Meta-analysis: are 3 questions enough to detect unhealthy alcohol use?

Ann Intern Med. 2008 Dec 16;149(12):879-88. doi: 10.7326/0003-4819-149-12-200812160-00007.


Background: Both the 10-item Alcohol Use Disorders Identification Test (AUDIT) and its abbreviated 3-item version (Alcohol Use Disorders Identification Test-Consumption [AUDIT-C]) are considered to detect unhealthy alcohol use accurately.

Purpose: To examine whether the AUDIT-C is as accurate as the full AUDIT for detecting unhealthy alcohol use in adults.

Data sources: MEDLINE, EMBASE, CINAHL, Web of Science, PsycINFO, and BIOSIS Previews from 1998 to July 2008.

Study selection: Three independent reviewers selected studies that administered both the AUDIT and the AUDIT-C, applied a valid reference standard, avoided verification and incorporation bias, and reported relevant data. No language restrictions were applied.

Data extraction: Two reviewers extracted study characteristics and outcome data, which were cross-checked by a third reviewer. One reviewer assessed methodological quality with a standardized checklist.

Data synthesis: Fourteen studies were found. Most involved primary care patients in Europe and the United States. Sample sizes ranged between 112 and 13 438 patients, and sex and age distributions varied considerably. No statistically significant differences were found between the overall accuracy of the AUDIT and the AUDIT-C for detecting risky drinking, alcohol use disorders, or unhealthy alcohol use in primary care. Hierarchical summary receiver-operating characteristic curve analysis yielded pooled positive likelihood ratios of 6.62 for the AUDIT and 2.99 for the AUDIT-C, respectively, for detecting risky drinking; 4.03 and 3.82, respectively, for detecting any alcohol use disorder; and 4.82 and 3.91, respectively, for detecting risky drinking or any alcohol use disorder. Findings from a few studies on general population samples and inpatients suggested but did not prove that the AUDIT might be better than the AUDIT-C for identifying severe conditions, such as alcohol dependence.

Limitation: Studies used different reference standards and had heterogeneous findings.

Conclusion: Available evidence is inconclusive but suggests that the full AUDIT may be superior to the AUDIT-C for identifying unhealthy alcohol use in adults in some settings.

Publication types

  • Meta-Analysis
  • Research Support, Non-U.S. Gov't

MeSH terms

  • Age Factors
  • Alcohol-Related Disorders / diagnosis*
  • China
  • Europe
  • Female
  • Humans
  • Male
  • Primary Health Care / methods*
  • ROC Curve
  • Sample Size
  • Sex Factors
  • Surveys and Questionnaires / standards*
  • United States