Public health implications of allergic rhinitis information on YouTube and Bilibili: a cross-cultural analysis of content quality, engagement, and seasonal trends Authors: Dongling Lian, Yingying Zhang, Zeyang Liu, Ying Lin, Maomao Ai, Yiwen Chen, Jian Wu, Xin Wang, Tao Lin, Huilong Guo, Feng Yu Published: 2026 ABSTRACT AbstractBackgroundAllergic Rhinitis (AR) affects over 500 million people globally, posing a significant public health burden. Video-sharing platforms like YouTube and Bilibili have become primary sources of health information. This study aimed to compare content quality, user engagement, and alignment with seasonal search trends of AR-related videos on these two culturally distinct platforms.MethodsWe retrieved the top 200 AR-related videos from each platform (keywords: "allergic rhinitis" for YouTube, "过敏性鼻炎" for Bilibili) published between January 2022 and January 2025. After applying predefined exclusion criteria (irrelevance, non-English YouTube videos, out-of-timeframe, advertisements), 240 videos (91 YouTube, 149 Bilibili) were retained. Quality was assessed using the Patient Education Material Assessment Tool (PEMAT), Video Information Quality Index (VIQI), Global Quality Scale (GQS), and the Modified DISCERN Scale (mDISCERN). We also integrated Google Trends and Baidu Index data to analyze video characteristics, engagement, and correlations with seasonal search trends.ResultsAR-related video quality was generally low, though YouTube scored significantly higher than Bilibili in median PEMAT-Total (76.5 vs. 71.4), GQS (4 vs. 3), and mDISCERN (3 vs. 1) (all P < 0.001). YouTube featured more content from medical professionals (33.3%, n = 30/91), whereas Bilibili's was predominantly from non-professionals (61.4%, n = 92/149). Bilibili showed higher user engagement, with greater interactions through comments and donations. Both platforms showed spring and autumn search peaks, coinciding with allergen seasons. A moderate positive correlation emerged between Bilibili videos and Baidu Index (r = 0.37; P = 0.03), while YouTube videos correlated strongly with Google Trends (r = 0.63; P = 0.03). Professional content scored higher on both platforms, but treatment-related Bilibili videos correlated negatively with search volume (r = − 0.69; P = 0.01), signaling potential misinformation risks during peak periods.ConclusionsFor AR-related videos, YouTube offers better content quality, while Bilibili excels in user engagement. Both platforms need improved content quality and coverage. A seasonal "demand-content" loop exists between AR search trends and video content, carrying misinformation risks during peak periods. We recommend year-round promotion of evidence-based content, adding medical warnings to misleading information, and encouraging collaboration between medical professionals and social media creators.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12889-026-28015-7.Keywords: Allergic rhinitis, Baidu Index, Bilibili, Cross-sectional study, Google Trends, Public health, Public engagement, Search engine, Seasonal trend, Social media, YouTube Allergic rhinitis (AR), a highly prevalent chronic inflammatory disease worldwide, has undergone a significant epidemiological surge over the past three decades [1]. In Europe, adult prevalence rates have risen from 19% to 32%, while China reported a 6.5% increase in incidence over the past six years [2]. Characterized by nasal obstruction, rhinorrhea, and comorbidities such as asthma, AR severely compromises patients' quality of life. Despite evidence-based frameworks provided by international guidelines, including the Allergic Rhinitis and its Impact on Asthma (ARIA) [3] initiative and the International Consensus Statement on Allergy and Rhinology (ICAR) [4, 5], persistent gaps in public awareness and the proliferation of non-scientific claims remain critical barriers to effective AR management. The internet has become the primary source of health information for the public. Studies indicate that a substantial majority of adults seek health information online, with search engines serving as the initial gateway [6, 7]. For instance, a large-scale survey revealed that in 2018, nine out of ten American adults used the internet, and 75% of them conducted online searches for medical content [8]. This paradigm shift highlights the profound influence of digital information on health beliefs and behaviors. Consequently, platforms such as Google Trends (global) and Baidu Index (China) have emerged as real-time tools for epidemiological surveillance and intervention studies due to their capacity to capture dynamic health-seeking behaviors [9, 10]. Within this digital landscape, YouTube and Bilibili represent two dominant yet culturally distinct video-sharing platforms for health communication [11, 12]. YouTube dominates global health education and serves as an archetype of Western/international digital ecosystems, despite concerns that approximately 30% of its medical content may be misleading [13–15]. In China, Bilibili has emerged as a leading video platform, particularly among younger demographics, engaging users through interactive features like bullet comments, yet systematic evaluations of its health-related content quality remain scarce [16]. The comparative analysis of these two platforms is therefore well-suited to examine how cultural and systemic differences shape health information environments. The quality of such widely consumed video content is a paramount public health concern. Effective patient education materials must be not only accurate but also understandable and actionable to accommodate varying levels of health literacy [17–19]. Recognizing this, institutions like the U.S. National Institutes of Health and the American Medical Association recommend that patient education materials be written at or below a sixth-grade reading level [20]. Tools like the Patient Education Materials Assessment Tool (PEMAT) are designed to evaluate these exact dimensions of understandability and actionability [17]. Furthermore, evidence suggests that patients with better knowledge of their condition's etiology, pathophysiology, and treatment are more likely to engage effectively in their care and adhere to management plans, leading to improved health outcomes [21]. Therefore, the quality of information on platforms like YouTube and Bilibili has direct implications for patient empowerment, self-management, and clinical outcomes. Despite the recognized role of video platforms in health communication [11], several key questions regarding AR information remain underexplored, particularly in cross-platform and cross-cultural contexts. Previous studies have either focused on a single platform or assessed content quality in isolation from real-world public demand [22–24]. A critical gap exists in understanding how the quality and engagement dynamics of AR videos differ across major cultural platforms, and how these digital content patterns interact with seasonal fluctuations in public information-seeking behavior. Therefore, examining these two culturally distinct platforms can provide critical insights for crafting tailored health communication approaches that resonate within different digital landscapes. To address this, our study employs an integrated, cross-sectional design to: (1) compare the quality, source characteristics, and user engagement of AR-related videos between YouTube and Bilibili; (2) analyze the temporal relationships between video publication/engagement metrics and seasonal search trends on their corresponding search engines; and (3) identify potential disparities between information demand and content quality, particularly during peak allergy seasons. Thus, this synthesis aims to bridge the identified gaps between information supply, quality, and public demand during critical seasonal periods. Materials and methods Study design and data sources This cross-sectional study analyzed publicly available data from two major video-sharing platforms (YouTube and Bilibili) and their corresponding dominant search engines (Google and Baidu). The design facilitated a comparison of AR-related information ecosystems across cultural contexts. YouTube and its associated search engine, Google Trends, were selected to represent globally accessible, English-predominant sources. Bilibili and Baidu Index represent the dominant video platform and search engine in mainland China, allowing for a parallel analysis within that regional context. Video collection and screening Video retrieval was conducted on January 7, 2025. To minimize the influence of personalized algorithms and search history biases, all browser cookies and histories were cleared prior to data collection, and searches were performed using the incognito mode of the Google Chrome browser. We systematically retrieved the top 200 most-viewed AR-related videos (sorted by view count, the default sorting metric) from YouTube (English keyword: "allergic rhinitis") and Bilibili (Chinese keyword: "过敏性鼻炎"). This process yielded a preliminary sample of 400 videos. This sample size was chosen to capture the most viewed and thus potentially most influential content while maintaining analytical feasibility. The study period was set from January 1, 2022, to January 1, 2025, to encompass multiple seasonal allergy cycles and recent trends. On January 8, 2025, a subsequent screening of the collected videos was conducted based on the following exclusion criteria (Fig. 1): (1) videos irrelevant to AR (not covering pre-defined topics such as treatment or symptoms). (2) The YouTube video is not in English or cannot be translated into English subtitles (to ensure consistent evaluation). (3) Videos published outside the specified study period. (4) videos primarily intended as advertisements (identified by tags such as “sponsored” or explicit promotional content). Data collection and pre-processing workflow for allergic rhinitis-related research on YouTube, Bilibili, Baidu Index, and Google Trends After applying these criteria, 240 videos (91 from YouTube, 149 from Bilibili) were retained for final analysis. Changes in search index and video publishing volume To examine the alignment between public information demand and video content availability over time, we collected complementary search trend data. This allowed for the analysis of potential "demand-content" dynamics. Search volume indices were obtained from Google Trends and Baidu Index, corresponding to the primary user bases of YouTube and Bilibili, respectively. Search on Google Trends using the English keyword "allergic rhinitis" and select by the following filter (Fig. 1): (1) worldwide, (2) Custom time range is 2022.01.01–2025.01.01, (3) All categories, (4) Web search. Similarly, the Chinese keyword "过敏性鼻炎" was queried on Baidu Index with nationwide (China) coverage across both PC and mobile platforms for the identical period. Both platforms provide normalized indices rather than absolute search counts. Weekly data points were extracted and subsequently aggregated into monthly and quarterly values for temporal analysis alongside video metrics. The temporal alignment and statistical correlation between these search trends and video metrics (e.g., publication volume, engagement) were then analyzed. Figure 1 outlines the integrated data collection and preprocessing workflow. Video and engagement metrics extraction For each included video, standardized metadata and engagement metrics were extracted to enable cross-platform comparison. The collected data encompassed basic video features and user interaction indicators. Basic features included view count, publication date, and video duration. Engagement metrics consisted of likes and comments. For