Abstract
BACKGROUND: Individuals with lower health literacy are more vulnerable to misinformation and may face challenges in disease prevention, treatment, and healthcare utilization6. Accordingly, accurately assessing the level of individuals’ health literacy and implementing educational and policy-based interventions to improve it are essential public health goals. This study assessed the psychometric properties of a health literacy assessment tool, using data from the 2023 Korea National Health and Nutrition Examination Survey (KNHANES). Specifically, the study tested measurement invariance across sociodemographic groups and identified latent health literacy subgroups using multiple-group latent class analysis (LCA). METHODS: Data from 5,677 adults aged 19 years and older who participated in the 2023 KNHANES were analyzed. A confirmatory factor analysis (CFA) was conducted to validate the factor structure of the tool, and measurement invariance was tested across gender, age, educational level, income, and spousal cohabitation using multiple-group analyses. In addition, an LCA was performed to classify subgroups based on health literacy levels and to identify item-level response differences. RESULTS: The CFA results supported a unidimensional structure with excellent model fit. The measurement invariance testing confirmed configural, metric, scalar, and strict invariance across all of the sociodemographic groups. The LCA identified a two-class model as the best fit, with a significant between-group difference in the response to the item assessing confidence in knowing what to do first in an emergency. CONCLUSION: The findings reaffirm the validity and reliability of the health literacy tool across diverse subpopulations. Moreover, the LCA proved useful in identifying subpopulations with limited health information utilization. These results provide a valuable empirical basis for the development of targeted health literacy interventions and public health policy planning. TRIAL REGISTRATION: Clinical trial number: not applicable.