Using the Geriatric Depression Scale-Short Form with cognitively normal and cognitively impaired individuals: an IRT-based DIF analysis

Introduction: The Geriatric Depression Scale - short form (GDS-SF; Sheik & Yesavage, 1986) is a largely employed 15-item instrument to assess depression among elderly. Nonetheless, it has been argued that the scale should be used cautiously in case of cognitive impairment. Indeed, the items of the GDS-SF might work differently for cognitively normal and cognitively impaired individuals. Item Response Theory (IRT)-based Differential Item Functioning (DIF) analyses may help to disentangle this point. Objectives: We aimed at studying DIF in the GDS-SF as a function of cognitive impairment. The final purpose was to ascertain if the scale could be used to detect depression in both normal and impaired old people. Design/Methodology: Data from a large longitudinal Italian multi-centric clinical-based study focused on cognitive impairment and dementia in old people were employed. All subjects underwent a comprehensive evaluation including clinical examination, laboratory screening, neuroimaging, and cognitive and behavioral assessments. They were classified as cognitively normal (N=594; Male = 42%, Mean age=75.3, SD=6.6) and cognitively impaired (N= 529; Male = 40%, Mean age=78.9, SD=6.1). The items of the GDS-SF were tested for DIF applying IRT and test information curves for the two groups were compared. Results: One item (i.e., the one referring to problems with memory) showed uniform DIF. Nonetheless, the test information curves of the two groups were nearly overlapped, indicating the item differential functioning did not produce differential test functioning. Conclusions: Results suggest the clinical utility as a screening test of the GDS-SF for both cognitively normal and impaired old people.