About this item:

551 Views | 0 Downloads

Author Notes:

Dr. Kyle Steenland, Emory University, Rollins School of Public Health, 1518 Clifton Road, Atlanta, GA 30322 (USA), Tel. +1 404 712 8277, E-Mail nsteenl@sph.emory.edu

Subjects:

Research Funding:

This work was supported by an NIH-NIA Center Grant for the Emory Alzheimer's Disease Research Center (P50AG025688).

Keywords:

  • Alzheimer's disease
  • Mild cognitive impairment
  • Cognition

Analyses of Diagnostic Patterns at 30 Alzheimer's Disease Centers in the US

Tools:

Journal Title:

Neuroepidemiology

Volume:

Volume 35, Number 1

Publisher:

, Pages 19-27

Type of Work:

Article | Post-print: After Peer Review

Abstract:

Background The US Alzheimer's Disease Centers (ADCs) (n = 30) recently created a uniform data set. We sought to determine which variables were most important in making a diagnosis, and how these differed across ADCs. Methods A cross-sectional analysis of first visits to ADCs via polytomous logistic regression. We analyzed subjects with complete data (n = 7,555, 89%), and also used multiple imputation to infer missing data. Results There were 8,495 subjects; 50, 26, and 24% were diagnosed as normal, having mild cognitive impairment (MCI), or mild Alzheimer's disease [Clinical Dementia Rating (CDR) score <1], respectively. The model using 7,555 subjects was 86% accurate in predicting diagnosis. Important predictors were physician-reported decline and the CDR sum of boxes, followed by 4 cognitive tests (Mini Mental State Examination, Category Fluency Tests, Logical Memory Test, Boston Naming Test). Multiple imputation revealed Trail Making Test B to be additionally important. Consensus versus single-clinician diagnoses were 2–3 times more likely to result in MCI than normal diagnoses. Excluding clinical judgment variables, functional assessment and psychiatric symptoms were important additional predictors; model accuracy remained high (78%). There were significant differences between centers in the use of different cognitive tests in making diagnoses. Conclusions We recommend creating a hypothetic data set to use across ADCs to improve diagnostic consistency, and a survey on the use of raw or adjusted cognitive test scores by different ADCs.

Copyright information:

© 2010 by S. Karger AG, Basel

Export to EndNote