Quality, guideline concordance, and dangerous misinformation in pediatric febrile seizure videos on Chinese short-video platforms: a cross-sectional study

Abstract Background Febrile seizures affect 2–5% of children under 5 years and, although clinically benign, are among the most frightening events a caregiver can witness. Parents increasingly seek first-aid information on Chinese short-video platforms, where content is surfaced by engagement-driven algorithms. Existing evaluations rely on generic video-quality instruments not designed to detect disease-specific content or dangerous first-aid demonstrations. We evaluated pediatric febrile seizure videos on three major Chinese platforms and tested whether engagement tracks clinical safety. Methods We conducted a cross-sectional analysis of 269 pediatric febrile seizure videos retrieved in March 2026 from Douyin ( n = 96), Kuaishou ( n = 87), and Bilibili ( n = 86). Two pediatric reviewers independently assessed each video using five established generic video-quality instruments, a 14-item guideline-anchored Febrile Seizure Content Index (FSCI), and an 11-item Febrile Seizure Dangerous Misinformation Checklist (FSDMC) with Grade A/B/C severity. Kruskal–Wallis and chi-square tests, Spearman correlation analyses, and multivariable regression models with sensitivity analyses were performed. Results Dangerous misinformation was highly prevalent: 55.0% of videos contained at least one dangerous item, and 52.4% contained a Grade A item carrying immediate physical-harm risk—most commonly placing objects in the child's mouth (46.1%), pressing the philtrum or slapping (41.6%), restraining the child (38.7%), and forcing water or oral medication (32.3%). FSCI was essentially uncorrelated with the Global Quality Scale ( ρ = − 0.03) and modified DISCERN ( ρ = 0.02), and only weakly with JAMA benchmark scores ( ρ = 0.16). Engagement correlated positively with generic quality but negatively with FSCI (likes ρ = − 0.22; shares ρ = − 0.20). Douyin videos scored highest on generic instruments yet lowest on FSCI and carried the highest Grade A burden (63.5%). Medical professionals uploaded 79.2% of videos, yet 50.7% of these still contained at least one dangerous item. Conclusions Pediatric febrile seizure videos on Chinese short-video platforms frequently contain dangerous first-aid misinformation, generic quality instruments fail to detect this risk, and platform engagement is inversely associated with guideline-based content completeness—a quality–safety paradox. Evaluations of short-video health content for acute pediatric presentations should routinely pair generic quality instruments with disease-specific content and misinformation assessment anchored to current clinical guidelines, and clinical credentials alone should not be treated as a marker of content safety.