(P61) Evaluating for Cognitive Assessment and Racial Disparities

Introduction/

Objective: More than 50% of nursing home residents experience some degree of cognitive impairment. The Brief Interview for Mental Status (BIMS) has been used as part of the Minimum Data Set 3.0 to screen nursing home residents for cognitive impairment. The Center for Medicare and Medicaid Services (CMS) mandated the BIMS to be used across all post-acute care settings starting in October 2010. Though used as the current standard of care, studies have shown that the BIMS is not sensitive in detecting milder forms of cognitive impairment, whereas the SLUMS has been shown to be comparable in reliability and validity to the Montreal Cognitive Assessment (MoCA). We also considered possible implicit bias regarding race in assessments of cognitive impairment between the MoCA and BIMS. We evaluated 37 geriatric patients across two nursing homes in North Carolina using the standard SLUMS, MoCA, and BIMS. Of these 37 research subjects, 18 identified as African American, and 19 identified as White.

Design/Methodology: We performed a literature review regarding the accuracy of SLUMS and BIMS compared to the MoCA. We compiled cognitive assessment results from various EMRs across two nursing home settings and performed several cognitive assessments in person. We then compared BIMS and SLUMS scores with MoCA scores for each patient and even compared score accuracy to reveal probable racial disparities. Given that each cognitive assessment has their own individual categories and score ranges, and BIMS does not have a “mild cognitive impairment” category, we chose to mark BIMS as assessing for mild cognitive impairment if the patient’s assessment score was between 13 and 14, with 15 being a perfect “normal” cognitive assessment score. Additionally, we combined “moderate” and “severe” score ranges from the MoCA as a comparison to the “dementia” score range of the SLUMS.

Results: Our data analysis demonstrated that BIMS was inaccurate in assessing cognitive impairment 81% of the time when using MoCA as a gold standard, whereas SLUMS was only inaccurate 27% of the time. We also found that BIMS was inaccurate in assessing mild cognitive impairment 100% of the time while SLUMS was only inaccurate 50% of the time. Finally, when scrutinizing for racial bias, it does not initially appear that the BIMS is biased towards predicting a more severe level of cognitive impairment in patients who identify as African American (p=0.858). Yet, when looking at “moderate” compared to “severe” categories, patients who are AA were more likely to score as severely cognitively impaired (70%) on the BIMS cognitive assessment compared to patients who are white (20%, P=0.02). The SLUMS also revealed racial bias, showing patients who are African American were more likely to score as moderate/severe cognitively impaired (89%) on the SLUMS compared to patients who are white (58%, p=0.053).

Conclusion/

Discussion: Though limited by a small population size, our findings suggest significant deficits in the BIMS as a tool for assessing cognitive impairment. Additionally, our data suggests that both the BIMS and the SLUMS are limited by racial biases. Although in need of improvements regarding racial disparities, our data indicates that SLUMS is superior to BIMS both in accuracy and in ability to detect mild cognitive impairment.