Bilibili, platform-specific interactive features—integral to its culture—were also collected: shares, collections, "coins" (a virtual tipping currency), and bullet comments ("danmu"). The methodology for locating these data points is detailed in Multimedia Appendix 1. To account for the wide variation in view counts and enable fair comparison, relative engagement rates were calculated as follows: like rate (user approval) = likes/views × 100%; comment rate (depth of participation) = comments/views × 100%; share rate (secondary dissemination) = shares (Bilibili)/views × 100%. Based on the normalization logic of social media interaction rates, an improved video propagation index (mVPI) [25] was proposed, with the formula: (likes + comments)/views × 100%, aiming to integrate active interaction behaviors and eliminate the impact of view-based differences on the assessment of propagation effectiveness. Video and uploader categorization To systematically analyze the source and thematic focus of the content, uploader metadata (including creator ID, follower count, certification status, and professional background) were collated on January 7, 2025, and further classified based on the creator’s professional attributes. Video uploaders were classified into the following categories based on their professional background and official platform certification status: professionals (specifically including physicians, other medical staff or medical students, non-profit organizations such as hospitals/medical associations/departments, for-profit organizations, and government-regulated official media such as BBC) and non-professionals (such as self-media); the video topics were also divided into anatomy, symptom, diagnosis, treatment, etiology/prevention, prognosis or epidemiology. Those not falling under the above content themes were excluded. Multiple theme videos were coded with multiple codes; similarly, the video style was subdivided into monologue narration, question-and-answer format, PPT/lecture style, animation/action sequence, clinical scene, or documentary/TV segment. The detailed certification criteria can be found in Multimedia Appendix 1. Quality assessment of videos Video quality was assessed using four validated instruments, each measuring distinct yet complementary dimensions. We implemented a dual-reviewer process to ensure scoring reliability. Two independent evaluators, DL.L. and YY.Z., performed the initial scoring. To address discrepancies in ratings, a third arbitrator (Y.L.) was tasked with finalizing the consensus scores. Inter-rater agreement between the primary reviewers was quantified using Cohen’s kappa coefficient (κ). The assessment tools [26] were employed as follows: The Patient Education Materials Assessment Tool for Audiovisual Materials (PEMAT-A/V) [27] was used to evaluate understandability (PEMAT-U) and actionability (PEMAT-A), yielding percentage scores for each dimension and an overall score (PEMAT-T). The Global Quality Scale (GQS) [28] provided a single, holistic 5-point rating (1 = Poor, 5 = Excellent) of a video’s overall educational utility and flow from a viewer’s perspective. The Video Information and Quality Index (VIQI) [29] assessed content accuracy and presentation quality across four 5-point items (e.g., information flow, visual quality). A modified DISCERN instrument (mDISCERN) [30, 31] was adapted to evaluate the reliability and completeness of video information, focusing on source transparency, balance, and coverage of uncertainties. Detailed descriptions of each tool and their scoring criteria are available in Multimedia Appendices 2 & 3. Statistical analysis Data processing utilized Python 3.10 (pandas, numpy, scipy libraries) and MATLAB 2023b. Normality was assessed via Shapiro–Wilk tests, with parametric data reported as mean ± SD and non-parametric data as median (min–max; IQR). Between-group comparisons employed Mann–Whitney U tests (significance threshold: P < 0.05). Categorical variables were analyzed using chi-square, continuity correction, or Fisher’s exact tests, selected based on sample size and expected frequencies. Spearman’s rank correlation quantified monotonic associations between video metrics (quality, engagement) and search trends. The r is defined as Spearman's rank correlation coefficient. r > 0: positive correlation; r < 0: negative correlation. |r|< 0.2: no relationship; 0.21 <|r|< 0.4: weak relationship; 0.41 <|r|< 0.6: moderate relationship; 0.61 <|r|< 0.8: strong relationship; |r|> 0.81: very strong relationship. A value of P < 0.05 was considered statistically significant. Results Video characteristics The analysis included 91 YouTube and 149 Bilibili videos meeting the inclusion criteria. While video length showed no significant difference (YouTube: median 220 s; Bilibili: 150 s; P = 0.06), YouTube videos had longer publication histories (median 677 vs. 518 days; P = 0.02). Modified Video Propagation Index (mVPI) scores suggested comparable platform interaction efficiency (Bilibili: 2.39 vs. YouTube: 2.16; P = 0.32). However, Bilibili significantly outperformed YouTube in view counts (median 36,000 vs. 15,303; P < 0.001), likes (median 804 vs. 343; P < 0.001), and comment rates (0.28% vs. 0.12%; P < 0.001). Conversely, YouTube generated higher absolute comment numbers (median 114 vs. 105; P < 0.001). Bilibili’s participatory design—marked by unique features like coin donations (median 48), bullet comments (median 28), and shares (median 227)—highlighted its engagement-driven architecture (Table 1). Table 1Characteristics of videos about allergic rhinitis on YouTube/BilibiliPlatformYouTube(N1 = 91)Bilibili(N2 = 149)P-valueCharacteristicMmin–maxP25-P75Mmin–maxP25-P75Video length(s)2205–137159–392.2515012–91477.5–225.750.06aDuration(day)6777–1075405.25–84551831–1023252.75–8050.02aViews15,3032804–18801976503–43548.7536,00015,000–1271000021,000–77250 < 