For a sample of 1156 patients (520 alcoholics and 636 nonalcoholics), discriminant function analyses were performed on the total score, a nine-item version, and a two-item version of the Self-Administered Alcoholism Screening Test (SAAST). With sensitivities set at 90 and 95%, specificities for the total score and nine-item versions ranged from 96.4 to 99.4%. Cross-validation of the nine-item versions with the 'jackknife' procedure resulted in only one additional misclassification of the 1156 subjects. Separate analyses of the male and female samples revealed that more items entered the discriminant function for women than for men and resulted in a higher, although clinically nonsignificant, percentage of correct classification for women. The results strongly support the use of either the total score or the nine-item version of the SAAST in large-scale screening for alcoholism in a medical patient population.
|Original language||English (US)|
|Number of pages||5|
|Journal||Alcoholism: Clinical and Experimental Research|
|State||Published - 1987|
ASJC Scopus subject areas
- Medicine (miscellaneous)