Improving identification of cognitive impairment in primary care.

Objectives: To compare the relative level and predictors of accuracy of a brief cognitive screen, the Mini-Cog, with spontaneous detection of cognitive impairment by subjects' primary care physicians.

Participants: A heterogeneous community sample (n=371) of predominantly ethnic minority elderly assessed by standardized research protocol, 231 of whom met criteria for dementia or mild cognitive impairment (MCI).

Results: The Mini-Cog detected cognitively impaired subjects much more effectively than did subject's own physicians (p<0.0001), correctly classifying 83% of the sample and 84% of cognitively impaired subjects. Physicians correctly classified 59% of all subjects but identified only 41% of cognitively impaired subjects. The Mini-Cog's advantage over physicians was greatest when impairment was mildest (screen vs physician recognition at CDR 0.5, 58% vs 6%; at CDR 1, 92% vs 41%). Additional subject variables associated with missed detection by physicians were non-Alzheimer type dementia and low education, low literacy, and non-English speaking, factors that had little or no effect on the performance of the Mini-Cog. Ethnic differences, also observed for physician recognition, were not significant in final regression equations. The number and recency of primary care visits, and duration of the primary care relationship, were not associated with physicians' recognition of cognitive impairment.

Conclusion: This study demonstrates that recognition of cognitive impairment by primary care physicians is adversely influenced by important patient and disease characteristics. Results also show that use of the Mini-Cog would improve recognition of cognitive impairment in primary care, particularly in milder stages and in older adults subject to disparities in health care quality due to sociodemographic factors.