Abstract: Objective To translate the Rowland Universal Dementia Assessment Scale (RUADS) into a Chinese version, and to conduct a reliability and validity test and determine a cut-off value. Methods After being authorized by Dr. Rowland, the Brislin method was used in the translation of RUDAS, back translation, cross-cultural adaptation and pre-test to form a Chinese version. Convenience sampling was used to survey subjects who match the inclusion criteria in three nursing homes in the community. SPSS 23.0 and AMOS 21.0 software were used for statistical analysis of the data. Results The Cronbach's α coefficient of the Chinese version of RUDAS was 0.818, and the test-retest reliability was 0.942, and the inter-rater reliability was 0.977, and the content validity was 0.97, and the criterion-related validity was 0.886. The exploratory factor analysis obtained a single factor analysis model, with a cumulative contribution rate of 60.423%, and the corresponding principal component loadings of 6 items were all> 0.4. Confirmatory factor analysis showed a ratio of chi-square value to degrees of freedom (χ²/df)=1.686, a goodness of fit index (GFI)=0.962, a root mean square of approximate error (RESEA) of 0.074, and a comparative fit index (CFI) of 0.981, and the results suggested that the model fit well. Average Variance Extracted (AVE)=0.621 and composite reliability (CR)=0.88 indicated good aggregation (convergence) validity. According to the Youden Index (YI) maximization principle, 21 was are the best cut-off value for the Chinese version of RUDAS. Conclusions As a multicultural cognitive assessment scale, the Chinese version of RUDAS has concise entries and good reliability and validity. It is suitable for the screening of high-risk populations of dementia with low education or illiteracy in clinical and community.

Key words: dementia, Rowland Universal Dementia Assessment Scale, reliability, validity