Short videos platforms as sources of health information about cervical cancer screening: A content and quality analysis

Abstract

Background

Cervical cancer is a significant global health concern with over 662,000 new cases and approximately 349,000 deaths in 2022. Despite the clear benefits of screening, a portion of the population remains unaware of its importance. In China, short video platforms such as Kuaishou, Bilibili, and TikTok host numerous related videos, but the quality varies significantly.

Method

Using the keyword “cervical cancer screening,” the top 100 videos on each platform were searched (totaling 300), with 259 meeting the criteria. A comparative analysis was conducted on video duration, engagement metrics (likes, favorites, comments, shares), follower count, uploader identity, and video type. The Global Quality Score (GQS) and modified DISCERN tool were used for evaluation.

Results

The study included 82 Kuaishou videos, 93 Bilibili videos, and 84 TikTok videos. Bilibili had the longest median video duration (109 s), while Kuaishou had the shortest (54.5 s). Kuaishou outperformed TikTok and Bilibili in engagement metrics. TikTok had a higher proportion of videos on the importance, process, considerations, and timing of screening. Professional uploaders (obstetricians and gynecologists whose expertise directly pertains to cervical cancer screening) were most prevalent on TikTok (74%). TikTok videos had the highest quality scores in GQS and mDISCERN, followed by Bilibili and Kuaishou. Significant differences in mDISCERN scores were found among the platforms (all pairwise comparisons p < .001). Spearman rank correlation analysis showed that higher-quality videos (measured by GQS and mDISCERN) were more likely to achieve higher audience engagement. Still, video duration did not affect quality or engagement.

Conclusion

Social media platforms provide accessible health information, but the quality and reliability of cervical cancer screening videos vary significantly. Professionally uploaded videos generally have higher engagement and information reliability. Content creators should prioritize high-quality, accurate videos, and platforms should enhance content quality control to prevent misinformation dissemination.

Introduction

Cervical cancer is the fourth most common cancer affecting women's health globally, with over 662,000 new cases diagnosed and approximately 349,000 deaths in 2022. ¹ Unlike many other cancers, cervical cancer is considered one of the most preventable, primarily due to its strong association with high-risk human papillomavirus (HPV) and the availability of effective screening and vaccination strategies. ² In support of global elimination goals, the World Health Organization (WHO) has proposed that 70% of women aged 35 to 45 should undergo effective screening by 2030. However, participation remains suboptimal in many regions, hindered by factors such as limited health literacy, stigma, accessibility issues, and misinformation. In recent years, short-video platforms such as TikTok, Kuaishou, and Bilibili have become popular among younger audiences in China, who increasingly turn to them for health-related information.^3,4 While there are many videos about cervical cancer screening available, the quality and professionalism of these videos vary significantly. Health-related content on cervical cancer screening needs to cover specific elements, such as the importance of screening, the process, potential risks, and the timing of screening. ⁵ In particular, screening methods such as the Papanicolaou test (Pap smear), HPV testing, and visual inspection with acetic acid (VIA) are commonly recognized techniques recommended for early detection of cervical abnormalities. High-quality health content on short video platforms ideally provides substantive information on key aspects of cervical cancer screening, such as recognized methods, procedures, interpretation of results, and appropriate screening timing. However, the extent to which such content is available and reliable on these platforms remains unclear. Some creators upload videos about cervical cancer screening on social media, facilitating sharing and discussion among the public. ⁶ However, some of the information conveyed may be incorrect, subjective, or unprofessional.^7–9 Exposure to such misinformation can delay diagnosis and affect health outcomes. Despite the ubiquity of health-related content on social media platforms, there are limited studies assessing the quality of cervical cancer screening videos on these platforms.^8,9

On these short video platforms, cervical cancer screening-related content originates from a diverse range of uploaders. Some videos are posted by individual users, including young women sharing their personal screening experiences, aiming to document their health journeys or offer peer advice. Previous studies have shown that user-generated health content can influence peer attitudes and foster online discussion communities. ⁶ In contrast, other videos are produced by healthcare professionals, such as obstetricians and gynecologists, who use these platforms to disseminate professional medical knowledge, which is often perceived as more reliable and trustworthy. ¹⁰ Additionally, some content comes from official media accounts operated by public health authorities or hospitals, providing institution-backed health education. The intentions behind these uploads vary from personal sharing and emotional support to professional health education or public awareness campaigns. Previous research suggests that the source of information may significantly influence viewers’ trust, interpretation, and health-related behaviors, underscoring the need to assess not only the content itself but also the characteristics of the uploaders. ⁹

In recent years, short video platforms such as TikTok, Kuaishou, and Bilibili have rapidly gained popularity and become prominent sources of health-related information, particularly among younger demographics. Their visually engaging and easily accessible content formats enable the rapid dissemination of public health knowledge, often in a more relatable and digestible manner than traditional media. In China, these platforms are increasingly used for health education, disease awareness campaigns, and patient engagement, offering a novel channel for reaching underserved populations. Given their algorithm-driven nature and high user engagement, these platforms hold great potential for promoting health literacy, but also raise concerns about the accuracy and quality of the shared content.

