Activity Number:
|
486
- Missing Data Analysis
|
Type:
|
Contributed
|
Date/Time:
|
Thursday, August 6, 2020 : 10:00 AM to 2:00 PM
|
Sponsor:
|
Biometrics Section
|
Abstract #313802
|
|
Title:
|
A Comparison of Imputation Methods for Missing Multi-Item Scales in Mental Health Settings
|
Author(s):
|
Panteha Hayati Rezvan* and Thomas R. Belin
|
Companies:
|
University of California Los Angeles and UCLA
|
Keywords:
|
Missing data;
Multi-item scale;
Multiple imputation;
Fully conditional specification;
GAD-7;
PHQ-9
|
Abstract:
|
The 7-item Generalized Anxiety Disorder scale (GAD-7) and the 9-item Patient Health Questionnaire (PHQ-9) are widely used to measure severity of anxiety and depression symptoms. Item-level missingness is not uncommon in GAD-7 and PHQ-9, leading to missing scale scores. Previous studies have shown that a preferred multiple imputation (MI) strategy is to incorporate items as auxiliary variables in item-level imputation models. Given large imputation models may lead to convergence problems various strategies have been proposed to stabilize predictors. Yet such methods have not been widely adopted in practice; users often exclude subjects with any missing items or use ad-hoc methods such as person’s mean imputation. Drawing on experience from applied research, we envision a scenario where GAD-7 and PHQ-9 are candidate predictors of a health outcome along with demographic variables and risk factors. Using MI strategies with fully conditional specification we investigate four variations on item-level imputation that differ in the way they make use of composites of groups of variables. The statistical properties will be compared with a complete case analysis and person’s mean imputation.
|
Authors who are presenting talks have a * after their name.