[The Depression List for mood evaluation in psychogeriatrics: psychometric analysis and scalability]

The Depression List is a Dutch self-report measure of depressive symptoms in persons with dementia. Fifteen items differ to the extent that they provide reliable measurements of latent constructs such as feeling depressed, tired or lonely. Item scalability and construct reliability were studied by administering the questionnaire to 599 consecutive visitors of a psychogeriatric day care department. Using confirmatory factor analysis three competing hypothetical models were tested for model fit. A four-factor model composed of latent constructs for feeling depressed, tired, lonesome and appetite or sleep disturbance provided an adequate fit to the clinical data, which did not apply to a two-factor model (distinguishing a construct of loneliness from a more general mood construct) and a one-factor model, that hypothesized a single general mood construct for all fifteen items. Nine items carried relatively small proportions of error variance. These reliable items were selected for the construction of three scales that satisfied the assumptions of Mokken's criteria for homogeneity. The resulting measures of feeling depressed, tired or lonely showed evidence of convergent validity for a latent affective construct. The affective measures were not associated with severity of cognitive impairment (discriminant validity). The three scales allow a reliable measurement of differences in affective function in cognitively impaired subjects.