Previous studies evaluating the quality of health-related videos on short video platforms have primarily focused on general health topics, such as Helicobacter pylori, ¹¹ breast cancer, ¹² or lung cancer, ¹³ with limited attention to cervical cancer screening specifically. Some studies analyzed video content from a single platform (e.g. TikTok or Bilibili) or focused solely on video content characteristics without considering the influence of uploader types. Moreover, few studies systematically applied established quality assessment tools, such as the Global Quality Scale (GQS) and the mDISCERN instrument, to evaluate the reliability and educational value of cervical cancer screening videos.

To address these gaps, the current study adopts a differentiated approach by:

1. Simultaneously analyzing cervical cancer screening-related videos across three major Chinese short video platforms (TikTok, Kuaishou, and Bilibili) to allow cross-platform comparisons.

2. Using a dual assessment framework (GQS and mDISCERN tools) to systematically evaluate both the overall quality and reliability of the videos.

3. Incorporating a detailed classification of uploader types, distinguishing between healthcare professionals (specialists), nonprofessionals, and individual users, to investigate how information source affects video quality and audience engagement.

By combining content analysis with uploader characterization, our study provides a more comprehensive understanding of both the quality and source-related factors influencing cervical cancer screening information dissemination via short video platforms. Our intended audience includes young women, individuals with limited access to traditional health education, and those who regularly consume health content on short video platforms.

In this study, “short videos for health information” refer to brief, publicly available videos—typically less than 5 min in duration—distributed via social media platforms, which aim to disseminate factual knowledge, promote health behaviors, or raise awareness about specific health topics, such as cervical cancer screening.

However, only a limited number of studies have systematically evaluated the quality of cervical cancer screening information available on short-video platforms. This gap is concerning because these platforms are increasingly influential in shaping public understanding and attitudes toward health. The dissemination of low-quality or misleading health content may result in misconceptions, reduce trust in evidence-based screening programs, and ultimately hinder screening participation. Given that cervical cancer screening uptake remains suboptimal in many regions, ensuring the accuracy and reliability of publicly accessible health information is essential for improving preventive behaviors. This study aims to address this gap by conducting a multiplatform analysis of cervical cancer screening-related videos using validated evaluation tools, thereby providing insights into the quality, engagement patterns, and source credibility of this emerging form of health communication.

Materials and methods

Study design, time, and location

This study was designed as a cross-sectional content analysis, aiming to evaluate the quality and reliability of cervical cancer screening videos across major Chinese short-video platforms. The study was conducted from 1 March 2025 to 30 April 2025. All video data were systematically collected from three widely used platforms in China: Kuaishou, Bilibili, and TikTok. Data collection and analysis were carried out jointly by researchers from the Physical Examination Center and Department of Obstetrics and Gynecology of Hebei General Hospital, the Department of Obstetrics and Gynecology of the Fourth Hospital of Shijiazhuang, and the Department of Obstetrics and Gynecology of the Second Hospital of Hebei Medical University, all located in Shijiazhuang, Hebei Province, China.

Ethical considerations

The data utilized in this study were exclusively sourced from publicly accessible videos on platforms including Kuaishou, Bilibili, and TikTok. TikTok emphasizes professionally created authoritative content, Bilibili is popular among younger demographics and known for longer educational videos, while Kuaishou typically prioritizes user engagement with shorter, more entertainment-focused content. This study was reviewed by the Institutional Ethics Committee of Hebei General Hospital. Given that all analyzed data were obtained from publicly accessible online videos without involving human participants, personally identifiable information, or sensitive data, the committee waived the requirement for formal ethical approval.

Video collection and selection

In China, Douyin (Chinese TikTok) incorporates a medical professional certification system and actively promotes content from verified healthcare professionals and official health authorities, especially for sensitive public health topics. This regulatory framework contributes to the platform algorithm's tendency to prioritize authoritative, professionally created health content. In contrast, Kuaishou's decentralized algorithm emphasizes content engagement (likes, comments, shares) rather than professional authority, resulting in greater visibility for self-media creators and individual users.^10,14 Bilibili, as a long-form video and knowledge-sharing platform, hosts both professional medical education content and popular self-media videos but lacks explicit prioritization of certified professional accounts.

For this study, we selected TikTok (Douyin), Kuaishou, and Bilibili based on their large user bases, high relevance for health information dissemination, and public accessibility. Other platforms such as WeChat or Xiaohongshu were excluded due to their semi-private nature, limited search functionality, or primarily image- and text-based formats.

In this context, a “short video for health information” is defined as a user- or organization-generated video clip of no more than 5 min in length, intended to provide educational, preventive, or promotional information related to health, particularly cervical cancer screening.

These videos were primarily intended for nonprofessional audiences, including women of screening-eligible age (typically 25–65 years old), general platform users interested in health-related content, and members of the public with limited access to conventional health education. Videos specifically targeting medical professionals or containing technical academic presentations were excluded.

These distinctions are specific to the Chinese versions of these platforms, and may differ from their international counterparts, such as TikTok in the United States.

The video data for this study were collected from three prominent Chinese short-video platforms: Kuaishou, Bilibili, and TikTok (Douyin). The search was conducted from 1 March to 30 April 2025, using the simplified Chinese keyword “宫颈癌筛查” (cervical cancer screening). All videos were in Mandarin Chinese or had Chinese subtitles; content in other languages was excluded.

