In this cross-sectional study based on electronic media, the following search terms were used: “中风” “脑卒中” (Chinese for “stroke”), “脑梗死” (Chinese for “ischemic stroke”), “脑出血” (Chinese for “hemorrhagic stroke”), and “预防” (Chinese for “prevention”). These terms were used to retrieve the top 100 videos automatically recommended by the default overall ranking on 4 popular Chinese-language online video-sharing platforms: Bilibili (Bilibili Inc/Shanghai Hode Information Technology Co, Ltd/Sony Group Corporation) [19], Douyin (ByteDance Ltd) [20] (the Chinese counterpart of TikTok), Haokan Video (Baidu, Inc) [21], and Xigua Video (ByteDance Ltd) [22]. A brief introduction to these 4 Chinese-language video platforms is provided in Table 1.

The selection of search keywords was based on both medical terminology and commonly used lay terms for stroke within the Chinese context to ensure relevance and comprehensiveness. This approach aimed to capture a broad spectrum of videos that users might realistically encounter. For example, we included both formal terms such as “脑卒中” (formal medical terminology for “stroke” in the Chinese language) and more colloquial expressions such as “中风” (colloquial description of “stroke” in the Chinese language), which are widely used in public discourse and online media. In addition, we referred to trending tags and frequently searched phrases related to stroke on Chinese video-sharing platforms to enhance the validity of the search keyword set.

With regard to the automatic recommendation and ranking algorithm, consultation with the customer service agents from the selected platforms indicated that the algorithms are primarily driven by video relevance, followed by a comprehensive evaluation based on popularity metrics such as views and likes. However, more detailed information about the algorithm’s formulas and methodologies is unavailable due to concerns related to commercial confidentiality.

Video retrieval was conducted on a Windows 10 (Microsoft Corporation) personal computer using a newly installed Mozilla Firefox browser (version 121.0.1; Mozilla Foundation and Mozilla Corporation) in Suzhou, Jiangsu, China, within a single day (December 28, 2023) to minimize potential bias from newly uploaded content. To reduce the influence of personalized recommendation algorithms, all cookies, browsing history, and temporary internet files were cleared, and platform accounts were logged out before and between search queries, which were performed in incognito mode.

After conducting the searches, the top 100 videos automatically recommended based on the default overall ranking on each video platform were selected and prepared for initial screening. This decision was informed by previous literature indicating that most amateur health seekers rarely browse beyond the first 2 pages of search results [23]. Additionally, most Chinese-language video-sharing platforms limit the number of videos displayed per page to no more than 30. Specifically, Bilibili displays 30 videos per page, Haokan Video 10, Xigua Video 20, and Douyin 20. Therefore, we believe that our sample size (n=100) exceeds the number of videos typically displayed on the first 2 pages (Bilibili, n=60; Haokan Video, n=20; Xigua Video, n=40; and Douyin, n=40), which audiences are most likely to engage with, thereby ensuring representativeness. Furthermore, a sample size of 100 has been widely adopted in similar previous studies [24-26], further supporting the rationale for selecting the top 100 automatically recommended videos for subsequent filtering and analysis.

The exclusion criteria were defined as follows: (1) videos not directly related to stroke prevention; (2) duplicated videos repeatedly uploaded to the same platform; (3) videos not in Chinese (including Mandarin, Cantonese, and other dialects); and (4) commercial advertisements. In this study, 78 off-topic videos and 9 duplicate videos uploaded to the same platform were identified and excluded, resulting in a final sample of 313 eligible videos for further analysis (Figure 1). All eligible videos were produced in Mandarin Chinese, and no highly similar or fully duplicate videos were observed across different platforms.

For the included videos, general information, including URLs; authorship; number of views, likes, shares, and comments; upload date; and video length (in seconds), was collected and recorded in an Excel spreadsheet (Microsoft Corporation) by an independent researcher (RGG).

The included videos were categorized based on the identity of their producers as follows: (1) general users (eg, amateurs); (2) health professionals (eg, medical practitioners); (3) science communicators (eg, citizen science storytellers); (4) news agencies (eg, CCTV [China Central Television]); (5) medical organizations (eg, hospital); (6) for-profit organizations (eg, commercial health care–related enterprise); and (7) nonprofit organizations (eg, voluntary patient communities).

