PSYCHOMETRIC EVALUATION OF A NOVEL PATIENT-PROVIDER TOOL FOR ASSESSING DIABETES SELF-MANAGEMENT KNOWLEDGE AND BEHAVIOR IN ADULTS WITH TYPE 2 DIABETES MELLITUS
Author(s)
Arman Arabshomali, PharmD1, Minsoo Kang, PhD2, Meagen Rosenthal, PhD3;
1University of Mississippi, Department of Pharmacy Administration, PhD Candidate, Oxford, MS, USA, 2University of Mississippi, Department of Health, Exercise Science, and Recreation Management, University, MS, USA, 3University of Mississippi, Department of Pharmacy Administration, University, MS, USA
1University of Mississippi, Department of Pharmacy Administration, PhD Candidate, Oxford, MS, USA, 2University of Mississippi, Department of Health, Exercise Science, and Recreation Management, University, MS, USA, 3University of Mississippi, Department of Pharmacy Administration, University, MS, USA
Presentation Documents
OBJECTIVES: Effective Type 2 Diabetes Mellitus (T2DM) management requires robust self-management, yet existing assessment tools have limitations for guiding personalized Diabetes Self-Management Education and Support (DSMES). A novel 55-item tool, DiSMET (Diabetes Self-Management Evaluation Tool), assessing T2DM knowledge and behavior across eight DSMES domains was previously developed and validated using Classical Test Theory. This study aims to conduct a psychometric evaluation of DiSMET using Rasch measurement theory.
METHODS: A secondary analysis was conducted using cross-sectional survey data from 306 U.S. adults with T2DM recruited via Amazon MTurk. Rasch dichotomous analysis was performed separately for knowledge and behavior domains. Model-data fit was assessed using Infit and Outfit mean square statistics (acceptable range 0.5-1.5). Unidimensionality was evaluated using principal component analysis of residuals, and local independence was examined through standardized residual correlations. Item-person maps were used to assess targeting, and item and person separation indices and reliabilities were calculated. Differential item functioning (DIF) by gender was assessed using the Mantel-Haenszel method.
RESULTS: Five knowledge items and one behavior item were removed based on misfit and expert content review. The remaining 24 knowledge items and 25 behavior items demonstrated good model fit, supported unidimensionality, and met assumptions of local independence. No items exhibited meaningful gender-based DIF. Item separation reliability was excellent for both domains (knowledge = 0.97, behavior = 0.97), indicating stable item hierarchies. However, person separation reliability was moderate (knowledge = 0.70, behavior = 0.71), suggesting limited precision in distinguishing participants’ levels of ability. Item-person maps revealed notable ceiling effects in both domains, with items generally easier than participants’ ability levels.
CONCLUSIONS: Rasch analysis confirms strong psychometric properties of refined DiSMET and supports its potential for assessing T2DM knowledge and behavior. However, limited item targeting and moderate person separation suggest further refinement is needed, including adding harder items and testing in more targeted, representative samples.
METHODS: A secondary analysis was conducted using cross-sectional survey data from 306 U.S. adults with T2DM recruited via Amazon MTurk. Rasch dichotomous analysis was performed separately for knowledge and behavior domains. Model-data fit was assessed using Infit and Outfit mean square statistics (acceptable range 0.5-1.5). Unidimensionality was evaluated using principal component analysis of residuals, and local independence was examined through standardized residual correlations. Item-person maps were used to assess targeting, and item and person separation indices and reliabilities were calculated. Differential item functioning (DIF) by gender was assessed using the Mantel-Haenszel method.
RESULTS: Five knowledge items and one behavior item were removed based on misfit and expert content review. The remaining 24 knowledge items and 25 behavior items demonstrated good model fit, supported unidimensionality, and met assumptions of local independence. No items exhibited meaningful gender-based DIF. Item separation reliability was excellent for both domains (knowledge = 0.97, behavior = 0.97), indicating stable item hierarchies. However, person separation reliability was moderate (knowledge = 0.70, behavior = 0.71), suggesting limited precision in distinguishing participants’ levels of ability. Item-person maps revealed notable ceiling effects in both domains, with items generally easier than participants’ ability levels.
CONCLUSIONS: Rasch analysis confirms strong psychometric properties of refined DiSMET and supports its potential for assessing T2DM knowledge and behavior. However, limited item targeting and moderate person separation suggest further refinement is needed, including adding harder items and testing in more targeted, representative samples.
Conference/Value in Health Info
2026-05, ISPOR 2026, Philadelphia, PA, USA
Value in Health, Volume 29, Issue S6
Code
PCR44
Topic
Patient-Centered Research
Topic Subcategory
Instrument Development, Validation, & Translation, Patient-reported Outcomes & Quality of Life Outcomes
Disease
SDC: Diabetes/Endocrine/Metabolic Disorders (including obesity)