The development of a short depression screening tool in older adults

Dear Editor, We read with great interest the study of Gokcekuyu et al., which has just been published in your journal. In this study, the authors aimed to validate a fivequestion version of the Geriatric Depression Scale (GDS-5) in the elderly Turkish population, but actually the GDS-5 was previously revalidated in Turkish by Dokuzlar et al. The study of Gokcekuyu et al. seems valuable; however, some methodological considerations need to be taken into account when validating the short forms of scales used for screening. Just like in this study, Dokuzlar et al.’s aimed to validate the GDS-4 and GDS-5 in Turkish elderly patients, and outpatients to geriatric clinics were included in both studies. However, due to the fact that these studies were conducted in two different sociocultural regions of Turkey, the sensitivity of the GDS-5 was higher (96.4% vs 85.8%) and the specificity was lower (67.7% vs 92.1%) depending on the differences in the education levels of the participants. However, the important finding common to both studies is that the Cronbach’s α value is below 0.7, that is, the internal consistency coefficient is at a questionable level, which means that the GDS-5 or some of its items do not represent respondents with depression. To solve this problem, by performing Kappa consistency analysis, the inter-rater reliability of each of the items in the long form of the GDS can be checked, and community-specific short forms can be produced with those items having the highest kappa values.We also applied thismethod to ensure the validity in developing the Turkish GDS-5 (TGDS-5). Thus, we found that the TGDS-5 and the original GDS-5 have only one common item (‘Do you often get bored?’). Other items differed from the GDS-5; however, the Cronbach’s α of the TGDS-5 increased to 0.807, that is, the internal consistency coefficient was ‘good’. In our study, although the specificity and positive predictive value (PPV) of theGDS-5were high, its sensitivity valuewas significantly lower, and low a sensitivity value ​​for a screening tool limits its use. Due to the mentioned drawbacks and the low internal consistency of the GDS-5 (according to the Cronbach’s α analysis), the use of the GDS-5 in Turkish older adults is not appropriate. Since the sensitivity and negative predictive values of the short TGDS-5 are higher than its specificity and PPV, the TGDS-5 is more effective for excluding depression rather than for diagnosing depression. Thus, the TGDS-5 may be useful as a first-stage screening test in a daily practice. As a result, the contribution of Gokcekuyu et al.’s study to the literature is significant. However, it seems to be beneficial to develop appropriate scales for each population, according to their sociocultural characteristics. Thus, the TGDS-5might be amore appropriate test for depression in the Turkish elderly than the GDS-5.