Other studies have found significan

Other studies have found significant associations between plaque presence, number, and total plaque area and cognition,34,35 but we did not. This may be because of acoustic shadow in the assessment of carotid ultrasounds affecting the accuracy of our plaque measurements. Also, those with carotid plaques may have received primary prevention treatment that attenuated the natural history of the disease and subsequently the effect of atherosclerosis on cognition. Plaque may not be an effective marker of cognitive performance in our cohort.There were no statistically significant findings in our longitudinal analysis, except for episodic memory, which paradoxically exhibited less decline on average with greater cIMT. This may be explained by the healthy cohort effect because participants who were available for a second NPE are likely healthier than those who did not complete the second NPE. Competing risk of mortality is also a concern in this older cohort. Because our cohort was older at the initial NPE, pathology affecting cognition may have already progressed substantially; so a change in cognition may not be readily detected. Evidence suggests that midlife risk factor burden is more important for predicting cognitive performance than late-life risk factors.36 Therefore, our longitudinal analysis may not be sensitive to changes in cognition in this elderly cohort and should be interpreted with caution.Strengths of the current study include its population-based design in a mostly Hispanic/Latino community, multiple time points of domain-specific cognitive assessments, and various measures of subclinical carotid atherosclerosis. The limitations of this study are important to note. The current study was conducted within a subcohort of NOMAS participants, and the findings may not generalize to the full study cohort, especially given the survival bias and competing risk of mortality that is common in longitudinal cohort study designs. Although our ultrasound protocol specifies the use of multiple angles of insonation to improve the accuracy of plaque area measurements, acoustic shadowing may still decrease the accuracy of these measurements. Longitudinal cognitive data were missing for 340 participants in this study, but the missingness of follow-up NPE was unrelated to cIMT and unrelated to APOE ε4 allele status, limiting the likelihood of selection bias for the analysis of cognitive decline.