Diagnosis of Alzheimer's disease in epidemiologic studies by staged review of clinical data

We explored the inter-rater agreement and validity of diagnoses of Alzheimer's disease (AD) and other dementias made in an epidemiological study. A previously described protocol for cognitive screening and clinical assessment was applied to a large registry of twins. An expert panel then reviewed results from the assessment of 41 subjects whose screening results suggested the presence of AD. After review of the information at each of four stages of data collection, we assessed inter-rater agreement among the experts as well as their individual agreement with the final consensus diagnosis. We investigated these measures to assess the amount and quality, respectively, of new and diagnostically useful information that was revealed at each stage. A new scheme of weighted differences among the available diagnostic categories was developed for these analyses. As expected, incremental information from successive stages of data collection enabled the panel to increase their diagnostic agreement and rates of 'correct' diagnoses. Over half of the total information was available, however, after review of only the initial telephone screening results (stage 1). A brief standardized videotape segment of the mental status and neurologic examinations provided substantial additional information. We were able to compare the final consensus diagnoses with autopsy results from seven individuals who had consensus clinical diagnoses of Probable or Possible AD (n = 6) or 'demented, questionable etiology' (n = 1). All these subjects had Definite AD.

Duke Scholars

Author Kathleen Anne Welsh-Bohmer Psychiatry & Behavioral Sciences, Behavioral Medicine & Neur ...

Author David Carl Steffens Psychiatry & Behavioral Sciences, Adult Psychiatry & Psychol ...

Author James Robert Burke Neurology, Behavioral Neurology