Objectives: The aim of this study was to identify natural subgroups of older adults based on cognitive performance, and to establish each subgroup’s characteristics based on demographic factors, physical function, psychosocial well-being, and comorbidity. Methods: We applied latent class (LC) modeling to identify subgroups in baseline assessments of 1345 Einstein Aging Study (EAS) participants free of dementia. The EAS is a community-dwelling cohort study of 70+ year-old adults living in the Bronx, NY. We used 10 neurocognitive tests and 3 covariates (age, sex, education) to identify latent subgroups. We used goodness-of-fit statistics to identify the optimal class solution and assess model adequacy. We also validated our model using two-fold split-half cross-validation. Results: The sample had a mean age of 78.0 (SD=5.4) and a mean of 13.6 years of education (SD=3.5). A 9-class solution based on cognitive performance at baseline was the best-fitting model. We characterized the 9 identified classes as (i) disadvantaged, (ii) poor language, (iii) poor episodic memory and fluency, (iv) poor processing speed and executive function, (v) low average, (vi) high average, (vii) average, (viii) poor executive and poor working memory, (ix) elite. The cross validation indicated stable class assignment with the exception of the average and high average classes. Conclusions: LC modeling in a community sample of older adults revealed 9 cognitive subgroups. Assignment of subgroups was reliable and associated with external validators. Future work will test the predictive validity of these groups for outcomes such as Alzheimer’s disease, vascular dementia and death, as well as markers of biological pathways that contribute to cognitive decline. (JINS, 2018, 24, 511–523)

Longitudinal administration of neuropsychological instruments are often used to assess age-related changes in cognition. Informative loss to follow-up may bias the results of these studies. Herein, we use auxiliary data to adjust for informative loss to follow-up. In the Einstein Aging Study, memory was assessed annually in a community sample of adults age 70+, free of dementia at baseline, using the free recall from the Free and Cued Selective Reminding Test, and via telephone using the Memory Impairment Screen for Telephone (the auxiliary data). Joint linear mixed models were used to assess how the effect of the APOE ε4 genotype may be affected by informative missingness in the in-person data. A total of 620 EAS participants contributed 2085 person years of follow-up to the analyses. Memory decline rates estimated in joint models were 19% greater in ε4 negative participants and 27% greater in ε4 positive participants compared to traditional approaches; the effect of APOE ε4 on memory decline was 37% greater. Joint modeling methods can help address bias caused by informative missing data in the estimation of the effect of risk factors on cognitive change, and may be applicable to a broader range of outcomes in longitudinal aging studies. (JINS, 2014, 20, 1–6)

This study examined the relationship between cognitive function and sleep onset/maintenance difficulties (SO/MD) in nondemented older adults. We hypothesized that SO/MD negatively impacts cognition and that older adults with lower education would be especially vulnerable to its effects. The sample comprised 549 older adults from the Einstein Aging Study (EAS), a community-based cohort. Participants completed neuropsychological assessment and a sleep questionnaire. Univariate ANCOVAs were performed with cognitive performance as a dependent variable, SO/MD (present or absent) and education (lower: ⩽12 years; higher: >12 years) as between-subjects factors, and age, ethnicity, gender, depression, and cardiovascular comorbidies as covariates. Participants were an average age of 79.7 ± 5.0 years (range = 71–97 years). Fifty-seven percent (n = 314) of the sample met criteria for SO/MD. Among participants with SO/MD, those with lower education performed more poorly on a test of category fluency than participants with higher education (means: 35.2 vs. 41.0; p < .001); among older adults without SO/MD, educational attainment had no measurable effect on cognition (SO/MD × education interaction (F(1,536) = 14.5; p = .00)). Consistent with the cognitive reserve hypothesis, older adults with lower education appear selectively vulnerable to the negative effects of sleep onset/maintenance difficulties on tests of verbal fluency. (JINS, 2012, 18, 461–470)

Individuals with amnestic mild cognitive impairment (aMCI) show deficits on traditional episodic memory tasks and reductions in speed of performance on reaction time tasks. We present results on a novel task, the Cued-Recall Retrieval Speed Task (CRRST), designed to simultaneously measure level and speed of retrieval. A total of 390 older adults (mean age, 80.2 years), learned 16 words based on corresponding categorical cues. In the retrieval phase, we measured accuracy (% correct) and retrieval speed/reaction time (RT; time from cue presentation to voice onset of a correct response) across 6 trials. Compared to healthy elderly adults (HEA, n = 303), those with aMCI (n = 87) exhibited poorer performance in retrieval speed (difference = −0.13; p < .0001) and accuracy on the first trial (difference = −0.19; p < .0001), and their rate of improvement in retrieval speed was slower over subsequent trials. Those with aMCI also had greater within-person variability in processing speed (variance ratio = 1.22; p = .0098) and greater between-person variability in accuracy (variance ratio = 2.08; p = .0001) relative to HEA. Results are discussed in relation to the possibility that computer-based measures of cued-learning and processing speed variability may facilitate early detection of dementia in at-risk older adults. (JINS, 2012, 18, 260–268)

Impaired recall for early items (primacy) and late items (recency) on word list recall tests are seen in Alzheimer's disease (AD). We compared conventional scoring on the Telephone Instrument for Cognitive Status (TICS) recall list with scorings based on retention-weighted recall (RWR: each item weighted by its serial position) in older adults participating in a community-based aging study. Subjects with mild AD (N = 18) did not differ from those without dementia (N = 231) with respect to recency (46% vs. 59%, p = 0.2), but had impaired primacy (2% vs. 39%, p < .001) on word recall on the TICS. RWR scoring improved the effect size (1.52 SD) compared to conventional scoring (1.08 SD). With a fixed sensitivity of 85%, specificity was lower using conventional scoring (56%) than RWR (76%) scoring. Our findings suggest that optimized RWR scoring of word list free recall can improve detection of mild AD compared to conventional scoring. (JINS, 2006, 12, 436–440.)