Development and Validation of a Cognitive Evaluation tool for Multilingual Populations: The Visual Cognitive Assessment Test (VCAT)

Abstract Background With increasing research into novel disease modifying therapies for mild cognitive impairment (MCI) and mild dementia, there is a need for large scale early detection of cognitive disorders. However, existing cognitive evaluation tools largely cater for native English speakers. We designed and validated a novel visual based cognitive evaluation tool that could be applied to multilingual populations worldwide without the need for translation. Methods The development of the VCAT included pilot testing of individual test items, test refinement and development of a field version of VCAT. The test items covered domains of episodic memory, executive function, visuospatial function, semantic language and attention. The initial validation was carried out in a single centre cohort from Singapore comprising of cognitively normal (CN), MCI and dementia participants. Subsequent validation was carried out in centres across four Southeast Asian countries including Singapore, Malaysia, Indonesia and Philippines. Additional evaluation of construct validity against comprehensive neuropsychological test battery has been performed. Further validation in Canada, Brazil, India and South Korea is on‐going. Results In the initial single centre validation in a cohort of 206 participants with a mean age of 67.8 (8.86) years and mean years of education of 10.5(6.0), the area under the curve (AUC) of VCAT for detection of cognitive impairment, CI (MCI and mild dementia) was 93.3 (95% CI 90.1 to 96.4). The Se and Sp of VCAT for the diagnosis of CI were 85.6% and 81.1%, respectively. In a wider Southeast Asian multicentre cohort of 284 participants, VCAT had an AUC of 90.5 (95% CI 87.0‐94.0) in discriminating CN from CI. The multiple languages used to administer VCAT in the four countries did not significantly influence test scores. VCAT and its subdomains demonstrated good construct validity in terms of both convergent and divergent validity and good internal consistency (α = .74). Mean time‐to‐complete VCAT was 15.7 ± 7.3 min. Conclusions The VCAT without the need for translation in multilingual populations has demonstrated high accuracy for the detection of MCI and mild dementia.