Latent profile analysis of eHealth literacy and its sociodemographic correlates: a cross-sectional study

Abstract

Objective This study used Latent Profile Analysis (LPA) to classify Chinese adults with access to digital health information into distinct eHealth literacy subgroups and depict their multidimensional ability traits. We further examined cross-sectional sociodemographic and health-related correlates of subgroup affiliation, generating descriptive data that may serve as a preliminary reference for future stratified digital health outreach research. Methods A cross-sectional online survey was conducted among Chinese residents between June and July 2023. The widely used Chinese eHealth literacy scale (eHEALS) measured three dimensions of digital health literacy via 1–5 Likert items. Optimal latent class number was determined using AIC, BIC, entropy and BLRT. Weighted multinomial logistic regression, adjusted by LPA posterior probabilities, was adopted to explore associated factors. LPA was performed in R 4.2.1, and regression analysis in Stata 17.0. Results In total, 1,391 valid questionnaires were collected for analysis. The five-class model was selected as optimal, identifying five eHealth literacy latent profiles: Low eHealth Literacy (11.29%), Moderate-Low Balanced eHealth Literacy (18.26%), Moderate-High Balanced eHealth Literacy (30.91%), High Balanced eHealth Literacy (29.69%), and Application-Preferred Moderate-High eHealth Literacy (9.85%). Regression results revealed subgroup-specific correlational patterns. Higher educational attainment was positively correlated with membership in the Application-Preferred Moderate-High eHealth Literacy profile (OR range: 3.774–4.552, all p < 0.01), while advancing age was negatively associated with membership in this profile (OR = 0.936, p < 0.001). Participants with chronic diseases or commercial medical insurance were more likely to be categorized into the Application-Preferred Moderate-High eHealth Literacy subgroup (OR = 3.408, p < 0.001). Urban residence was negatively associated with the Application-Preferred Moderate-High profile (OR = 0.579, p = 0.044), while non-employed status was negatively associated with the Moderate-High Balanced profile (OR = 0.535, p = 0.007). Conclusion Different eHealth literacy latent profiles exhibited unique cross-sectional correlational patterns with multiple sociodemographic and health characteristics among digitally accessible Chinese adults. Educational attainment consistently showed positive correlational links with higher self-reported digital health capacity, particularly for the Application-Preferred Moderate-High profile, while age and non-employment showed divergent correlations across subgroups. Household income demonstrated a profile-specific correlational pattern: lower household income was only associated with increased odds of belonging to the Application-Preferred Moderate-High subgroup versus the Low eHealth Literacy reference group.