Education has a profound effect on older adults’ cognitive performance. In Hong Kong, some dementia screening tasks were originally designed for developed population with, on average, higher education.

We compared the screening power of these tasks for Chinese older adults with different levels of education. Community-dwelling older adults who were healthy (N = 383) and with very mild dementia (N = 405) performed the following tasks: Mini-Mental State Examination, Alzheimer's Disease Assessment Scale-Cognitive subscales, Verbal Fluency, Abstract Thinking, and Visual/Digit Span. Logistic regression was used to examine the power of these tasks to predict Clinical Dementia Rating (CDR 0.5 vs. 0).

Logistic regression analysis showed that while the screening power of the total scores in all tasks was similar for high and low education groups, there were education biases in some items of these tasks.

The differential screening power in high and low education groups was not identical across items in some tasks. Thus, in cognitive assessments, we should exercise great caution when using these potentially biased items for older adults with limited education.