Quality and reliability of Alzheimer's disease videos on Douyin and Bilibili: A cross-sectional content analysis study

Background Alzheimer's disease (AD) poses a significant public health challenge to China's aging population. Patients and their families increasingly turn to short-video platforms such as Douyin and Bilibili for information. However, there is currently a lack of systematic analysis regarding the quality and reliability of advertising content on these platforms, creating a critical gap in understanding this emerging information ecosystem. Aim Systematically evaluate the quality and reliability of videos on Douyin and Bilibili, analyzing the relationship between content themes, upload sources, and user engagement metrics. Methods Using “Alzheimer's disease” as the keyword, we retrieved the top 100 videos from multiple platforms. Videos were categorized by uploader type and content. Two qualified researchers assessed their reliability and quality using the JAMA, the modified DISCERN instrument (mDISCERN), and Global Quality Score (GQS) scale. Data analysis employed nonparametric statistical methods. Apply relevance and logistic regression analysis to discuss factors that may influence video quality. Results This study analyzed a total of 171 videos. Results indicate that compared to Douyin, videos on the Bilibili platform scored higher across multiple quality evaluation metrics (GQS: 2.0(1.0–2.0) vs 1.0(1.0–2.0); mDISCERN: 2.0(2.0–2.0) vs. 2.0(2.0–2.0); JAMA: 2.0(1.0–2.0) vs. 1.0 (1.0–2.0); p  < 0.001). This disparity may be attributed to Bilibili's longer video format, which allows for more in-depth content, and its user base that tends to favor detailed, knowledge-oriented media. Regarding uploader identity, videos posted by professionals (e.g. physicians) demonstrated superior quality compared to nonprofessional sources (e.g. patients). However, patient-uploaded videos exhibited stronger engagement metrics (e.g. likes, comments). Content-wise, videos focusing on disease prevention and treatment consistently achieved the highest overall quality (all comparisons p  < 0.05). Correlation analysis indicated that while interaction metrics showed strong internal correlations, they did not significantly correlate with JAMA, mDISCERN, or GQS scores. Ordered logistic regression analysis indicates that uploader identity, content classification, and presentation format are the three key factors influencing video quality. Conclusion This study reveals a pronounced “quality-dissemination paradox” in AD content across mainstream short-video platforms: While scientifically rigorous content published by medical professionals receives high quality ratings, it significantly underperforms in user engagement metrics compared to nonprofessional content centered on patient narratives and lived experiences. This highlights a severe disconnect between scientific rigor and public participation within algorithmic dissemination ecosystems. To address this, platforms should optimize algorithms to enhance the visibility of authoritative content, encourage collaboration between professional and nonprofessional creators to boost content appeal, and strengthen health media literacy education for the public—particularly older adults—to improve their ability to discern information.