What this review uncovers about cancer information on new media
As digital platforms have become primary sources for health information, understanding the quality of cancer-related content on new media is vital. A systematic review and meta-analysis of 75 studies published between 2014 and 2023 reveals that while new media offer valuable support and information sharing, the overall quality of cancer-related content remains moderate to moderate-low across several domains. This has important implications for patients, caregivers, and clinicians who rely on social media, video channels, and AI chatbots for cancer knowledge and decision-making.
Key characteristics of the evidence base
The included studies spanned text-based platforms (Twitter/X, Facebook, Reddit, WeChat), video platforms (YouTube, TikTok, Xigua), and generative AI-chatbot content (e.g., ChatGPT). Over time, the focus shifted from text- and image-based media to video content, driven by rising engagement metrics and policy changes that shape data access. A growing minority of studies began examining AI-generated information, marking a new frontier in information quality research. Across cancer types, breast, prostate, skin, and colorectal cancers dominated the literature, with rarer cancers receiving less high-quality scrutiny.
How information quality was assessed
Researchers used tools such as DISCERN, GQS (Global Quality Score), PEMAT (Understandability and Actionability), JAMA Benchmark Criteria, and Health on the Net (HON) codes to evaluate content. The review found that:
- Overall quality was moderate: DISCERN around 43.6/100 and GQS near 50/100.
- Understandability was generally better than actionability, with PEMAT-U around 67% and PEMAT-A around 37% for actionability.
- Misinformation affected about 27% of posts, with higher risk on certain platforms and content types.
- Completeness, or coverage of topics, was often limited, with only about one-third of essential topics being defined in many studies.
- Harmful content and commercial bias were present, though not uniformly across platforms or cancer topics.
AI-chatbot information, while promising in some cases for accuracy, still shows gaps in understandability and occasional hallucinations. The evidence base highlighted strong variation by platform, cancer type, and the expertise of content creators, with medical professionals and institutions generally delivering higher-quality information than non-professional sources.
What drives higher quality conclusions (RQ2)
Several factors consistently predicted higher-quality information conclusions. Text-based, information-rich content on common cancers tended to yield better quality assessments than video-based or content on rare cancers. Processes that included clear search strategies and explicit rating criteria, multiple raters with expertise, and use of structured quality tools were more likely to report higher-quality findings. Conversely, reliance on single search sources, languages other than English, or multi-platform analyses often correlated with lower-quality conclusions.
Patterns across media types and cancer topics (RQ3)
Video platforms attracted high engagement (views, likes, comments) but frequently delivered lower-quality content than text-based sources. In contrast, text-based posts—especially those from medical professionals and institutions—tended to score higher on DISCERN and related measures. AI-generated content showed potential for quality gains but also raised concerns about accuracy and readability. Actionability remained a notable weakness across formats, with many posts offering little practical guidance for patients making treatment decisions.
Clinical and public health implications
These findings underscore the need for clinicians to address patients’ use of new media during consultations. Encouraging healthcare institutions to maintain an active, evidence-based presence on popular platforms can help improve information quality. For researchers, standardizing quality assessments (using DISCERN, PEMAT, GQS, and JAMA criteria) will enable more robust meta-analyses and clearer cross-platform comparisons. For platform designers and AI developers, prioritizing transparency, source citation, readability, and actionable guidance is essential to reduce misinformation and cognitive burden on patients and caregivers.
Future directions and practical recommendations
To improve cancer information quality on new media, the review recommends a three-pronged approach: (1) expand high-quality content production by medical professionals on video platforms, (2) advance AI-chatbot accuracy and explainability with clear sourcing and disclaimers, and (3) implement standardized, transparent quality checks across platforms, languages, and cancer types. Moreover, ongoing monitoring of platform changes and user behavior will be critical to detect temporal shifts in information quality and reach.
Bottom line
New media increasingly shapes how people learn about cancer. While there is meaningful high-quality content, substantial gaps remain in accuracy, completeness, actionability, and bias. A collaborative effort among clinicians, researchers, platform providers, and AI developers is required to harness the benefits of digital health information while safeguarding patients against misinformation and harmful content.
