Marie Germund Nielsen1, Eva Ørnbøl2, Mogens Vestergaard3, Per Bech4, Kaj Sparle Christensen5. 1. Research Unit for General Practice, Department of Public Health, Aarhus University, Denmark. Electronic address: mm@ph.au.dk. 2. Research Clinic for Functional Disorders and Psychosomatics, Aarhus University Hospital, Denmark. 3. Research Unit for General Practice, Department of Public Health, Aarhus University, Denmark; Section for General Medical Practice, Department of Public Health, Aarhus University, Denmark. 4. Psychiatric Research Unit, Psychiatric Centre North Zealand, Copenhagen University Hospital, Denmark. 5. Research Unit for General Practice, Department of Public Health, Aarhus University, Denmark.
Abstract
OBJECTIVE: We aimed to assess the measurement properties of the ten-item Major Depression Inventory when used on clinical suspicion in general practice by performing a Rasch analysis. METHODS: General practitioners asked consecutive persons to respond to the web-based Major Depression Inventory on clinical suspicion of depression. We included 22 practices and 245 persons. Rasch analysis was performed using RUMM2030 software. The Rasch model fit suggests that all items contribute to a single underlying trait (defined as internal construct validity). Mokken analysis was used to test dimensionality and scalability. RESULTS: Our Rasch analysis showed misfit concerning the sleep and appetite items (items 9 and 10). The response categories were disordered for eight items. After modifying the original six-point to a four-point scoring system for all items, we achieved ordered response categories for all ten items. The person separation reliability was acceptable (0.82) for the initial model. Dimensionality testing did not support combining the ten items to create a total score. The scale appeared to be well targeted to this clinical sample. No significant differential item functioning was observed for gender, age, work status and education. The Rasch and Mokken analyses revealed two dimensions, but the Major Depression Inventory showed fit to one scale if items 9 and 10 were excluded. CONCLUSION: Our study indicated scalability problems in the current version of the Major Depression Inventory. The conducted analysis revealed better statistical fit when items 9 and 10 were excluded.
OBJECTIVE: We aimed to assess the measurement properties of the ten-item Major Depression Inventory when used on clinical suspicion in general practice by performing a Rasch analysis. METHODS: General practitioners asked consecutive persons to respond to the web-based Major Depression Inventory on clinical suspicion of depression. We included 22 practices and 245 persons. Rasch analysis was performed using RUMM2030 software. The Rasch model fit suggests that all items contribute to a single underlying trait (defined as internal construct validity). Mokken analysis was used to test dimensionality and scalability. RESULTS: Our Rasch analysis showed misfit concerning the sleep and appetite items (items 9 and 10). The response categories were disordered for eight items. After modifying the original six-point to a four-point scoring system for all items, we achieved ordered response categories for all ten items. The person separation reliability was acceptable (0.82) for the initial model. Dimensionality testing did not support combining the ten items to create a total score. The scale appeared to be well targeted to this clinical sample. No significant differential item functioning was observed for gender, age, work status and education. The Rasch and Mokken analyses revealed two dimensions, but the Major Depression Inventory showed fit to one scale if items 9 and 10 were excluded. CONCLUSION: Our study indicated scalability problems in the current version of the Major Depression Inventory. The conducted analysis revealed better statistical fit when items 9 and 10 were excluded.