Abstract

Objectives: To explore whether it is possible to detect decline in global scores of cognitive function in the proximity of death whilst simultaneously investigating potential risk profiles. Methods: Using the Mini Mental State Examination in a population study of the oldest old in which 99% of participants have died, a linear and quadratic time-to-death repeated measures random effects models were used to detect decline and potential factors which might indicate individual variation. Results: Decline and acceleration of this decline were detectable in the period before death. Some between person variation was detected in this pattern, which included differences in cognitive performance by age at death (-0.2 (SE = 0.02)), sex (-1.2 (SE = 0.2)), initial cognitive impairment (-7.5 (SE = 0.2)) and mobility (-0.6 (SE = 0.2)), in rate of decline by age at death (-0.04 (SE = 0.005)), sex (-0.1 (SE = 0.06)), initial cognitive impairment (-0.3 (SE = 0.07)) and mobility (-0.1 (SE = 0.05)) and differences in change in rate of decline by sex (-0.008 (SE = 0.004)), initial cognitive impairment (-0.02 (SE = 0.04)) and mobility (-0.01 (SE = 0.003)). Conclusion: Using an extension of existing methods for exploring terminal decline, the phenomenon of decline in global cognition measures in the proximity of death was clearly detected as well as potential variables which could influence that pattern. Further work is required to explore whether similar methods can be used to detect the onset of the acceleration of this decline in each individual together with the potential to identify individual level factors that can allow clinicians to distinguish between the normal and preterminal phases of change in extreme old age. Copyright (C) 2010 John Wiley Sons, Ltd.