Aims: Excessive alcohol consumption is a common problem in society and medical practice. There is a need for a diagnostic tool with both high sensitivity and specificity for the detection of excessive alcohol consumption in unselected medical populations. Therefore, we evaluated the diagnostic accuracy of carbohydrate-deficient transferrin (CDT) in the detection of excessive alcohol consumption.
Methods: Computerised literature searches in Medline, Embase and Current Contents databases (01/1966-06/2003) and reference checking. Articles on the detection of excessive alcohol consumption reporting CDT levels and self-reported alcohol consumption as a reference test were selected (n = 101). Studies concerning treatment, relapse detection and traffic offenders were excluded. Quality assessment and data-extraction was done by two reviewers independently. Only studies scoring positive on core validity criteria by Lijmer were eligible for quantitative analysis (n = 29).
Results: Only two CDT-assays (CDTect and CDTriTIA) were evaluated in more than two high validity studies fulfilling the criteria for inclusion in the statistical analysis. Sensitivity of CDTect (14 data points) ranged from 20 to 85%, whereas specificity ranged from 77 to 95%. A summary ROC curve was computed for CDTect. Sensitivity of CDTTriTIA (4 data points) ranged from 10 to 67%, and specificity ranged from 90 to 100%. No summary measure could be computed for CDTTriTIA. The heterogeneity of results could not be explained clinically.
Conclusions: The validity of CDT as a diagnostic tool is still questionable. If the higher values for sensitivity that some studies report can be confirmed by others it is a useful diagnostic tool in unselected populations. However, more methodologically sound, comparable studies need to be performed before firm conclusions can be drawn.