Premium
Putting the Alzheimer's cognitive test to the test II: Rasch Measurement Theory
Author(s) -
Hobart Jeremy,
Cano Stefan,
Posner Holly,
Selnes Ola,
Stern Yaakov,
Thomas Ronald,
Zajicek John
Publication year - 2013
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1016/j.jalz.2012.08.006
Subject(s) - rasch model , item response theory , psychology , cognition , scale (ratio) , scrutiny , psychometrics , clinical psychology , developmental psychology , psychiatry , geography , cartography , political science , law
Background The Alzheimer's Disease Assessment Scale—Cognitive Behavior section (ADAS‐Cog) is the most widely used measure of cognitive performance in AD clinical trials. This key role has rightly brought its performance under increased scrutiny with recent research using traditional psychometric methods, questioning the ADAS‐Cog's ability to adequately measure early‐stage disease. However, given the limitations of traditional psychometric approaches, herein we use the more sophisticated Rasch Measurement Theory (RMT) methods to fully examine the strengths and weaknesses of the ADAS‐Cog, and identify potential paths toward its improvement. Methods We analyzed AD Neuroimaging Initiative (ADNI) ADAS‐Cog data (675 measurements across four time‐points over 2 years) from the AD participants. RMT analysis was undertaken to examine three broad areas: adequacy of scale‐to‐sample targeting; degree to which, taken together, the ADAS‐Cog items adequately perform as a measuring instrument; and how well the scale measured the subjects in the current sample. Results The 11 ADAS‐Cog components mapped‐out a measurement continuum, worked together adequately, and were stable across different time‐points and samples. However, the scale did not prove to be a good match to the patient sample supporting previous research. RMT analysis also identified problematic “gaps” and “bunching” of the components across the continuum. Conclusion Although the ADAS‐Cog has the building blocks of a good measurement instrument, this sophisticated analysis confirms limitations with potentially serious implications for clinical trials. Importantly, and unlike traditional psychometric methods, our RMT analysis has provided important clues aimed at solving the measurement problems of the ADAS‐Cog.