Objectives: Objective cognitive impairment determined by neuropsychological test performance is a core criterion for the diagnosis of mild cognitive impairment (MCI), yet no consensus has been reached on how this criterion should be operationalized. The aims of this study were to investigate the effect of varying the criteria used to determine cognitive impairment (CI) on prevalence and case definition and to examine comparability of different criteria.
Design: Cross-sectional study.
Setting: Sydney Memory and Ageing Study, Australia.
Participants: Nine hundred eighty-seven nondemented community-dwelling adults aged 70-90 years were enrolled in this study.
Measurements: Participants received a comprehensive neuropsychological test battery measuring four cognitive domains. They were classified as normal or cognitively impaired by applying two types of “impairment” rule that varied the statistical threshold for impairment and the criteria used to determine impairment for each cognitive domain. Prevalence of four MCI cognitive subtypes was determined according to nine different criteria and two types of normative data. Rates of CI were compared in persons of English-speaking and non-English-speaking backgrounds (NESB).
Results: Prevalence of CI ranged from 4 to 70% depending on the impairment criteria used. Agreement between different criteria was poor to moderate. This lack of consistency had greatest impact on MCI subtype classifications with many being reclassified as “normal” or into a different subtype when stringency of the criteria was increased or decreased. Higher rates of impairment were found in persons of NESB across all cognitive domains.
Conclusions: The prevalence of CI was strongly affected by the choice of neuropsychological assessment parameters. Guidelines for operationalizing CI are required.