Reliability and quality assessment of internet videos as guidance for dietary weight loss intervention: a cross-sectional study in China

Background Obesity, a chronic condition affecting multiple physiological systems, poses major public health challenges. Dietary interventions are widely recognized as effective strategies for weight management. With over 4 billion Internet users worldwide, an increasing number of individuals rely on online platforms for health information. In China, TikTok, Bilibili, and Kwai are major channels for disseminating health-related content. However, the quality of dietary weight loss information on these platforms remains unclear. Objective This study aims to assess the reliability and quality of the information in Chinese videos on dietary weight loss shared on the BiliBili, TikTok, and Kwai, three video–sharing platforms. Methods We identified the top 100 dietary weight-loss videos on each platform in February 2024, resulting in a total of 300 videos. Video information quality and reliability were assessed using the Global Quality Score (GQS) and modified DISCERN (mDISCERN). Correlations between video quality and video characteristics were also analyzed. Results The average GQS scores for BiliBili, TikTok, and Kwai were 2.04, 1.81, and 1.70, respectively, while the average mDISCERN scores were 2.01, 1.81, and 1.73. Median scores for both tools across all platforms were 2. BiliBili showed significantly higher GQS scores than TikTok and Kwai ( p < 0.01 and p < 0.05, respectively). Regarding mDISCERN, BiliBili scored significantly higher than TikTok ( p < 0.05), while the difference with Kwai was not statistically significant ( p = 0.08). Nevertheless, none of the platforms achieved scores above 3, indicating generally low information quality and reliability. Significant positive correlations were found between video duration and both GQS ( r = 0.41, p < 0.01) and mDISCERN ( r = 0.32, p < 0.01). Additionally, strong correlations were observed between likes and saves ( r = 0.90, p < 0.01), likes and comments ( r = 0.92, p < 0.01), and saves and comments ( r = 0.86, p < 0.01). Conclusion While acknowledging limitations regarding cross-sectional design and specific sampling of single keyword and Chinese platforms, our findings highlight that the prevalence of low-quality videos on Chinese social media exposes viewers to significant risks of misinformation and inappropriate dieting. These findings underscore the need to promote digital health literacy and improve strategies for digital health communication.