A Psychometric Evaluation of the Cornell Scale for Depression in Dementia in a Frail, Nursing Home Population

Abstract

OBJECTIVE The authors conducted a psychometric evaluation of the Cornell Scale for Depression in Dementia (CSDD) through factor analysis and assessment of criterion validity in an older, frail nursing home population, with a secondary analysis of pre-intervention data from a longitudinal clinical trial aimed at reducing restraints in nursing homes. METHODS The sample for the present study was 642 nursing home residents (mean [SD] age 84.3 [7.6] years; range: 61–105; 82% women) with completed CSDD scores, who were interviewed immediately before the intervention. Nursing home residents' scores from the 19-item CSDD were subjected to exploratory factor analysis and criterion-validity analysis. RESULTS The factor analysis resulted in four distinct clinically interpretable domains: Depression, Somatic/Vegetative, Disturbed Sleep, and Anxiety. Sixteen items were retained in these domains, and summated score indices and a global score were constructed. The global score and the four indices demonstrated adequate internal consistency and reliability. The indices generated by the factor analysis correlated as expected with criterion variables. CONCLUSION Results suggest that in frail, institutionalized older adults with high rates of dementia, medical illness, and functional disability, depression measurement methods that are less dependent on items highly sensitive to comorbid conditions and not necessarily associated with depression may be more appropriate. Authors recommend further validity testing of the CSDD with similar populations of frail, institutionalized older adults. The authors conducted a psychometric evaluation of the Cornell Scale for Depression in Dementia (CSDD) through factor analysis and assessment of criterion validity in an older, frail nursing home population, with a secondary analysis of pre-intervention data from a longitudinal clinical trial aimed at reducing restraints in nursing homes. The sample for the present study was 642 nursing home residents (mean [SD] age 84.3 [7.6] years; range: 61–105; 82% women) with completed CSDD scores, who were interviewed immediately before the intervention. Nursing home residents' scores from the 19-item CSDD were subjected to exploratory factor analysis and criterion-validity analysis. The factor analysis resulted in four distinct clinically interpretable domains: Depression, Somatic/Vegetative, Disturbed Sleep, and Anxiety. Sixteen items were retained in these domains, and summated score indices and a global score were constructed. The global score and the four indices demonstrated adequate internal consistency and reliability. The indices generated by the factor analysis correlated as expected with criterion variables. Results suggest that in frail, institutionalized older adults with high rates of dementia, medical illness, and functional disability, depression measurement methods that are less dependent on items highly sensitive to comorbid conditions and not necessarily associated with depression may be more appropriate. Authors recommend further validity testing of the CSDD with similar populations of frail, institutionalized older adults.