YouTube, TikTok, and Bilibili as sources of medical information on lumbar spinal stenosis

Abstract

Lumbar spinal stenosis (LSS) is a common degenerative condition that impairs mobility and quality of life, particularly in older adults. Video-sharing platforms are increasingly used for health information, yet the quality and reliability of LSS-related videos and cross-platform differences remain unclear. This study compares the content, uploader characteristics, and information quality of LSS-related videos on YouTube, TikTok, and Bilibili and examines associations between video quality and audience engagement. A cross-sectional search of YouTube, TikTok, and Bilibili was conducted on February 1, 2026, using relevant Chinese and English keywords. Of the retrieved videos, 253 met the eligibility criteria. Reliability, overall quality, understandability, actionability, and audiovisual quality were assessed using the modified DISCERN instrument, Global Quality Score, Patient Education Materials Assessment Tool understandability and Patient Education Materials Assessment Tool actionability score scores, and Video Information and Quality Index (VIQI). The sample comprised 89 YouTube, 80 Bilibili, and 84 TikTok videos, with median durations of 389, 171, and 103 seconds, respectively. TikTok videos were the shortest and showed relatively high engagement across available metrics. YouTube videos more frequently covered etiology and prevention and were predominantly uploaded by hospitals/departments and verified physicians; Bilibili and TikTok videos focused mainly on treatment, and TikTok had the highest proportion of verified physician uploaders. YouTube videos had higher modified DISCERN instrument and Global Quality Scores. Total VIQI scores were higher on YouTube and TikTok than on Bilibili, whereas Patient Education Materials Assessment Tool actionability scores did not differ significantly across platforms. Associations between video quality and audience engagement varied by platform, with VIQI showing the most consistent positive correlations. The quality and dissemination characteristics of LSS-related videos differed across platforms. YouTube videos were more reliable, TikTok videos showed greater dissemination appeal, and Bilibili showed no clear advantage in most quality metrics. High audience engagement did not necessarily indicate medical reliability. Platform recommendation systems should consider uploader qualifications, content reliability, and presentation quality to promote high-quality LSS information.

1. Introduction

1.1. Lumbar spinal stenosis

Lumbar spinal stenosis (LSS) is a spinal disorder primarily caused by degenerative changes in the intervertebral discs, facet joints, and ligamentum flavum.^[1]Its typical clinical symptoms include low back pain, radiating pain in the lower extremities, numbness, and neurogenic claudication.^[2] These symptoms severely limit the ability of older patients to perform activities of daily living and significantly reduce their quality of life. Epidemiological studies have shown that the prevalence of LSS is approximately 11% in the general population and may reach 25% to 39% among clinical populations presenting with low back and leg pain, with the prevalence increasing significantly with age.^[3]

In recent years, video content has gradually replaced traditional text-based information and has become a dominant form of communication on social media. In particular, during the coronavirus disease 2019 pandemic, the influence of social media on public health awareness increased substantially, and its role in patients’ access to health information became increasingly important.^[4] Against this background, the rapid evolution of the digital health information environment has made video-sharing platforms represented by YouTube, TikTok, and Bilibili important channels through which the public obtains knowledge about LSS and related diseases.^[5] With ecosystems primarily based on user-generated content and strong interactivity, video-sharing platforms such as YouTube, TikTok, and Bilibili have become important channels for the public to actively search for and learn medical knowledge.^[6,7] However, this model of autonomous information acquisition is also accompanied by substantial risks. Health information on these platforms varies widely in quality, and inaccurate or misleading content may lead patients to make inappropriate health decisions, thereby adversely affecting their clinical outcomes.^[8,9] Therefore, evaluating and improving the quality of health-related video content is essential for guiding the public toward reliable knowledge and promoting effective health communication.^[10]

Video duration and presentation style may affect viewers’ attention, information comprehension, and engagement behaviors, whereas uploader type may influence viewers’ judgments regarding the credibility of the information source. Therefore, this study used the modified DISCERN instrument (mDISCERN), Global Quality Score (GQS), Patient Education Materials Assessment Tool understandability score (PEMAT-U) and Patient Education Materials Assessment Tool actionability score (PEMAT-A), and Video Information and Quality Index (VIQI) to comprehensively evaluate LSS-related videos in terms of reliability, overall quality, understandability, actionability, and audiovisual quality.

