Identification of heavy drinkers using a combination of laboratory tests

Objective: This study derived and evaluated a model that used results of commonly performed laboratory tests to identify men who are heavy drinkers. Method: The results of 40 commonly available laboratory tests were obtained on a diverse sample of 426 heavy drinkers and 188 light drinkers. A logistic regression equation for identifying heavy drinkers was derived in a training data set of 411 subjects and tested in a validation data set of 203 subjects. Results: Ten laboratory measurements were included in the final regression equation: chloride, sodium, ratio of direct to total bilirubin level, blood urea nitrogen, high density lipoproteins, monocyte count, phosphorus, platelets, aspartate aminotransferase, and mean corpuscular hemoglobin. In the validation data this model correctly identified 98% of the 161 heavy drinkers and 95% of the 42 light drinkers. Other models reported in previous literature were applied to these subjects and did not perform as well. The model performed better for subjects of lower socioeconomic status. Conclusions: The laboratory tests in our model may help identify heavy drinkers. The performance of models to identify heavy drinkers depends on the demographic characteristics of the subjects. (C) 1997 Elsevier Science Inc.