HAI Book 2025 - Flipbook - Page 564
Sai, Isha
127
Impacts of testing language and rates of amyloid positivity among
underrepresented groups in the A4 Study
Isha Sai1, Joshua D. Grill2, Elizabeth C. Mormino1,3, Christina B Young1
1
Department of Neurology and Neurological Sciences, Stanford University, Stanford, CA, US
Institute for Memory Impairments and Neurological Disorders, Departments of Psychiatry and Human Behavior and
Neurobiology and Behavior, University of California, Irvine, Irvine, CA, US
3
Wu Tsai Neurosciences Institute, Stanford University, Stanford, Stanford, CA, US
2
Objective: To determine whether abnormal amyloid (A+) PET rates in cognitively unimpaired (CU) participants
screened for a secondary prevention trial differs across racial/ethnic/testing language groups and identify
potential screening biases related to cognition/functioning.
Methods: Data from 6669 participants screened in their primary language for the A4/LEARN Study were divided
into race/ethnicity/testing language groups (Table 1). Among individuals that passed initial screening and
completed the PET visit, differences in A+ rates between race/ethnicity/testing language groups were compared
with logistic regression. To probe potential biases in the screening process, we examined rates of screen-failure
due to medical/other reasons, cognition/functioning, and normal amyloid PET status (A-). In those who screenfailed due to low cognitive/functioning, we further examined differences between race/ethnicity/language
groups in screening and non-screening cognitive tests using logistic, multinomial logistic, and linear regression.
Results: Underrepresented groups [non-Hispanic Black tested in English (NHB-E), non-Hispanic Asian tested in
English (NHA-E), and non-Hispanic Asian tested in Japanese (NHA-J)] had significantly lower rates of A+
compared to the non-Hispanic White group tested in English (NHW-E; Figure 1A). Of all screened participants,
underrepresented groups were significantly more likely to be excluded for medical/other reasons and
cognitive/functioning performance prior to the PET visit (Figure 1B). Of those who screen-failed due to
cognitive/functional performance, the pattern of failure on the three screening tests differed across
race/ethnicity/language groups (Figure 1C). When probing specific MMSE and CDR items in those who were
screened out due to low cognition/functioning, patterns of missed items and endorsed functional domains
differed between race/ethnicity/language groups (Figure 1D-E). When examining non-screening cognitive tests,
performance was comparable between those who screen-failed due to low cognition/functioning and A+
individuals who passed initial screening within NHA-E and NHA-J groups (Figure 2).
Conclusion: Screening procedure biases may disproportionately reduce the number of participants from
underrepresented groups who are eligible for randomization.
HAI2025 - 564
