Updated: July 9, 2025

Media Monitoring Methods

The Health Information and Trust Initiative aims to equip health communicators across media, policy, health, and community roles with timely insights into how health information is discussed both to and by the public. By tracking trends across both traditional and social media, the program identifies false, misleading, or unverifiable claims early, helping communicators stay informed about public discourse and respond more effectively. To achieve this, the program uses a media monitoring and social listening tool to monitor health-related narratives in near real time across news and major social media platforms.

Boolean Search Construction

The program uses a Boolean keyword search strategy to identify common (frequently repeated), emerging (recently appearing or accelerating), or high-engagement narratives (widely shared or reacted to), beliefs, claims, or other forms of information shared around health topics of interest through the media monitoring tool. These searches are designed to capture the different ways people discuss key topics of interest across platforms—using different spellings, phrases, and related terminology. Boolean logic (AND, OR, NOT) allows us to combine multiple terms and exclude irrelevant content to ensure the results are comprehensive and relevant. Queries are built with combinations of: Queries are tailored per topic and adjusted as narratives evolve. Search logic is tested iteratively to balance comprehensive capture, precision, and relevance.

Narrative Thresholds and Spikes

Thresholds represent the level at which data becomes noteworthy, and these thresholds are determined using KFF’s editorial judgment. They are flexible markers of engagement, based on baseline data over time and typical engagement patterns. When data falls below a certain threshold, it may lack meaningful insights or patterns. When it surpasses this threshold, it indicates something noteworthy and may be considered a “spike.”

A “spike” is defined as a sudden, large increase in online mentions of a set of specific search terms, brand, or topics, constructed through the above-described Boolean search logic. The relative importance of a spike is determined by comparing current mention volume with historical baselines for similar periods of time. For topics that typically experience large volumes of media coverage, such as COVID-19, KFF sets a higher threshold. In contrast, health topics with less widespread coverage have lower thresholds, meaning that even a smaller increase in mentions can generate meaningful data.

However, a high number of posts on a topic or narrative does not necessarily mean it is widely believed. Rather than indicating belief or consensus, such patterns are used to identify signals of public confusion, salience, or potential areas of concern. Additionally, prioritizing high-engagement content highlights widely shared posts but doesn’t always reflect broad public belief, since amplification may come from a small number of accounts or platform algorithms.

Narrative Inclusion Criteria

Each item included in the Monitor is reviewed through an editorial lens that considers not only data signals—such as volume, engagement, and velocity—but also the broader context, source credibility, and potential public health implications. This process involves careful human judgment to assess what is most relevant and actionable for our audience. Narratives are generally excluded if they are low in impact, unlikely to affect public discourse, or cannot be addressed effectively without risking further amplification. The editorial process ensures that inclusion is based on relevance and salience—not just the volume of online discussion.

Mitigating Amplification Risks

In addition to applying a weight-of-evidence approach, KFF takes steps to avoid amplifying unverified or misleading claims and protect privacy. This includes not linking to sources known to contain false information, using contextual language to accurately frame narratives, and limiting repetition of false or misleading claims to what is necessary for clarity or analysis. In line with evidence-based best practices, the program avoids presenting false narratives without also providing accurate context to correct or clarify the claim.

The program also filters out fringe or coordinated narratives that have not broken into broader discourse or shown signs of meaningful reach. This helps ensure that the program is not inadvertently elevating low-traction content and allows us to focus on claims with demonstrated impact or relevance. When quotes are included, they are selected to illustrate the tactics, language, or themes that contribute to the spread of misinformation and individual privacy is maintained. However, quotes that dehumanize or attack someone’s identity (e.g., based on race, gender, sexuality, religion, disability, or other factors) are excluded. In these cases, the program extracts and describes the rhetorical strategies being used—such as scapegoating, fear-mongering, or coded language—without repeating the harmful language directly. This helps avoid normalizing stigmatizing content while still enabling analysis of how it spreads.

View all KFF Monitors