Approaches to Harmonizing Two Functional Assessment Instruments in a Longitudinal Data Set of Aging Adults in the Unites States Open Access

Holloway, JaNae (Spring 2021)

Permanent URL:


Background: Declining functional ability in aging adults can be an early predictor of cognitive impairment. Individuals with mild cognitive impairment (MCI) are at a higher risk of developing dementia, which severely effects memory, language, problem solving skills, and other thinking abilities. It is essential to have a valid and reliable tool to measure functional ability. While there are multiple tools available to clinicians, there is no standardized harmonization to measure the trajectory of functional impairment over time using all available patient data.

Methods: We compared the use of linear regression and item response theory (IRT) in harmonizing two functional ability instruments, the Functional Assessment Questionnaire (FAQ) and the Modified ADL/IADL (MAIADL) Scale, for n=179 patients in the Emory Cognitive Neurology Data Set (NeuCog). We evaluated agreement between the two measurements by calculating the concordance correlation coefficients (CCC) and evaluated model performance by calculating R^2 and root mean square error (RMSE). Lastly, we conducted IRT on FAQ scores from n=24,038 participants in the National Alzheimer's Coordinating Center (NACC) Uniform Data Set (UDS) and calculated new weighted FAQ scores. We performed survival analysis to clinically evaluate these new scores.

Results: We found that the agreement was best between factor scores (CCC=0.8585) from our IRT model compared to original scores (CCC=0.7320) and weighted scores (0.7589). Additionally, the RMSE was lowest for the linear regression model using factor scores (6.96) compared to original (17.51) and weighted (17.37) scores, however R^2 was highest for the model using weighted scores (0.7983). After controlling for demographic and clinical covariates, we found that FAQ significantly contributed to both time to death (HR=0.99, 95% CI: 0.99, 0.99) and time to dementia (HR=0.98, 95% CI: 0.98, 0.98), however the new weighted FAQ did not provide additional information regarding risk for time to death or time to dementia compared to the original FAQ scores.

Conclusion: In this paper we demonstrated the usefulness of IRT in measuring functional ability compared to the standard method of using total FAQ and MAIADL Scale scores. We believe these results can be applied in a clinical setting to assist in diagnosing patients with MCI or dementia.

Table of Contents

Introduction Materials and Methods Data Sources Neurology-Cognitive Data Set Uniform Data Set Instrumental Activities of Daily Living Scales Harmonization Analyses Linear Regression Item Response Theory Survival Anlysis Results Patient Demographics Linear Regression Item Response Theory Comparison of Harmonization Techniques Clinical Evaluation of Weighted FAQ Scores Discussion Conclusion



About this Master's Thesis

Rights statement
  • Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.
Subfield / Discipline
  • English
Research Field
Committee Chair / Thesis Advisor
Committee Members
Last modified

Primary PDF

Supplemental Files