There is growing interest in identifying Parkinson's disease (PD) patients with mild cognitive impairment (PD-MCI), but widely disparate criteria have been used. We assessed 143 PD patients and 50 matched controls on 20 measures across 4 cognitive domains (executive function, attention and working memory, learning and memory, visuoperception). Twenty-four patients met criteria for dementia (PD-D); nondementia patients were classified as either with normal cognition or MCI for 12 neuropsychological criteria. We compared the influence of these criteria on the distribution of global cognitive performance in the resulting PD-MCI groups relative to the control and PD-D groups. Different criteria produced substantial variation in the proportion of PD-MCI cases identified. Fourteen percent PD-MCI was found when using 2 scores in 1 domain at 2 standard deviations (SD) below normative scores, with no controls identified as MCI, through to 89% PD-MCI with 1 score in 1 domain at 1 SD below normative scores, when 70% of controls were identified as MCI. The balance of cases with impaired cognition but not those with generally intact cognition was better served by using criteria that required 2 specific deficit scores or deficits across 2 domains. As comparisons with external normative data may have greater applicability across centers, we suggest that 2 scores at -1.5 SD within any single domain (30% PD-MCI) or 1 score at -1.5 SD in each of 2 domains (37% PD-MCI) provide suitable criteria to minimize the inclusion of cognitively well patients. Clinical dementia rating did not improve the relative identification of cognitively impaired and unimpaired nondementia PD patients.
Read full abstract