Objective: Compare diagnostic characteristics of brief cognitive screening tests in residential care/assisted living (RC/AL) residents.
Design: Cross-sectional study involving a comprehensive clinical examination to ascertain a consensus diagnosis of probable dementia, no significant cognitive impairment, or mild cognitive impairment (MCI), including both amnestic and non-amnestic subtypes.
Setting: Fourteen RC/AL facilities in North Carolina.
Participants: Participants were 146 RC/AL residents, aged 65 years or older, who did not have diagnosed cognitive impairment.
Measurements: Diagnostic characteristics of the Mini-Cog, the Mini-Mental State Exam (MMSE), and a new 50-point test based on expanding selected MMSE items (MMX).
Results: Overall, 55/146 (38%) participants were diagnosed with probable dementia, and 76 (52%) met criteria for MCI (most non-amnestic). Both the Mini-Cog and the MMSE showed high sensitivity and negative predictive value for dementia, but had relatively low sensitivity and negative predictive value for MCI. The Mini-Cog had low specificity and was less accurate as a dementia screen than either the MMSE or MMX. Reliability and validity data for the MMX were satisfactory, and it performed better as a screening test for MCI than either the MMSE or Mini-Cog.
Conclusion: Although the MMSE and Mini-Cog are both sensitive to dementia, modest specificity and positive predictive value may limit their utility as screening tools. Preliminary MMX data suggest it improves screening for MCI compared to the Mini-Cog or MMSE, while providing a similar level of screening for dementia. Further work is needed to identify suitable instruments for cognitive screening across the range of MCI and dementia.