF printed them in 2018. The last two do not match each other, and both are shown with their own dates rather than one being treated as silently correcting the other.
| Document | How it was read | As of |
|---|---|---|
| Babor TF, Higgins-Biddle JC, Saunders JB, Monteiro MG. AUDIT: The Alcohol Use Disorders Identification Test – Guidelines for Use in Primary Care, 2nd edition. WHO/MSD/MSB/01.6a. Geneva: World Health Organization; 2001. Official source | full PDF | 2001 |
| Bush K, Kivlahan DR, McDonell MB, Fihn SD, Bradley KA. The AUDIT alcohol consumption questions (AUDIT-C): an effective brief screening test for problem drinking. Arch Intern Med. 1998;158(16):1789-1795. doi:10.1001/archinte.158.16.1789 Official source | abstract | 1998-09-14 |
| Bradley KA, DeBenedetti AF, Volk RJ, Williams EC, Frank D, Kivlahan DR. AUDIT-C as a brief screen for alcohol misuse in primary care. Alcohol Clin Exp Res. 2007;31(7):1208-1217. doi:10.1111/j.1530-0277.2007.00403.x Official source | abstract | 2007-07 |
| Lee JH, Kong KA, Lee DH, Choi YH, Jung KY. Validation and proposal for cut-off values of an abbreviated version of the Alcohol Use Disorder Identification Test using the Korean National Health and Nutrition Examination Survey. Clin Exp Emerg Med. 2018;5(2):113-119. Official source | full text | 2018-06-29 |
| Lee KW, Choi YH, Lee JH. Cut-off points for screening at-risk drinking by AUDIT-C Korean version at emergency department. Turk J Emerg Med. 2018;18(2):57-61. doi:10.1016/j.tjem.2018.03.001 Official source | full PDF | 2018-03 |
| US Preventive Services Task Force. Screening and Behavioral Counseling Interventions to Reduce Unhealthy Alcohol Use in Adolescents and Adults: US Preventive Services Task Force Recommendation Statement. JAMA. 2018;320(18):1899-1909. doi:10.1001/jama.2018.16789 Official source | full text | 2018-11-13 |
| NHS. Calculating alcohol units. Official source | full page | checked 2026-08-01 |
| UK Chief Medical Officers’ Low Risk Drinking Guidelines. Department of Health, August 2016. Official source | full PDF, 11 pages | 2016-08 |
| National Institute on Alcohol Abuse and Alcoholism. What Is a Standard Drink? (drink size poster). Official source | full PDF | checked 2026-08-01 |
| National Institute on Alcohol Abuse and Alcoholism. Patient Education: What is a ‘standard’ drink? (sources given as RethinkingDrinking.niaaa.nih.gov and DietaryGuidelines.gov). Official source | full PDF | checked 2026-08-01 |
| National Institute on Alcohol Abuse and Alcoholism. Understanding Alcohol Drinking Patterns. Official source | full page | checked 2026-08-01 |
| Lee S, Kim JS, Jung JG, Oh MK, Chung TH, Kim J. Korean Alcohol Guidelines for Moderate Drinking Based on Facial Flushing. Korean J Fam Med. 2019;40(4):204-211. doi:10.4082/kjfm.19.0059 Official source | abstract and Table 3 | 2019-07 |
One number here is a derivation rather than a published constant. The ethanol density used throughout, 0.789 g/ml, is not quoted from a chemistry table: it is what NIAAA’s own two-part definition implies, since 14 grams divided by 0.6 US fluid ounces (17.744 ml) gives 0.789. The United Kingdom meanwhile calls a unit both 10 ml and 8 grams, which implies 0.80 g/ml. Those two conventions differ by 1.4 per cent, so each intake limit is converted to grams using its own authority’s gram figure and no comparison is made across a rounding the source did not make. Data file built 2026-08-01, re-verification due 2027-02-01.
The score runs 0 to 12 and is a plain sum. Each of the three items scores 0 to 4 and they are added with no weighting. Box 4 marks the “Never” response to Question 1 with a skip instruction, so a respondent who never drinks is not asked Questions 2 and 3 at all; the tool follows that instruction and fixes the total at zero rather than allowing an incoherent combination.
