Older adults are often evaluated in a range of settings where brief instruments are useful, such as primary care and residential aged settings as well as acute care contexts (Pachana et al., Reference Pachana, Helmes, Byrne, Edelstein, Konnert and Pot2010, Reference Pachana, Mitchell, Pinsker, Morriss, Lo and Cherrier2016). Often, clinicians in both clinical and research contexts turn to reviews of brief instruments to see which tool has the best reliability, validity and efficacy in the populations or clinical conditions of interest. Dementia screening tools, for example, have proliferated, and so such research comparing assessment tools is extremely useful (e.g. Milne et al., Reference Milne, Culverwell, Guss, Tuppen and Whelton2008). And in the search for tools that are time-efficient, applicable to population or longitudinal research and minimally demanding on client or provider time, short versions of established instruments can be attractive (Pachana et al., Reference Pachana, Mitchell, Pinsker, Morriss, Lo and Cherrier2016).

But a proliferation of versions of a single scale, whether in translations or in varying versions of the original, is a more unusual circumstance. This is the case with the Geriatric Depression Scale (GDS; Yesavage et al., Reference Yesavage1982). There are many short versions of the scale (e.g. ranging from 15 items to a single item), and these have been examined for diagnostic accuracy (e.g. Pocklington et al., Reference Pocklington, Gilbody, Manea and McMillan2016). There have been versions (e.g. a 15-item version) where various authors have selected slightly different item sets to compose their own “short version” resulting in a proliferation on nonidentical short versions of similar item length, making for complex choices for researchers (e.g. Dow et al., Reference Dow2018). In essence, with so many versions with varying item pools, one must be careful not only which version to use, but in interpreting a given study using any one version. Is this version the same as in say another published report? If the specific items are not listed, one can be led astray.

For example, Pocklington et al. Reference Pocklington, Gilbody, Manea and McMillan(2016) state that meta-analysis of other brief versions of the GDS was not possible due to an insufficient number of studies with standardized items. In translations of the GDS, some researchers have offered large numbers of options for short forms being useful without giving more specific guidance about which might be of greatest utility, either from a clinical or a methodological perspective (Chin et al., Reference Chin, Liu, Lee and Chu2014). Often, both researchers and clinicians may be puzzled as to the best choices to make regarding GDS versions.

In their paper, Sacuiu et al. Reference Sacuiu(2022) have tackled issues of diagnostic accuracy for major depressive disorder (MDD) across 12 brief versions of the GDS. This cross-sectional study across five Swedish hospitals examined 70 older adults with and without cognitive impairment, who were assessed at home 1-year postsuicide attempt. The 15-item GDS was administered, allowing for the other 11 brief versions of the scale to be calculated from this item pool.

The 7-item version of the GDS by Broekman et al. Reference Broekman, Niti, Nyunt, Ko, Kumar and Ng(2011) was found to be the most accurate in this study. This version, with a cutoff of 3, identified MDD with 88% sensitivity and 81% specificity. Broekman et al. Reference Broekman, Niti, Nyunt, Ko, Kumar and Ng(2011) in part chose the items in their version to minimize the effects of gender, age, ethnicity and the presence of a chronic illness with item performance. In the current study by Sacuiu et al., this version of the GDS was the best-performing even in the presence of cognitive impairment, which was not the case with the longer 15-item GDS. In their review, Pocklington et al. Reference Pocklington, Gilbody, Manea and McMillan(2016) mention that further research is required regarding the effect of cognitive impairment on diagnostic accuracy; the paper by Sacuiu et al. begins to address this concern. However, given the proliferation of versions of the GDS and the inability of any one study to tackle issues of reliability and validity across multiple settings, patient populations and contexts (e.g. Adams, Reference Adams2011; Sutcliffe et al., Reference Sutcliffe2000), further research and perhaps a rationalization of GDS short versions may be in order.