When it comes to understanding how conspiratorial health content spreads, few platforms are as important, or as difficult to study, as Telegram. With its millions of users worldwide, Telegram became a hub for health updates, policy debates, and alternative viewpoints during the pandemic. Its importance lies in how quickly information circulates through large group chats and channels with minimal content moderation. But unlike more transparent platforms, Telegram’s closed group chats and encrypted features make it challenging to track. This combination allows conspiratorial health content to spread quickly while remaining largely hidden from traditional monitoring efforts.

Nimblemind’s paper, "User Archetypes and Information Dynamics on Telegram: COVID-19 and Climate Change Discourse in Singapore", explores how conspiratorial health content spreads on Telegram, and what roles different types of users play in shaping these conversations.

Why Telegram Matters for Public Health

Unlike Facebook or Twitter, Telegram’s groups and channels allow large communities to share content quickly and often without oversight. Around the world, Telegram has become a venue for everything from health advice to political movements. This environment creates both opportunity and risk; while most users are passive or genuinely curious, a small, highly active minority can amplify conspiracy theories and alternative health narratives, influencing public opinion in ways that may undermine public health efforts.

Archetypes Behind the Messages

We analyzed more than 200,000 messages across public Telegram groups focused primarily on COVID-19. By combining behavioral features (like number of posts or link-sharing) with linguistic features (like sentiment or topic similarity), and applying clustering techniques, three archetypes emerged in COVID-19 health discussions:

• Inquisitive Moderates: the majority of users; they ask questions, share occasionally, but rarely push strong narratives.

• Critical Examiners: prolific, analytic contributors relying on data and policy discussions.

• Conspiratorial Amplifiers: a vocal minority spreading alternative theories, often through heavy link-sharing.

Even a small set of conspiracy-driven messages can undermine vaccination, delay diagnosis, and increase unnecessary clinic visits.

What We Learned from Telegram Conversations

Most users are passive. Across COVID-19 groups, nearly 96% of participants were Inquisitive Moderates. Only 3.4% were Conspiratorial Amplifiers, but their outsized activity shaped much of the contrarian discourse, particularly around vaccines, treatments, and public health measures.

Links drive health narratives: Conspiratorial Amplifiers leaned heavily on external links to spread alternative health narratives, from unproven remedies to distrust in vaccines, suggesting that counter-messaging efforts should focus on these high-link-density users.

Automation can safeguard health communication: Our decision-tree classifier achieved high performance in assigning users into archetypes. This shows that real-time monitoring and triage of high-risk health messaging is not only feasible but could be deployed to intervene before harmful narratives gain traction.

Why This Matters

By surfacing these archetypes, our research provides a framework for understanding how conspiratorial health content and skepticism spread in closed platforms. For public health leaders, this means designing more targeted interventions that can counter conspiracy-driven content and provide timely, accurate information to those who are curious but undecided.

For healthcare systems and providers, the implications are equally important. Reducing skepticism and conspiratorial health content can increase trust in clinical guidance, improve adherence to treatment plans, and ultimately lead to higher patient engagement. That engagement translates into fewer missed appointments, better uptake of preventive care, and more consistent use of services—improving both patient outcomes and revenue capture. 

For a deeper dive into our methods, findings, and implications, check out the full paper on arXiv →