Abstract / Summary
BackgroundDigital diabetes services are expanding rapidly in China, and older patients’ uptake of them depends on their eHealth literacy. This construct is usually reported as a single mean score, which conceals differences between patients sharing the same total. We aimed to identify latent profiles of eHealth literacy in Chinese older adults with type 2 diabetes and the characteristics associated with profile membership.MethodsA cross-sectional survey was conducted in the endocrinology departments of two tertiary hospitals in Jinzhou, Liaoning Province, between May and June 2026. Overall, 544 patients older adult(s) 60 years and over participated (mean age 71.4 ± 6.74 years, 52.21% female). Latent profile analysis was performed on the eight item-level scores of the Chinese eHealth Literacy Scale. Predictors for multinomial logistic regression were specified a priori from the Lily Model, van Dijk’s resources-and-appropriation theory and social cognitive theory, and retained regardless of univariate p-values. Multicollinearity was acceptable (VIF 1.10–1.69) and coefficients were checked against 1,000 bootstrap resamples.ResultsA three-profile solution fitted best (entropy 0.895): Skill-Dependent (n = 309, 56.8%), Comprehensive Deficit (n = 82, 15.1%) and Comprehensive Advancement (n = 153, 28.1%), whose mean total scores (26.88, 17.76 and 34.72) diverged despite a sample mean of 27.71 ± 5.93. With Comprehensive Advancement as reference, older age (OR 8.50 per 10 years, 95% CI 3.45–20.94; OR 3.56, 95% CI 1.76–7.21) and greater fear of disease progression (OR 4.60 per 5 points, 95% CI 3.07–6.87; OR 1.97, 95% CI 1.46–2.67) were associated with Skill-Dependent and Comprehensive Deficit membership, respectively. Higher diabetes self-efficacy, education of high school or above, smartphone use and longer daily internet time were associated with lower odds of both lower-literacy profiles (all p < 0.001). Complications were associated with Skill-Dependent membership only (OR 6.85, 95% CI 2.04–22.95, p = 0.002).ConclusioneHealth literacy was distributed across three qualitatively distinct profiles, so a single mean score describes this population incompletely. Age, educational attainment, device access, daily internet time, diabetes self-efficacy and fear of disease progression were associated with profile membership. Being cross-sectional, these findings describe associations rather than causal effects; profile-tailored education needs prospective evaluation.