For videos classified under health professionals, a further subcategorization was conducted as follows: (1) doctors specializing in neurology within modern evidence-based medicine; (2) doctors specializing in other areas of modern evidence-based medicine; (3) doctors of traditional Chinese medicine (TCM); and (4) other health-related professionals, further specified according to their specialty. Detailed definitions of the authorship categories are provided in Table S1 in Multimedia Appendix 1.

In this study, the Shapiro-Wilk test was used to assess the normality of the quantitative data extracted from the included videos. For nonnormally distributed data, the median and IQR were used for descriptive analysis, whereas for normally distributed data, the mean and SD were reported. Categorical data were presented as absolute numbers and percentages.

Comparisons between 2 groups were performed using the nonparametric Mann-Whitney U test, while comparisons among 3 or more groups were conducted using the Kruskal-Wallis H test. The Spearman correlation coefficient was then used to examine associations between video variables and quality and reliability scores. For interpretation, a correlation coefficient less than 0.10 was considered negligible; between 0.10 and 0.39, weak; between 0.40 and 0.69, moderate; between 0.70 and 0.89, strong; and greater than 0.90, very strong [39]. Finally, a linear regression model was applied to explore potential temporal trends in the quality and reliability of stroke prevention–related videos over the upload period. A P value of <.05 was considered statistically significant. All analyses were performed with the GraphPad Prism software (version 9.5.1; GraphPad Software, Inc).

This study did not involve human participants or experimental animals; therefore, ethics approval and consent to participate were not applicable. Our study involved the analysis of publicly available data or the use of secondary data that do not contain any personal identifiers, thus ensuring the anonymity and confidentiality of the individuals concerned. Compliance with ethical standards was maintained throughout the research process.

As shown in Figure 1, after the search and screening process, a total of 313 eligible online videos on stroke prevention education were included for further analysis. Of these, 68 were from Bilibili, 74 from Douyin (the Chinese counterpart of TikTok), 86 from Haokan Video, and 85 from Xigua Video.

Overall, as shown in Table 4, the median values among the included videos were as follows: video length, 137 seconds; views, 779; likes, 18; shares, 67; and comments, 1. Regarding the year of video upload, the number of uploaded videos has steadily increased over time, with the majority uploaded in 2023, accounting for 106 out of 313 (33.9%) total videos uploaded. When analyzed by video authorship, health professionals, including doctors specializing in neurology and other fields of modern medicine, doctors of TCM, and pharmacists, collectively created and uploaded 113 stroke prevention–related videos (n=113, 36.1%), followed by news agencies (n=95, 30.4%) and general users (n=40, 12.8%). More detailed descriptive results by video platforms are presented in Table 4.

With regard to the overall quality and reliability of the included online videos, the kappa coefficients for the modified DISCERN and GQS scores, as assessed by 2 independent raters, were 0.71 and 0.73, respectively, indicating substantial interrater agreement. The observed discrepancies were primarily due to 1 rater slightly overestimating the quality and reliability of certain stroke prevention–related videos. However, most of these minor disagreements were resolved through discussion between the 2 raters. In cases where disagreement persisted, a senior researcher was assigned to make the final decision. Taken together, with the median scores for the modified DISCERN and GQS being 2 and 3, respectively, our results suggest that the included videos exhibited relatively poor reliability but moderate quality.

The distribution of video uploads over time revealed a clear upward trend. As shown in Table 4, the number of stroke prevention–related videos increased steadily each year. Only a small proportion of videos were uploaded before 2020, with a marked rise beginning in 2021. Notably, the highest number of videos (n=106, 33.9%) was uploaded in 2023 alone, suggesting growing public and institutional engagement with stroke-related health education on video platforms in recent years. This increase may reflect heightened awareness of cerebrovascular disease prevention, evolving content production practices, or rising user engagement with online platforms across China. These temporal patterns underscore the relevance and urgency of evaluating video-based health information, as both the volume and impact of such content continue to grow.

When examined by specific online video platforms, as shown in Figure 2A and B, our results suggest that videos from Douyin (median modified DISCERN score of 3) exhibited higher median reliability score compared with those from other platforms (median modified DISCERN scores of 2.5, 2, and 2 for Bilibili, Haokan Video, and Xigua Video, respectively). However, this difference was statistically significant only between Douyin and Haokan Video (P<.001). Meanwhile, in terms of video quality, the median GQS scores were highly consistent across the 4 video platforms (GQS=3 for all 4 platforms). A statistically significant difference was observed only between Bilibili and Haokan Video (P=.03).

