Evaluation of dementia: a systematic study of the usefulness of the American Academy of Neurology's practice parameters.

OBJECTIVE: This study examined the usefulness of the "Practice Parameters for the Evaluation of Dementia," published by the Quality Standards Subcommittee of the American Academy of Neurology (1994). The Practice Parameters are stratified according to three levels of certainty (standards, guidelines, and options), and suggest indications for the use of neuroimaging studies.

METHODS: We reviewed 119 consecutive cases referred for assessment of memory loss to a university-affiliated interdisciplinary clinic. We assessed the diagnostic value of laboratory, neuropsychological, and neuroimaging studies above the standard history, neurologic, and mental status examinations. We also assessed the sensitivity and specificity of four clinical indicators (i.e., early symptom onset, noninsidious course, focal neurologic signs or symptoms, and gait disturbance) for predicting the diagnostic utility of neuroimaging studies.

RESULTS: The largest changes in diagnostic categories between the standard and the comprehensive diagnostic process was a 9% decrease in the diagnosis of Alzheimer's disease, a 6% increase in the diagnosis of mixed dementia (due largely to laboratory studies), and a 4% increase in the diagnosis of vascular dementia (due to neuroimaging). The clinical indicators were 82% sensitive and 50% specific in predicting that neuroimaging studies would change the diagnosis. In six cases, meaningful neuroimaging findings were not associated with any clinical indicator (5% false negatives). In 43 cases, neuroimaging provided no significant clinical findings despite the presence of an indicator (36% false positives).

CONCLUSIONS: In this convenience sample, diagnostic accuracy was improved to a comparable degree by laboratory and neuroimaging studies, although at a significant difference in cost. Use of the four clinical indicators would have reduced the frequency of neuroimaging studies by 33% but missed clinically meaningful information in 5%. Although imperfect, the Practice Parameters represent a first step toward improving the cost effectiveness of the dementia work-up.