Drawing a line: Differentiating mild from moderate dementia using the functional activities questionnaire

BACKGROUND: Accurate differentiation between mild and moderate dementia is increasingly important, particularly as amyloid-targeting therapies are restricted to early disease stages. Functional impairment in instrumental activities of daily living is a hallmark of progression beyond mild dementia. The informant-based Functional Activities Questionnaire (FAQ) is widely used, but empirically validated cut-offs distinguishing mild from moderate dementia remain insufficiently defined. METHODS: The optimal cut-off score was derived from the entire National Alzheimer's Coordinating Center (NACC) Uniform Data Set as a discovery cohort (n = 34,513) and validated in two independent multicentric cohorts (Alzheimer's Disease Neuroimaging Initiative (ADNI) n = 381 and Frontotemporal Lobar Degeneration Neuroimaging Initiative (FTLDNI) n = 74). The dementia staging was based on the Clinical Dementia Rating (CDR) global score. Functional impairment was assessed using the 10-item FAQ. Receiver operating characteristic analyses in NACC identified optimal thresholds, which were applied unchanged in validation cohorts. Sensitivity, specificity, positive predictive value, and negative predictive value were calculated, and discordant cases were examined to identify factors associated with misclassification. RESULTS: In NACC, the FAQ demonstrated excellent discrimination of moderate dementia (AUC=0.947, 95% CI 0.944-0.949). A cut-off value of ≥18 maximized discrimination (sensitivity 96%, specificity 87%). A higher threshold of ≥23 improved specificity (92%), while maintaining sensitivity (83%). In ADNI and FTLDNI, the sensitivity of ≥18 threshold yielded 92% and 94%, respectively. Moreover, older age and lower cognitive performance were associated with higher odds of misclassification. CONCLUSIONS: The FAQ robustly differentiates mild from moderate dementia across diverse cohorts. A threshold of ≥18 prioritizes sensitivity, whereas ≥23 favors specificity, supporting context-dependent functional staging in clinical and research settings. Individuals with FAQ scores between 18 and 22 may benefit from more detailed clinical staging.