Quality assessment of exercise videos for chronic diseases on social media platforms: a systematic review and meta-analysis

Abstract Background Social media platforms have become important sources of exercise information for people with chronic diseases, but the evidence on the quality of such videos has never been quantitatively synthesized. This study aimed to evaluate the educational quality of exercise videos for chronic disease populations on social media and to examine how video characteristics, engagement metrics and uploader type relate to that quality. Methods A systematic review and meta-analysis was conducted following PRISMA guidelines. Seven sources (Web of Science, PubMed, SportDiscus (EBSCO), Embase, Scopus, CNKI and Google Scholar) were searched from inception to September 2025; eligibility was restricted to English- and Chinese-language studies assessing the quality of exercise videos for chronic disease management on social media. Two reviewers independently screened and extracted data. Study-level random-effects meta-analyses (restricted maximum likelihood) were performed in Stata 19 (SE), reporting 95% confidence intervals (CI) and, for the principal instruments, 95% prediction intervals (PI); quality strata were compared by meta-regression. Study quality was appraised with the eight-item D’Souza framework. Results Seventeen studies comprising 1,515 videos were included; 13 (76.5%) were conducted in Turkey, and YouTube was the dominant platform (16/17, 94.1%). Overall quality was low to moderate. Pooled mean scores were 2.96 of 5 for the Global Quality Scale (95% CI 2.63–3.29; 95% PI 1.70–4.23), 2.53 of 5 for the modified DISCERN (95% CI 2.18–2.90; 95% PI 1.24–3.84) and 1.49 of 4 for the JAMA benchmark criteria (95% CI 0.65–2.33), with substantial heterogeneity (I² 93.0–98.6%). Video duration, view count, likes, dislikes and comments did not differ across quality strata. Videos from individual physician accounts were over-represented in the poor-quality stratum (34.2% versus 3.6% in the good stratum, p = 0.03), whereas those from academic or institutional channels were more often of good quality ( p = 0.068). Conclusion Exercise videos for chronic disease populations are widely available but of suboptimal educational quality, and engagement metrics do not indicate quality. Higher-quality content is associated with academic and institutional channels rather than individual physician accounts. These conclusions rest on English- and Chinese-language, largely Turkish and YouTube-based evidence, and the wide prediction intervals warrant caution in generalizing them. Trial Registration PROSPERO CRD420261345615; https://www.crd.york.ac.uk/PROSPERO/view/CRD420261345615 .