Detecting Alzheimer's risk with cognitive dispersion on neuropsychological testing.

Background: Cognitive dispersion refers to the within‐person variability across tasks. Relative to the normative approach, cognitive dispersion may be more sensitive to subtle changes in preclinical Alzheimer's disease. Greater dispersion has been associated with increased risk for conversion to mild cognitive impairment and faster rates of medial temporal lobe atrophy. This metric can be derived from a regular neuropsychological assessment and complement traditional outcome approaches, particularly in those underrepresented in normative data. This study evaluated cognitive dispersion as an independent predictor of clinical status across racial and ethnic groups. Method: Baseline data on 1811 cognitively unimpaired older adults with high vascular risk from the U.S. POINTER multisite lifestyle intervention study were examined. Sample characteristics included mean age of 68 (SD = 5.2), 70% female, 30% less‐than‐college education, and 29% underrepresented groups (301 Black, 231 Other). Correlations between cognitive dispersion and other brain health risks (vascular, subjective cognitive decline (SCD), physical activity) were tested. Cognitive dispersion was defined as the standard deviation of individuals' z‐transformed scores of 8 cognitive tests. Vascular risk burden was the sum of 5 conditions (hypertension, diabetes, hypercholesterolemia, heart disease, obesity). Self‐report SCD was measured by the Cognitive Function Instrument (CFI) and Measurement of Everyday Cognition (ECOG12). Logistic Regression Models were fitted to predict Clinical Dementia Global Ratings (CDR 0.5 vs. 0) to evaluate cognitive dispersion as an independent predictor in the whole sample, White participants and Black participants, controlled for demographics, vascular risk, and physical activity. Logistic Regression Models were also evaluated with CFI or ECOG12 as additional covariate. Result: Greater cognitive dispersion was correlated with older age, greater vascular burden and less physical activity, but not SCD. Cognitive dispersion was significantly higher in men, Black participants, and participants with less education. Cognitive dispersion predicted CDR in Black participants (OR = 4.7, p<0.01), and trended in the whole sample (OR = 1.6, p = 0.05), but was not significant in White participants (OR = 1.2, p = 0.63). Results hold when controlled for SCD (Table 1). Conclusion: Results supported cognitive dispersion as a sensitive objective metric, in addition to self‐report SCD, to identify individuals showing the earliest clinical signs of decline, particularly among those underrepresented in traditional norms. [ABSTRACT FROM AUTHOR]