Related Experiment Video
Updated: Aug 5, 2026
Using Motion Capture Technology in the Instrumented Timed Up and Go Test to Detect the Risk of Falling in Aged Adults
Published on: October 25, 2024
Content quality and source credibility of fall prevention videos for older adults on Chinese short-video platforms: A
Peng Liu1, Wenjing Wang2, Fuzhi Wang1
1School of Medical Information and Engineering, Bengbu Medical University, Bengbu, China; Bengbu Medical University's Health Information Resource Governance and Application Research Innovation Team(12202403), China.
Background:
Falls constitute a leading cause of injury-related mortality among older adults globally. Chinese short-video platforms collectively reach over 900 million users, presenting unprecedented opportunities for health education, but the quality of fall prevention content and its relationship with user engagement have not been systematically evaluated.
Objective:
To evaluate fall prevention video quality across major Chinese short-video platforms, identify content creator characteristics associated with higher-quality information, and examine whether user engagement metrics correlate with video quality.
Methods:
We conducted a cross-sectional analysis of 216 fall prevention videos from five platforms (Douyin, Kuaishou, Bilibili, Xiaohongshu, Xigua Video) during October-November 2025. Two independent medical-school graduates with formal medical education and research expertise in medical informatics assessed video quality using the modified DISCERN instrument (mDISCERN; range 5-25) and Global Quality Scale (GQS; range 1-5). Interrater agreement was quantified using both intraclass correlation coefficients (ICC) and Cohen's weighted κ. User engagement metrics were extracted and analyzed using both Pearson and Spearman correlations.
Results:
Interrater reliability was excellent for mDISCERN (ICC=0.890; weighted κ=0.890) and good for GQS (ICC=0.723; weighted κ=0.722). Mean mDISCERN score was 17.61 (SD 2.87), with 48.1% achieving high quality. Uploader type demonstrated the strongest quality association (ε²=0.64): healthcare professionals substantially outperformed self-media creators (Cohen d=3.42). Platform verification strongly predicted quality (88.7% vs 9.1% high-quality; φ=0.79). Engagement metrics showed weak association with quality in this sample (Spearman ρ=0.149 for likes, explaining only 2.2% of variance), with detection power constrained by severe right-skewness and floor effects (e.g., 30.1% of videos had zero comments).
Conclusions:
Content creator credentials and platform verification effectively discriminate video quality, while engagement metrics show only weak association in this sample. These findings support platform policies prioritizing verified professional content and indicate that engagement-based metrics, despite their algorithmic prominence, do not reliably signal health information quality in this dataset.
