Performance of the Rowland Universal Dementia Assessment Scale for the Detection of Mild Cognitive Impairment and Dementia in a Diverse Cohort of Illiterate Persons From Rural Communities in Peru.

The accurate diagnosis of neurocognitive disorders in illiterate Peruvian populations is challenging, largely owing to scarcity of brief cognitive screening tools (BCST) validated in these diverse populations. The Peruvian version of the Rowland Universal Dementia Assessment Scale (RUDAS-PE) is a BCST that relies minimally on educational attainment and has shown good diagnostic accuracy in an urban illiterate population in Peru, yet its psychometric properties in illiterate populations in rural settings of the country have not been previously investigated. To establish the diagnostic accuracy of the RUDAS-PE compared to expert clinical diagnosis using the Clinical Dementia Rating (CDR) Scale in healthy and cognitively impaired illiterate persons living in two culturally and geographically distinct rural communities of Peru. A cross-sectional, population-based study of residents ≥ 50 years of age living in the Peruvian rural communities of Santa Clotilde and Chuquibambilla. A total of 129 subjects (76 from Santa Clotilde and 53 from Chuquibambilla) were included in this study. Gold standard diagnostic neurocognitive evaluation was based on expert neurological history and examination and administration of the CDR. Receiver operating characteristics, areas under the curve (AUC), and logistic regression analyses were conducted to determine the performance of RUDAS-PE compared to expert gold standard diagnosis. Compared to gold standard diagnosis, the RUDAS-PE was better at correctly discriminating between MCI and dementia than discriminating between MCI and controls in both sites (97.0% vs. 76.2% correct classification in Chuquibambilla; 90.0% vs. 64.7% in Santa Clotilde). In Chuquibambilla, the area under the curve (AUC) of the RUDAS to discriminate between dementia and MCI was 99.4% (optimal cutoff at <18), whereas between MCI and controls it was 82.8% (optimal cutoff at <22). In Santa Clotilde, the area under the curve (AUC) of the RUDAS to discriminate between dementia and MCI was 99.1% (optimal cutoff at <17), whereas between MCI and controls it was 75.5% (optimal cutoff at <21). The RUDAS-PE has acceptable psychometric properties and performed well in its ability to discriminate MCI and dementia in two cohorts of illiterate older adults from two distinct rural Peruvian communities.