Is Geriatric Depression Scale-15 a suitable instrument for measuring depression in Brazil? Results of a Rasch analysis

Abstract

Depressive symptoms are the most prevalent mental health condition in older adults. Since it cannot be measured directly, the use of instruments is mandatory. The 15-item Geriatric Depression Scale (GDS) is one of the most widely used scales to measure depression in the elderly. It is recognized that the cultural context is a major determinant of the instrument's psychometric performance. Up to the present, this scale has mainly been investigated through classical psychometric approaches. The present study aims to explore whether the 15-item GDS is a suitable instrument in a Brazilian sample. In addition, it explores the potential improvement in the psychometric performance by item refinement. Four hundred twenty-four elderly adults selected through convenience sampling completed the 15-item GDS. Data were analyzed by the Rasch Measurement Model. The Rasch analysis is a powerful modern approach to explore psychometric performance of instruments in health sciences. It examines both the scale and the individual item performance in depth. The 15-item GDS proved not to be suitable in a Brazilian sample. Item misfit and differential item functioning were responsible for considerable misperformance. Scale reduction led to a 10-item structure. This refined format presented adequate psychometric performance and no differential item functioning. The present study offers an alternative and more adequate version of the GDS to be applied in Brazilian subjects. It is also in line with the need for shorter, valid scales in clinical settings. Further investigations are needed to develop a set of cultural-invariant items, which could then be applied in transcultural investigations free of bias.

Keywords:

• measurement
• scale
• elderly
• Rasch
• psychometrics

Acknowledgments

All authors were involved in designing the study and writing the article. E. Chachamovich also supervised the data collection and carried out the statistical analysis. M. Power collaborated in the statistical analysis.