0.001aLikes3430–8700093–910.2580426–93,000323.5–2093.5 < 0.001aLike rate (%)1.890–13.041.18–3.011.910.15–12.730.95–3.390.95aComments1140–13715.25–941056–377051.75–260.5 < 0.001aComment rate (%)0.120–1.30.05–0.290.280–3.120.13–0.54 < 0.001amVPI (%)2.160.08–13.391.28–3.282.390.19–14.301.26–4.150.32aShares---2272–31,00055–734-Share rate (%)---0.510.01–6.420.16–1.49-Collections---4948–134,000195.25–1465.75-Coins---480–1200016–216-Bullet comments---280–19906–71.25-Originality Statement-93.29%-Abbreviations: M Median; min–max, minimum to maximum, P25-P75 25th to 75th percentile, s seconds, mVPI modified Video Propagation IndexItalics indicate that the P <0.05aMann-Whitney U test Characteristics of videos about allergic rhinitis on YouTube/Bilibili Abbreviations: M Median; min–max, minimum to maximum, P25-P75 25th to 75th percentile, s seconds, mVPI modified Video Propagation Index Italics indicate that the P <0.05 aMann-Whitney U test Changes in search index and video publishing volume The monthly analysis indicates that from 2022 to 2025, Baidu's AR-related search index (left axis, blue line) exhibited marked seasonal fluctuations, with mean peaks clustered in spring (February–March) and autumn (August–September), characterized by substantial volatility and wider standard deviations. In contrast, Google’s search index (right axis, green line) displayed a smoother trend, peaking in February–March and September–October with narrower standard deviations. Bilibili’s monthly AR video uploads (right axis, yellow bar chart) showed pronounced seasonal peaks during spring and autumn, while YouTube’s uploads (right axis, red bar chart) remained stable overall, with minor seasonal increases. Cross-year trends revealed consistent annual growth in both search indices and video volumes since 2022, with Baidu’s autumn fluctuations being particularly pronounced (Fig. 2). The trend comparison of search volume and video release volume for allergic rhinitis (2022–2025). a Monthly trend charts of search volume and video releases related to allergic rhinitis on Google Trends, Baidu Index, YouTube, and Bilibili platforms. b Quarterly trend charts of search volume and video releases related to allergic rhinitis on Google Trends, Baidu Index, YouTube, and Bilibili platforms The quarterly analysis further highlights Baidu’s seasonal volatility, peaking in Q3 (July–September) with wider standard deviations and intra-quarter variability. Google’s index maintained stability overall, with slight peaks in Q1 (January-March) and Q4 (October-December) and narrower deviations. Bilibili’s video uploads aligned with Baidu’s Q3 peaks, while YouTube’s uploads showed minimal quarterly variation, peaking marginally in Q4 (October-December) (Fig. 2). Video classification and uploader characteristics Analysis of uploader profiles revealed distinct platform-specific dynamics. 75 unique creators contributed to YouTube's 91 AR-related videos, while Bilibili's 149 videos originated from 114 uploaders. YouTube creators had significantly larger follower bases (median 41,600 vs. 9,394; P < 0.001) and were predominantly physicians (33.3%, n = 25/75) or nonprofit organizations (14.7%, n = 11/75). In contrast, Bilibili uploaders were overwhelmingly self-media creators (61.4%, n = 70/114), with physicians comprising only 23.9% (n = 27/114). Identity verification rates were comparable (Bilibili: 34.2%, n = 39/114 vs. YouTube: 24%, n = 18/75; P = 0.14), but certified YouTube uploaders primarily included physicians (27.7%, n = 5/18) and nonprofits (22.2%, n = 4/18), whereas Bilibili’s certified accounts were dominated by physicians (61.5%, n = 24/39) and medical staff (10.3%, n = 4/39) (Table 2). Table 2Characteristics of video uploaders and categorization of allergic rhinitis videos on YouTube/BilibiliPlatformYouTube(N1 = 91)Bilibili(N2 = 149)P-valueNumber of uploaders75114-Followers, Median [P25, P75]41600[8307.5,272000]9393.5[1160,47000] < 0.001aType of uploaders, n (%)Doctor25 (33.3%)27 (23.9%)0.15bOther medical worker/student2 (2.7%)7 (6.1%)0.45cNon-profit organization11 (14.7%)4 (3.5%)0.006bCompany with profit15 (20%)3 (2.6%) < 0.001bOfficial media1 (1.3%)3 (2.6%)0.93cSelf-media21 (28%)70 (61.4%) < 0.001bDoctor of TCM, n (%)2 (2.7%)39 (34.2%) < 0.001bCertification, n (%)18 (24%)39 (34.2%)0.14bType of certified uploaders, n (%)Doctor5 (27.7%)24 (61.5%)0.02bOther medical worker/student1 (5.6%)4 (10.3%)0.94cNon-profit organization4 (22.2%)1 (2.5%)0.05cCompany with profit3 (16.7%)3 (7.7%)0.57cOfficial media1 (5.6%)3 (7.7%)0.79cSelf-media4 (22.2%)4 (1.3%)0.42cNumber of topics per video, Median [P25, P75]3[2, 4]3[2,3.25]0.56aType of topics, n (%)Anatomy16 (17.6%)8 (5.4%)0.002bEtiology/Prevention56 (61.5%)80 (53.7%)0.004bPathology24 (26.4%)5 (3.4%) < 0.001bEpidemiology1 (1.1%)28 (18.8%) < 0.001bSymptoms65 (71.4%)120 (80.5%)0.10bExaminations/Diagnosis33 (36.3%)32 (21.5%)0.01bTreatment72 (79.1%)136 (91.3%)0.007bPrognosis8 (8.8%)20 (13.4%)0.28bTCM, n (%)2 (2.2%)51 (34.2%) < 0.001bStyle of video shooting, n (%)Solo narration62 (68.1%)112 (75.2%)0.24bQuestion & Answer9 (9.9%)11 (7.4%)0.50bPPT or Class9 (9.9%)5 (3.4%)0.04bAnimation/Action11 (12.1%)11 (7.4%)0.22bMedical scenarios6 (6.6%)12 (8.1%)0.68bTV show/Documentary4 (4.4%)5 (3.4%)0.68cOthers1 (1.1%)5 (3.4%)0.51cAbbreviations: P25-P75 25th to 75th percentile, TCM Traditional Chinese MedicineItalics indicate that the P <0.05aMann-Whitney U testbChi-squared testcContinuity correction Characteristics of video uploaders and categorization of allergic rhinitis videos on YouTube/Bilibili Abbreviations: P25-P75 25th to 75th percentile, TCM Traditional Chinese Medicine Italics indicate that the P <0.05 aMann-Whitney U test bChi-squared test cContinuity correction Both platforms prioritized treatment-focused content (YouTube: 79.1%, n = 72/91; Bilibili: 91.3%, n = 136/149) and symptom-related themes (YouTube: 71.4%, n = 