¹Tungs' Taichung MetroHarbor Hospital, Department of Psychiatry, Taichung City, Taiwan, ²Tungs' Taichung MetroHarbor Hospital, Research Assistant in Psychiatry, Taichung City, Taiwan, ³Tungs' Taichung MetroHarbor Hospital, Department of Nursing, Taichung City, Taiwan

Background: Carbohydrate-deficient transferrin (%CDT) is a highly specific biomarker for chronic heavy alcohol consumption and has been incorporated into legal and medical regulations for relicensing drunk-driving offenders in several European countries, typically using cut-off values between 1.7% and 2.1%. However, in Taiwan, there is currently no validated %CDT threshold for forensic application in alcohol-use monitoring or driving license regranting programs.

Aims & Objectives: To determine the optimal %CDT cut-off value that differentiates heavy drinkers from non-drinkers and to assess its sensitivity, specificity, and diagnostic accuracy (AUC) for forensic use in license regranting programs in Taiwan.

Method: A total of 50 individuals were recruited from a mandatory alcohol treatment and evaluation program supervised by the Taichung Motor Vehicle Office between 2020 and 2025. Participants were classified into heavy drinkers and non-drinkers. Heavy drinking was defined as ≥50–60 g of ethanol per day for a minimum of 2 consecutive weeks, confirmed by AUDIT and clinical interview. Based on this definition, 25 participants were enrolled who scored a 4 ("Daily or almost daily") on the 3rd question of the AUDIT questionnaire: "How often do you have six or more drinks on one occasion?"The control group consisted of 25 non-drinkers with an AUDIT score of less than 8, or an AUDIT-C score of 0, indicating little to no alcohol consumption. ROC curve analysis was used to determine the area under the curve (AUC), optimal cut-off based on Youden’s index, and additional thresholds prioritizing specificity for forensic purposes.

Results: The %CDT levels were significantly higher among heavy drinkers compared to non-drinkers(2.67% vs. 1.55%, p < 0.001). ROC curve analysis demonstrated an AUC of 0.89, indicating good diagnostic accuracy. The optimal cut-off value determined by Youden’s index was 1.9, yielding a sensitivity of 0.72 and specificity of 0.96. For forensic applications, a threshold of ≥2.03% achieved 100% specificity with sensitivity of 0.64.

Discussion & Conclusions: 

Cut-off (%CDT)	Sensitivity	Specificity	Application
1.90%	0.72	0.96	Clinical (optimal balance; Youden index)
2.03%	0.64	1.00	Forensic use (100% specificity, zero false-positives)

%CDT demonstrated high diagnostic accuracy for identifying heavy alcohol use in Taiwanese drunk-driving offenders. A dual-threshold model is recommended: ≥1.90% for clinical screening and ≥2.03% for forensic decision-making.These findings provide the first locally validated reference for %CDT in Taiwan. Larger-scale studies are recommended to enhance generalizability.