Using a mathematical model originally developed overseas to measure achievement test results, three Illinois researchers have put together a "yardstick" they say could help doctors trying to diagnose Alzheimer’s disease.

"Alzheimer’s disease affects nearly 4 million people in this country, but it can be diagnosed only by ruling out all other possible causes for the dementia," says A. Kyle Perkins, an SIUC professor of linguistics who specializes in language testing. "We wanted to produce something that would be valuable to a general practitioner or Alzheimer’s specialist in determining more quickly whether a particular patient has Alzheimer’s."

To do that, Perkins, along with Larry Hughes, a research professor at SIU’s School of Medicine in Springfield, and Benjamin Wright, founder of the University of Chicago’s Measurement Evaluation Statistical Analysis Psychometric Laboratory, turned to a Rasch analysis computer program. Rasch analysis is a measurement tool that makes it possible to show information about both people and things on the same scale.

In building their yardstick, the researchers used data from 600 Illinois patients who had visited the SIU School of Medicine’s Center for Alzheimer Disease and Related Disorders. Because these patients have already been diagnosed—a third of them probably have Alzheimer’s, a third of them might have it, and a third of them do not—the researchers were able to use those diagnoses to doublecheck their yardstick’s accuracy.

The computer first combined information from the patients’ medical histories and mental ability tests with facts about their personal attributes, such as their ages and genders. From there, it produced an Alzheimer’s scale—basically, a mathematical depiction of the odds of having Alzheimer’s.

"No measure is ever gold-standard certain," says Wright, a key figure in the American school of Rasch analysis. "A measure implies a region of uncertainty." The Alzheimer’s scale consists of a line of interval measures ranging from zero to 100. Zero represents the complete absence of Alzheimer’s disease, while 100 stands for its presence full-blown.

"Based on the mathematics, I might choose 36 and 50 as the edges of my diagnosis," Wright says. "With a measure below 36, I would hazard that Alzheimer’s is unlikely, while with a measure above 50, I would hazard it likely."

The computer analysis also showed that three common tests of mental functioning worked best as indicators of the disease. The Activities of Daily Living scale, commonly abbreviated as ADL, assesses memory and performance of everyday activities. The Short Blessed Dementia scale assesses concentration and recall. The Mini-Mental State Exam, or MMSE, assesses memory and orientation.

"Overall, the ADL seems to be the best predictor," Perkins says. "If a patient has an ADL score of 5 or above, there’s a high probability of Alzheimer’s. A score of 16 or higher on the Short Blessed is associated with high probability, as is an MMSE score of 20 or less. 

"If a doctor has a patient with one or two of these scores, he or she could look on the (Alzheimer’s) scale and very quickly see where the patient falls."

Hughes stresses that the Rasch measurement model he and his colleagues produced does not by itself predict or diagnose the disease. "It helps us focus the existing information a little more thoroughly," he says. "You have to sort through a lot of data in determining what’s Alzheimer’s disease and what’s not. That requires a judgment on the part of the physician and a fair amount of data evaluation.

"That’s where I see this making a contribution, in simplifying data-handling procedures and giving us something more easily understood by the physician. It’s one more tool in the arsenal for them to solve one more piece of the puzzle."
 

Fall 2001 Contents | Perspectives Home | SIUC Home

Comments: Perspectives Webmaster
Copyright © 2001, Board of Trustees, Southern Illinois University | Privacy Policy