The Addenbrooke's Cognitive Examination Revised is as effective as the original to detect dementia in a French-speaking population.

INTRODUCTION: This paper presents the validation of the French version of the Addenbrooke's Cognitive Examination Revised (ACE-R).

METHODS: The variability of the 3 versions of the ACE-R (A, B and C), performed by the same observer, hence mainly 2 or 3 times on 119 patients showing no progression, was first calculated by Cronbach's alpha coefficient, t test and linear regression. The alpha coefficients of the 3 versions were obtained showing that the ACE-R versions can be considered as one, and an analysis of the interobserver variability was performed by Cohen's kappa coefficient, t test and linear regression on 12 patients. Eventually, we performed a receiver operating characteristic (ROC) analysis to compare the sensitivities and specificities to detect dementia of the ACE, the ACE-R and Mini Mental State Examination on 319 consecutive patients.

RESULTS: The ROC areas of sensitivities and specificities of the ACE and ACE-R were very similar. Two cutoffs were identified at 83/100 and 89/100 with a specificity to normality of 98.6% if the ACE-R score was ≥83 and a sensitivity to dementia of 98.4% if the ACE-R score was ≤89.

CONCLUSION: ACE-R in French is as reliable and valid as the original version to detect dementia.