Comparing Global measures of cognitive decline in amnestic mci subjects in the National Alzheimer's Coordinating Center (NACC) uniform data set

Abstract

Many studies have investigated factors associated with the rate of cognitive decline and evolution from mild cognitive impairment to Alzheimer's disease (AD) dementia in elderly patients. However, measures of cognitive decline vary greatly among studies. In this analysis we compared the rates of cognitive decline estimated from three common global measures of cognition: Mini Mental Status Exam (MMSE) score, Clinical Dementia Rating sum of boxes score (CDR-SB), and a neuropsychological tests composite score (CS). A total of 1,375 subjects in the Uniform Data Set (UDS) age 65+ years diagnosed at 29 Alzheimer's disease Centers in the U.S. with amnestic mild cognitive impairment (aMCI) were included in this analysis. All subjects underwent 3-6 cognitive evaluations, approximately 1 year apart, resulting in a total of 4,924 subject-visits. MMSE, CDR-SB, and CS were standardized by the mean and standard deviation of the baseline scores. Population-averaged cognitive decline rates were estimated and compared using Generalized Estimating Equations (GEE). The impact of ApoE e4 allele, age, race, sex, Hachinski Ischemic score (HI), and body mass index (BMI) on each measure was also calculated and compared. Cognitive scores decreased over time for MMSE (ß=-0.28, < .001), CDR-SB (ß=-0.52, < .001), and CS (ß=-.06, < .001). However, the average rate of decline differed significantly among the measures over time: the difference between the CDR-SB and MMSE scores increased by .25 standard deviations (SD) after every year of observation, while the difference between the CS and MMSE score increased by .20 SD. The strength of the relationship between the average rate of cognitive decline and age (p=.024), HI (p=.033), and BMI (p< .001) differed between the CDR-SB and MMSE scores. Similar differences were observed for BMI (p< .001), ApoE (p=.013), and race (p< .001) when the CS was compared to the MMSE score. These results suggest that both the rate of cognitive decline and the effects of AD risk factors on cognitive decline can vary depending on the cognitive measure used. Thus, estimates of cognitive decline and their correlates can be measurement-dependent, reflecting the different aspects of cognition and functioning that they attempt to capture.