The role of subjective cognitive concerns (SCC) as a diagnostic criterion for MCI remains uncertain and limits the development of a universally (or widely)-accepted MCI definition. The optimal MCI definition should define an at-risk state and accurately predict the development of incident dementia. Questions remain about operationalization definitions of self- and informant-reported SCCs and their individual and joint associations with incident dementia.

The Einstein Aging Study is a longitudinal study of community-residing individuals, aged 70+. In-person standardized neuropsychological evaluations were completed at baseline. Informant-reported SCC at baseline were assessed via the CERAD clinical history questionnaire. Self-reported SCC were measured using the CERAD, items from the EAS Health Self-Assessment, and the single memory item from the Geriatric Depression Scale. Cox proportional hazards models examined the association of different operationalization of SCC with Petersen and Jak/Bondi MCI definitions on the risk of dementia. Time-dependent sensitivity and specificity at 2, 3, 5, 7y for each definition, and Youden’s index were calculated as an accuracy measure. Controlling for age, sex, education, and race/ethnicity was further considered in the Cox models. Cox proportional hazards models were also used to evaluate the associations of combinations of self- and informant-reported SCC with the risk of incident dementia.

Participants' (N=1097) had a mean age of 78.6y, age ranged from 70 to 100 (mean = 78.6 ± 5.4) years, the sample was 62.4% female, and educational achievement averaged 13.7 ± 3.5 years, with 45.1% obtaining 12 years or fewer years of education. Most participants identified as White (70.6%), though Black participants were well-represented (29.4%). 91% of the sample endorsed at least one SCC. Youden’s index showed that not including SCC in either Jak/Bondi or Petersen classifications had the best balance between sensitivity and specificity across follow-up. A subset of individuals with informants, on average, had a lower proportion of non-Hispanic Blacks and 94% endorsed at least one self-reported SCC. Both informant-reported and self-reported SCC were significantly associated with incident dementia.

Our findings suggest that the SCC criterion may not improve the predictive validity for dementia of various definitions of MCI. Consistent with some prior research, informant-reported SCC was more related to risk of incident dementia than self-reported SCC. Given that requiring informant report as a diagnostic criterion may unintentionally exclude health disparate groups, additional consideration is needed to determine how best to utilize informant-report in MCI diagnosis.