When analyzed by specific video creator categories, as shown in Figure 2C and D, videos produced by health care professionals (median modified DISCERN score of 3) and science communicators (median modified DISCERN score of 3) demonstrated greater reliability compared with those created by amateur individual users (median modified DISCERN score of 2). However, no statistically significant difference was observed among videos created by different organizational authors (P>.99 in all cases). In terms of video quality, the findings were consistent with those for video reliability. Stroke prevention–related videos created by science communicators received higher GQS scores (median GQS score of 4) than those by amateur individual users (median GQS score of 3). No significant difference was observed between videos created by health care professionals and those by amateur users (P=.06). Among organizational video creators, videos from nonprofit organizations (median GQS score of 3.5) demonstrated higher quality compared with those from amateur users and news agencies (median GQS score of 3 for both cases).

When the health care professional authorship category was further subdivided (Figure 2E and F), no statistically significant differences were found among videos produced by neurologists (P>.99), doctors from other fields of modern medicine (P>.99), doctors of TCM (P>.99), and pharmacists (P>.99), regardless of whether the assessment was based on reliability or quality. However, despite the lack of statistical significance (P>.99), videos created by doctors specializing in modern medicine received higher reliability scores than those produced by TCM doctors (median modified DISCERN scores of 3 for both neurologists and other modern medical specialists vs 2 for TCM doctors). Interestingly, pharmacists, though contributing a relatively small proportion of the videos, produced content with moderate quality (median GQS score of 3), possibly reflecting their specialized knowledge in pharmacological prevention strategies, such as the use of antiplatelet agents for stroke risk reduction.

These findings suggest that professional background may influence the rigor of content, even if the differences are not substantial, and underscore the potential value of encouraging more contributions from neurologists and other clinical specialists in stroke-related video creation.

Regarding the stroke prevention strategies recommended in the included online videos, the vast majority focused on primary prevention (n=262), targeting the first onset of stroke. By contrast, relatively few addressed secondary prevention aimed at preventing stroke recurrence (n=28), and only a small number of videos covered both primary and secondary prevention simultaneously (n=23).

To be more specific, among primary stroke prevention measures, the most frequently mentioned strategy was controlling abnormal blood pressure (158/313, 50.5%), followed by adopting healthy dietary habits (145/313, 46.3%) and engaging in physical activity (144/313, 46.0%). These lifestyle-oriented recommendations reflect prevailing public health messaging trends and align with international stroke prevention guidelines. By contrast, for secondary prevention, the most frequently recommended measure was also managing abnormal blood pressure (24/313, 7.7%), followed by controlling abnormal blood glucose (21/313, 6.7%), regulating abnormal blood lipid levels (21/313, 6.7%), and using oral antiplatelet agents (21/313, 6.7%). This distribution suggests that, where addressed, secondary prevention videos tended to emphasize pharmacological and laboratory-monitored interventions rather than lifestyle-based recommendations.

Collectively, these findings suggest that while primary prevention dominates the landscape of online stroke-related education in China, a considerable gap exists in public-facing content addressing secondary prevention. Enhancing the visibility and clarity of such content, particularly among stroke survivors, may represent a valuable opportunity for public health intervention.

As shown in Table 5, significant correlations were observed among several general video information metrics. In particular, the number of views was positively associated with video length, number of likes, number of shares, and number of comments (r=0.142, P=.03; r=0.838, P<.001; r=0.897, P<.001; and r=0.593, P<.001, respectively). Similarly, videos that received more likes were more likely to have higher numbers of shares and comments (r=0.900, P<.001 and r=0.816, P<.001, respectively).

Nevertheless, when considering video quality and reliability, video length was found to be positively correlated with both the modified DISCERN and GQS scores (r=0.151, P=.007 and r=0.441, P<.001, respectively). Additionally, a greater number of comments was associated with higher video reliability, although this correlation was relatively weak (r=0.137, P=.01).

According to Figure S1 in Multimedia Appendix 1, an analysis of video upload time revealed a slight but statistically significant positive correlation between more recent upload dates and both video quality (P=.002) and reliability (P=.01). Specifically, Pearson correlation coefficients indicated that videos uploaded more recently tended to have higher GQS scores (r=0.113, P=.002) and higher modified DISCERN scores (r=0.086, P=.01). Although the strength of these associations was relatively weak, the findings suggest a gradual improvement in the overall quality and reliability of stroke-related video content over time. This trend may reflect growing awareness among content creators regarding evidence-based health communication, as well as improved platform policies aimed at enhancing the accuracy of health information.

