Performance of three cognitive screening tools in a sample of older New Zealanders

Background: With the ubiquitous Mini-Mental State Exam now under copyright, attention is turning to alternative cognitive screening tests. The aim of the present study was to investigate three common cognitive screening tools: the Montreal Cognitive Assessment (MoCA), the Rowland Universal Dementia Assessment Scale (RUDAS), and the recently revised Addenbrooke's Cognitive Assessment Version III (ACE-III). Methods: The ACE-III, MoCA and RUDAS were administered in random order to a sample of 37 participants with diagnosed mild dementia and 47 comparison participants without dementia. The diagnostic accuracy of the three tests was assessed. Results: All the tests showed good overall accuracy as assessed by area under the ROC Curve, 0.89 (95% CI = 0.80-0.95) for the ACE-III, 0.84 (0.75-0.91) for the MoCA, and 0.86 (0.77-0.93) for RUDAS. The three tests were strongly correlated: r(84) = 0.85 (0.78-0.90) between the ACE-III and MoCA, 0.70 (0.57-0.80) between the ACE-III and RUDAS; and 0.65 (0.50-0.76) between the MoCA and RUDAS. The data derived optimal cut-off points for were lower than the published recommendations for the ACE-III (optimal cut-point 76, sensitivity = 81.1%, specificity = 85.1%) and the MoCA (20, sensitivity = 78.4%, specificity = 83.0%), but similar for the RUDAS (22, sensitivity = 78.4%, specificity = 85.1%). Conclusions: All three tools discriminated well overall between cases of mild dementia and controls. To inform interpretation of these tests in clinical settings, it would be useful for future research to address more inclusive and potentially age-stratified local norms.