Confirming Diagnosis of Hazardous and Harmful Alcohol

Diagnostic accuracy of a computer assisted diagnostic system compared to conventional markers of alcoholism. Netherlands (van Lambalgen, professor of logic and artificial intelligence). Acknowledgements All laboratory tests, except carbohydrate-deficient transferrin (CDT) were performed in the ABSTRACT Context Conventional tests for alcoholism have good screening properties for detecting hazardous and harmful alcohol use (HHAU), but they fail to confirm this diagnosis accurately and objectively. Objective To validate a Bayesian Alcoholism Test (BAT) for confirming the diagnosis of hazardous and harmful alcohol use Design and Subjects BAT is based on studies on the prevalence of HHAU and other diseases causing similar abnormalities, and on conditional probabilities of these disorders and associated biochemical markers and clinical signs. BAT was compared to carbohydrate-deficient transferrin (CDT), and gamma-glutamyltransferase (GGT) in three populations: 67 treatment seeking alcoholics, 68 non-treatment seeking heavy drinkers and 47 controls. The diagnostic criterion was reported alcohol consumption. Results Comparing alcoholics and controls, sensitivity of BAT (94%) was significantly higher than CDT (63%) and GGT (73 %). The specificity of BAT (98 %) was not significantly higher than CDT (93 %) and GGT (92%). The area under the curve for BAT (0,989) was significantly higher than the area under the curve for CDT (0,909) and area under the curve for GGT (0,902). In the group of heavy drinkers, BAT could differentiate significantly better than GGT between heavy drinkers above a harmful level, heavy drinkers with a hazardous use and heavy drinkers below a hazardous consumption. The difference with CDT was not statistically significant. 3 Using pooled data of all 182 subjects included in the study, the amount of drinking had a significant better correlation coefficient with BAT (0.795) than with CDT (0.657), and GGT (0.604).