New model may help detect dementia early

Researchers have developed a new statistical model that could help determine whether memory loss in older adults is benign or a stop on the way to Alzheimer's disease.
The model developed by researchers from Johns Hopkins University used scores obtained from cognitive tests.
The risk of developing dementia increases markedly when a person is diagnosed with mild cognitive impairment, a noticeable and measurable decline in intellectual abilities that does not seriously interfere with daily life.
But physicians have no reliable way to predict which people with mild cognitive impairment are likely to be in the 5 to 10 per cent a year who progress to dementia.

In the new study, researchers analysed records of 528 people age 60 and over, who were referred to the Johns Hopkins Medical Psychology Clinic for cognitive testing as part of a dementia work-up between 1996 and 2004.
The results were compared to those of 135 healthy older adults who participated in a study of normal ageing. Both groups completed tests of memory, language, attention, processing speed and drawing abilities from which 13 scores were recorded.

Since each person is naturally more skillful in some areas than in others, the scores of healthy adults showed a symmetrical, bell-shaped range: Most of their scores were high, a few were a bit lower, and a few were even lower.

By grouping the patients into cohorts based on the severity of their dementia, the researchers found a trend in the test scores that is likely to mimic the deterioration of an individual's scores over time.
At the outset, Alzheimer's disease subtly disrupts some mental abilities, while leaving others intact. Thus, well before a person develops clear cognitive impairment, his or her performance declines slightly on a few measures.
When shown on a graph, these changes cause the healthy symmetric, bell-shaped curve to shift and become asymmetrical.
Regardless of how low a person's test scores were, the researchers determined that lopsidedness in their score distribution correlated with dementia.

They predicted that people with low scores that were evenly distributed were not likely to develop dementia. But those with clearly lopsided test score distributions on the measures administered were already experiencing varying levels of dementia.
"Departures from the normal bell-shaped pattern of variability on cognitive tests might determine which people with low scores develop dementia," said David J Schretlen, a professor of psychiatry and behavioral sciences at the Johns Hopkins University School of Medicine.
The study was published in the journal Neuropsychology.