📄 medicine

Characteristics and Quality Assessment of Influenza-related Video on the Chinese Version of TikTok During an Epidemic Season: A Cross-Sectional Study

This cross-sectional study of 92 influenza-related videos on the Chinese version of TikTok during an epidemic season found that while overall content quality was moderate-to-high with news organizations outperforming physicians, user engagement metrics failed to predict quality, whereas video length and uploader follower count emerged as significant positive predictors.

Original authors: Xiaofang Jiang, Feng Yu, Jingbang Liu, Jiahan Hu, Meng Wang, Shina Qiao, Jinhang Xu, Lili Wu

Published 2026-06-25

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Xiaofang Jiang, Feng Yu, Jingbang Liu, Jiahan Hu, Meng Wang, Shina Qiao, Jinhang Xu, Lili Wu

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the internet as a massive, bustling marketplace where people go to find answers about the flu. In this marketplace, TikTok (the Chinese version) is like a giant, fast-moving stage where short videos are performed every second. This study is like a team of quality inspectors who walked onto that stage during the height of flu season to see what kind of performances were happening and how good they actually were.

Here is what they found, broken down into simple terms:

1. The Cast of Characters

The researchers looked at the top 100 most popular flu videos. They found two main groups of performers:

The Doctors: About 60% of the videos were made by doctors. They came from all sorts of specialties, not just flu experts, but also heart doctors, pediatricians, and even Traditional Chinese Medicine practitioners.
The News Teams: About 26% were made by news organizations (like TV stations or online news outlets).

The Analogy: Think of the doctors as skilled mechanics who know how to fix the car (the body), but the news teams are like the professional broadcast crews who know how to tell the story clearly and quickly.

2. The "Speed" of the Show

Flu season is fast, and the videos were fast too.

The videos were very fresh. The median age of a video was only 9 days old.
This is like a daily newspaper rather than a textbook. The information was constantly updating to match the current "outbreak" of the flu.

3. The Quality Report Card

The researchers used four different "report cards" to grade the videos. Here is the verdict:

Overall Grade: The videos were generally "B" to "A-" quality. They were easy to watch, technically clear, and mostly useful.
The Big Problem: Almost all videos failed the "Source Check."
- The Metaphor: Imagine a chef serving you a delicious soup but refusing to tell you where the ingredients came from or if they are fresh. Most videos didn't say where their information came from, didn't list references, and didn't say when the video was recorded.
- The Result: Because they lacked these "receipts," it was hard to know if the advice was trustworthy or outdated.

4. Who Made the Best Videos?

Surprisingly, the News Organizations scored higher than the individual doctors on almost every quality metric.

Why? The news teams were better at citing their sources, being transparent, and keeping their content balanced. They were like the professional broadcast crews who follow strict rules, whereas individual doctors were sometimes more casual.
Content Matters: Videos about "Hot News" and "Policy" (like government rules) were shorter and higher quality than videos just trying to teach general prevention tips.

5. The "Popularity" Trap

This is the most important finding: Being popular does not mean being good.

The Analogy: Think of a viral dance video. Just because 1 million people liked it, shared it, or commented on it, doesn't mean the dancer is a professional gymnast.
The Data: The researchers found zero connection between how many "likes" or "shares" a video got and how accurate or high-quality the medical advice was. A video could be wildly popular but full of bad advice, or it could be high-quality but have fewer views.

6. What Predicts a Good Video?

The study found two things that did predict a higher quality score:

Length: Longer videos tended to be better.
- Analogy: A 30-second video is like a quick soundbite; a 2-minute video allows the speaker to explain the "why" and "how" more thoroughly, like a mini-lecture instead of a headline.
Follower Count: Accounts with more followers tended to make better videos.
- Analogy: It's like trusting a famous, established restaurant over a random food truck. The accounts with huge followings (often news orgs or senior experts) seemed to have a "reputation mechanism" that kept their quality high.

The Bottom Line

During flu season, TikTok is a busy, fast-paced place full of helpful information, but it's also a bit of a "wild west."

The Good: Doctors and news teams are trying hard to help.
The Bad: Most videos forget to show their "receipts" (sources), making it hard to trust them.
The Warning: Don't just watch the videos with the most "likes." Popularity is not a guarantee of truth.
The Fix: To make things better, creators should act more like the news teams: clearly state where their info comes from and when they made the video. And for the apps, they should stop just recommending videos based on "likes" and start looking for actual quality.

Technical Summary: Characteristics and Quality Assessment of Influenza-related Video on the Chinese Version of TikTok

Problem Statement
Influenza remains a globally monitored priority infectious disease where public awareness is critical for effective prevention. While video-based social media platforms like TikTok have become primary channels for public health information access, the quality of influenza-related content on these platforms, particularly during epidemic seasons, remains under-evaluated. There is a significant risk that misinformation or low-quality content could lead to inaccurate health decisions. Existing studies have evaluated disease-specific videos on social media, but systematic assessments of short-form video content on TikTok during critical influenza seasons are limited. This study addresses the gap in understanding the source characteristics, content themes, engagement metrics, and overall quality of influenza-related videos on the Chinese version of TikTok.

