Predictive validity of Alcohol Use Disorder Identification Test – Consumption (AUDIT‐C) for register‐based alcohol‐attributable events among general‐population men and women of different ages

Background and aims: Originally developed for clinical screening for hazardous alcohol use, the Alcohol Use Disorder Identification Test - Consumption (AUDIT-C) is also widely used in epidemiological research and monitoring. However, its cutoff values may not suit all purposes, and evidence on its predictive validity is limited. We examined how AUDIT-C scores relate to subsequent register-based alcohol-attributable events across age and sex and identified optimal thresholds for predicting these outcomes.

Design: Five general population surveys conducted in 2011-2017, linked to nationwide registers with follow-up until the end of 2023 (total follow-up 879 964 person-years).

Setting: Nationwide, Finland.

Participants: Adults aged ≥20 years (n = 103 567).

Measurements: AUDIT-C to assess exposure.

Outcome: incident alcohol-attributable events in any of the registers capturing care, deaths and prescription medicines (n = 1444).

Findings: The hazard of register-based alcohol-attributable events increased approximately exponentially with increasing AUDIT-C score. It rose more strongly among women than men [hazard ratio (HR) = 1.61, 95% confidence interval (CI) = 1.55-1.67 for women; HR = 1.45, 95% CI = 1.40-1.49 for men; P < 0.001 for the sex × AUDIT-C interaction]. The risk of alcohol-attributable events was statistically significantly raised at 2 points among women and 3 points among men with 1 point as a reference level. The suggested optimal cutoff values were 2 points for women aged ≥80, 3 points for women aged 65-79 and men aged ≥80, 5 points for women aged 20-64 and men aged 65-79 and 6 points for men aged 20-64.

Conclusions: Using Alcohol Use Disorder Identification Test - Consumption (AUDIT-C) scores, optimal cutoff values for predicting alcohol-attributable harm may differ by age and sex, with lower cutoffs for older individuals and women. The ≥65 age group appears to be heterogeneous with regard to vulnerability to alcohol-attributable harm, which should be considered when establishing cutoffs or guidelines for alcohol use.