Across 128 videos (56 Douyin, 72 Bilibili) analyzed via modified DISCERN, Global Quality Score, and JAMA benchmarks, Bilibili content scored significantly higher on both reliability metrics (DISCERN p<0.001; JAMA p=0.003) and hosted in-depth material comparable to clinical patient education. Douyin, by contrast, mirrored the uniformly low quality documented on Western short-video platforms like TikTok and Instagram. Critically, both platforms harbored content promoting counterfeit products exploiting GLP-1 brand recognition — a hazard specific to China's unregulated domestic supply chain and amplified by the approaching 2026 semaglutide patent cliff.
This cross-sectional study arrives at a consequential moment. Western regulators have already flagged counterfeit semaglutide as a patient safety emergency, but the Chinese context adds a distinct layer: mazdutide's June 2025 domestic approval and looming patent expiry create fertile ground for fraudulent branding at scale. The Bilibili finding is genuinely surprising — platform architecture (longer format, community-driven credibility norms) appears to moderate misinformation in ways that short-form algorithmic feeds structurally cannot. That said, cross-sectional design captures a single temporal snapshot, and content quality scores do not measure actual audience behavior or health outcomes. The study also analyzed only 128 videos, limiting statistical power for subgroup conclusions. Still, the platform-architecture-as-public-health-variable thesis is underexplored and actionable: regulators and clinicians should treat platform design as a lever in combating GLP-1 misinformation, not just content moderation.