Comprehensive Demographic Correction Improves Sensitivity and Reduces Bias in Cognitive Assessment

Area of Science:

• Neuropsychology
• Psychometrics
• Biostatistics

Background:

• Standard neuropsychological assessments use age, education, and gender (AEG) corrections.
• Demographic factors like race/ethnicity and crystallized ability also impact test performance.
• AEG corrections can lead to systematic misclassification of cognitive impairment.

Purpose of the Study:

• To develop and validate a Comprehensive (C-) model scoring algorithm for neuropsychological tests.
• To improve the accuracy of cognitive assessment by including additional demographic predictors.
• To reduce demographic disparities in the classification of cognitive impairment.

Main Methods:

• Developed a C-model incorporating AEG plus vocabulary, age², race/ethnicity, SES, and medication use.
• Utilized stability-selection LASSO to identify robust predictors in 1,914 adults assessed with the California Cognitive Assessment Battery (CCAB).
• Validated the C-model's generalizability in two distinct cohorts using cross-sample frozen-coefficient methods.

Main Results:

• The C-model approximately doubled explained variance compared to the AEG model (r²=0.50 vs 0.25).
• Racial disparities in mild cognitive impairment (MCI) classification were substantially reduced (Black-vs-White ratio from 5.6 to 1.8).
• Sensitivity improved for high-functioning individuals, reducing MCI classification ratios in low-vs-high vocabulary quartiles (from 11.3 to 2.1).

Conclusions:

• The C-model offers a more accurate and equitable approach to neuropsychological scoring.
• It significantly reduces racial bias and enhances detection of cognitive decline in diverse populations.
• Parallel use of C- and AEG-models can increase diagnostic confidence and provide additional clinical information.