Testing the Classification Efficiency of the Geriatric Depression Scale Items and Five Alternative Depression Items

Background and Justification: Although the Geriatric Depression Scale (GDS) is one of the best known and most widely used screens for depression in older adults, the literature is not in agreement about score cut-offs, raising questions about its adequacy as a screening tool for some presentations of depression or subthreshold depression. These discrepancies suggest that it may be appropriate to consider minor revisions to the GDS itself. One of its obvious limitations is that the GDS does not differentiate “core” symptoms from other symptoms; previous work has suggested that items pertaining to withdrawal, apathy and lack of vigor (WAV) are poor discriminators because they are highly endorsed by non-depressed respondents. In addition, the recent clinical literature emphasizes presentations of geriatric depression not found on the GDS, such as unexplained somatic symptoms, “feeling a burden,” “grieving an important loss,” and “thinking of death.” To improve the measurement properties of the GDS, this study examined the discriminatory abilities of GDS items and selected additional items, using two categorization methods.

Method: Residents of congregate housing facilities average age approximately 82 years (N = 178; 22% African American) completed the thirty item GDS, along with five other yes/no indicators of geriatric depression, as part of a self-administered survey. 166 of these respondents were also classified using modules from the MINI diagnostic interview (Sheehan et al., 1998) administered by phone, as either Depressed (major depressive episode), Subthreshold Depression, or Depression-free. Classification efficiency of each GDS item and the 5 new items was estimated using two depression criteria—1.) GDS cut-off of 11 points or more and, 2.) MINI diagnostic interview indicative of either subthreshold or criterion depression—by calculating the items' positive and negative hit rates. These item hit rates (the probability of being in a depression category given the presence—positive hit rate (Hp)—or absence—negative hit rate (Hn)—of a particular symptom) were examined and total combined hit rates were compared to the base rate, the proportion of respondents who scored above criterion.

Findings: In this sample with base rates of depression: GDS, 18.9% and MINI, 31.9%, each individual item's total hit rates exceeded these base rates, meaning each item improved the scale's discriminatory ability for depression. However, positive hit rates were generally far lower than negative hit rates and Withdrawal/Apathy/Lack of Vigor items had the lowest positive hit rates, from 26-39%, and total hit rates around 55%. The best performing new item, “I don't feel like myself” had Hp = 65.2 and Hn of 88.8, followed by “I am a burden to others,” Hp = 64.7, Hn = 85.9. These two improved discrimination of depression over 83% of GDS items.

Implications: The low positive hit rates of withdrawal- and apathy-related items provide further evidence that these may not adequately represent depression. The GDS may be improved by the substitution of selected depression indicators that more accurately represent the clinical presentation of geriatric depression among clients served by geriatric social workers.