MACE for Diagnosis of Dementia and MCI: Examining Cut-Offs and Predictive Values

Area of Science:

• Neurology
• Medical Diagnostics
• Biostatistics

Background:

• Test cut-off definitions critically influence diagnostic accuracy measures like sensitivity and specificity.
• Revising established cut-offs is often viewed cautiously due to potential bias introduction.

Purpose of the Study:

• To identify optimal test cut-offs for diagnosing dementia and mild cognitive impairment (MCI) using the Mini-Addenbrooke's Cognitive Examination (MACE).
• To evaluate MACE's diagnostic performance using receiver operating characteristic (ROC) and precision-recall (PR) curves.

Main Methods:

• Analysis of pragmatic test accuracy data from a large study on the MACE.
• Determination of optimal cut-offs via maximal Youden index and maximal correct classification accuracy.
• Calculation of MACE predictive values across varying disease prevalences.

Main Results:

• Optimal MACE cut-offs were identified at a point lower than those in the original index studies.
• MACE demonstrated strong performance metrics, including area under the ROC curve and Cohen's d effect size for both dementia and MCI.
• PR curves indicated superior performance for MCI diagnosis, and high negative predictive values were observed across all tested prevalences.

Conclusions:

• Revised MACE cut-offs enhance diagnostic accuracy for dementia and MCI.
• The MACE exhibits robust diagnostic capabilities, particularly its high negative predictive value, which can effectively rule out dementia and MCI.