Abstract
BACKGROUND: Short-video platforms have become major channels for public access to health information in the digital era. However, the low barriers to content creation and the increasing use of AI-generated content have accelerated the spread of health misinformation, underscoring the need to better understand how users identify health misinformation in short videos. METHODS: Grounded theory was applied to analyze 47 in-depth interviews and extract core factors influencing users' recognition of health misinformation in short videos. Based on the derived factor structure, a questionnaire survey was conducted and 279 valid samples were collected. Partial least squares structural equation modeling (PLS-SEM) was used to test the proposed relationships, and fuzzy-set qualitative comparative analysis (fsQCA) was further employed to identify the causal configurations through which different factor combinations contribute to users' health misinformation discernment. RESULTS: The results identified three key categories: information quality, user characteristics, and external environments. The PLS-SEM model demonstrated acceptable explanatory power (R(2) = 0.478) for users' health misinformation discernment in short videos. Among the seven proposed hypotheses, content logic (p < 0.05), narrative expression (p < 0.05), information structure (p < 0.01), cognitive level (p < 0.05), and external influences (p < 0.05) were statistically supported, while information reliability and psychological needs showed non-significant effects. The fsQCA further revealed three distinct causal configurations leading to effective discernment. When content logic functioned as the core condition, users tended to rely on central, analytical processing; whereas when external influences were dominant, users were more likely to depend on heuristic processing rather than message logic. DISCUSSION: The findings highlight three distinct ways users process health misinformation in short videos, including primarily analytical evaluation, peripheral reliance on content cues, and peripheral reliance on cognitive cues. These results suggest practical strategies for mitigating health misinformation on short-video platforms, emphasizing interventions at individual, platform, and policy levels.