To minimize personalization bias from platform algorithms, all searches were performed in a nonlogged-in browser, with cache and history cleared prior to each session. Videos were retrieved using each platform's default sorting mode, which typically reflects a mix of popularity, recency, and relevance, although the exact weighting is proprietary.

The top 100 videos per platform were screened based on predefined inclusion criteria to ensure content relevance. Although this method aimed to simulate a typical user experience, we acknowledge that the use of a single keyword may have limited the completeness of the dataset. Future studies should incorporate synonyms, lay terms, and alternative phrasing to broaden search sensitivity and capture a wider range of content (Figure 1).

Figure 1.

Search strategy for videos on cervical cancer screening.

Video characteristics

Video characteristics refer to the measurable and descriptive attributes of a video that provide information about its content, structure, popularity, and technical quality. These include features such as video length, source type, format (e.g. narration, animation), number of likes/comments/shares, and presence of audiovisual enhancements (e.g. subtitles, background music).

Uploaders characteristic

The uploaders of the videos were categorized into the following groups: doctors, other medical workers/student, official media, self-media, and doctors of traditional Chinese medicine (TCM).

Professionals: Obstetricians and gynecologists were classified as professionals, as cervical cancer screening directly falls within their clinical expertise and scope of practice. Nonprofessionals: This group included other licensed physicians (from nongynecology specialties), other medical professionals or students, official media accounts (operated by public health authorities, Centers for Disease Control (CDC), or tertiary hospitals), and doctors of TCM. Although possessing medical qualifications or institutional endorsement, they were classified as nonprofessionals as their expertise does not directly pertain to cervical cancer screening. Official media refers to accounts operated by recognized public institutions, such as the CDC, governmental health departments, or officially certified hospital accounts. These accounts disseminate institution-backed health information and are generally verified by the platforms. Personal users and self-media uploaders were classified as individual users.

Personal users refer to private individuals without any formal healthcare qualifications or institutional affiliation, such as ordinary users who share personal experiences or opinions. Self-media uploaders are a specific subset of individual users who create content professionally or semiprofessionally on social media for commercial or personal branding purposes but lack formal medical certification or official institutional endorsement. These include independent creators who operate under self-branded accounts but are not affiliated with recognized health organizations.

Video review and categorization

Two authors (Luyang Su and Jingrun Yao) independently reviewed the videos and excluded those that were duplicates or near-duplicates (i.e. videos that were identical or substantially similar in content, even if uploaded by different accounts or presented with minor modifications). This step was taken to avoid redundancy and ensure that each analyzed video represented unique content.

Videos were also excluded if they were irrelevant to cervical cancer screening based on predefined thematic categories. These included: (1) Focused solely on unrelated commercial promotion. (2) Presented general gynecological or health topics without specific reference to cervical cancer screening. (3) Mentioned cervical cancer screening only superficially (e.g. in hashtags or captions) without delivering substantive content. (4) Consisted exclusively of personal emotional narratives (e.g. expressing anxiety or fear) without conveying any factual information regarding the screening process or its importance. These videos were considered irrelevant in the context of this study's aim of evaluating the quality and reliability of educational information. Therefore, they were excluded from analysis.

Content analysis process

We conducted a cross-sectional content analysis of the top 100 videos retrieved from each of the three platforms (TikTok, Kuaishou, and Bilibili), using the Chinese keywords for “cervical cancer screening.” Two trained researchers (Luyang Su and Jingrun Yao) independently reviewed all videos to determine their eligibility. Duplicate, near-duplicate, and irrelevant videos (e.g. unrelated commercial content, superficial mentions without substantive information, or solely emotional narratives) were excluded based on predefined criteria.

Eligible videos were categorized according to six thematic domains derived from prior literature and clinical guidelines: (1) importance of screening, (2) screening methods (e.g. Pap smear, HPV test, VIA), (3) process of screening, (4) precautions, (5) appropriate screening time, and (6) interpretation of results. Videos were also coded based on their production style (e.g. solo narration, Q&A, animations), uploader type (e.g. professional, self-media), and engagement metrics.

For quality evaluation, we applied two validated tools: the GQS and an mDISCERN instrument. Both reviewers rated each video independently, and a third reviewer (Xiaohang Ai) resolved discrepancies. Interrater reliability was assessed using Cohen's kappa coefficient. The final scores were used for further analysis of platform differences and correlation with viewer engagement.

Style of video shooting

The style of video shooting was classified as solo narration, questions and answers (Q&A), PPT/Class, Animation/Action, and others.

