| Women have been underrepresented in substance abuse research (Blumenthal, 1998). While alternative screening methods have been called for due to gender differences, no brief screening instrument for substance dependence has been developed and validated on women (Russell, Chan, & Mudar, 1997). This study aimed to fill this gap by developing three brief screening tools based on subsets of items from the Substance Abuse Subtle Screening Inventory-3 (SASSI-3) that would predict substance dependence among a sample of women. One hundred fifty women from five sites completed the Substance Abuse Module (SAM), SASSI-3, CAGE-AID, and BDI-II. Logistic regression analysis identified the optimal item predictors from the SASSI-3, using SAM diagnosis as the criterion variable. These optimal predictors were then combined into three subsets, which were found to have area under the ROC curve (c) values of .93, .84, and .95. The optimal subsets demonstrated good reliability. Jackknifing demonstrated these values of c were stable across 75% resamples of the original data. Each of the optimal subsets exhibited accuracy that was superior to the CAGE-AID (c value of .83), while two of the optimal subsets exhibited accuracy superior to the full SASSI-3 ( c value of .86). The results demonstrate that the SASSI-3 optimal subsets can be used to screen women for substance dependence. The results also support the utility of face valid items in screening for substance dependence (Clements, 2002). As a supplemental aim, the study examined the utility of the SASSI-3's Defensiveness (DEF) scale in predicting clinical depression, as measured by the BDI-II. Of the 32 participants (22.5%) with low DEF scale scores, 22 (69%) could be considered depressed. These results support the DEF scale's ability to serve as a rough screen for depression (Miller & Lazowski, 1999). |
