Background: The validity of the Cornell Scale for Depression in Dementia is seldom studied in institutions.
Method: Two reliability studies, with 103 and 32 patients, and a validity study with 231 patients in nursing homes and in hospital were performed. They were assessed by the Cornell Scale, Clinical Dementia Rating scale and Self-Maintenance scale. A psychiatrist 'blind' to the Cornell assessment diagnosed depression according to the ICD-10, DSM-IV-TR and the Provisional Criteria for Depression in Alzheimer's Disease (PCD-dAD).
Results: Cronbach's alpha values were 0.81 and 0.95, and the mean kappa for the Cornell items were 0.91 and 0.57, respectively. In the validity study, 164 subjects had dementia; 105 (45.5%) had depression according to the ICD-10, 68 (29.9%) according to the DSM-IV-TR and 88 (53.3%) of the demented patients had depression according to the PCD-dAD. In the receiver operating characteristic analysis, the DSM-IV-TR criteria produced the highest area under the curve, i.e. 0.81 (95% CI: 0.75-0.87). A range of cutoff points for a depressive disorder was found for the various clinical criteria.
Conclusion: The Cornell Scale is reliable and a range of cutoff points should be used for various clinical criteria of depression.