Introduction: This study aims to discover the criterion validity of the Blessed Dementia Rating Scale (BDRS) for the diagnosis of Alzheimer's disease. Different cut-off scores and corresponding diagnostic sensitivities and specificities were established. Test-retest reliability and internal consistency of the BDRS were also analyzed.
Sample: 451 subjects were studied (254 controls, 86 subjects with mild cognitive impairment and 111 patients with Alzheimer's disease).
Instruments: scores from different sections of the Blessed score were obtained. The global score (BBRS-Total) is the result of the sum of the three sections, A (changes in every day activities), B (changes in habits) and C (changes in personality). The sum of parts A and B (BDRS-Mod) were also quantified.
Statistics: ROC curves, intraclass correlation coefficient and Cronbach's alpha.
Results: The best cut-off score for the BDRS-Total was 3.5 (sensitivity: 87.39%, and specificity: 90%). For the BDRS-Mod, the best cut-off score was 1.5 (sensitivity: 90%, and specificity: 89%). Area under the ROC curve was 0.964 and 0.963 respectively. Intraclass correlation coefficient was 0.98 and Cronbach's alpha was 0.925.
Conclusions: The BDRS has good discriminative validity in terms of sensitivity, specificity and predictive value. It also has good test-retest reliability and internal consistency.