Abstract
OBJECTIVES: We evaluated the use of the Cornell Scale for Depression in Dementia (CSDD) as a proxy measure. Study questions were: How do residents' self-reports on the CSDD compare with the nurse proxy CSDD ratings of the resident? How do characteristics of depression as rated by the resident CSDD and the nurse CSDD compare? To what extent are demographic and clinical variables associated with resident CSDD, nurse CSDD, and the discrepancy between resident and nurse CSDD scores? METHODS: Residents and nurse proxy pairs (n=395 pairs) from 28 nursing homes (NHs) participated. We calculated discrepancy scores for total and subscale CSDD scores, examined correlations between resident and nurse CSDD scores, and described rates of clinical depression using each of the scores. We conducted multivariate analyses to examine factors associated with resident and nurse CSDD and discrepancy scores. RESULTS: On average, participants had mild cognitive impairment, were White, and female. Associations between resident and nurse CSDD were low (r=0.16). The mean discrepancy score was -2.03 (SD=5.28, p<0.001), indicating that nurses evaluated residents as less depressed than residents evaluated themselves. Discrepancy scores were not associated with residents' cognitive status, but were associated with a measure of self-report reliability. Regression analyses indicated that depression diagnosis accounted for a small but significant association with resident CSDD, but was not significantly associated with nurse CSDD. CONCLUSION: These findings underscore the importance of obtaining resident input when assessing depression in NH residents with dementia, and educating NH nurses in the most effective ways to assess depression.