Quality, reliability, and content coverage of pleural effusion–related short videos on Douyin (TikTok) and Bilibili: a cross-sectional study

Pleural effusion is a common clinical finding with heterogeneous causes, including infection, malignancy, heart failure, tuberculosis, pulmonary embolism and systemic disease. Because patients often encounter pleural effusion as an imaging finding rather than as a diagnosis, short-video platforms may shape early understanding, anxiety and care-seeking behaviour. Evidence on the quality of this content is lacking. We conducted a cross-sectional content analysis of publicly available pleural effusion–related short videos on Douyin and Bilibili. The Mandarin search term “胸腔积液” (pleural effusion) was entered on 15 May 2026 and the first 150 results returned under each platform’s default ranking were screened. Video duration, engagement metrics, uploader category, coverage of six content domains, Global Quality Score (GQS), modified DISCERN (mDISCERN) and Journal of the American Medical Association (JAMA) benchmark scores were extracted. Each video was scored independently by two reviewers, with discrepancies resolved by consensus with a third senior reviewer. Platform comparisons used two-sided Mann–Whitney U tests; uploader comparisons used Kruskal–Wallis tests with Holm-adjusted pairwise Mann–Whitney U tests. Proportional-odds ordinal logistic regression was used to estimate platform and uploader effects, with adjustment for uploader type and additional adjustment for log video duration and the number of coded content domains. All pairwise correlations were adjusted for multiplicity using Benjamini–Hochberg false-discovery-rate and Bonferroni corrections. The study is reported in accordance with the STROBE checklist for cross-sectional studies. The final sample comprised 200 videos (100 per platform). Overall quality was moderate and reliability limited, with median GQS 3 (IQR 2 to 3), mDISCERN 2 (IQR 1 to 2) and JAMA 2 (IQR 1 to 3). Specialists uploaded 51.00% of videos. Treatment (57.00%) and diagnosis (48.00%) were the most frequently covered content domains, whereas epidemiology (6.00%) and follow-up (5.50%) were the least; 14 videos (7.00%) covered none of the six coded domains. Unadjusted platform comparisons showed no difference in GQS or mDISCERN ( P = 0.961 and P = 0.603) and a higher JAMA score on Douyin ( P = 0.008). After adjustment for uploader type, however, Douyin videos had significantly lower overall quality (GQS adjusted OR 0.39, 95% CI 0.20 to 0.75, P = 0.005) and the apparent JAMA advantage was no longer present (adjusted OR 0.81, 95% CI 0.45 to 1.46, P = 0.479). The GQS difference was fully attenuated after further adjustment for video duration and content coverage (adjusted OR 1.06, 95% CI 0.48 to 2.32, P = 0.891), whereas log duration (OR 1.67, P = 0.002) and the number of coded content domains (OR 3.56, P < 0.001) remained associated with GQS. Uploader category showed the largest quality gradient (GQS epsilon-squared 0.423, 95% CI 0.294 to 0.557). After adjustment for multiple comparisons, none of the engagement metrics was associated with any quality instrument: likes, collections and shares showed no association, and the weak negative associations of comments with GQS (rho = − 0.20) and mDISCERN (rho = − 0.21) did not survive Bonferroni correction. Pleural effusion–related short videos on Douyin and Bilibili are widely accessible but medically incomplete. Platform identity was not independently associated with reliability or transparency once uploader composition had been accounted for, and the residual platform difference in overall quality was explained by shorter video duration and narrower content coverage rather than by the platform itself. Engagement metrics should not be read as indicators of information quality. Future short-video health education on pleural effusion should emphasise differential diagnosis, red-flag symptoms, thoracentesis and imaging pathways, malignancy and infection risk, and follow-up after initial treatment.