Abstract
BACKGROUND: Oral frailty (OF) is an intermediate and reversible state between oral health and oral hypofunction. Digital health interventions have great potentiality, but eHealth literacy plays the key role. This study aimed to explore the dose-response relationship between OF and eHealth literacy and evaluate the potential effect of improving eHealth literacy on OF. METHODS: A multicentric cross-sectional study design was applied. A total of 636 older adults were recruited by convenience sampling from 5 medical institutions in Changning District, Shanghai, China from July to November 2024. The eHealth Literacy Scale (eHEALS) and the OF Index-8 (OFI-8) were used to assess the participants’ eHealth literacy and OF. Binary logistic regression models were used to adjust potential confounders and to explore the independent relationship between OF and eHealth literacy, threshold effect analysis was performed based on it. Restricted cubic spline (RCS) was applied to examine the non-linear association and visualize the dose-response relationship. Moreover, subgroup analyses concerning sociodemographic and health condition characteristics were conducted in this study. RESULTS: 636 participants were finally included, with 432 (67.9%) OF patients. Regarding eHealth literacy, in the fully adjusted model, an increased level of eHealth literacy was significantly associated with a lower odds ratio of OF, with OR = 0.89, 95% CI: 0.85–0.93, p < 0.001. Threshold effect analysis showed that eHealth literacy was significantly associated with OF when it exceeded 22.40, with OR = 0.75, 95% CI: 0.66–0.86, p < 0.001. Restricted cubic spline confirmed a non-linear association between eHealth literacy and OF (p for non-linear = 0.011). Subgroup analyses further confirmed that the negative association between eHealth literacy and OF was consistent across groups. Notably, age (p for interaction = 0.003) and education level (p for interaction = 0.005) were found to modulate the strength of this association. CONCLUSIONS: Our study indicated that eHealth literacy is negatively associated with OF in older adults, with a non-linear dose-response relationship and a data-driven threshold effect. The threshold should not be interpreted as a fixed “optimal range” or a lower bound for association, but rather a reference for targeted interventions. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s12877-026-07243-x.