The quality and reliability of Chinese short-video information on spinal robotic navigation surgery on TikTok and Bilibili: a cross-sectional content analysis

Abstract Objective This cross‑sectional content‑analysis study aimed to systematically evaluate the quality and reliability of Chinese‑language short‑videos concerning spinal navigation robots on TikTok and Bilibili, and identify predictors of video information quality. Methods Short‑videos were retrieved from TikTok and Bilibili using predefined search terms on July 19, 2026. Videos were screened according to pre‑established inclusion and exclusion criteria. Two blinded attending physicians scored eligible videos using the Global Quality Score (GQS) and modified DISCERN (mDISCERN) scale. Uploader type, content theme, video metadata and engagement metrics were coded. Inter‑rater reliability was assessed by Cohen’s κ. Non‑parametric tests, Spearman’s rank‑order correlation and GEE‑based robust Poisson regression were applied for statistical analysis. Results A total of 52 eligible short‑videos were finally enrolled, including 38 from TikTok and 14 from Bilibili. Inter‑rater reliability was substantial for both GQS (κ = 0.708) and mDISCERN (κ = 0.741). Significant inter‑platform differences existed in video duration, engagement indicators, uploader composition and content themes. Videos produced by clinicians achieved significantly higher GQS and mDISCERN scores. Benefit‑risk analysis‑oriented videos obtained the highest GQS, whereas pure technical‑device‑oriented videos showed poorer quality. Likes and comments were positively correlated with GQS, while only comment count was positively associated with mDISCERN. Multivariable GEE robust Poisson regression demonstrated that content type was an independent predictor for both quality scales. Technology‑equipment‑focused content independently predicted lower GQS (adjusted IRR = 0.614, 95%CI 0.442‑0.853, P = 0.004). Clinician‑released videos were independently associated with higher mDISCERN scores compared with science‑popularization bloggers (adjusted IRR = 1.713, 95%CI 1.014‑2.892, P = 0.044). After adjustment, platform origin exerted no independent effect on quality outcomes. Conclusions The overall quality and reliability of spinal‑navigation‑robot‑related short‑videos remain unsatisfactory on the two platforms. Uploader background and content theme are more critical determinants of information quality than platform source or user‑engagement performance. High‑interaction metrics cannot represent medically balanced and trustworthy health information. More clinician‑participated popular‑science production, stricter creator‑qualification audit and better exposure for balanced benefit‑risk‑analysis content are required to optimise online patient educational resources for robotic spinal surgery.