Abstract
BACKGROUND: The Dementia Severity Rating Scale (DSRS) is an informant report, dementia staging tool that is quick to administer and has previous been shown to differentiate between people with dementia and healthy controls. However, it is not clear how accurate the tool is screening against diagnostic criteria in middle-income settings. METHODS: Embedded within the STRiDE programme, older adults (aged ≥65 years) and their informants were randomly recruited from four sites across Indonesia and South Africa. All informants were asked to complete DSRS. We report the tool's psychometric properties and accuracy against the 10/66 short diagnostic algorithm. RESULTS: Between September and December 2021, data was collected from 2110 older adults in Indonesia and 408 in South Africa. Overall, the DSRS scores significantly differed between those with and without dementia, as identified on the 10/66 short algorithm (p < .05). The difference between groups remained significant after controlling for key factors related to older adult and informant demographics. A score >2 on the DSRS had the greatest agreement with the 10/66 short algorithm and had excellent discriminative properties in both Indonesia (Area Under Curve (AUC) = .75, 95% CIs = .72-.77) and South Africa (AUC = .82, 95% CIs = .76-.88). CONCLUSIONS: The DSRS has potential as a screening tool for dementia in middle-income countries, with high sensitivity and specificity against a standardized diagnostic algorithm.