Previous studies have assessed the quality of disease-related health information on social media platforms. For example, Tu et al evaluated the reliability and quality of knee osteoarthritis-related videos on TikTok and Bilibili; Jang et al conducted a cross-sectional analysis of the reliability, quality, and educational suitability of scoliosis exercise videos on TikTok; and Hong et al analyzed the sources, content characteristics, and associations with quality indicators of YouTube educational videos on cervical spondylosis.^[11–13] However, systematic evaluations of LSS-related videos remain limited, particularly direct comparisons across YouTube, TikTok, and Bilibili. Therefore, this study aimed to systematically evaluate the content characteristics, uploader characteristics, and information quality of LSS-related videos on these 3 platforms, compare differences among platforms, and analyze the association between video quality and audience engagement. The findings may provide a basis for the public to identify reliable LSS-related health information and may offer a reference for content creators and platform managers to optimize the production and dissemination strategies of medical videos.

2. Methods

2.1. Ethical review

This study was an observational content analysis based on publicly available online information, and all data were obtained from publicly available video content on YouTube, Bilibili, and TikTok. No private or sensitive personal data were collected. Only publicly displayed video- and account-level information, including uploader category, follower count, and verification status, was analyzed. No individual uploader was identified in the reported results. Because this study analyzed only publicly available data and did not involve any interaction or intervention with users, ethics committee approval was not required, in accordance with the Declaration of Helsinki and relevant research ethics standards stating that studies using publicly available data may be exempt from ethical review.

2.2. Search strategy

All video data were obtained from YouTube, TikTok, and Bilibili. These 3 platforms were selected because they are widely used video-sharing platforms and represent different language environments and video formats, making them suitable for cross-platform comparisons of LSS-related health information. The data collection cutoff date was February 1, 2026, and searches on all 3 platforms were completed on the same day. To minimize the influence of personalized recommendation algorithms, separate independent accounts were created for each platform, and browsing and search histories were cleared before the searches. No additional filters were applied during the search process, and the default ranking algorithm of each platform was used. The Chinese keyword “腰椎椎管狭窄” was used on TikTok and Bilibili, whereas the English keyword “LSS” was used on YouTube. The top 100 videos ranked by default on each platform were retrieved for screening.

2.3. Inclusion and exclusion criteria

The inclusion criteria were as follows: publicly available videos; videos directly related to LSS; and videos covering at least 1 aspect of disease introduction, etiology or anatomy, symptoms, diagnosis, treatment, prevention, or prognosis. The exclusion criteria were duplicate videos, videos unrelated to LSS, videos lacking valid medical information, and videos published within 7 days before the search date. This 7-day exclusion criterion was applied during the original screening of the top 100 videos retrieved from each platform because the audience engagement metrics of recently published videos, such as views, Thumbs-up, and comments, might not yet have stabilized.

2.4. Video quality assessment

To assess the reliability, overall quality, understandability, actionability, and audiovisual quality of information related to LSS presented in each video, video quality was evaluated using mDISCERN, GQS, VIQI, PEMAT-U, and PEMAT-A. The mDISCERN was used to assess the reliability of video information and consists of 5 items, with 1 point assigned for each “yes” response and 0 points for each “no” response, yielding a total score of 0 to 5.^[14] The GQS uses a 5-point scale to assess the overall quality of videos, with total scores ranging from 1 to 5.^[6] The VIQI includes 4 dimensions: information flow, information accuracy, video quality, and information precision, with each dimension scored from 1 to 5 and a total score ranging from 4 to 20.^[15] PEMAT-U and PEMAT-A were used to assess video understandability and actionability, respectively, with scores presented as percentages.^[16] After standardized training and preliminary scoring, 3 evaluators with licensed physician qualifications (Rui Guo, Jiachen Liang, and Yiliu Xu) independently assessed all included videos. Interrater agreement for scale scores was assessed using the intraclass correlation coefficient (ICC).^[17] Discrepancies in assessment results were first resolved through discussion; if consensus could not be reached, the final decision was made by a chief physician (Qing Sun) who was not involved in the initial independent scoring.

2.5. Uploader characteristics

Publicly displayed uploader account characteristics, including follower count, verification status, and uploader category, were systematically collected. Uploader categories and professional qualifications were determined based on publicly available account information and platform verification information. Uploaders were classified as physicians, hospitals/departments, commercial institutions, official media, self-media accounts, or individual users. Accounts verified as licensed physicians were classified as physicians; official accounts of hospitals or clinical departments were classified as hospitals/departments; and commercial institutions and official media were classified according to their publicly disclosed institutional identities. Unverified accounts primarily engaged in the creation and dissemination of health-related content were classified as self-media accounts, whereas the remaining unverified personal accounts were classified as individual users.

