Objective:

This study aimed to extend and externally validate published mappings between two cognitive assessments, the Mini-Mental State Examination (MMSE) and the Clinical Dementia Rating Sum of Boxes (CDR-SB) scale, to support translation of measurements between these tools across all Alzheimer’s disease (AD) stages.

Background:

CDR-SB is commonly used in specialist research settings and clinical trials. Shorter assessments, such as the MMSE, are commonly employed in clinical practice. A targeted literature review identified existing MMSE-to-CDR-SB mappings. However, these lacked the granularity for accurate translation across all AD stages.

Design/Methods:

Optimal functional form for mapping was determined using power-transformed univariate regressions, and the modeled mapping was extended to capture all AD stages. The mapping was externally validated using the US National Alzheimer’s Coordinating Centers (NACC) database, stratified by age, sex, education and AD medication use. Individuals’ most recent visit with valid MMSE and CDR-SB scores, and concordant AD staging based on clinical diagnoses and global CDR scores, were included. Calibration and discrimination metrics adhered to best practices for clinical prediction tool validation.

Results:

A linear MMSE-to-CDR-SB mapping (equation: CDR-SB=-0.6809 MMSE + 20.1982) demonstrated the optimal functional form and high goodness-of-fit (adjusted R² =0.989). Externally validated on the NACC population (n=25,768), the mapping satisfied all validation metrics (Somers’ D=0.705, R²=0.739; surpassing the prespecified threshold of 0.7), and deciles of patients by CDR-SB were generally well calibrated. The mapping demonstrated acceptable discrimination between no/mild versus moderate AD dementia (AUC=0.885; 95% CI: 0.879;0.891) and moderate versus severe AD dementia (AUC=0.756; 95% CI: 0.743;0.769). The mapping accurately predicted AD stage across overall NACC database and by subgroups.

Conclusions:

This extended MMSE-to-CDR-SB mapping demonstrated robust performance, especially in less severe AD stages. The findings can support translation of AD severity assessments between CDR-SB and MMSE datasets, bridging a common information gap between clinical trials and real-world studies.