Citation

Hillers VN, Alldredge JR, Massey LK. Alcohol Alcohol. 1986; 21(2): 199-205.

Copyright

(Copyright © 1986, Oxford University Press)

DOI

unavailable

PMID

3741553

Abstract

Combinations of automated blood tests were used to screen for level of alcohol consumption through the use of quadratic discriminant analysis. When 15 variables were used, quadratic discriminant analysis correctly placed 91% of persons who reported consumption of less than or greater than four drinks per day (48 g of ethanol). When alcohol intake of greater than or less than 24 g ethanol per day was used as the division point, 86% of persons were correctly placed. As the number of blood tests in the discriminant function was decreased, the percentage of persons correctly classified also decreased. When the number of variables was increased beyond 15, the discriminatory power also appeared to increase; however, the percentage correctly classified may be inflated due to an excessive number of blood tests for the number of subjects.

Language: en