2.6. Video style assessment

To compare the presentation characteristics of LSS-related information across different platforms, video styles were classified into the following categories: personal narration, question-and-answer format, PowerPoint presentation (PPT)/classroom presentation, real medical scenarios, animation/action, show/documentary TV, and other formats.

2.7. Statistical analysis

All statistical analyses were performed using the Statistical Package for the Social Sciences, version 27.0. The normality of continuous variables was assessed using the Shapiro–Wilk test. Data with a normal distribution were presented as mean ± standard deviation, whereas non-normally distributed data were presented as median and interquartile range M [P₂₅, P₇₅]. Pairwise comparisons among the 3 platforms were performed using the Mann–Whitney U test with Bonferroni correction, with P < .0167 considered statistically significant.^[18] Categorical variables were expressed as frequencies (n) and percentages (%), and comparisons between groups were performed using the χ² test or Fisher exact test. Interrater agreement was assessed using ICC for scale scores. Spearman correlation analysis was used to evaluate the relationships between video quality scores and audience engagement metrics across platforms. The absolute value of the Spearman correlation coefficient was used to determine the correlation strength: |r| ≤ 0.20 indicated no or very weak correlation; 0.20 < |r| ≤ 0.40 indicated weak correlation; 0.40 < |r| ≤ 0.60 indicated moderate correlation; 0.60 < |r| ≤ 0.80 indicated strong correlation; and |r| > 0.80 indicated very strong correlation. To further evaluate the robustness of the correlation results, partial Spearman correlation analyses were performed after controlling for video duration. Multiple correlation tests were also adjusted using the Bonferroni correction. Unless otherwise specified, a 2-sided P < .05 was considered statistically significant.

3. Results

3.1. Basic video characteristics

After excluding duplicate and irrelevant content, 89 YouTube videos, 80 Bilibili videos, and 84 TikTok videos were ultimately included in this study (Fig. 1). All YouTube videos were in English or had English subtitles, whereas all Bilibili and TikTok videos were in Chinese or had Chinese subtitles. The basic characteristics of videos across the 3 platforms are shown in Table 1. The Shapiro–Wilk test showed that all continuous variables were non-normally distributed.

Figure 1.

Search strategy and video screening process.

Table 1.

Characteristics of videos across YouTube, Bilibili and TikTok.

Variable	YouTube	Bilibili	TikTok	PY-B	PB-T	PY-T
Video length	389 (198, 691)	171 (91, 388)	103 (50.5, 173.25)	<.001	<.001	<.001
Views	49,134 (4112.5, 129543)	2114.5 (512.5, 7802.25)	–	<.001	–	–
Thumbs up	419 (31, 1879)	26 (7, 129.75)	871.5 (298.5, 2701.5)	<.001	<.001	.0023
Comments	36 (1, 154)	2 (0, 14)	50.5 (13.25, 109.5)	<.001	<.001	<.001
Collections	–	36 (6.25, 221)	339 (102.25, 1398.5)	–	<.001	–
Shares	–	12 (2, 67)	138 (46.5, 817.5)	–	<.001	–
Followers	89,200 (5305, 868,000)	15,000 (5010, 24,000)	123,000 (66,500, 381,250)	<.001	<.001	.073

Pairwise comparisons were performed using the Mann–Whitney U test with Bonferroni correction.

M = median, P25 = 25th percentile, P75 = 75th percentile, P_B-T = Bilibili versus TikTok, P_Y-B = YouTube versus Bilibili, P_Y-T = YouTube versus TikTok.

TikTok videos had the shortest median duration, at 103 (50.5–173.25) seconds, followed by Bilibili videos at 171 (91–388) seconds and YouTube videos at 389 (198–691) seconds. After Bonferroni correction, all pairwise comparisons of video duration among the 3 platforms showed statistically significant differences. Audience engagement metrics also differed across platforms; however, some indicators could not be directly compared across all 3 platforms because of limitations in platform data availability. Specifically, TikTok did not provide view count data, and YouTube did not provide collection or share counts. Therefore, among the comparable indicators, YouTube videos had higher view counts than Bilibili videos, whereas TikTok videos had higher numbers of Thumbs-up, comments, collections, and shares than Bilibili videos. Overall, Bilibili showed relatively lower audience engagement across the available engagement metrics.

