ABSTRACT
In Alzheimer’s dementia (AD), greater declines in semantic fluency (SF) relative to letter fluency (LF) have been assumed to reflect semantic disintegration. However, the same pattern is observed in typical aging and neurodegenerative disorders besides AD. We examined this assumption by comparing different aspects of SF and LF performance in older adults with and without dementia, and identifying which verbal fluency measures most clearly distinguish AD from typical aging. Verbal fluency data were compared from 109 individuals with AD and 66 typically aging adults. Correct items, clusters, and errors were analyzed using both raw counts and proportions. Regression analyses examined Task-by-Group interactions and the impact of demographic variables on verbal fluency measures. ROC analyses examined the sensitivity and specificity of the different outcome measures. In regressions, interactions were found for raw but not proportional data, indicating that different group patterns were driven largely by the number of correct items produced. Similarly, in ROC analyses, raw SF totals showed stronger discriminability between groups than either raw discrepancy scores (SF–LF) or discrepancy ratios (SF/LF). Age and cognitive status (MMSE) were the strongest individual predictors of performance. Findings suggest that AD entails quantitative declines in verbal fluency, but qualitatively similar patterns of performance relative to typically aging adults. Thus, SF declines in AD seem to be at least partially attributable to an exaggeration of the underlying mechanisms common to typical aging, and do not necessarily implicate semantic disintegration.
Acknowledgments
This research was partially supported by a grant from the John Templeton Foundation and the University of Chicago. Additional thanks go to the developers of and contributors to DementiaBank, especially Davida Fromm for her help with data extraction, and to the research assistants who helped code the data.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Supplementary material
Supplemental data for this article can be accessed online at https://doi.org/10.1080/13825585.2022.2079602
Notes
1. Although we did not have specific hypotheses about sex differences, we included this variable because some prior research has found sex differences in verbal fluency (e.g., Laws, Citation2004; Scheuringer et al., Citation2017; Tombaugh et al., Citation1999; Vaughan et al., Citation2016).
2. Following the suggestion of a reviewer, analyses were also replicated using outcome measures residualized on MMSE scores to factor out the impact of dementia severity. These models included the same predictors as the original analogous models, except for the main effect of MMSE and interactions involving MMSE scores.
3. Results of an additional analysis, using total words and discrepancy scores residualized on MMSE, are also shown in Supplementary Table S2. These models show similar findings to the original models, with reduced group effects.