Video quality assessment tools

The quality and reliability of the collected videos were systematically evaluated using two well-established assessment tools: the DISCERN instrument^15–17 and the GQS.^16–18

1. mDISCERN

   • What it measures: Adapted from the original DISCERN instrument (designed for written consumer health information), it evaluates five key dimensions: (1) clarity of purpose, (2) relevance of information, (3) accuracy of evidence, (4) balance and bias, and (5) overall reliability.
   • How scores are assigned: Each dimension is scored on a 5-point Likert scale (1 = poor, 5 = excellent). A total score is calculated (range 5–25), with higher scores indicating superior quality. The same dual-rater plus arbitration process was used.
   • Evidence of validity/reliability: The modified version has demonstrated strong content validity (CVI = 0.89) and inter-rater reliability (κ = 0.78–0.92) in prior assessments of short-video health content ¹⁴ (Supplemental Table S2).
2. GQS

   • What it measures: Overall educational value of a video, including accuracy, clarity, comprehensiveness, and usefulness for a lay audience.
   • How scores are assigned: Each video is rated on a 5-point Likert scale (1 = poor quality, 5 = excellent quality). Two independent reviewers (LS and JY) scored all videos; disagreements were resolved by a third arbitrator (XA).
   • Evidence of validity/reliability: GQS has been validated in multiple studies of social media health content, reporting interrater reliability (Cohen's κ) between 0.75 and 0.90. ¹³ In this study, κ values ranged from 0.81 to 0.89, indicating excellent agreement (Supplemental Table S1).

Two independent reviewers (Luyang Su and Jingrun Yao), both with expertise in public health and medical research, conducted the evaluations. To ensure consistency and minimize bias, the reviewers underwent a training session to familiarize themselves with the scoring criteria and application of the tools. A third arbitrator (Xiaohang Ai) assigned the final score if the two raters’ scores were inconsistent. Interrater reliability was assessed using Cohen's kappa coefficient (κ), which measures the level of agreement between reviewers. The κ values ranged from 0.75 to 0.89, indicating strong agreement across different categories and platforms. Any discrepancies in scoring were resolved through discussion and consensus.

The evaluation results were interpreted based on the mean scores for each dimension of the DISCERN tool and the GQS. Videos with an average DISCERN score ≥ 4.0 and a GQS score ≥ 4.0 were classified as high-quality, while those with scores below 3.0 were considered low-quality. Scores between 3.0 and 3.9 indicated moderate quality. Previous studies have validated the aforementioned tools, particularly in the context of social media platforms.^13,16,17 Detailed descriptions of these tools are provided in Additional File 1 and Additional File 2 in supplemental material.

Statistical analysis

Given the nonparametric distribution of the data, continuous variables were expressed as median and interquartile range (IQR). For comparisons between two groups, the Mann–Whitney U test was employed, while the Kruskal–Wallis H test was used for comparisons involving three or more groups. Interrater reliability between the two independent reviewers was assessed using Cohen's kappa coefficient (κ), with values interpreted as follows: < 0.20, poor; 0.21–0.40, fair; 0.41–0.60, moderate; 0.61–0.80, good; and > 0.81, excellent. The P values were all corrected by Bonferroni. The effect size, Cramer's V, ranges from 0 to 1. The interpretation criteria are as follows: 0.00–0.10: Weak association; 0.10–0.30: Moderate correlation; 0.30–0.50: Strong correlation; 0.50 and above: Strong correlation. Spearman's rank correlation analysis was conducted to evaluate the relationships between video variables (e.g. duration, likes, comments, shares, favorites, and follower count) and their association with video quality scores. A two-tailed p-value < .05 was considered statistically significant. All statistical analyses were performed using IBM SPSS Statistics 25, and data visualization was conducted using GraphPad Prism 9.

Results

Video characteristics

Our study included 82 Kuaishou videos, 93 from Bilibili, and 84 from TikTok, after excluding duplicate or near-duplicate videos and irrelevant content (Figure 1). All videos on Kuaishou, Bilibili, and TikTok were in Chinese or had Chinese subtitles. The characteristics of the videos from Kuaishou, Bilibili, and TikTok are detailed in Table 1. The analysis focused on several key metrics, including video duration, likes, collections, comments, shares, and follower counts. The median video duration varied significantly across platforms. Bilibili had the longest median video duration at 109.00 s (IQR: 65.00, 246.00), significantly higher than TikTok (77.00 (53.00, 110.00) and Kuaishou 54.50 (36.00,106.75). Kuaishou videos consistently exhibited higher engagement metrics across likes, collections, comments, shares, and follower counts compared to TikTok and Bilibili.

Table 1.

Characteristics of videos about cervical cancer screening on Bilibili/Kuaishou/TikTok.

Variables M (Q1, Q3)	Total (n = 259)	Bilibili (n = 93)	Kuaishou (n = 82)	TikTok (n = 84)	p
Likes	185.00 (16.50, 1216.00)	8.00 (1.00,41.00)	1064.00 (353.25,2716.75)	318.50 (58.00,1773.50)	<.001
Collections	58.00 (5.00, 392.00)	5.00 (1.00,44.00)	243.00 (85.25,568.50)	71.00 (7.50,608.75)	<.001
Comments	16.00 (1.00, 138.00)	0.00 (0.00,4.00)	73.50 (25.00,487.00)	27.00 (4.75,224.00)	<.001
Shares	58.00 (5.00, 456.50)	4.00 (1.00,29.00)	337.00 (90.00,719.50)	73.50 (7.00,629.00)	<.001
Video duration(s)	78.00 (47.50, 130.50)	109.00 (65.00,246.00)	54.50 (36.00,106.75)	77.00 (53.00,110.00)	<.001
Followers (10,000)	2.20 (0.17, 23.70)	0.13 (0.02,1.80)	13.45 (3.28,39.58)	5.55 (0.92,41.10)	<.001

