WASHINGTON (American Psychological Association) -- Diagnosticians would do well to raise the bar when testing high-functioning people for pre-clinical signs of Alzheimer's disease, according to a new study. Higher test cutoffs, rather than the standard group average, more accurately predicted how many highly intelligent people would deteriorate over time. This finding is reported in the January issue of Neuropsychology, which is published by the American Psychological Association (APA).

Early diagnosis of Alzheimer's has taken on growing importance, given new medical and psychological interventions that can slow (but not stop) the course of the disease. In addition, highly intelligent people have been found, on average, to show clinical signs of Alzheimer's later than the general population. Once they do, they decline much faster. Thought to reflect their greater mental reserves, this different pattern may call for a different approach to diagnosis.

Says lead author Dorene Rentz, PsyD, "Highly intelligent elders are often told their memory changes are typical of normal aging when they are not. As a result, they would miss the advantages of disease-modifying medications when they become available."

Rentz, of Brigham and Women's Hospital's Department of Neurology and Harvard Medical School, led her co-authors in a study of 42 older people with IQ's of 120 or more, drawn from a longitudinal study of aging and Alzheimer's Disease. Rentz and her colleagues analyzed participant performance on measures of cognitive ability, such as word generation, memory and visuospatial processing, with scores gathered at the start of the study and three-and-a-half years later. Then, they asked which of two different test norms forecast problems: The standard norm, derived from a large cross-section of the population, or an adjusted high-IQ norm that measured changes against the individual's higher ability level?

The raised cutoffs worked better. For memory scores obtained at baseline, raised cutoffs predicted that 11 of the 42 individuals were at risk for future decline -- compared with standard cutoffs, which indicated they were normal. But, what's normal for people of average IQ isn't normal for people of higher IQ: True to the former prediction, three and a half years later, nine of those 11 people had declined. Six of those went on to develop mild cognitive impairment (MCI), a transitional illness from normal aging to a dementia (of which one type is Alzheimer's). Five of these individuals have since received a diagnosis of Alzheimer's disease, two years after this study was submitted.

"With standard norms, people with higher levels of ability would be classified as normal for up to three years before they began demonstrating a decline on standardized tests," says Rentz. "In this case, they could be at risk for not receiving early treatment intervention."

The statistical approach was simple. High-IQ people were scored against an average that was "normal" for them, proportionately higher than the cross-sectional average of 100. Scores were considered abnormal if, according to standard practice, they were 1.5 standard deviations or more below the (adjusted) norm.

Rentz and her co-authors believe that by the same reasoning, adjusted norms could help people at the other end of the scale. "People with below-average intelligence have the potential for being misdiagnosed as 'demented' when they are not, because they score below the normative cutoffs," explains Rentz. Adjusting norms to match the ability level of the person being evaluated, she believes, could have the greatest predictive power in detecting the early signs of an incipient Alzheimer's-type dementia.

Rentz points out that for women, adjusting for IQ may be particularly useful compared with traditional adjustments for education. "Education-based methods often underestimate ability in women who did not have the same educational advantages as men, particularly in this aged cohort," she says.

The authors also report that IQ-adjusted norms might help to control for some inaccuracies that have filtered into normative data, as these norms were derived from cross-sectional populations who might not have been adequately screened for preclinical Alzheimer's disease. Because it could take a long time to develop new databases from longitudinal studies, the authors say that the "use of an IQ-adjusted method may provide a temporary solution for clinicians and research investigators evaluating older highly intelligent individuals at risk for Alzheimer's disease."