Diagnostic accuracy of the Thai version of the Mini‐Addenbrooke's Cognitive Examination as a mild cognitive impairment and dementia screening test

Abstract

Aim To examine the diagnostic accuracy of the Thai version of the Mini‐Addenbrooke's Cognitive Examination (MACE‐T) for detecting mild cognitive impairment and dementia. Methods Participants consisted of 60 controls, 40 people with mild cognitive impairment (MCI) and 48 people with dementia. All participants completed the Thai version of the Addenbrooke's Cognitive Examination – III (ACE‐T) and the MACE‐T scores were obtained from items within the ACE‐T. Results A cut‐off score of 21/22 detected MCI with a sensitivity of 95% (95% confidence interval (CI): 86.1–100) and specificity of 85% (95% CI: 77.8–93.4). A cut‐off score of 16/17 detected dementia with a sensitivity and specificity of 95.8% (95% CI: 88.2–100) and 85% (95% CI: 76.4–89.7), respectively. Using Bayes theorem, the MACE‐T demonstrated high negative predictive values (> 94%) for diagnosing MCI and dementia, independent of their prevalence rates. Education was the only significant predictor of performance in the MACE‐T. Conclusion The MACE‐T is a good short cognitive screening test with high diagnostic accuracy for screening for MCI and dementia.

Adaptation and Validation of the Mini-Addenbrooke’s Cognitive Examination in Dementia in Arabic Speakers in Egypt

“…In addition, although our sample size was not large, it was comparable to other validation studies of the mini-ACE in dementia such as the Brazilian version (dementia = 23 and controls = 25) [20], the Thai version (dementia = 48 and controls = 60) [22] and the Greek version (dementia = 40 and controls = 38) [23]. In addition, our sample size was adequate to provide high sensitivity (92%), high specificity (95%), and high accuracy (94%) for the cutoff score calculated that was comparable to other studies.…”

“…Our optimal cutoff point was 18 out of 30, which was lower than the upper cutoff score provided by the English language version of the m-ACE (which was 25) but comparable with the lower cutoff score (which was 21) [4]. Other language versions’ cutoff scores ranged from 16 to 23 out of 30, so our cutoff score of 18 can be considered midrange: 16/30 for the Spanish version (86.7% sensitivity and 87% specificity) [19]; 20/30 for the Brazilian version (95.7% sensitivity and 90.2% specificity) [20]; 21–22/30 for the Chinese version (96% sensitivity and 87% specificity) [21]; 16–17/30 for the Thai version (95.8% sensitivity and 85% specificity) [22]; and 23/30 for the Greek version (97.4% sensitivity and 94.7% specificity) [23].…”

“…Possible explanations for the variation between our cutoff score with other language versions of the m-ACE may be due to the type samples investigated in each study with some studies (including ours) using heterogeneous samples that included different types of dementia [4,19], while others used homogeneous samples (e.g., Alzheimer patients only) [20][21][22]. In addition, some studies only investigated patients with mild dementia [20,21], so with less severity of dementia, a better performance might be expected, which can lead to higher cutoff scores.…”

Validation of the Mini-Addenbrooke’s Cognitive Examination in Mild Cognitive Impairment in Arabic Speakers

“…Our optimal cut-off point was 21 out of 30, with comparable sensitivity and specificity to other language versions: 21–22 out of 30 for the Thai version (95% sensitivity and 85% specificity) [8], 25–26 out of 30 for the Chinese version (88% sensitivity and 72% specificity) [9], 25–26 out of 30 for the Japanese version (77% sensitivity and 72% specificity) [10], and 27 out of 30 for the Brazilian version (75% sensitivity and 60% specificity) [11]. A possible explanation for variations in cut-off scores might be differences in education levels between the different samples from different countries.…”

“…The m-ACE was found to be superior to the Mini-Mental State Examination (MMSE) in diagnostic utility [5] and has been translated into several languages [6-12], including Arabic [13]. It has also been validated in MCI in several populations [8-11] but not in Arabic speakers.…”