identification of homogeneous groups would ultimately help in advancing the assembling of cohorts needed for meaningful trails. P3-468 CONSTRUCT VALIDITY OF THE VISUAL COGNITIVE ASSESSMENT TEST (VCAT): AN INTERNATIONAL LANGUAGE- NEUTRAL COGNITIVE SCREENING TOOL Audrey Low 1 , Levinia Lim 1 , Linda Lay Hoon Lim 1 , Benjamin Wong 1 , Eveline Silva 1 , Kok Pin Ng 2 , Nagaendran Kandiah 1 , 1 National Neuroscience Institute, Singapore, Singapore; 2 McGill University Research Centre for Studies in Aging, Verdun, QC, Canada. Contact e-mail: Audrey_WN_LOW@nni.com.sg Background: As most well-established cognitive assessments tend to be developed in English-speaking countries, translation is unavoidable when administering tests to those not fluent in English, or from a different culture. This results in inherent problems with regard to construct validity. For instance, to test for verbal fluency, the Montreal Cognitive Assessment (MoCA) requires patients to name words begin- ning with the letter F, which is not possible in cultures without letter- equivalent linguistic units, such as Mandarin. Therefore, the Mandarin MoCA developed in Hong Kong requires patients to name animals instead, transforming a test of lexical fluency into a measure of seman- tic fluency. The VCATwas developed to overcome these challenges by eliminating the need for translation or adaptation. As such, it provides the opportunity for meaningful cross-cultural research both within multilingual countries and across countries differing in language and culture. Methods: To evaluate the construct validity of the VCAT, we recruited 829 participants from the community (n¼109) and from a specialist outpatient memory clinic in Singapore (n¼720) who completed the VCAT and at least one other cognitive measure. Construct validity was determined by the VCAT’s correlation with es- tablished neuropsychological measures of cognitive functioning across the domains of episodic memory, attention, working memory, execu- tive function, visuospatial functioning, and language. Results: The VCAT was significantly correlated to all administered measures and domains of cognitive functioning: Global Cognition, Mini-Mental State Examination (MMSE; .736, p<.001); Episodic Memory Domain, WMS-IV Immediate (.621, p<.001) and Delayed Logical Memory Task (.637, p<.001); Attention Domain, Digit Span Forward (.302, p<.001), Digit Span Backward (.492, p<.001); Working Mem- ory Domain, the ADAS-Cog Immediate (-.638, p<.001)and Delayed Word Recall Task (-.652, p<.001); Executive Function Domain, Color Trails Test 1 (-.580, p<.001), Color Trails Test 2 (-.634, p<.001); Table 1 Prevalence of MCI at baseline in the five classes. Class 1 Class 2 Class 3 Class 4 Class 5 N (%) 107 (8.0) 457 (34.0) 539 (40.1) 118 (8.8) 124 (9.2) aMCI (%) 23 (21.5) 70 (15.3) 24 (4.5) 14(11.9) 1 (0.8) naMCI (%) 29 (27.1) 39 (8.5) 6(1.1) 29 (24.6) 0 Dysexecutive function (%) 93 (86.9) 30 (6.6) 2 (0.4) 102 (86.4) 0 Note. Class 1 ¼ High Risk. Class 2 ¼ Low Average/poor Memory. Class 3 ¼ Average. Class 4 ¼ Dysexecutive Function. Class 5 ¼ Elite. Table 1 Correlation between VCAT score and cognitive tests Correlation with VCAT (Pearson’s r) b Adjusted p-value Global Cognition MMSE .736** .632 <0.001 Episodic Memory WMS-IV Logical Memory (Delayed) .637** .565 <0.001 ADAS-Cog Delayed Word Recall -.652** -.581 <0.001 Attention ADAS-Cog Immediate Word Recall -.638** -.592 <0.001 Digit Span Forward .302** .232 <0.001 WMS-IV Logical Memory (Immediate) .621** .572 <0.001 Color Trails Test 1 -.580** -.520 <0.001 Working Memory Digit Span Backward .492** .383 <0.001 Executive Function Color Trails Test 2 -.634** -.531 <0.001 Visuospatial Rey-O Figure Copy .666** .601 <0.001 Language Verbal Fluency (Animals) .628** .550 <0.001 **p<.001 b ¼ standardized beta coefficient of multiple linear regression, controlling for gender, age, race, years of education and language of administration Adjusted p-value ¼ Adjusted for gender, age, race, years of education and language of administration Figure 1. Patterns of scores across neuropsychological domains of Language, Episodic Memory, Working Memory, Executive Function, and Processing Speed, for each of the classes. Note. Class 1 ¼ High Risk. Class 2 ¼ Low Average/ poor Memory. Memory. Class 3 ¼ Average. Class 4 ¼ Dysexecutive Function. Class 5 ¼ Elite. Language ¼ CAT; letter fluency; Boston Naming Task. Primary Memory ¼ Digit Span. Secondary Memory ¼ Free Recall; Logical Memory. Processing Speed ¼ Trail Making Test A; Digit Symbol Coding. Executive Function ¼ Trail Making Test B; Block Design. Poster Presentations: Tuesday, July 24, 2018 P1299