3.2. Uploader characteristics and video styles

The distribution of uploader categories differed markedly across platforms (Fig. 2). YouTube had 62 uploaders, Bilibili had 52, and TikTok had 60. Uploaders on YouTube and TikTok had higher median follower counts than those on Bilibili (P < .001). On YouTube, hospitals/departments and verified physicians were the predominant uploader types. On Bilibili, physicians constituted the largest uploader group, followed by individual users. TikTok videos were also predominantly uploaded by physician accounts, and TikTok had the highest verification rate (95.2%). In addition, YouTube had the most diverse range of uploader types, including hospitals/departments, verified physicians, official media outlets, for-profit institutions, and individual users.

Figure 2.

Proportions of uploader types for lumbar spinal stenosis-related videos across different platforms. Professional entities include for-profit enterprises, hospitals/departments, licensed physicians, and official media outlets. Nonprofessional entities include individual users and self-media accounts.

3.3. Content themes

The distribution of video content themes is presented in Table 2 and Figure 3A. Overall, Treatment-related content was common across platforms. YouTube videos most frequently addressed the etiology/prevention category, whereas Bilibili and TikTok videos focused primarily on treatment, followed by examinations/diagnosis. YouTube videos more frequently covered etiology and prevention, with a higher proportion than those on Bilibili and TikTok. In contrast, videos on Bilibili and TikTok focused more on treatment-related content, and diagnosis/examination-related content was also common. TikTok videos mainly addressed treatment, diagnosis/examination, and symptom-related information and more frequently used real medical scenarios as the presentation format. Epidemiological content appeared only in a small number of YouTube videos, whereas prognosis-related content accounted for a relatively low proportion across all 3 platforms.

Table 2.

Content classification of lumbar spinal stenosis-related videos on YouTube, Bilibili, and TikTok.

Platform	YouTube	Bilibili	TikTok	PY-B	PB-T	PY-T
Video content themes,n(%)
Anatomy	0 (0%)	7 (8.7%)	5 (5.9%)	.005 †	.492*	0.025†
Etiology/prevention	37 (41.6%)	4 (5%)	5 (5.9%)	<.001 *	1.000†	<.001 *
Epidemiology	2 (2.2%)	0 (0%)	0 (0%)	.498†	1.000†	.498†
Symptoms	9 (10.1%)	5 (6.2%)	9 (10.7%)	.363*	.306*	.897*
Examinations/Diagnosis	5 (5.6%)	16 (20%)	20 (23.8%)	.005 *	.556*	<.001 *
Treatment	33 (37.1%)	44 (55%)	42 (50%)	.019*	.522*	.087*
Prognosis	3 (3.4%)	4 (5%)	3 (3.5%)	.709†	.715†	1.000†
Style of video shooting, n(%)
Solo narration	53 (59.6%)	14 (17.5%)	23 (27.4%)	<.001 *	.130*	<.001 *
Question & answer	4 (4.5%)	6 (7.5%)	6 (7.1%)	.520†	.930†	.527†
PPT/classroom presentation	10 (11.2%)	20 (25%)	6 (7.1%)	.019*	.002 *	.353*
Animation/action	8 (9%)	15 (18.8%)	9 (10.7%)	.065*	.146*	.703*
Medical scenarios	10 (11.2%)	24 (30%)	40 (47.6%)	.002 *	.021*	<.001 *
TV show/documentary	3 (3.4%)	0 (0%)	0 (0%)	.248†	1.000†	.246†
Others	1 (1.1%)	1 (1.3%)	0 (0%)	1.000†	.488†	1.000†

P_B-T = Bilibili versus TikTok, PPT = PowerPoint presentation, P_Y-B = YouTube versus Bilibili, P_Y-T = YouTube versus TikTok, TV = television.

Bold values indicate statistical significance after Bonferroni correction (P < .0167).

^*Pearson chi-squared test.

^†Fisher’s exact test (two-sided).

Figure 3.

(A) Distribution of content themes in lumbar spinal stenosis videos on YouTube, Bilibili, and TikTok. (B) Video style distribution of lumbar spinal stenosis-related videos on YouTube, Bilibili, and TikTok.

Video presentation styles also differed across platforms (Table 2 and Figure 3B). Solo narration was the predominant format on YouTube, whereas real medical scenarios were more commonly used on Bilibili and TikTok. PPT/classroom-style presentations were more common on Bilibili than on TikTok.