65/91; Bilibili: 80.5%, n = 120/149), with solo narration as the predominant style (YouTube: 68.1%, n = 62/91; Bilibili: 75.2%, n = 112/149). Notably, Bilibili featured significantly more traditional Chinese medicine (TCM) content (34.2%, n = 51/149 vs. 2.2%, n = 2/91; P < 0.001) (Table 2). Quality assessment of videos Video quality assessments revealed significant disparities between platforms and creator types. YouTube videos consistently outperformed Bilibili across key metrics, including the overall understandability score (PEMAT-T) (median 76.5 vs. 71.4; P < 0.001) and reliability scores (mDISCERN) (median 3 vs. 1; P < 0.001) (Table 3). Table 3Quality assessment for AR videos on YouTube/Bilibili and comparison between professional and non—professional uploadsPlatformYouTube (N1 = 91)Bilibili (N2 = 149)Professionals (N1 = 141)Non-professionals (N2 = 99)P-valueScoresMmin–maxP25-P75Mmin–maxP25-P75Mmin–maxP25-P75Mmin–maxP25-P75PY-BPP-NPPEMAT-T76.514.3–10068.9–84.371.411.1–10064.3–73.873.314.3–10071.4–81.866.711.1–10060–73.3 < 0.001a < 0.001aPEMAT-U7518.2–10064.4–88.569.211.1–10063.6–7572.718.2–10069.2–86.263.611.1–10058.3–74.40.002a < 0.001aPEMAT-A1000–10066.7–10066.70–10066.7–1001000–10066.7–10066.70–10066.7–1000.01a < 0.001aVIQI-sum148–1912–15158–1912.8–16158–1913–16.3138–1711–150.02a < 0.001aVIQI-143–53–444–54–443–54–443–54–4 < 0.001a0.02aVIQI-241–53–440–53–440–53–430–52–40.67a < 0.001aVIQI-331–42–331–52–3.331–53–431–42–30.012a0.001aVIQI-441–53–441–53–442–54–431–53–40.10a0.005aGQS41–53–431–53–441–53–431–53–40.30a < 0.001amDISCERN-sum30–52–410–51–330–52–3.310–41–2 < 0.001a < 0.001amDISCERN-185(93.4%)133 (89.3%)140(99.3%)78(78.8%)0.28b < 0.001bmDISCERN-276(83.5%)65(43.6%)114(80.9%)27(27.3%) < 0.001b < 0.001bmDISCERN-357(62.6%)58(38.9%)92(65.2%)23(23.2%) < 0.001b < 0.001bmDISCERN-425(27.5%)4(2.7%)22(15.6%)7(7.1%) < 0.001b0.046bmDISCERN-519(20.9%)15(10.1%)34(24.1%)0(0%)0.02b < 0.001bAbbreviations: M Median, min–max minimum to maximum, P25-P75 25th to 75th percentile, PEMAT Patient Education Material Assessment Tool, PEMAT-U PEMAT Understandability, PEMAT-A PEMAT Actionability, PEMAT-T PEMAT Total, VIQI Video Information and Quality Index, VIQI-sum total VIQI score (sum of items 1–4); VIQI-1 to VIQI-4, individual VIQI item scores, GQS Global Quality Scale, mDISCERN modified DISCERN scale, mDISCERN-sum total mDISCERN score (sum of items 1–5); mDISCERN-1 to mDISCERN-5, individual mDISCERN item scoresPY-B: YouTube versus Bilibili; PP-NP: Professionals versus Non-professionals; Italics indicate that the P <0.05aMann-Whitney U testbChi-squared test Quality assessment for AR videos on YouTube/Bilibili and comparison between professional and non—professional uploads Abbreviations: M Median, min–max minimum to maximum, P25-P75 25th to 75th percentile, PEMAT Patient Education Material Assessment Tool, PEMAT-U PEMAT Understandability, PEMAT-A PEMAT Actionability, PEMAT-T PEMAT Total, VIQI Video Information and Quality Index, VIQI-sum total VIQI score (sum of items 1–4); VIQI-1 to VIQI-4, individual VIQI item scores, GQS Global Quality Scale, mDISCERN modified DISCERN scale, mDISCERN-sum total mDISCERN score (sum of items 1–5); mDISCERN-1 to mDISCERN-5, individual mDISCERN item scores PY-B: YouTube versus Bilibili; PP-NP: Professionals versus Non-professionals; Italics indicate that the P <0.05 aMann-Whitney U test bChi-squared test Professionally created content (58.8% [n = 141/240] of total videos) demonstrated superior quality compared to non-professional content, with higher mDISCERN scores (median 3 vs. 1; P < 0.001), global quality ratings (GQS) (median 4 vs. 3; P < 0.001), PEMAT-T scores (median 73.3 vs. 66.7; P < 0.001), and accuracy scores (VIQI) (median 15 vs. 13; P < 0.001) (Table 3). Correlations between video quality and audience engagement However, correlations between video quality and audience engagement were limited. On YouTube, only VIQI scores showed mild to moderate associations with views (r = 0.451; P < 0.001), likes (r = 0.295; P = 0.005), and comments (r = 0.342; P < 0.001). For Bilibili, higher GQS scores paradoxically correlated with reduced views (r = − 0.324; P < 0.001), while other quality metrics showed negligible relationships with engagement (Table 4). Table 4Spearman correlation between video quality and audience interaction on YouTube/Bilibilir,P-valueViewsLikesCommentsCollectionsSharesYouTube (N1 = 91)PEMAT0.0440.68−0.0170.87−0.2350.03----VIQI0.451 < 0.0010.2950.0050.342 < 0.001----GQS0.1700.110.0240.820.1400.19----mDISCERN0.0580.58−0.0080.940.0100.93----Bilibili (N2 = 149)PEMAT−0.0770.350.0980.24−0.0730.370.1030.210.1000.22VIQI−0.1930.020.0340.680.0800.340.0230.780.1030.21GQS−0.324 < 0.001−0.0280.74−0.0740.37−0.0440.590.0120.88mDISCERN0.0020.980.0890.280.0700.400.0700.400.1760.03Abbreviations: r Spearman's rank correlation coefficient, PEMAT Patient Education Material Assessment Tool, VIQI Video Information and Quality Index, GQS Global Quality Scale, mDISCERN modified DISCERN scaleItalics indicate that the P <0.05 Spearman correlation between video quality and audience interaction on YouTube/Bilibili Abbreviations: r Spearman's rank correlation coefficient, PEMAT Patient Education Material Assessment Tool, VIQI Video Information and Quality Index, GQS Global Quality Scale, mDISCERN modified DISCERN scale Italics indicate that the P <0.05 Search trends and video feature correlations Significant platform-specific correlations emerged between video features and search trends. Bilibili video counts showed a weak positive correlation with Baidu Index monthly searches (r = 0.37; P = 0.03). YouTube video counts exhibited a strong positive correlation with Google Trends quarterly searches (r = 0.63; P = 0.03). On Bilibili, engagement metrics demonstrated strong positive correlations with Baidu Index quarterly search volumes, including comment rate (r = 0.60; P = 0.04), like rate (r = 0.64; P = 0.02), and modified Video Propagation Index (r = 0.69; P = 0.01). Conversely, treatment-themed videos on Bilibili exhibited a strong negative correlation with search trends (r = − 0.69; P = 0.01), while symptom-related content showed a moderate negative correlation with quarterly search volumes (r = − 0.58; P = 0.046). YouTube’s reliability scores (mDISCERN) demonstrated a strong positive correlation with Google Trends quarterly searches (r = 0.61; P = 0.04), though other quality metrics (GQS, VIQI) showed no significant associations (Fig. 3). Correlation between search trend dynamics and video characteristics. The values within each color box represent r (P-value). r, Spearman's rank correlation coefficient. Different colors represent different r values, as shown in the color navigation diagram on the upper right. The red series indicates a positive correlation, the blue series indicates a negative correlation, and the darker the color, the larger the absolute value of r. Italics indicate that the P <0.05 Discussion Principal findings This cross-sectional study aimed to compare the ecosystem of allergic rhinitis (AR) information on two major video platforms, YouTube and Bilibili, by integrating content quality assessment, engagement analysis, and search trend data. It is important to emphasize upfront that this study focused exclusively on video media, which represents only one of the many digital channels through which the public obtains health information. Other digital resources—including authoritative medical websites, official health apps, and evidence-based AI tools developed by clinical institutions—can also provide reliable health information aligned with international guidelines. However, accessing and evaluating these high-quality resources requires adequate digital health literacy skills [32]. The findings indicate that while the overall quality of AR-related videos was limited on both platforms, significant differences existed. YouTube content achieved higher scores in reliability and understandability, corresponding with a greater proportion of uploads from professional sources. Bilibili, in contrast, demonstrated superior user engagement metrics, supported by its interactive features, but its content landscape was dominated by self-media creators. Importantly, a seasonal alignment was observed, where video publication volumes and certain engagement metrics on each platform showed positive correlations with search interest on their corresponding regional search engines. A notable exception was the treatment-related content on Bilibili, which exhibited a negative correlation with search volume. This inverse relationship raises concern that, during peak demand periods, the information landscape may be saturated with lower-quality treatment advice. Platform ecosystems, content quality, and cross-cultural context The distinct profiles of AR content observed on YouTube and Bilibili may stem, in part, from deeper divergences in their platform governance, core user bases, and embedded cultural contexts [33–35]. YouTube's environment, which hosted a relatively higher share of content from healthcare professionals and institutions, may align with its global reach and ongoing efforts to surface authoritative health information [14]. In contrast, Bilibili's ecosystem, deeply rooted in Chinese youth digital culture, thrives on community interaction and creator-led content, naturally elevating the visibility of self-media creators [36]. It is noteworthy that among verified accounts, the proportion of physicians was higher on Bilibili (61.5%, n = 24/39) than on YouTube (27.7%, n = 5/18), yet the overall quality scores on Bilibili were lower. This finding could indicate that the impact of these professional voices may be diluted within the platform's vast pool of non-professional content. The superior user engagement observed on video platforms, particularly Bilibili, reflects the inherent advantages of short-form video as a health communication medium. Compared with text-based resources, videos use visual demonstrations, narrative storytelling, and fragmented content delivery to lower the threshold for accessing complex medical information, aligning with the fragmented reading habits of modern digital users [37]. Additionally, platform recommendation algorithms are optimized to prioritize content that generates high user interaction, creating a feedback loop that amplifies the reach of trending health topics. This algorithmic amplification effect explains why video content can exert a disproportionate influence on public health beliefs and behaviors, even when its scientific quality is suboptimal [33]. Furthermore, the substantial presence of Traditional Chinese Medicine (TCM) content on Bilibili—accounting for 34.2% of all videos—constitutes a key cultural and therapeutic divergence between the platforms. This finding mirrors the formal integration of TCM within China's mainstream healthcare system and its widespread public acceptance [38]. The prevalence of TCM-focused treatment advice, which may be based on theoretical frameworks and evidence standards distinct from those of Western medicine, could partially explain the lower overall quality scores observed on Bilibili [39]. Moreover, it may contribute to the observed negative correlation between treatment-themed videos and search trends: during seasonal peaks, users may be confronted with a mix of conventional and TCM-based recommendations, potentially leading to confusion or the dissemination of practices unsupported by international AR guidelines. These differences underscore the importance of a cross-cultural perspective in digital health communication research. Platforms are not neutral conduits but are co-shaped by cultural factors, regulatory environments, and user practices. Comparative frameworks suggest that information ecosystems reflect