Abstract
INTRODUCTION: Cut-offs derived from baseline cognitive assessments, stratified by intellectual disability (ID) level, have been proposed to diagnose symptomatic Alzheimer's disease (AD) in Down syndrome (DS). However, discrepancies in ID classification risk misclassification when applying cut-offs across sites. METHODS: This dual-center cohort study included 673 adults with mild to moderate ID at different AD stages. We assessed ID classification discrepancies across sites and the impact on Cambridge Cognitive Examination for Older Adults with Down's Syndrome (CAMCOG-DS) cut-offs for AD dementia diagnosis derived from receiver operating characteristic analysis. RESULTS: Inter-rater agreement for ID level classification was 95% within sites but 60% between sites. While CAMCOG-DS score distributions in the whole cohort were similar across sites, ID classification discrepancies caused higher cut-offs in Barcelona for mild and moderate ID compared to Munich. Applying site-specific cut-offs to another cohort reduced sensitivity and specificity. DISCUSSION: Standardizing ID classification is critical for generalizable cut-offs to accurately diagnose AD dementia based on neuropsychological assessments in DS. HIGHLIGHTS: CAMCOG-DS cut-offs by intellectual disability level classify dementia in Down syndrome. ID classification discrepancies between sites impact CAMCOG-DS diagnostic cut-offs. Applying site-specific cut-offs to other cohorts reduces sensitivity and specificity. Standardized ID classification is essential for generalizable cognitive cut-offs. Use site-specific cut-offs until ID classification is standardized.