To quantitatively synthesize data on the accuracy of the Rowland Universal Dementia Assessment Scale (RUDAS) in different sociocultural settings and compare its performance to other brief screening instruments for the detection of dementia.

Systematic review and meta-analysis. Literature searches were performed in PubMed, EMBASE, and CINAHL from January 1, 2004 until September 1, 2019.

Community, outpatient clinic, and hospital settings in high-, and low- and middle-income countries.

Twenty-six studies reporting diagnostic accuracy of the RUDAS were included with almost 4000 participants, including approximately 1700 patients with dementia.

Procedures for translation and cultural adaption of the RUDAS, and influence of demographic variables on diagnostic accuracy, were compared across studies. Bivariate random-effects models were used to pool sensitivity and specificity results, and diagnostic odds-ratios and the area under the hierarchical summary receiver operator characteristic curve were used to present the overall performance.

The pooled sensitivity and specificity for the detection of dementia were .82 (95% CI, .78–.86) and .83 (95% CI, .78–.87), respectively, with an area under the curve of .89. Subgroup analyses revealed that the RUDAS had comparable diagnostic performances across high-, and low- and middle-income settings (z = .63, P = .53) and in samples with a lower and higher proportion of participants with no formal education (z = −.15, P = .88). In 11 studies making direct comparison, the diagnostic performance of the RUDAS was comparable to that of the Mini-Mental State Examination (z = −.82, P = .41), with areas under the curve of .88 and .84, respectively.

The RUDAS has good diagnostic performance for detecting dementia in different sociocultural settings. Compared to other brief screening instruments, advantages of the RUDAS include its limited bias in people with limited or no formal education and a minimal need for cultural or language adaptation.