ABSTRACT Objective This study extends the Montefiore Einstein Robust Geriatric (MERGER) norms by developing and validating standardized regression-based (SRB) change formulas for a co-normed neuropsychological battery in a robust sample of older adults. We also examine their clinical utility for differentiating mild cognitive impairment (MCI) and dementia and provide multivariate base rates of decline. Methods Standardized regression-based equations were derived from a subsample of MERGER participants (n = 320) using backward regression to predict Time 2 scores from Time 1 performance and relevant demographic factors. Equations were validated in an independent subset (n = 100) and applied to 63 individuals with MCI and 32 with dementia who completed at least two study visits but were excluded from normative analyses. Clinical utility was examined by comparing predicted and observed performance across diagnostic groups. Categorical and multivariate base-rate analyses were used to establish clinically meaningful thresholds. Results Baseline performance was the strongest predictor of follow-up scores, with demographic factors adding measure-specific variance. SRB equations performed well in the validation sample. Exploratory clinical validation analyses indicated that decline was uncommon in cognitively unimpaired (CU) participants but more frequent in MCI and dementia, particularly on memory and executive tasks. Multivariate analyses showed that decline on ≥3 measures was rare in CU participants but observed in nearly half of participants with MCI and most individuals with dementia. Conclusions MERGER SRB change formulas offer a practical, individualized method for tracking cognitive change in older adults. Incorporating multivariate base rates of decline further improves differentiation between normal aging, MCI, and dementia, supporting more accurate longitudinal interpretation and clinical decision-making.
Freilich et al. (Tue,) studied this question.