Cognitive Function Remains Associated With Functional Impairment in Profound Dementia: Alzheimer Disease and Dementia...

The Baylor Profound Mental Status Examination (BPMSE) was developed to assess cognitive function in the profound stage of dementia. The Clinical Dementia Rating (CDR) scale has been widely used in measuring functional performance in dementia. We aimed to determine whether cognitive function is related to overall functional impairment in profound dementia.

We selected 864 patients with probable Alzheimer disease (AD) and 25 patients with possible dementia with Lewy Bodies (DLB) cases with profound dementia by Mini-Mental Status Examination or/and clinical global impression. We used BPMSE to measure cognitive function and the CDR sum-of-boxes (CDR-SB) score to determine overall functional status. We used Spearman rank order correlation to examine the univariate association between CDR-SB and BPMSE in the 2 diagnostic groups separately and multivariable regression analysis to investigate whether BPMSE remained associated with functional status after adjustment for age, sex, education, and APOE ε4 genotype. We expected to see an inverse correlation between BPMSE and CDR-SB scores based on the directionality of the rating scale scoring.

In both AD and DLB, total BPMSE scores had a significant inverse correlation with CDR-SB scores (AD: r = -0.453, p < 0.001; DLB: r = -0.489, p = 0.013). It is of interest that in DLB, the "attention" domain of BPMSE had the strongest association with CDR-SB (r = -0.700, p < 0.001) compared with other domains. The multivariable regression models showed that higher BPMSE scores (i.e., better cognitive function) remained significantly correlated with lower CDR-SB scores (i.e., better global function) in AD (CDR-SB: β = -0.340, p < 0.001), but the regression coefficient for BPMSE did not reach significance in the DLB model (CDR-SB: β = -0.298, p = 0.174).

In patients with AD and DLB who enter the profound dementia stage, cognitive function is associated with the severity of functional impairment. The lack of significance for DLB in multivariable regression could be due to small sample size because the correlation magnitude is similar to that in AD.