Positive means screen positive, not diagnosis. A positive AUDIT-C says that a fuller assessment is warranted. The USPSTF is explicit that when a patient screens positive on a brief instrument, clinicians should ensure follow-up with a more in-depth risk assessment to confirm unhealthy alcohol use and determine next steps, and that the sensible sequence is a sensitive brief instrument first and a more specific longer one, such as the full AUDIT, second.
Which cut-off applies changes the answer, and by more than one point. For a male respondent under 65, the applicable published thresholds run from 4 to 6. A score of 4 or 5 is therefore positive under one validated rule and negative under another, and neither is wrong. The reason is that the studies were validated against different reference standards: a diagnostic interview in the United States study, the full AUDIT score in both Korean studies. Those are different targets, and a threshold optimised for one cannot be read as optimal for the other. The practical consequence is that a recorded AUDIT-C result is only interpretable if the threshold set used is recorded alongside it.
There is an item-level reading as well as a total. The WHO manual states that in general a score of 1 or more on Question 2 or Question 3 indicates consumption at a hazardous level. Both of those items sit inside the AUDIT-C, so that rule is applied and shown separately from the total, and it can flag a respondent whose sum falls below every cut-off on the page.
The converter exists because the questions are ambiguous without it. Questions 2 and 3 are asked in “drinks”, and the manual itself tells the interviewer to code answers in terms of standard drinks and to explain what is meant using local examples. A standard drink is 8 grams of alcohol in the United Kingdom, 10 grams in the WHO definition, and 14 grams in both the United States and the Korean Academy of Family Medicine guideline — a spread of 75 per cent. An identical answer of “5 or 6” therefore describes 40 grams of alcohol to a British respondent and 70 grams to an American one, on the same score.
Consider a male outpatient aged 52 who drinks two to four times a month (Question 1 scores 2), typically three or four drinks on a drinking day (Question 2 scores 1), and has six or more drinks on an occasion less than monthly (Question 3 scores 1). The AUDIT-C total is 4 out of 12.
That single score produces three different published answers. Under Bradley 2007 the cut-off for men is 4, so the screen is positive. Under Lee KW 2018 the emergency-department cut-off for men is 5, so it is negative. Under Lee JH 2018 the Korean population cut-off for men is 6, so it is negative there too. The tool prints “positive by 1 of 3” rather than picking a winner. Separately, because Question 2 and Question 3 each scored 1, the WHO manual’s own item-level rule is triggered and shown, independently of any total.
Now consider the drink side. Suppose that respondent’s week is ten 12 fl oz cans of regular beer at 5 per cent. Twelve US fluid ounces is 354.9 ml, of which 5 per cent is 17.7 ml of ethanol, or 14.0 grams at 0.789 g/ml — one US standard drink, and 1.77 UK units by the NHS’s own formula of ABV multiplied by millilitres divided by 1,000. Ten of them is roughly 140 grams of pure alcohol a week. That is 17.7 UK units against the UK Chief Medical Officers’ ceiling of 14 a week, and 10 US drinks against NIAAA’s heavy-drinking threshold of 15 a week for men. The same week is over one authority’s line and under another’s, which is the entire reason the comparison is shown as a table of named authorities rather than as a single verdict.
A screening calculator earns trust by naming what it cannot do. These are the boundaries for this one, and in most of them the tool declines to print a figure rather than printing a plausible-looking one.
And a note on the beverage presets. Only servings an authority actually publishes are offered. The moment a volume or an ABV is edited, the entry is relabelled as custom, so no serving is ever attributed to a body that did not publish it. The Korean guideline’s serving equivalents are stated without an ABV in the source, so they are reported exactly as written rather than recomputed for a particular bottle.
This page is an educational reference implementation of a published screening instrument and of published intake limits, written for clinicians and for informed readers who want to see where each threshold comes from. It is decision support, not medical advice; it creates no doctor–patient relationship, it makes no diagnosis, and it must not be the sole basis for any clinical decision. Every threshold on the page carries the name, population and year of the study or guideline it came from precisely so that it can be checked against that source. A reader with concerns about their own drinking, or about a screening result recorded in their notes, should raise them with a clinician rather than acting on a calculator.