Subjective cognitive decline prediction of mortality: Results from the Einstein aging study

Background: The relation of pre-dementia stages to mortality has not been fully explored. Previouswork examining subjective cognitive decline (SCD) and mortality is limited and mixed regarding methods used and consistency of findings. Objective: To examine SCD and mortality in a longitudinal, community-based cohort, using item response theory (IRT) methodology to form a compositeSCDmeasure. Also, to assess whether this relationshipwas independent of clinical cognitive status. Methods: The Einstein Aging Study is a diverse longitudinal cohort of adults aged ≫70. SCD items were extracted from baseline CERAD questionnaires and a composite score was formed using IRT methodology. A total of 1,741 participants with complete data were clinically diagnosed as cognitively normal, or as having amnestic mild cognitive impairment (aMCI), nonamnestic mild cognitive impairment (naMCI), or dementia. 645 deaths occurred over a period of 8,912 person-years of follow-up. Cox proportional hazard models predicted time to death adjusting for covariates. Results: A one standard deviation unit increase in level of SCD was associated with >20% higher risk of mortality. However, when models were adjusted for clinical cognitive status, the association was no longer significant. Both dementia and aMCI predicted mortality. Furthermore, when analyses focused only on those without cognitive impairment, SCD level did not predict mortality. Conclusions: The association of SCD with mortality may be due to the association of SCD with clinical cognitive status. Thus, SCD may be used as a community-based screen to initially identify those with cognitive impairment who may be at greatest risk for death.