We have compared the individual sensitivity, specificity and predictive value of the CAGE questionnaire, plasma levels of ethanol and acetate, mean corpuscular volume (MCV), gamma glutamyl transferase (GGT), and glycosylated haemoglobin (Hb A1c) in a group of 50 healthy non-alcoholic controls and 31 patients with non-alcoholic liver disease (Group I), and in a second group of 40 alcoholic patients (Group II). Taken individually, the CAGE questionnaire was the most efficient (96%sensitive and 92%specific), followed by plasma levels of acetate (74%sensitive and 85%specific), MCV (64%sensitive and 91%specific) and GGT (72%sensitive and 80%specific). Hb A1c did not show any statistically significant difference between alcoholics and non-alcoholics and thus is of no use as a screening test for the diagnosis of alcoholism. Furthermore, we attempted to design a discrimination procedure to separate alcoholics from controls and patients with non-alcoholic hepatic diseases using a combination of the most promising tests. The most powerful discrimination model was constructed with the four questions of the CAGE questionnaire. The percentage of correct classifications using this model was 99%from Group I (specificity) and 90%from Group II (sensitivity). The CAGE questionnaire was itself so useful as a discriminant in our sample that no increased diagnostic efficacy was noticed on adding any of the other tests. Using objective variables (MCV, acetate and GGT) as discriminants, we could correctly classify 96%of subjects from Group I (specificity) and only 64%from Group II (sensitivity).
展开▼