varying cultural values, such as trust in authority, and differing norms around content creation and consumption [40]. Therefore, our findings highlight that strategies for improving online health information cannot be one-size-fits-all. On creator-driven platforms like Bilibili, the focus might be on effectively integrating professional medical knowledge into the existing high-engagement community culture [33]. On more institutionally-oriented platforms like YouTube, efforts might better focus on further refining algorithms to recognize and recommend authoritative sources [41]. Seasonal dynamics, search trends, and algorithmic mediation Our study identified marked seasonal peaks in AR-related information searches during spring and autumn, which is consistent with known allergen cycles [23, 42]. A notable finding is the observed temporal alignment between fluctuating public demand and video platform activity. For instance, Bilibili's video upload volume and user engagement metrics showed positive correlations with Baidu search indices. While causal direction cannot be established from our data, this alignment is consistent with a potential reinforcing dynamic, whereby periods of high public search interest coincide with increased content publication and user interaction. One plausible explanation for this alignment involves platform recommendation algorithms. These systems, designed to maximize engagement, may prioritize content relevant to trending search topics [37, 43]. Such a mechanism could inadvertently favor videos with attention-grabbing titles over those with nuanced scientific merit [44]. This offers a potential explanation for the paradoxical finding in our study: on Bilibili, videos with higher Global Quality Scale (GQS) scores were associated with lower view counts (r = − 0.324, P < 0.001). Well-produced, evidence-based content may suffer from lower visibility if it is less optimized for algorithmic promotion. Furthermore, the surge of similar content during peak seasons may contribute to information overload [45, 46], leading users to gravitate towards simpler, more direct solution claims, thereby further marginalizing nuanced, high-quality information. Clinical and public health implications: from information quality to patient management A key public health consideration arising from this study is the strong negative correlation observed between treatment-themed video volume and search trends on Bilibili. This creates a specific risk window during seasonal peaks when symptom burden is high and patient demand for actionable treatment information is greatest. Our findings indicate that during these periods, the Bilibili ecosystem may become saturated with content from non-professional sources, which risks promoting unverified remedies while potentially neglecting guideline-endorsed first-line treatments like intranasal corticosteroids [15, 26]. The potential impact of such misleading information merits attention. Research in other health domains suggests that exposure to low-quality information online can influence patient behavior, potentially leading to reduced adherence to evidence-based care [47]. For a chronic condition like AR, where treatment adherence is crucial, such behavioral shifts could contribute to suboptimal clinical outcomes [48]. Therefore, this study connects identified flaws in the digital health information environment to tangible challenges in clinical management [49]. We suggest that clinicians caring for AR patients, especially during predictable seasonal peaks, should consider proactively inquiring about patients' online health information-seeking behaviors. Besides, clinicians could proactively recommend authoritative online resources (e.g., ARIA or ICAR guideline summaries) during consultations, especially before seasonal peaks, to counteract potential misinformation. This provides an opportunity to address potential misconceptions directly and to reinforce clear, evidence-based guidance, thereby transforming online information search from a potential risk into a chance for effective patient education [32]. To mitigate these risks, coordinated efforts across stakeholders are advisable. Video platform operators could refine content discovery algorithms by incorporating quality and authority signals, particularly during seasonal health events. The implementation of contextual labels, such as warning notices or fact-check banners on videos containing unsubstantiated claims, may also aid user discernment [50]. Healthcare institutions and professionals are encouraged to actively produce and disseminate authoritative content aligned with clinical guidelines. Collaborative initiatives with credible content creators on platforms like Bilibili could further bridge expertise with public engagement styles [51]. Public health agencies may consider launching pre-emptive digital communication campaigns ahead of seasonal peaks, partnering with platforms to promote evidence-based messages and enhance public digital health literacy [52]. Improving digital health literacy is particularly critical, as it empowers individuals to critically evaluate online health information, distinguish evidence-based content from unsubstantiated claims, and navigate multiple digital channels to access reliable resources. This dual approach—improving platform content quality and enhancing user literacy—can collectively mitigate the risks of health misinformation. Strengths, limitations, and future directions This study benefits from a cross-cultural comparative design, the application of multiple validated quality assessment instruments, and the integration of video data with temporal search trend analytics in the context of allergic rhinitis. These methodological choices provide a multifaceted perspective on the AR information ecosystem across two major digital platforms. Several limitations