3.4. Video quality scores

The results of the video quality assessment are presented in Table 3. YouTube videos received higher mDISCERN and GQS scores, indicating better information reliability and overall quality. Total VIQI scores were higher on YouTube and TikTok than on Bilibili, with no significant difference between YouTube and TikTok. PEMAT-A scores did not differ significantly across the 3 platforms. Bilibili videos did not demonstrate clear advantages in most video quality assessment metrics. Interrater agreement for all scale scores ranged from good to excellent. The ICC values for mDISCERN, GQS, the total VIQI score, PEMAT-U, and PEMAT-A were 0.896, 0.911, 0.961, 0.998, and 0.998, respectively.

Table 3.

Quality scores of lumbar spinal stenosis-related videos across YouTube, Bilibili, and TikTok.

Platform	YouTube		Bilibili		TikTok		P-value
Scores	M	P25-P75	M	P25-P75	M	P25-P75	PY-B	PB-T	PY-T
PEMAT-U	73	60–77	69	55–77	75	55–85	.362	.098	.241
PEMAT-A	50	50–75	57	43–67	64	40–76	.074	.083	.744
VIQI-sum	10	9–13	9	6–11	11	9–14	<.001	<.001	.294
VIQI-1	4	3–4	2	1–3	3	3–4	<.001	<.001	.191
VIQI-2	3	2–3	2	2–3	3	2–3	<.001	.019	.319
VIQI-3	2	1.5–2	2	1–3	3	2–3	.182	.018	<.001
VIQI-4	3	2–3	2	2–3	3	2–4	.036	.013	.514
GQS	3	3–4	3	2–3	3	2–3	<.001	.011	.016
mDISCERN-sum	4	3–4	3	2–3	3	2–3	<.001	.108	<.001

Pairwise comparisons were performed using the Mann–Whitney U test with Bonferroni correction. Bold values indicate statistical significance after Bonferroni correction (P < .0167).

GQS = global quality score, M = median, mDISCERN = modified DISCERN instrument, P₂₅ = 25th percentile, P₇₅ = 75th percentile, P_B-T = Bilibili versus TikTok, PEMAT-A = patient education materials assessment tool actionability score, PEMAT-U = patient education materials assessment tool understandability score, P_Y-B = YouTube versus Bilibili, P_Y-T = YouTube versus TikTok, VIQI = video information and quality index, VIQI-1 = information flow, VIQI-2 = information accuracy, VIQI-3 = video quality, VIQI-4 = information precision, VIQI-sum = total video information and quality index score.

3.5. Correlation analysis

Spearman correlation analysis showed that the associations between video quality assessment measures and audience engagement metrics varied across platforms (Table 4). VIQI demonstrated the most consistent positive correlations with audience engagement metrics, reaching statistical significance for multiple engagement metrics across all 3 platforms. GQS was also positively correlated with some audience engagement metrics, particularly on YouTube and TikTok. PEMAT-U and PEMAT-A showed mainly weak to moderate positive correlations with audience engagement metrics on TikTok, whereas the correlations between mDISCERN and audience engagement metrics were relatively weak and inconsistent. After Bonferroni correction and adjustment for video duration, most correlations involving VIQI remained statistically significant, particularly on Bilibili and TikTok (Supplementary Table S1, Supplemental Digital Content 1).

Table 4.

Spearman correlations between video quality assessment measures and audience interaction metrics across platforms.

Platform	Assessment measure	Views	Thumbs-up	Comments	Collections	Shares
YouTube	PEMAT-U	0.154, 0.149	0.315, 0.003	0.285, 0.007	–	–
	PEMAT-A	0.323, 0.002	0.481,<0.001	0.445,<0.001	–	–
	VIQI	0.507,<0.001	0.684,<0.001	0.504,<0.001	–	–
	GQS	0.409,<0.001	0.543, <0.001	0.497,<0.001	–	–
	mDISCERN	0.149, 0.162	0.364, <0.001	0.263, 0.013	–	–
Bilibili	PEMAT-U	0.074, 0.512	0.195, 0.084	0.081, 0.473	0.223, 0.047	0.188, 0.096
	PEMAT-A	0.074, 0.516	0.238, 0.034	0.090, 0.429	0.269, 0.016	0.221, 0.049
	VIQI	0.471,<0.001	0.578,<0.001	0.454,<0.001	0.659,<0.001	0.596,<0.001
	GQS	0.194, 0.084	0.311, 0.005	0.197, 0.079	0.342, 0.002	0.295, 0.008
	mDISCERN	0.085, 0.451	0.165, 0.143	0.174, 0.123	0.105, 0.353	0.100, 0.377
TikTok	PEMAT-U	–	0.481, <0.001	0.365, <0.001	0.477, <0.001	0.485, <0.001
	PEMAT-A	–	0.474, <0.001	0.368, <0.001	0.483, <0.001	0.478, <0.001
	VIQI	–	0.664, <0.001	0.590, <0.001	0.639, <0.001	0.639, <0.001
	GQS	–	0.548, <0.001	0.482, <0.001	0.574, <0.001	0.577, <0.001
	mDISCERN	–	0.445, <0.001	0.348, <0.001	0.400, <0.001	0.406, <0.001

Data are presented as Spearman’s r and P-value.

