To estimate within-person change thresholds for cognitive and functional measures at diagnosis of incident mild cognitive impairment (MCI) -i.e., a clinically meaningful outcome- in community-dwelling cognitively unimpaired (CU) participants of the Mayo Clinic Study of Aging (MCSA).

Research is limited on meaningful score changes for individual patients in clinical outcome assessments frequently used in Alzheimer’s disease clinical trials, particularly for community-based populations early in the disease course.

The study comprised 3930 CU participants [mean age (standard deviation): 77.3 (9.8); 49.5% female] of the population-based MCSA. A subset with positive brain amyloid-PET was analyzed as a sensitivity analysis. Anchor-based methods were used to estimate thresholds of meaningful score changes for the outcomes of interest.  Distribution-based estimates were also analyzed and included calculations for standard error of measurement and standard error of difference.

Six-hundred forty-six participants (16.4%) developed incident MCI to be used as an anchor. Mean (SD) annualized estimates of change were: Clinical Dementia Rating Scale – Sum of Boxes (CDR-SB) 0.49 (0.81), Mini-Mental State Examination (MMSE) -1.01 (1.36), global cognitive z-score -0.32 (0.43), memory z-score -0.30 (0.47), attention z-score -0.37 (0.68), functional activities questionnaire (FAQ) score 0.51 (1.56). In the amyloid-positive subset, N=133, estimates were comparable: CDR-SB 0.41 (0.66), MMSE -1.09 (1.19), global z-score -0.39 (0.41), memory z-score -0.39 (0.50), attention z-score -0.42 (0.74), FAQ score 0.81 (1.97).  Distribution-based estimates were similar and in combination, provide a range of meaningful score changes associated with MCI onset. Applying the anchor-based estimates as thresholds for progression to the entire cohort, we observed progression in 256 (6.6%) participants in CDR-SB, 574 (14.6%) in MMSE, 698 (17.8%) in global z-score, 956 (24.3%) in memory z-score, 756 (19.2%) in attention z-score, 275 (7.4%) in FAQ.      

Community-based estimates for within-person change thresholds are important to understanding the natural history of disease and applying early interventions.