In this study, by searching and screening online videos related to stroke prevention across 4 popular Chinese-language video platforms, we systematically described and analyzed the general characteristics, as well as the overall quality and reliability, of the included health-related online videos. To the best of our knowledge, this may be one of the first studies to specifically evaluate and report on the quality and reliability of Chinese-language online video platforms as information sources for stroke prevention education in Mainland China. We hope that our findings will provide valuable data for real-world policy makers and medical practitioners, ultimately contributing to improved stroke prevention and management efforts in the future.

Overall, using the modified DISCERN and GQS instruments, the included online videos related to stroke prevention were evaluated as relatively unreliable and of moderate quality. This suggests that, although video creators may offer some useful advice on primary and secondary stroke prevention measures, they often fail to provide supporting references and frequently lack fairness and objectivity in presenting information. These findings are consistent with previous research in other medical domains, which has similarly shown that health-related content on new media, particularly online video-sharing platforms, is often unsatisfactory and, at times, misleading [40-43]. This phenomenon may be largely attributed to the competition for audience attention in the digital marketplace, where accurate scientific information, often complex and less engaging for the general public, is easily overshadowed by sensationalized or simplified claims [44]. Consequently, entertainment-oriented online platforms are more likely to recommend light, engaging content over scientifically rigorous materials, which may be more difficult for general audiences to understand. Nevertheless, a growing number of stroke-related educational videos are being created and uploaded each year, suggesting that stroke prevention is becoming an increasingly visible topic on Chinese social media, particularly on video-sharing platforms. Notably, a positive temporal trend was observed in the quality and reliability of stroke-related videos uploaded between 2017 and 2023. We speculate that this improvement may be driven by increased participation from health care professionals in content creation, thereby directly contributing to the enhanced quality of stroke prevention information available online.

When analyzed by specific video platforms, videos from Douyin appeared to be more favorable, exhibiting higher content quality and reliability compared with those from the other 3 platforms, although statistical significance was not consistently observed. This finding may be explained by the fact that approximately 81% of eligible videos on Douyin were created and uploaded by qualified health care professionals, who generally receive more comprehensive and systematic evidence-based medical education and training than general users. Therefore, this phenomenon is relatively understandable. However, when the data for health care professionals were further stratified into more specific subgroups, videos created by doctors specializing in modern medicine—regardless of their field—tended to be more reliable than those produced by doctors of TCM, despite the absence of statistically significant differences.

Unfortunately, alongside the potentially beneficial health-related recommendations found in stroke prevention videos, concerns have also been raised about the prevalence of misinformation in the included study videos. Notably, some general individual users without a medical education background may share personal opinions in their videos that lack solid scientific evidence. For example, one video created by amateurs claimed that “regularly receiving intravenous injections can prevent stroke onset because it may help clear blood vessels.” This statement has been identified as a widely circulated misconception among the Chinese population, particularly among older adults. Additionally, some video creators were observed to use emotionally charged and sensationalized titles and content, rather than objective and evidence-based information, in an effort to attract public attention and maximize viewership. However, this approach may lead to severe misunderstandings that could potentially jeopardize public health at the community level. Furthermore, the risk of misinformation is not limited to nonprofessionals, as certain content produced by health care professionals also warrants scrutiny. Specifically, current stroke prevention recommendations and suggestions from TCM doctors are highly heterogeneous, with some approaches not yet completely validated by formal clinical trials and largely based on empirical knowledge. For instance, some TCM practitioners propose that massage on specific body points (commonly referred to as “acupoints”) may help prevent cerebral stroke. While this perspective is rooted in traditional practice, there is currently insufficient evidence from randomized controlled trials to substantiate its efficacy. Nevertheless, despite TCM’s historically experience-driven approach, it remains an integral component of China’s health care system and continues to serve as a complementary and alternative option in stroke prevention and management. In recent decades, TCM practitioners have made concerted efforts to integrate traditional practices with modern scientific research, contributing to its ongoing development and potential role in future stroke care [45].