Uploader characteristics

As detailed in the video characteristics section, a total of 259 eligible videos were analyzed across the three platforms. To avoid redundancy, we summarize uploader types here and refer readers to Figure 2 and Table 2 for full distributions. TikTok had the highest proportion of uploads from professionals (74%), while Bilibili had the largest proportion of self-media creators (44%). Kuaishou showed a more balanced mix of professionals (53%), nonprofessionals (25%), and individual users (22%). The statistical test results show that there are significant differences in the distribution of uploader types across different platforms (p < .001), with the Cramer's V value for doctors being 0.32, for official media being 0.31, and for self-media being 0.41. Further post-hoc tests indicate that: Doctors: Kuaishou and TikTok are significantly higher than Bilibili. Official media: TikTok is significantly higher than Kuaishou and Bilibili. Self-media: Bilibili is significantly higher than Kuaishou and TikTok (Table 2).

Figure 2.

(A) Numbers of video uploaders about cervical cancer screening on Bilibili/Kuaishou/TikTok (all the authors). (B) Numbers of video uploaders about cervical cancer screening on Bilibili/Kuaishou/TikTok (the certified authors).

Table 2.

Characteristics of video uploaders about cervical cancer screening on Bilibili/Kuaishou/TikTok.

Type of uploaders, n (%)	Bilibili (n = 93)	Kuaishou (n = 82)	TikTok (n = 84)	p-value	Cramer's V	Post-hoc test
Doctor	28(30.1%)	53(64.6%)	53(63.1%)	<.001	0.32	K > B;T > B
Other medical worker/student	13(14.0%)	6(7.3%)	5(6.0%)	.14	0.12
Official media	5(5.4%)	1(1.2%)	19(22.6%)	<.001	0.31	T > K;T > B
Self-media	41(44.1%)	18(22.0%)	2(2.4%)	<.001	0.41	B > K > T
Doctor of TCM	6(6.5%)	5(6.1%)	4(4.8%)	.88	0.03

Note: B: Bilibili; T: TikTok; K: Kuaishou.

The p-values were all corrected by Bonferroni.

The effect size, Cramer's V, ranges from 0 to 1.

The interpretation criteria are as follows: 0.00–0.10: weak association; 0.10–0.30: moderate correlation; 0.30–0.50: strong correlation; 0.50 and above: strong correlation.

Table 3 presents the video quality assessment for Bilibili, TikTok, and Kuaishou using the GQS and mDISCERN tools. Scores for each subscale (e.g. Content Accuracy, Usefulness, Engagement Quality) are included to improve interpretability. TikTok ranked highest in all GQS dimensions, especially in Content Accuracy and Engagement Quality, with an overall GQS score of 4, largely due to professional uploaders. Bilibili showed a more varied performance, with an average GQS of 3, as videos came from both professionals and self-media creators. Kuaishou, though scoring lowest in content accuracy, excelled in user engagement, reflecting its algorithm's emphasis on engagement.

Table 3.

Quality assessment of videos about cervical cancer screening on Bilibili, Kuaishou, and TikTok (with subscale scores).

Platform	GQS (content accuracy)	GQS (usefulness)	GQS (engagement quality)	mDISCERN (clarity of purpose)	mDISCERN (relevance of information)	mDISCERN (overall reliability)
Bilibili	3 (1–5)	2–4	3	2 (1–5)	2–4	2
TikTok	4 (1–5)	2–4	4	4 (1–5)	2–4	4
Kuaishou	3 (1–3)	2–4	3	2 (1–5)	2–4	2

Note: GQS: Global Quality Score.

The results indicate that videos uploaded by professionals and nonprofessionals generally had higher quality in terms of both engagement and information reliability compared to those uploaded by individual users (Table 4). Figures 3 and 4 further confirm this result.

Table 4.

Quality comparison between the videos uploaded by professionals/nonprofessionals/individual user.

	Professionals (N = 84)			Nonprofessionals (N = 93)			Individual users(N = 82)			p-value
Scores	M	Min-Max	Q1-Q3	M	Min-Max	Q1-Q3	M	Min-Max	Q1-Q3
GQS	3	1–5	2.25, 5	3	2–5	2，4	3	1–5	2, 3	.45
mDISCERN	3	1–5	2, 4	3	1–5	2，4	2	1–5	1, 3	.15

Note: GQS: Global Quality Score.

Figure 3.

Comparison of videos from different videos. (A): GQS score; (B): mDISCERN score.

GQS: Global Quality Score.

Figure 4.

Comparison of videos from different sources. (A): GQS score; (B): mDISCERN score.

GQS: Global Quality Score.

Video categorization

The categorization of videos is shown in Table 5. Compared with Bilibili and Kuaishou, TikTok had a higher proportion of videos focusing on the importance of screening (70.2%), the process of screening (71.4%), precautions for screening (85.7%), and the time for screening (67.9%). In terms of screening methods, Bilibili (79.6%) had a higher proportion than the other two platforms. Kuaishou had a slightly higher proportion of videos on the interpretation of screening results (34.1%) compared to the other two platforms. However, significant differences among all pairwise comparisons of the three platforms were only observed in the category of precautions for screening.