Bold values indicate significance after Bonferroni correction (YouTube: P < .0033; Bilibili: P < .0020; TikTok: P < .0025).

“–” indicates unavailable data.

GQS = global quality score, mDISCERN = modified DISCERN instrument, PEMAT-A = patient education materials assessment tool actionability score, PEMAT-U = patient education materials assessment tool understandability score, VIQI = video information and quality index.

4. Discussion

4.1. Principal findings

Social media videos have become an important means through which the public obtains health information.^[10] For patients with LSS, online videos can provide information on symptom recognition, diagnostic evaluation, treatment options, and daily management. However, substantial differences exist across platforms in video format, recommendation mechanisms, uploader composition, and modes of audience engagement, which may affect the quality and dissemination of medical information. This study systematically compared the quality and dissemination characteristics of LSS-related videos on YouTube, Bilibili, and TikTok. The results showed that the 3 platforms differed in video duration, uploader composition, video style, content themes, and quality scores. YouTube videos performed better in terms of mDISCERN and GQS, indicating relatively higher information reliability and overall quality. total VIQI scores were higher on YouTube and TikTok than on Bilibili, whereas PEMAT-A scores did not differ significantly across platforms.Bilibili exhibited a more mixed content ecosystem but did not demonstrate clear advantages in most quality metrics. In addition, the associations between video quality assessment measures and audience engagement varied across platforms, with VIQI showing the most consistent positive correlations with audience engagement metrics.

4.2. Platform characteristics and content differences

Differences in video characteristics across platforms may be related to platform positioning and user habits.^[19] YouTube and Bilibili are better suited to medium- and long-form videos, providing uploaders with more space to explain the etiology, anatomical basis, diagnostic methods, and treatment options for LSS. In contrast, TikTok primarily disseminates short-form videos and places greater emphasis on rapid browsing, immediate feedback, and algorithm-based recommendations. Consequently, its videos were shorter but performed better on the available engagement metrics. It should be noted that engagement metrics were not fully consistent across platforms. TikTok did not provide view counts, whereas YouTube did not provide collection or sharing data. Therefore, the relevant comparisons can only be interpreted on the basis of metrics that were both available and comparable across platforms, and it cannot simply be concluded that any single platform has an advantage across all dimensions of engagement.

Uploader composition is also an important factor affecting content quality. YouTube videos were mainly uploaded by hospitals/departments and verified physicians, which may partly explain their higher information reliability and overall quality scores. Professional uploaders are generally more likely to provide structured, balanced, and less exaggerated medical information.^[20] Although TikTok had a high proportion of verified physicians, the short-video format may limit the extent to which complex disease mechanisms, treatment indications, and risks and benefits can be fully explained.^[12] Bilibili had a more mixed uploader composition, including both physicians and individual users. This may contribute to greater content diversity but may also result in wider variation in medical accuracy and reliability.

In terms of content themes, YouTube more frequently addressed etiology and prevention, possibly because of its longer video format and higher proportion of institutional and professional uploaders. By contrast, Bilibili and TikTok focused more on treatment-related content, reflecting users’ practical needs regarding symptom relief, treatment options, and disease management. With respect to video style, solo narration predominated on YouTube, whereas Bilibili and TikTok more frequently used real medical scenarios. PPT/classroom-style presentations were also more common on Bilibili. These findings indicate that the presentation of medical information differs across platforms: long-form video platforms are more suitable for systematic explanations, whereas short-form video platforms are better suited to concise and contextualized communication.

4.3. Video quality and potential confounding factors

This study used mDISCERN, GQS, VIQI, PEMAT-U, and PEMAT-A to evaluate video quality from different dimensions. mDISCERN and GQS primarily reflect information reliability, balance, and overall educational quality; VIQI places greater emphasis on information flow, audiovisual presentation, and consistency between the title and content; and PEMAT-U and PEMAT-A assess the understandability and actionability of patient education materials, respectively. Therefore, differences in scale performance across platforms should not be interpreted simply as indicating that any single platform is superior across all dimensions of medical quality.

