Objectives: To determine the 2-year outcome from 16 different current classifications of mild cognitive impairment (MCI) in a population-based sample.
Design: Prospective cohort study: baseline and 2-year follow-up phases.
Setting: Large-scale multicenter study, United Kingdom.
Participants: : Thirteen thousand four individuals aged 65 and older from the Medical Research Council Cognitive Function and Ageing Study. From this, a subsample of 2,640 individuals was selected and completed a more-detailed cognitive assessment. Individuals who underwent further assessment were asked to complete annual or 2-year follow-ups.
Measurements: Information on sociodemographic status, general health, cognitive impairment and functional ability were collected using a structured interview. Individuals were classified according to 16 different definitions of MCI. These were applied retrospectively.
Results: The dominant outcome across definitions was an impairment that was not classifiable or reversion to normality. Progression to dementia was variable and generally poor. Overall progression was highest in classifications in which impairment extended to memory and nonmemory domains. Predictability was age dependent in some but not all classifications.
Conclusion: Current classifications of MCI have variable outcomes in population-based samples. Progression to dementia is relatively rare and is dependent on age and definition. Selection criteria developed for the clinic are based on a "high risk" approach that leads to exclusion of a large percentage of the impaired population who are neither normal nor demented and for whom no intervention options are currently available. A refined definition of this construct is urgently needed if MCI is to be used to predict dementia in population-based studies.