Table 5.

Categorization of videos about cervical cancer screening on Bilibili/Kuaishou/TikTok.

	Bilibili (n = 93)	Kuaishou (n = 82)	TikTok (n = 84)	p-value	Cramer's V	Post-hoc test
The important of screening	56 (60.2%)	46 (56.1%)	59 (70.2%)	.15	0.12
Screening methods	74 (79.6%)	34 (41.5%)	60 (60.2%)	<.001	0.34	B > K;T > K
Process of Screening	33 (35.5%)	27 (39.2%)	57 (71.4%)	<.001	0.32	T > B > K
Precautions for screening	14 (15.1%)	32 (39.0%)	72 (85.7%)	<.001	0.59	T > K > B
The time for screening	41 (44.1%)	16 (19.5%)	57 (67.9%)	<.001	0.39	T > B > K
Interpretation of screening result	18 (19.4%)	28 (34.1%)	28 (33.3%)	.05	0.15
Style of video shooting, n (%)
Solo narration	47 (50.5%)	36 (43.9%)	53 (63.1%)	.06	0.16
Question and answer	3 (3.2%)	9 (11.0%)	8 (9.5%)	.12	0.13
PPT or class	8 (8.6%)	2 (2.4%)	8 (9.5%)	.15	0.12
procedure	19 (20.4%)	11 (13.4%)	6 (7.1%)	.04	0.16
Animation/action	10 (10.8%)	13 (15.9%)	3 (3.6%)	.03	0.16
Others	5 (5.4%)	11 (13.4%)	6 (7.1%)	.14	0.12

Note: The p-values were all corrected by Bonferroni.

The effect size, Cramer's V, ranges from 0 to 1.

The interpretation criteria are as follows: 0.00–0.10: Weak association; 0.10–0.30: Moderate correlation; 0.30–0.50: Strong correlation; 0.50 and above: Strong correlation.

Regarding the style of video shooting, solo narration was the most common style, and the procedure of cervical cancer screening was more prevalent in long-video platforms.

Correlation analysis

Spearman's rank correlation analysis was conducted to explore the relationships between video quality metrics GQS and mDISCERN and various engagement metrics (likes, collections, comments, shares) as well as video duration (Table 6). The results analysis indicates that higher-quality videos (as measured by GQS and mDISCERN) are more likely to receive higher engagement from viewers, as evidenced by likes, collections, comments, and shares. However, video duration does not appear to influence either video quality or viewer engagement. These findings highlight the importance of producing high-quality content to enhance viewer engagement and information dissemination on social media platforms.

Table 6.

Spearman correlation between video quality and audience interaction on Bilibili/Kuaishou/TikTok.

	1	2	3	4	5	6	7
1 GQS	1
2 mDISCERN	0.886***	1
3 Likes	0.292***	0.345***	1
4 Collections	0.213**	0.232***	0.939***	1
5 Comments	0.271***	0.313***	0.937***	0.886***	1
6 Shares	0.240***	0.272***	0.923***	0.939***	0.886***	1
7 Video duration	0.135*	0.063	−0.089	0.006	−0.074	−0.07	1

Note: GQS: Global Quality Score.

*p < .05, **p < .01, ***p < .001.

Discussion

Principal findings

Our study revealed that TikTok videos had the highest quality, primarily due to the high proportion of professional uploaders, which led to greater content reliability and higher engagement. Kuaishou, while showing higher user interaction, exhibited lower video quality due to its focus on engagement over authority. Bilibili presented a balanced mix of content quality and engagement but lagged behind TikTok in both respects. These findings highlight the tension between content quality and user engagement on social media platforms.

Comparison to prior work

Our results align with previous studies that emphasize the importance of professional content in health communication. However, unlike prior research focused on single platforms, our study offers a cross-platform comparison, shedding light on how different algorithms influence content visibility and engagement. The tradeoff between high engagement and content quality found in Kuaishou, in particular, has been underexplored in prior studies.

Cervical cancer screening and public health importance

Cervical cancer remains a major public health issue, with early screening proven to reduce mortality. However, many women remain unaware of the importance of regular screenings. In China, cervical cancer incidence is high, and the WHO's goal of 70% screening uptake among women aged 35–45 by 2030 faces challenges like limited health literacy and stigma.

Digital platforms’ potential in promoting screening

Short-video platforms such as TikTok, Kuaishou, and Bilibili are increasingly influential in spreading health information. These platforms offer engaging and accessible ways to raise awareness about cervical cancer screening. Our study shows that these platforms can educate users on screening methods and timing, especially among younger, tech-savvy demographics.

Risks of misinformation

Despite their potential, these platforms also pose risks of misinformation. Our findings indicate that videos on platforms like Kuaishou, while highly engaging, often prioritize entertainment over content accuracy. This can lead to the spread of unverified information, especially when algorithmic recommendations favor viral content, regardless of its reliability. This underscores the need for stronger content quality control to ensure the dissemination of accurate health information.

