Objectives: To estimate the sensitivity, specificity, and reliability of the Minimum Data Set Cognition Scale (MDS-COGS) in screening for undetected dementia when completed by direct care staff in residential care/assisted living (RC/AL) facilities and secondarily to determine the prevalence of dementia in the sample.
Design: A cross-sectional study in which staff were trained to complete the MDS-COGS. Research interviewers and a neuropsychologist obtained information on each participant. Two neurologists reviewed the data and examined the participant, rendering a probable diagnosis of dementia/non-dementia diagnosis. MDS-COGS results were compared with the neurologists' determination.
Setting: Fourteen RC/AL facilities in North Carolina.
Participants: Data were collected from 50 staff on 166 residents without a diagnosis of dementia.
Measurements: In addition to the MDS-COGS, measures included a comprehensive neuropsychological battery. Depression and other neuropsychiatric symptoms were also assessed.
Results: Neurologists determined that 38% of participants had probable dementia. An MDS-COGS cutpoint of 2 was highly specific (0.97) but not very sensitive (0.49) for dementia. Test-retest and interrater agreement for a negative screen were high (88% and 93%, respectively).
Conclusion: The MDS-COGS is a simple, brief screen that RC/AL staff can complete. It will identify with high specificity a subset of residents with undetected dementia, allowing rapid identification of those likely to need dementia care. Caution needs to be exercised in light of its low sensitivity, because some with milder dementia will not be detected. Further work is needed to determine whether staff can and will use the MDS-COGS as a trigger for more-thorough assessment and to guide care and improve outcomes.