YouTube received higher mDISCERN and GQS scores, which may be related to its longer video duration and the higher proportion of videos uploaded by hospitals/departments and verified physicians. Longer videos and professional uploaders may facilitate a more complete and balanced presentation of the disease background, diagnostic basis, and treatment options for LSS.^[20] Total VIQI scores were higher on TikTok than on Bilibili but did not differ significantly between TikTok and YouTube. Although TikTok had the highest median PEMAT-A score, the differences in PEMAT-A scores across platforms were not statistically significant. However, LSS involves complex content, including degenerative changes, nerve compression, imaging evaluation, conservative treatment, and surgical indications; therefore, excessively short videos may be unable to present complete medical information adequately.^[21]

It should be emphasized that the observed differences in video quality across platforms may have been influenced by factors such as video duration, uploader type, language environment, and differences in user populations. YouTube videos were predominantly in English or included English subtitles, whereas Bilibili and TikTok videos were mainly in Chinese or included Chinese subtitles. Therefore, the observed platform differences may also reflect variations in linguistic and cultural backgrounds, medical communication practices, and audience needs. Although video duration was controlled for in the correlation analyses, no multivariable model was established to simultaneously adjust for video duration, uploader type, language, and other potential covariates. Accordingly, differences in video quality across platforms should be interpreted as observational findings rather than causal evidence that the platforms themselves caused these differences.

4.4. Association between video quality and audience engagement metrics

The associations between video quality assessment measures and audience engagement metrics differed markedly across platforms. Consistent with previous studies of health-related videos on social media, video quality and audience engagement were not always aligned.^[19,20] This suggests that audience engagement metrics, such as views, Thumbs-up, comments, collections, and shares, may be associated not only with the quality of medical content but also with platform recommendation mechanisms, video presentation formats, user preferences, and modes of content dissemination. Among the video quality assessment measures, VIQI showed the most consistent correlations with audience engagement metrics. Spearman correlation analysis showed that VIQI was positively correlated with the principal audience engagement metrics across all 3 platforms. These associations included views, Thumbs-up, and comments on YouTube and Bilibili, as well as collections and shares on Bilibili and TikTok. After further adjustment for video duration, VIQI remained significantly and positively correlated with most audience engagement metrics across all 3 platforms, indicating a relatively robust association with the level of audience engagement.

In addition to VIQI, GQS was also positively correlated with some audience engagement metrics, and these associations were mainly observed on YouTube and TikTok. After adjustment for video duration, GQS remained significantly correlated with views, Thumbs-up, and comments on YouTube, as well as Thumbs-up, comments, collections, and shares on TikTok. This may be related to the fact that GQS reflects the overall quality, information completeness, and clarity of expression of a video. In contrast, the associations of PEMAT-U, PEMAT-A, and mDISCERN with audience engagement metrics were more platform-specific. After adjustment for video duration, PEMAT-U and PEMAT-A remained significantly correlated mainly with Thumbs-up, collections, and shares on TikTok, whereas mDISCERN remained significantly correlated only with Thumbs-up on TikTok. These findings suggest that audience engagement metrics may be more closely associated with video presentation, the overall viewing experience, and practical content, rather than changing in parallel with information reliability scores.

These findings indicate that audience engagement metrics on social media platforms cannot simply be equated with the quality of medical information.^[22] Although higher numbers of views, Thumbs-up, or comments may reflect greater dissemination potential and appeal, they do not necessarily indicate that the information is more reliable. Similar studies have also shown that the public often finds it difficult to judge the scientific validity and credibility of health-related videos based solely on engagement data.^[8] Therefore, when viewing medical videos with high engagement, users should continue to consider the uploader’s identity, the sources of information, and the evidence underlying the content.

For content creators, well-organized information, clear expression, and high-quality audiovisual presentation may be associated with greater audience engagement. Nevertheless, the core of medical educational videos should remain accurate, reliable, and understandable health information, rather than the pursuit of dissemination effects alone. For platforms, recommendation mechanisms should not rely solely on engagement metrics such as views, Thumbs-up, or comments, but should also consider the uploader’s professional background, content reliability, and the results of video quality assessments to reduce the risk of disseminating low-quality medical information.^[9]

It should be emphasized that this study was a cross-sectional content analysis, and the correlation results cannot be interpreted as causal relationships. Therefore, this study can only demonstrate statistical associations between video quality assessment measures and audience engagement metrics and cannot establish that higher video quality directly leads to more views, Thumbs-up, comments, collections, or shares.

