Information quality, presentation characteristics, and engagement of oral cancer videos on Chinese social media: a cross-sectional comparison of healthcare professional–generated and patient-experience videos

Video-sharing platforms are increasingly used for health information, but surgery-focused oral cancer content may differ by platform and uploader. We compared information quality, observable presentation characteristics, and raw engagement indicators on Douyin and Bilibili. We analyzed 154 videos retrieved using the combined Chinese-language terms “oral cancer” and “surgery.” Quality was assessed using a Global Quality Scale (GQS)-based measure, modified DISCERN (mDISCERN), Journal of the American Medical Association (JAMA) Benchmark criteria, and abbreviated Patient Education Materials Assessment Tool for Audiovisual Materials (PEMAT A/V) items. The study-specific four-domain content coverage indicator, visual focus, emotional tone, and raw engagement indicators were also compared across platforms and between healthcare professional (HCP)-generated and patient-experience videos. Douyin videos had higher GQS scores than Bilibili videos (median 23 vs. 20; P < 0.001) and higher understandability proportions (median 0.75 vs. 0.50; P < 0.001). On both platforms, HCP-generated videos had higher mDISCERN and JAMA scores than patient-experience videos, whereas patient-experience videos had higher scores on the study-specific four-domain content coverage indicator (all P < 0.05), although the Bilibili patient-experience subgroup was small ( n = 11). HCP videos more often used talking-head presentations and positive or educational tones, whereas patient-experience videos more often showed postoperative wounds or daily-life content and fearful or regretful tones. Primary models associated Douyin with higher raw engagement counts and patient-experience videos with lower raw counts on selected indicators (adjusted ratios, 0.20–5.75; all q < 0.05), but none remained significant in the upload-age–adjusted sensitivity analysis ( n = 101). Within this surgery-focused sample, information quality, presentation characteristics, and raw engagement indicators differed by platform and uploader. However, the engagement analyses were based on unstandardized interaction counts and should not be interpreted as measures of reach or exposure.