Determinants of medication adherence in community-dwelling older adults with polypharmacy: a cross-sectional study based on the COM-B model

Polypharmacy presents a significant challenge for medication adherence among community-dwelling older adults. This study aimed to investigate the current status of medication adherence in this population and to explore how eHealth literacy and social support relate to adherence, and whether attitudes toward aging play a statistically mediating role in these relationships, within the framework of the Capability-Opportunity-Motivation-Behavior (COM-B) model. A cross-sectional study was conducted from May to November 2025 across three community health service centers in China. Participants included 552 older adults (aged ≥ 60 years) who were taking five or more daily medications. Data were collected using validated instruments: the Medication Adherence Rating Scale, the eHealth Literacy Scale, the Attitudes to Aging Questionnaire, and the Social Support Rating Scale. Statistical analyses, including descriptive statistics, univariate and multivariate analyses, and mediation analysis, were performed using SPSS 27.0. Overall medication adherence was moderate-to-good. Better eHealth literacy, positive aging attitudes, and stronger social support were independently associated with higher adherence. Mediation models showed that attitudes toward aging partially explained the associations of both eHealth literacy (indirect effect: 0.054, 95% CI: 0.036–0.073; 30.2% of total effect) and social support (indirect effect: 0.093, 95% CI: 0.069–0.123; 47.9% of total effect) with medication adherence. Better digital health skills and social support are positively associated with medication adherence, and positive views on aging are linked to more favorable adherence outcomes. Future community health interventions for older adults with polypharmacy could consider incorporating psychological factors and patient-centered communication models, although the effectiveness of such approaches requires further validation.