To effectively address these issues, several targeted countermeasures can be implemented to enhance the quality and reliability of health-related online videos and information [46-48]. First and foremost, the review process for medical information videos uploaded to online platforms should be reinforced and supervised in collaboration with qualified medical professionals. Additionally, establishing and strengthening an online monitoring and correction system could help limit the spread of misinformation and reduce its long-term impact. Finally, general social media users should be encouraged to share controversial health information more selectively and cautiously, in order to curb the dissemination of false content and restrict accounts that produce or promote it.

Recently, Chinese governmental departments and relevant enterprises have begun taking active measures to combat the widespread dissemination of health-related misinformation online. For example, in collaboration with the Ministry of Industry and Information Technology, the Chinese National Health Commission issued a policy on May 27, 2024, aimed at urgently strengthening the regulation of health-related webcasting, video content, and promotional activities on digital platforms. Simultaneously, ByteDance (TikTok’s parent company) launched a centralized governance initiative in 2024 focused on identifying and removing poor-quality or even fraudulent sources of medical information. Beyond China, the World Health Organization (WHO) has proposed several countermeasures to limit online misinformation [49]. These include awareness campaigns targeting patients and health care professionals, the promotion of platforms that provide evidence-based data, the integration of scientific evidence into health-related content in mass media, and efforts to improve media and health literacy. Additionally, the WHO encouraged the active involvement of experts and health care professionals in debunking misinformation and guiding internet users toward credible, evidence-based information sources. However, the persistence of misinformation in the digital space, particularly in health care, continues to pose a significant threat to public well-being worldwide. This ongoing challenge underscores the urgent need for stronger guidelines and innovative strategies to mitigate its impact effectively.

In addition to evaluating the quality and characteristics of stroke-related videos, it is important to consider the platform-specific policies that may influence the dissemination of health content. Major Chinese video-sharing platforms, such as Douyin, have publicly issued guidelines encouraging the distribution of accurate and authoritative health information. For instance, these platforms typically require health-related content creators to verify their professional credentials and may implement content moderation mechanisms to flag or remove misleading medical information. However, the enforcement and transparency of these policies vary across platforms, and the extent to which they affect the quality and reach of health-related videos remains largely understudied. Future research could benefit from a more systematic analysis of how platform-level governance affects the credibility and visibility of online health communication.

Despite these findings, several limitations of this study should be acknowledged. First and most importantly, our research team retrieved and analyzed only Chinese-language videos targeting Chinese audiences, using a limited set of search keywords. Videos in other languages, such as English, Japanese, and Korean, were not included, indicating a need for broader investigation in future research. Second, although measures were taken to minimize potential bias introduced by platform algorithms during the initial video search, we cannot fully exclude the possibility that our results were influenced by the artificial intelligence algorithms used by commercial video platforms. To be more specific, video search engines tend to prioritize and recommend content with higher engagement metrics. Although we logged out of our accounts, deleted temporary files and cookies, and conducted searches using a newly installed internet browser, the inherent nature of these automatic algorithms amplifies disparities between videos—allowing popular content to gain increasing visibility, while less popular videos receive minimal exposure. Consequently, this unavoidable limitation may have influenced the collection of general video-related metrics, particularly likes and shares. Third, this study was based on a limited sample size (from only 4 Chinese domestic video platforms). Therefore, further research involving a larger number of videos and platforms with broader representativeness is urgently needed. Fourth, data collection was conducted on a single day, which may not fully capture the dynamic and evolving nature of content rankings on video platforms. Videos that consistently maintain high visibility over time might have been overlooked, and our findings may therefore represent only a snapshot of what users were most likely to encounter at that specific moment. Future studies could incorporate longitudinal sampling strategies to account for temporal variations in content prominence. Finally, as a cross-sectional study using a relatively subjective evaluation strategy, this study primarily focused on a descriptive assessment of the current state of online video-based patient education in Mainland China. Therefore, future prospective studies are warranted to evaluate the real-world effectiveness of physician-created and peer-reviewed video materials for patient education.

This study examined the overall quality and reliability of Chinese-language videos related to public patient education on stroke prevention. The findings suggest that the quality and reliability of health-related videos, as a source of information for stroke management, remain relatively unsatisfactory and should be approached with caution by Chinese audiences. Therefore, targeted countermeasures, such as reinforcing the video content review process, establishing online correction mechanisms, and encouraging users to selectively and carefully share controversial information, are urgently needed to enhance the credibility and impact of stroke-related educational content in the future.