a1 Department of Public Health and Primary Care, University of Cambridge, Cambridge U.K.
a2 MRC Biostatistics Unit, Cambridge, U.K.
Background: Cognitive assessment of older persons, particularly those with impairment, is hampered by measurement error and the ethical issues of testing people with dementia. A potential source of valuable information about end-of-life cognitive status can be gained from those who knew the respondent well – mostly relatives or friends. This study tested the association between last cognitive assessment before death and a retrospective informant assessment of cognition.
Methods: Data were analyzed from 248 participants from the Medical Research Council Cognitive Function and Ageing Study who were aged 71 to 102 years at death. Late-life cognition was assessed 0 to 8 years before death using the Mini-mental State Examination (MMSE) and the informant measure was taken 0 to 7 years after death using a Retrospective Informant Interview (RInI).
Results: Zero-inflated Poisson regression showed a strong association between MMSE scores and RInI scores – those scoring 29–30 on the MMSE had a RInI score four times lower than those who scored <18 (p < 0.001). The time between MMSE and death was also a significant predictor with each additional year increasing RInI scores by 12.4% (p < 0.001). The time between death and RInI was only a significant predictor when including measures that were taken four years or more after death.
Conclusions: Cognitive scores from retrospective informant interviews are strongly associated with late-life MMSE scores taken close to death. This suggests that the RInI can be used as a proxy measure of cognition in the period leading up to death.
(Received April 08 2010)
(Revised April 29 2010)
(Revised May 13 2010)
(Accepted May 16 2010)
(Online publication July 19 2010)
c1 Correspondence should be addressed to: Riccardo E. Marioni, Department of Public Health and Primary Care, University Forvie Site, Cambridge CB2 0SR, U.K. Phone: +44 1223 763837; Fax: +44 1223 330330. Email: email@example.com.