Validity of clinical criteria for the diagnosis of dementia with Lewy bodies.

OBJECTIVE: To assess the clinical validity of clinical diagnostic criteria for dementia with Lewy bodies (DLB).

METHODS: We assessed the sensitivity, specificity, and positive and negative predictive values of the clinical criteria of the Consortium on dementia with Lewy Bodies (CDLB) in 18 patients with autopsy-proven DLB and in 76 patients with dementia not associated with Lewy bodies, using postmortem diagnosis as a gold standard.

RESULTS: CDLB criteria had either high sensitivity or high specificity, but no set of criteria simultaneously provided both high sensitivity and high specificity. Clinical criteria had higher predictive validity in patients with pure DLB than in patients with DLB and AD. Seventy-eight percent of patients with pure DLB had two or more major criteria, compared with 44% of patients with DLB and AD (p<0.02). If the nine patients with DLB and AD were excluded from the DLB group, the CDLB criteria for probable DLB had sensitivity of 78% and specificity of 85%. CDLB criteria for probable DLB (two or more major criteria) distinguished DLB from AD with a sensitivity of 78% and a specificity of 64%.

CONCLUSIONS: The proposed CDLB criteria have high negative predictive value and thus do well at excluding patients with DLB. Positive predictive value of 75% can be achieved by a combination of any three major or minor criteria, providing the analysis is confined to patients with mild to moderate dementia. Criteria were most accurate if confined to patients with pure DLB who had mild to moderate dementia.