Prevalence of risk-drinking in critically ill patients, screened with carbohydrate-deficient transferrin and AUDIT-C score : A retrospective study

Background Studies demonstrate that up to one-third of intensive care unit (ICU) admissions are directly or indirectly related to alcohol. Screening for alcohol use is not routine. This study examined the prevalence of elevated %CDT (carbohydrate-deficient transferrin) and above risk-level AUDIT-C (Alcohol Use Disorders Identification Test, Consumption) in patients admitted to ICU. Methods We conducted a retrospective analysis of clinical and laboratory data from a single ICU where %CDT and AUDIT-C were included in routine assessment. After excluding readmissions, 2532 adult patients from a 21-month period were included. Admission values of %CDT were available for 2049 patients, and AUDIT-C was available for 1617 patients. The association of %CDT and AUDIT-C with short- and long-term outcome was studied by using univariate and multivariate logistic regression analysis. Results %CDT was above the reference value in 23.7% (486/2048) of patients with available %CDT. Of patients with available AUDIT-C, 33% (544/1617) had a risk-level AUDIT-C score. Patients with a risk-level AUDIT-C score were significantly younger than those with a lower score (51 vs 64 years, P <.0001). Increased %CDT was associated with higher severity of illness. AUDIT-C was associated independently with increased risk of long-term mortality in multivariate analysis (P = .007). Conclusion One in three of ICU patients are risk-level alcohol users as measured with AUDIT-C score, and one in four are analysed with %CDT. The prevalence varies according to the method used and any method alone may be insufficient to detect risk-level consumption reliably. Editorial Comment Alcohol overconsumption is associated with need for ICU admission and with less favorable outcomes. Diagnosis of alcohol overconsumption though is problematic due to low sensitivity in screening. In a pilot study, a biomarker and a screening tool are compared. The finding is that multiple tools are needed to achieve an adequate sensitivity for detection.Peer reviewe