A study evaluating measurement invariance in patient-reported outcome measures has demonstrated that patient demographic and clinical characteristics are associated with measurement discrepancies in the 7-item Seattle Angina Questionnaire among individuals with coronary artery disease.
Establishing evidence of measurement invariance is essential to confirm that scores derived from patient-reported outcome measures are comparable across distinct groups or across time. Traditional assessment methods are predominantly group-based approaches that require potential population differences and violations of invariance to be specified a priori.
To address scenarios where covariates related to invariance violations are not known beforehand, researchers investigated the implementation of a tree-based structural equation model, known as SEMTree, which utilizes recursive partitioning to evaluate measurement invariance in patient-reported instruments.
The investigation utilized clinical data from 4,027 patients with coronary artery disease who completed the 7-item Seattle Angina Questionnaire after undergoing a cardiac angiogram procedure. Within this study cohort, the median age was 64.0 years with an interquartile range of 15.1 years, and 3,172 participants, representing 78.8 percent of the sample, were male.
Structural equation modeling demonstrated that a 3-factor measurement model for the 7-item questionnaire achieved an acceptable fit to the observed data. The SEMTree recursive partitioning analysis then tested demographic and disease or comorbid characteristics as explanatory splitting variables to detect patient subgroups showing non-invariance.
The resulting SEMTree analysis indicated that the measurement model was not invariant across subgroups partitioned by patient age, disease indication—specifically distinguishing acute coronary syndrome from stable angina—sex, and comorbid hypertension. The authors noted that future research will incorporate computer simulations to assess the statistical power of SEMTree to detect measurement invariance under varied data distributions and design characteristics.
Subject of Research: Assessment of measurement invariance in patient-reported outcome measures using tree-based structural equation models
Article Title: Tree-based structural equation model for assessing measurement invariance in patient-reported outcome measures
Article References: Sajobi, T. T., Arimoro, O. I., Mayo, N. E., Sawatzky, R., Nielsen, L. K., Sebille, V., Liu, J., Bohm, E., Awosoga, O. A., Norris, C. M., Wilton, S. B., James, M. T., & Lix, L. M. (2026). Tree-based structural equation model for assessing measurement invariance in patient-reported outcome measures. Health and Quality of Life Outcomes. https://doi.org/10.1186/s12955-026-02583-z
Image Credits: AI Generated
DOI: 10.1186/s12955-026-02583-z
Keywords: measurement invariance, patient-reported outcome measures, structural equation modeling, SEMTree, Seattle Angina Questionnaire, coronary artery disease, recursive partitioning, latent variable models, cardiac angiogram, acute coronary syndrome, stable angina, hypertension, factorial validity, splitting variables, computer simulations, subgroup analysis
News Source: Ophelia Keating. (November 21, 2026). Tree-Based Models Identify Measurement Invariance Violations in Cardiac Outcome Questionnaires. Scienmag.



