Dimensionality, Reliability and Validity of the InterRAI Depression Rating Scale in a Canadian Palliative Care Population

TitleDimensionality, Reliability and Validity of the InterRAI Depression Rating Scale in a Canadian Palliative Care Population
Publication TypeJournal Article
Year of Publication2015
AuthorsFisher K.A, Seow H., Brazil K., Smith T.F, Guthrie D.M
JournalSocial Indicators Research
Date PublishedMay
ISBN Number0303-8300
Accession NumberWOS:000352228900016
Keywordsadvanced cancer-patients, confirmatory factor-analysis, dimensionality, factor analysis, hospital anxiety, in-home care, interrai depression rating scale, inventory, Minimum data set, mokken scale analysis, nursing-homes, older-adults, palliative care, psychometric properties, reliability and validity, symptoms

A reliable and valid instrument is needed to screen for depression in palliative patients. The interRAI Depression Rating Scale (DRS) is based on seven items in the interRAI Palliative Care instrument. This study is the first to explore the dimensionality, reliability and validity of the DRS in a palliative population. Palliative home care patients (n = 5,175) residing in Ontario (Canada) were assessed with the interRAI Palliative Care instrument. Exploratory factor analysis and Mokken scale analysis were used to identify candidate conceptual models and evaluate scale homogeneity/performance. Confirmatory factor analysis compared models using standard goodness-of-fit indices. Convergent and divergent validity were investigated by examining polychoric correlations between the DRS and other items. The "known groups" test determined if the DRS meaningfully distinguished among client subgroups. The non-hierarchical two factor model showed acceptable fit with the data, and ordinal alpha coefficients of 0.83 and 0.82 were observed for the two DRS subscales. Omega hierarchical (omega(h)) was 0.78 for the bifactor model, with the general factor explaining three quarters of the common variance. Despite the multidimensionality evident in the factor analyses, bifactor modelling and the Mokken homogeneity coefficient (0.34) suggest that the DRS is a coherent scale that captures important information on sub-constructs of depression (e.g., somatic symptoms). Higher correlations were seen between the DRS and mood and psychosocial well-being items, and lower correlations with functional status and demographic variables. The DRS distinguished in the expected manner for known risk factors (e.g., social support, pain). The results suggest that the DRS is primarily unidimensional and reliable for use in screening for depression in palliative care patients.

Short TitleSoc Indic Res
Alternate JournalSoc Indic Res