Strengths and limitations

Strengths include the use of robust assessment tools and comprehensive analysis. Limitations are the cross-sectional design and focus on Chinese platforms, limiting generalizability. It should be noted that certain videos retrieved under the cervical cancer screening keyword, but excluded from our analysis due to lack of substantive information, may still influence public attitudes and perceptions through emotional framing or incidental exposure. Future research should explore the potential psychological and behavioral effects of such noninformative or emotionally charged content, as these may indirectly affect screening participation or health literacy.

To improve public health communication, platforms should prioritize high-quality, authoritative health content, particularly on platforms like Kuaishou, where engagement tends to promote lower-quality videos. Collaborations between health authorities and platforms to certify content could help reduce misinformation and increase the reliability of health messages. Additionally, platform algorithms could be adjusted to favor content with higher informational value, balancing engagement with accuracy.

This study was cross-sectional and focused on Chinese platforms, limiting its generalizability. Future research should adopt a longitudinal approach and explore additional platforms and regions. Furthermore, investigating the psychological impact of emotionally charged or inaccurate content would be valuable, as such videos could influence public health behaviors. Long-term studies could also track how algorithmic changes affect content quality and user engagement over time.

Future directions

Future research should explore the impact of algorithmic changes and interventions to improve health information quality on social media.

The advent of short-video platforms such as Bilibili, Kuaishou, and TikTok has fundamentally reshaped how health information is disseminated to the public, particularly within younger demographics. ^{17 19–21} With their ability to convey complex health messages through engaging, concise formats, these platforms have become indispensable tools for public health communication. ²² Yet, their role in promoting health literacy—especially on critical topics like cervical cancer screening—remains a relatively underexplored area of research. In this study, a total of 300 cervical cancer screening-related videos were initially retrieved from the three platforms. After excluding irrelevant, duplicate, or near-duplicate videos, 259 eligible videos remained for further analysis. These videos were evaluated to investigate the quality of health information, levels of user engagement, and the influence of different types of content creators. Our findings reveal significant variations across platforms, offering new insights into both the potential benefits and challenges of using short-video platforms for health communication.

One of the key findings of our study is the significant variation in video quality across platforms. TikTok, with its higher median GQS and DISCERN score, had relatively better content quality in our sample. This can be partly attributed to the fact that a large majority of the videos (approximately three-quarters) were uploaded by medical professionals. However, we acknowledge that TikTok's algorithm is not publicly transparent, and prior studies have indicated that, like other platforms, it can also contribute to the spread of misinformation when such content garners significant engagement. Therefore, our findings reflect observational associations in the sampled data, rather than conclusive evidence about the algorithm's prioritization mechanisms. This aligns with previous research indicating that content created by healthcare professionals is perceived as more reliable and trustworthy by viewers. ¹⁴ In contrast, Kuaishou's algorithm, which emphasizes engagement over content quality, resulted in a higher level of user interaction (median likes = 15.2k), even though the videos had lower quality scores. This suggests that Kuaishou's user base values accessibility and entertainment over authoritative content, with nonprofessional creators generating significantly higher engagement than on TikTok or Bilibili. ^{17 23–25}

The quality of health content is further influenced by the type of content creator. Professional creators generally produced higher-quality videos, as indicated by their superior GQS and DISCERN scores. However, on Bilibili, the combination of professional and self-media creators resulted in more varied content, with an average GQS of 3. This indicates that the videos were of moderate quality, offering fair educational value but lacking in comprehensive coverage of cervical cancer screening topics. While the videos provided some useful information, they often missed important details or presented the material in a less clear or balanced manner. This suggests that, on average, the videos were adequate but could benefit from more thorough and accurate content to enhance their educational value. This variability highlights the challenge of balancing different creator types, as nonprofessional content may lack the accuracy required for health communication. ²⁶ Interestingly, despite their lower quality scores, nonprofessional creators on Kuaishou garnered more engagement, indicating that users on this platform may prioritize relatability and entertainment over factual accuracy.^23,27 The influence of platform algorithms is a double-edged sword. Although TikTok videos in our sample showed higher quality scores, this does not imply that the platform systematically promotes reliable health content.^28,29 TikTok's algorithm, like others, remains opaque and engagement-driven, making it susceptible to amplifying sensational or misleading information—particularly when such content garners high user interaction.^30,31 Similarly, on Kuaishou, 35% of the most-engaged videos contained unverified screening advice from nonprofessionals. These findings underscore the risks of algorithmic amplification and the need for greater transparency and accountability across all platforms. ³²

The findings underscore a key challenge in health communication: the tradeoff between content quality and user engagement. Users are more likely to engage with content that is emotionally compelling or entertaining, even if it lacks authoritative information. ²³ This tendency is evident on platforms like Kuaishou, where nonprofessional content receives substantial engagement despite its lower quality. The high engagement on nonprofessional videos, particularly on TikTok and Kuaishou, points to the importance of crafting health messages that resonate emotionally with audiences, while also ensuring they are factually accurate. This dynamic suggests that health advocacy on short-video platforms needs to strike a balance between entertainment value and content reliability. ²⁴ Platforms should consider algorithmic interventions that prioritize the quality of health content, such as incorporating metrics like DISCERN or GQS into recommendation algorithms, ensuring that authoritative content has greater visibility without sacrificing user engagement. Such adjustments could help address the challenges posed by algorithmic amplification of low-quality or misleading content. ²⁵