Methodology
The study employed a cross-sectional design conducted on November 14, 2025.

Data Collection: Researchers created a new account on the Chinese version of TikTok (version 36.5.0) to search for the keyword "流感" (influenza). To minimize algorithmic bias, the top 100 ranked videos were collected within 24 hours without filters. Videos unavailable at the time of assessment or unrelated to influenza were excluded, resulting in a final sample of 92 videos.
Data Extraction: Metadata included uploader characteristics (type, verification status, follower count), content features (theme, length, duration since upload), and engagement metrics (likes, comments, favorites, shares). Uploader types were categorized as physicians, news organizations, individuals, or institutions. Content themes were classified into "Rumor Refutation and Knowledge Popularization," "Prevention and Medical Guidance," and "Policy Interpretation and Hot News."
Quality Assessment: Two senior health communication experts independently evaluated all videos using four validated instruments:
1. Global Quality Score (GQS): A 5-point Likert scale assessing overall usefulness and fluency.
2. Modified DISCERN (mDISCERN): A 5-item binary scale evaluating reliability and source citation.
3. Video Information and Quality Index (VIQI): A 4-dimension, 5-point scale assessing production and informational quality.
4. Medical Quality Video Evaluation Tool (MQ-VET): A 15-item, 5-point scale evaluating content, technical aspects, and reliability.
Statistical Analysis: Analyses were performed using SPSS 25.0 and R 4.5.1. Non-parametric tests (Kruskal–Wallis with Dunn's post hoc) compared groups. Spearman correlation analyzed associations between characteristics and quality. Stepwise multiple linear regression identified predictors of the GQS. Inter-rater reliability was assessed using weighted Kappa and Intraclass Correlation Coefficients (ICC).

Key Results

Demographics and Engagement: Of the 92 videos, 59.78% were uploaded by physicians (spanning 21 specialties) and 26.09% by news organizations. The median video duration was 9 days, indicating high timeliness. News organizations had significantly higher follower counts but produced significantly shorter videos than physicians.
Quality Scores: The overall quality was moderate to high. The mean GQS was 3.47 (range 1–5), mean mDISCERN was 2.15 (range 1–4), mean VIQI was 16.87 (range 11–20), and mean MQ-VET was 53.04 (range 31–65).
Critical Deficiencies: A major shortcoming identified was the lack of transparency regarding information sources. Only one video cited other information sources (mDISCERN item 4), and fewer than 10 mentioned uncertainty (item 5).
Group Differences: Videos by news organizations scored significantly higher than those by physicians on mDISCERN, VIQI, and MQ-VET scales. "Policy Interpretation and Hot News" content also outperformed other themes on all quality scales. Among physicians, associate chief physicians achieved higher GQS scores than attending physicians.
Correlations: User engagement metrics (likes, shares, etc.) showed weak or negligible correlations with quality scores ( $r = 0.00–0.26$ ), indicating that popularity does not equate to quality.
Predictors: Stepwise regression identified video length ( $\beta = 0.005, P < 0.001$ ) and uploader follower count ( $\beta = 0.108, P = 0.005$ ) as independent positive predictors of GQS. Together, these variables explained 31.4% of the variance in quality scores ( $R^2 = 0.314$ ).

Key Contributions

First Systematic Evaluation: This is the first study to systematically evaluate the characteristics and quality of influenza-related short videos on the Chinese version of TikTok during an active epidemic season.
Identification of Quality Gaps: The study highlights a pervasive lack of source transparency and citation in health videos, even those produced by medical professionals, which undermines the verifiability of health information.
Decoupling Engagement from Quality: The research provides empirical evidence that high user engagement (likes, shares) is not a reliable proxy for content quality, challenging current platform recommendation logic.
New Predictive Variables: The study introduces "uploader follower count" and "video length" as significant, independent predictors of video quality, suggesting that account reputation and content depth are better indicators of quality than engagement metrics.

Significance and Claims
The authors claim that during influenza seasons, the TikTok ecosystem is dynamic and dominated by physicians and news organizations. While the overall quality is moderate, the absence of cited references and source transparency is a critical flaw. The study asserts that the involvement of news organizations significantly enhances video quality, likely due to their adherence to professional standards regarding attribution.

The paper concludes that relying on engagement metrics to curate health content is ineffective and potentially risky. Instead, it suggests that video length and uploader follower count could serve as auxiliary indicators for quality. The authors propose that health communicators should adopt the transparency practices of news organizations (citing sources and dates) and that platform operators should integrate evidence-based quality assessments into their recommendation algorithms to better guide public health communication strategies. The study explicitly notes that while AI-generated content was not observed in this dataset, the increasing use of AI in video production necessitates continued vigilance regarding information accuracy.