Item response analysis of the inventory of depressive symptomatology
Authors Ira H Bernstein, A John Rush, Thomas J Carmody, Ada Woo, Madhukar H Trivedi
Published 15 December 2006 Volume 2006:2(4) Pages 557—564
Ira H Bernstein1, A John Rush2, Thomas J Carmody2, Ada Woo1, Madhukar H Trivedi2
1Department of Psychology, The University of Texas at Arlington, Arlington, TX, USA; 2Department of Psychiatry, The University of Texas Southwestern Medical Center at Dallas, Dallas, TX, USA
Background: Both the clinician (IDS-C30) and self-report (IDS-SR30) versions of the 30-item Inventory of Depressive Symptomatology have acceptable psychiatric properties and have been used in various clinical studies. These two scales, however, have not been compared using item response theory (IRT) methods to determine whether the standard scoring methods are optimal.
Methods: Data were derived from 428 adult public sector outpatients with nonpsychotic major depressive disorder. The IDS-C30 and IDS-SR30 were compared using Samejima’s graded response model.
Results: A model was constructed jointly fitting the IDS-C30 and IDS-SR30. An improvement in scale performance was obtained by grouping selected items into domains (specifically sleep, psychomotor, and appetite/weight domains) analogous to the standard scoring of the 16-item Quick Inventory of Depressive Symptomatology.
Conclusions: For the IDS-C30 and IDS-SR30, standard scoring (ie, computing total score using all individual items) provides simplicity, comparability to published data, and a basis for clinical decision making. The revised scoring method, however, improves the utility of both scales when comparing groups as it provides explicit tests of item parameters.
Keywords: Inventory of Depressive Symptomatology, item response theory, Samejima graded response model, depressive symptoms, symptom ratings