4.5. Implications for health communication practice

Creators of medical educational content should balance scientific rigor with dissemination effectiveness. High-quality videos on LSS should not only provide clear visuals and understandable explanations but also clearly identify the uploader, avoid exaggerating treatment effects, accurately distinguish the indications for conservative and surgical treatments, and advise patients to seek individualized medical guidance based on their specific clinical conditions.

Different platforms should adopt targeted improvement strategies. YouTube could further strengthen the presentation of uploader qualifications, links to reference sources, and video chapter navigation. Bilibili should enhance the verification of physician and medical institution accounts and distinguish professional health education from personal experience sharing and commercial promotional content. TikTok should strengthen verified physician identification, medical content review, and the weighting of quality in recommendation algorithms to avoid amplifying the dissemination of medical information with low reliability solely on the basis of engagement metrics such as Thumbs-up, comments, collections, or shares.

Overall, recommendation mechanisms for medical and health-related videos should shift from relying solely on traffic and audience engagement toward a quality-oriented model that comprehensively considers professionalism, reliability, and understandability, thereby increasing the public’s opportunities to access reliable medical information.

5. Limitations

This study has several limitations. First, although mDISCERN, GQS, VIQI, PEMAT-U, and PEMAT-A were used and all videos were independently scored by 3 trained evaluators with physician qualifications, existing video assessment tools cannot fully replace item-by-item evaluations of medical accuracy based on clinical guidelines, and subjective scoring and systematic bias cannot be completely avoided. Second, because of limitations in platform data availability, TikTok did not provide view counts, YouTube did not provide collection or share counts, and the platforms lacked standardized negative feedback indicators. Consequently, some audience engagement metrics could not be fully compared across platforms. Third, YouTube videos were predominantly in English, whereas Bilibili and TikTok videos were mainly in Chinese; therefore, the findings may have been influenced by differences in linguistic and cultural backgrounds, medical communication practices, and audience needs. Fourth, this study did not include supplementary information such as textual descriptions on video pages or comment content, nor did it establish a multivariable model to simultaneously adjust for potential confounding factors such as video duration, uploader type, verification status, and language environment. Finally, this study was a single-time-point cross-sectional content analysis and reflected only a limited sample of LSS-related videos retrieved on February 1, 2026. Searches were not repeated on multiple dates, and no sensitivity analysis of temporal stability was conducted. Therefore, the effects of platform rankings, content updates, changes in recommendation algorithms, and temporal changes in audience engagement metrics on the results could not be evaluated. Future studies may adopt longitudinal designs or repeated sampling at multiple time points to further assess the stability of these findings.

6. Conclusion

This study demonstrated that LSS-related videos on YouTube, Bilibili, and TikTok exhibit platform-specific quality and dissemination characteristics. YouTube videos were generally more reliable, TikTok videos showed greater dissemination appeal, whereas Bilibili did not demonstrate clear advantages in most quality assessment metrics. High audience engagement does not necessarily indicate high medical reliability, highlighting the need to improve public digital health literacy and optimize platform recommendation mechanisms for medical content. When recommending medical content, platforms should comprehensively consider uploader qualifications, content reliability, and video presentation quality. Against the background of population aging, the demand for health information related to LSS is expected to continue increasing.^[23] Future research should prioritize longitudinal analyses to evaluate dynamic changes in platform algorithms, video quality, and engagement behaviors among older users, while also conducting intervention studies to examine the actual effects of high-quality LSS videos on patients’ disease awareness, treatment choices, and healthcare-seeking behaviors.

Acknowledgments

The authors thank all uploaders of lumbar spinal stenosis educational videos on YouTube, Bilibili, and TikTok for providing publicly available research materials for this study. The authors also thank Xi’an Hospital of Traditional Chinese Medicine for its institutional and research platform support, as well as all researchers who participated in video screening, data collection, and quality assessment.

Author contributions

Data curation: Rui Guo, Jiachen Liang.

Software: Rui Guo, Qiang Sun, Yiliu Xu.

Formal analysis: Jiachen Liang.

Investigation: Jiachen Liang, Han Li.

Project administration: Han Li, Yiliu Xu, Qing Sun.

Supervision: Han Li.

Resources: Qing Sun.

Writing – original draft: Rui Guo, Xinjuan Wen.

Writing – review & editing: Qing Sun.