Development and validation of a short version of the Cornell Scale for Depression in Dementia for screening residents in nursing homes. 0063-CI. Accepted May 19, 2016.

Objectives: To develop and validate a short version of the Cornell Scale for Depression in Dementia (CSDD-19) for routine detection of depression in nursing homes.Setting: Australian nursing homesParticipants & Measurements: A series of cross-sectional studies were conducted involving: 1) descriptive analysis of pooled data from five nursing home studies that used the CSDD-19 (n=671) to identify patterns of responses and missing data on individual CSDD items; 2) analysis of 4 of the 5 studies (n=556) to assess CSDD-19 for unidimensionality, item fit, and differential item functioning using Rasch modeling to develop a shorter version, the CSDD-4; 3) validation of the CSDD-4 against the DSMIV using the fifth study of 115 residents, and through expert consultations; and 4) evaluation of the clinical utility of CSDD-4 using an independent cohort of 92 nursing home residents.Results: Four items from the original CSDD-19 were found to be most suitable for depression screening: anxiety, sadness, lack of reactivity to pleasant events, and irritability. The CSDD-4 highly correlated with the original scale (n=474, r=.831, p<.001), with acceptable internal consistency (Cronbach's alpha=0.70). At the cut-offscore of less than 2, sensitivity and specificity of CSDD-4 were 81% and 51% respectively for the independent cohort (n=92), of whom 50% had dementia. The CSDD-4 had an AUC of 0.73 (z=3.47, p<.001), which was compatible with the CSDD- 19 (AUC=0.69, z=2.89, p<.01).Conclusion: The CSDD-4 is valid for routine screening of depression in nursing homes. Its adoption is feasible and practical for nursing home staff, and may facilitate more comprehensive assessment and management of depression in nursing home residents