# Diagnostic considerations in Alzheimer disease. Predictive value of intrusion errors as a differential diagnostic sign of Alzheimer disease

код для вставкиСкачатьLETTERS Diagnostic Considerations in Alzheimer Disease as defined for this study, should therefore be helpful in differentiating Alt from other dementias” should be considered tenuous until confirmed by further data from large samples. Predictive Value of Intrusion Errors As a Differential Diagnostic Sign of Alzheimer Disease Institute for Psychopharmacologic Research 1842 Beacon St Brookline. M A 02146 Roland J. Branconnier, MA I read with interest the paper by Drs Fuld et a1 entitled “Intrusions As a Sign of Alzheimer Dementia: Chemical and Pathological Verification” [2].The authors have drawn the conclusion that intrusion errors are a useful differential diagnostic sign for Alzheimer dementia (AD) because those errors have significant discriminative validity. However, the establishment of discriminative validity of a diagnostic test under experimental conditions is not sufficient to predict the usefulness of the test when applied to an unselected population [5j. This is because the predictive value of a positive test result is a function of the prevalence (a priori probability) of a disease in the population [31. Concerning intrusion errors, the critical question to be answered is: What is the a posteriori probability that a demented patient who makes such errors does, in fact, have AD? Such a posteriori probabilities or predictive values cannot be determined from experimental data alone but require the application of predictive value analysis, a series of statistical methods based on Bayes’ theorem [ l , 31. From the data presented in the paper it is possible to evaluate the sensitivity, specificity, and predictive value of intrusion errors as a differential diagnostic sign of AD. Sensitivity can be defined as the percentage of positive results obtained when the test is administered to persons who are known to have a disease. In this case, 19 of 21 patients with a diagnosis of AD made intrusion errors. Thus, the sensitivity of that sign is 19/21, or 90%. Specificity can be defined as the percentage of negative test results obtained from persons known to be free of a disease. Eleven of 17 patients with other dementias did not make intrusion errors. Thus, the specificity is 11/17, or 65%. Epidemiological data reveal that 60% of all cases of dementia are due to A D [ 4 ] .Since the a priori probability that any demented patient has A D is 60%, and because intrusion errors have a sensitivity of 90% and specificity of 65%, by application of Bayes’ theorem the a posteriori probability that a demented patient who makes an intrusion error has A D is 79%. While this predictive value may appear to be satisfactory, it can be shown that if the intrusion error test is replaced by flipping a coin (a coin has a sensitivity equal to 50% and specificity of 50%), and a “head” indicates AD, then the a posteriori probability that “heads” is a sign of A D is 60%. Since the gain in differential diagnostic accuracy of the intrusion error test is only 19% over random selection, it can be concluded that a substantial portion of the observed differential diagnostic accuracy of that test for A D results from the 60% prevalence rate of A D among dementias. Moreover, since only 38 patients were studied, the 75% confidence limits of the sensitivity and specificity of the test can be shown to be 77 to 100% and 53 to 7796, respectively. Therefore, the authors’ conclusion that “intrusions, RejerenceJ 1. Bayes T An essay toward solving a problem in the doctrine of chance. Phil Trans Royal Soc 53:370-418, 1763 2. Fuld PA, Katzman R, Davies P, Terry RD: Intrusions as a sign of Alzheimer dementia: chemical and pathological verification. Ann Neurol 11:155-159, 1982 3. Galen RS, Gambino SR: Beyond Normality: The Predictive Value and Efficiency of Medical Diagnoses. New York, Wiley, 1975 4. Mortirner JA, Schuman LM, French LR: Epidemiology of dementing illness. In Mortimer JA, Schuman LM (eds): The Epidemiology of Dementia. New York, Oxford University Press, 1981 5. Vecchio TJ: Predictive value of a single diagnostic test in unselected populations. N Engl J Med 274:1171-1173, 1966 Neuropathological Basis for Diagnosis of Alzheimer’s Disease Meta A. Neumann, MS,’ and Robert Cohn, M D t With regard to the article on intrusions as a sign of Alzheimer’s disease by Fuld et al, we were surprised that these eminent authors considered the presence of plaques in only three cortical areas sufficient to establish the diagnosis of AD. The presence of plaques was alluded to three times in the article, while neurofibrillary changes were never mentioned. The thioflavin S method that was used clearly depicts diseased neurons as well as plaques. In more than 25 years’ experience with the study of brains of all autopsied cases at a large mental hospital (Saint Elizabeths Hospital, Washington, DC) together with a review we conducted of the clinical histories of these patients, we observed that patients with argyrophilic plaques alone did not show the specific clinical correlates of AD [2, 31, although memory loss, usually for recent events only, was apparent. Since “intrusions” as defined by the authors must occur in conditions other than AD, and even in normal subjects, the test obviously is not specific. Having the patient perform arithmetical operations, draw various maps, or carry out simple geometrical problems generates a clearer and less equivocal index of intellectual deterioration when repeated at intervals of several months. It must be emphasized that no clinical test can distinguish between AD and other cerebral degenerative processes such as Pick’s disease or progressive subcortical gliosis [4]. In these insidious degenerative diseases, the age of onset can hardly be established; moreover, A D as a specific clinical-neuropathological entity is the same process, irrespective of the age at onset or age at death. ‘7221 Pyle Rd Bethesda, M D 2081 7 tHoward University College of Medicine Washington, DC 20060 317

1/--страниц