merit consideration. First, the cross-sectional design and correlational analyses prevent us from establishing causal relationships. Moreover, while bivariate tests were employed, we did not utilize multivariate models to adjust for potential confounders such as video length or uploader type. Consequently, our ability to discern the independent factors shaping video quality or engagement remains limited. Second, there was potential retrieval terminology bias. This study only used the professional medical term "allergic rhinitis" (English) and "过敏性鼻炎" (Chinese) for video retrieval, and did not include common lay terms such as "hay fever" and "nasal allergies". This may have led to incomplete inclusion of relevant videos on YouTube, particularly content targeting general audiences rather than medical professionals, and affected the comprehensiveness of the sample to a certain extent. Third, the platforms’ distinct regional accessibility means our comparison involves different user bases. Specifically, given YouTube’s restricted availability in mainland China and Bilibili’s primary use within Chinese-speaking communities, the study essentially contrasts two separate digital ecosystems—each with its own user demographics and cultural context—rather than tracking a single population across platforms. Fourth, we did not analyze the sentiment or detailed content of user comments, which could have provided further insight into public perceptions and misunderstandings. Last, this study only focused on video media, and did not evaluate the quality of AR-related health information from other digital resources such as websites, apps and AI tools. Therefore, the conclusions of this study cannot be generalized to all online health information ecosystems. The digital health information landscape continues to evolve rapidly. While this study focused on user-generated video content, the distinct characteristics of other prevalent sources warrant consideration to contextualize our findings. Generative AI platforms (e.g., ChatGPT) present a conversational paradigm but carry risks like “hallucinations” and lack transparent sourcing [53, 54]; traditional health websites remain widely used yet face persistent challenges in readability and reliability [26]. Future research should assess the quality of AI-generated AR content and empirically evaluate this study's practical recommendations (e.g., algorithmic adjustments, warning labels) for effectiveness. Conclusion This study demonstrates that content quality of allergic rhinitis (AR) information was generally higher on YouTube, whereas Bilibili achieved stronger user engagement through its interactive features. However, the overall quality of videos on both platforms remained limited, indicating a clear need for improvement in accuracy, clarity, and coverage. Importantly, a recurrent seasonal "demand–content" loop was observed, in which peaks in public search interest aligned with increased video uploads and interaction. This loop, however, may inadvertently amplify lower-quality or misleading treatment information during periods of highest patient need—a particular concern on Bilibili, where most content is produced by non-professional creators. To address this, we recommend consistently promoting evidence-based content year-round, attaching visible medical warning labels to videos with unsubstantiated claims, and encouraging collaborative production of accurate yet engaging materials between healthcare professionals and experienced creators. Implementing these tailored measures could help both platforms better serve as reliable sources of AR information, ultimately supporting informed patient self-management and adherence to guideline-based care. Supplementary Information Supplementary Material 1. Supplementary Material 2. Supplementary Material 3. Acknowledgements The authors would like to express their to the video uploaders for their contributions to public health. Authors’ contributions DL.L. led the study’s conceptualization and design. Data acquisition was coordinated by T.L. (video collection) and X.W. (metadata extraction). YY.Z., DL.L., and Y.L. independently conducted video quality assessments and scoring. ZY.L. and YW.C. performed statistical analyses and data interpretation. The initial manuscript was drafted by DL.L., with substantive revisions provided by MM.A. and J.W.. F.Y. and HL.G. contributed critical intellectual revisions to enhance scientific rigor. All authors participated in iterative editing and unanimously approved the final submitted version. Funding This work was supported by the Guangzhou Key Laboratory of Otorhinolaryngology Clinical and Basic Research (2023A03J0521) and Guangzhou Science and Technology Plan Project (2024A03J0564). Data availability No datasets were generated or analysed during the current study. Declarations Ethics approval and consent to participate All data in this study were derived from publicly accessible sources (YouTube, Bilibili, Google Trends, Baidu Index), and no personal identifiable information of any individual was collected or analyzed. This study did not involve human subjects, any intervention, or data collection from individual participants. This study was conducted in accordance with the Declaration of Helsinki. Ethical approval was formally waived by the Institutional Review Board of Guangzhou Red Cross Hospital of Jinan University (Approval No. 2026–165-01). The requirement for informed consent was also waived due to the anonymous and public nature of all data used. Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Huilong Guo, Email: huilongguo@126.com. Feng Yu, Email: fishwoo@sina.com. References Associated Data Supplementary Materials Supplementary Material 1. Supplementary Material 2. Supplementary Material 3. Data Availability Statement No datasets were generated or analysed during the current study.