The findings underscore a key challenge in health communication: the tradeoff between content quality and user engagement. Users are more likely to engage with content that is emotionally compelling or entertaining, even if it lacks authoritative information.^33,34 This tendency is evident on platforms like Kuaishou, where nonprofessional content receives substantial engagement despite its lower quality. The high engagement on nonprofessional videos, particularly on TikTok and Kuaishou, points to the importance of crafting health messages that resonate emotionally with audiences, while also ensuring they are factually accurate. This dynamic suggests that health advocacy on short-video platforms needs to strike a balance between entertainment value and content reliability. ³⁵ Platforms should consider algorithmic interventions that prioritize the quality of health content, such as incorporating metrics like DISCERN or GQS into recommendation algorithms, ensuring that authoritative content has greater visibility without sacrificing user engagement. Such adjustments could help address the challenges posed by algorithmic amplification of low-quality or misleading content. ³¹

Cultural differences in health communication should be considered when evaluating health communication strategies.^24,36,37 In some regions, health communication is more formal and professional, with users trusting content from recognized health organizations. In contrast, platforms in China like Kuaishou prioritize informal, relatable content, even when it lacks professional authority. ³⁸ This cultural distinction highlights the need to tailor health communication strategies to specific user preferences while maintaining accuracy. Understanding how these cultural differences affect content engagement and trust in health information is essential for designing effective global health communication strategies.

While this study provides valuable insights into the dynamics of health communication on short-video platforms, it is not without limitations.

First, this study was based on data collected from Chinese-language platforms (Kuaishou, Bilibili, and TikTok), which may limit the generalizability of the findings to other linguistic or cultural contexts. Health communication strategies on platforms such as YouTube or Instagram—commonly used in Western settings—may involve different recommendation algorithms, content presentation styles, and audience engagement patterns. Additionally, we used only a single Chinese search term (“宫颈癌筛查”) to identify relevant videos. This may have resulted in the exclusion of videos using alternative expressions or lay terminology. Future studies should consider incorporating synonyms and vernacular variations to improve search sensitivity and content inclusivity.

Second, the study employed a cross-sectional design, capturing a snapshot of health communication at one point in time. Given the dynamic nature of short-video platforms and the continuous evolution of platform algorithms and user behavior, longitudinal studies are warranted to monitor temporal trends and evaluate the potential impact of algorithmic modifications on the dissemination and quality of health-related content.

Third, experimental designs that manipulate algorithmic ranking mechanisms may provide valuable insights into how such changes influence user engagement patterns and exposure to high-quality content.

Fourth, although we observed statistically significant associations between video quality and audience engagement metrics, the strength of these correlations ranged from weak to moderate. These findings should be interpreted with caution. Correlation measures the strength and direction of the relationship between two variables, indicating whether they tend to increase or decrease together. However, it is important to note that correlation does not imply causality. In other words, even though two variables may be related, this does not mean that one causes the other. Additionally, correlation on its own does not provide predictive validity or indicate the nature of the relationship.

Fifth, as our findings are derived from cross-sectional data, the observed associations may be influenced by unmeasured confounding variables, such as algorithm-driven promotion, presentation style, or the demographic characteristics of platform users. Hence, the correlations observed should be regarded as indicative rather than conclusive evidence of the relationship between content quality and viewer interaction.

Finally, although we analyzed uploader type and follower count, information on the geographic origin of the accounts and their channel-level thematic focus (e.g. health-specific vs general content) was not consistently available. Future research may benefit from including these dimensions to better characterize uploader influence and content intent.

This study highlights both the potential and the pitfalls of using short-video platforms for health communication. While platforms like TikTok can leverage their algorithms to amplify high-quality content and professional expertise, Kuaishou's algorithm-driven engagement prioritizes virality over content reliability, which can lead to the spread of misinformation. Our findings suggest that the success of health advocacy on these platforms depends not only on content quality but also on understanding platform mechanics, user behavior, and the influence of content creators. The insights gained from this research can inform future strategies for enhancing health literacy in the digital age, emphasizing the importance of algorithmic accountability, creators’ certification, and user-centric design in promoting accurate and engaging health content. Despite its limitations, this study provides a foundation for future research on health communication in algorithm-driven environments, offering evidence-based pathways to harness the power of short-video platforms for public health.

Conclusion

This study highlights the complex dynamics of health communication in the era of short-video platforms. For example, TikTok stands out for its focus on authoritative content, resulting in higher content quality. In contrast, Kuaishou's engagement-driven algorithm tends to promote lower-quality content, which may inadvertently contribute to the spread of misinformation. These findings emphasize the need for a more nuanced approach to health communication on social media. The objective should not only be to engage users but also to ensure the accuracy and reliability of the information shared. To achieve this, public health authorities must work closely with platform developers to implement quality control measures, such as flagging or downranking misleading health content. By doing so, we can more effectively promote health literacy and prevent the dissemination of false information.

Abbreviations

GQS

Global Quality Score

mDISCERN

modified DISCERN

PPT

powerpoint

HPV

human papillomavirus