Same Medical Research, Different Conversations: Clinical Trials on Bluesky and X

The research was an exploratory comparison of language and participation, without formal statistical testing to establish clear contrasts or evidence that platform design caused them. It did not measure whether users joined trials, changed treatment decisions or received better information.

Finding information about a clinical trial on social media can mean encountering research updates, hopeful treatment announcements or personal doubts about medical care. A study titled 'Accessing clinical trial information via two social media platforms: the new normal for patients and stakeholders,' published in npj Scientific Reports as an article in press, explores how those conversations differed between Bluesky and X, formerly Twitter.

Researchers Marlena Krawczyk-Suszek, Arkadiusz Gaweł, Martin Braddock and Andrzej Adamski found a more research-focused discussion in the Bluesky sample and a more varied, event-driven conversation on X, raising questions about how patients and professionals encounter information about emerging treatments.

Two Platforms, Unequal Snapshots

The team examined 8,849 public English-language Bluesky posts published between January 18 and May 9, 2025, and 5,520 X posts published between June 25 and June 30, 2025. Posts were retrieved using the keyword "clinical trial," with duplicates, irrelevant material and content lacking meaningful text excluded. Computer-based text analysis identified common words, recurring phrases, terms appearing together and vocabulary associated with positive or negative sentiment and different emotions.

The unequal collection periods are central to understanding the findings. Bluesky allowed access to posts across several months, and restrictions on X limited collection to six days, meaning the samples could reflect different news events and public concerns. The research was an exploratory comparison of language and participation, without formal statistical testing to establish clear contrasts or evidence that platform design caused them. It did not measure whether users joined trials, changed treatment decisions or received better information.

Research Updates Meet a Broader Conversation

Bluesky posts frequently used words connected with research, patients, treatment, drugs and funding, with cancer research particularly prominent. Recurring phrases concerned trial phases, patient participation and diversity, alongside references to institutions and research support. Maps showing which words appeared together revealed a closely connected core of medical and scientific vocabulary, suggesting that clinical research provided a common focus across much of the sampled discussion.

X carried a stronger emphasis on vaccines, childhood immunisation and placebo-controlled studies, with medical vocabulary appearing among smaller, less connected groups of topics. Treatment discussions mixed with promotional language, informal commentary and material outside medicine. The difference concerned the organisation and wording of the sampled conversations; it did not establish that one platform consistently provides more accurate information or that all its users share the same interests.

The vocabulary comparison showed limited overlap: Bluesky contained 20,648 distinct text units, compared with 10,220 on X, and 4,229 appeared in both samples. Shared terms represented roughly 16% of the combined vocabulary, giving a Jaccard similarity score of 0.159. Bluesky's longer collection period and larger sample could account for some of its greater vocabulary range, preventing a firm conclusion that its conversations are inherently richer.

Most contributors on both platforms posted only occasionally. Two accounts stood out among X's leading contributors, but the measure covering all authors showed more evenly distributed activity on X. The Gini coefficient, which rises as posting becomes more concentrated, was 0.150 for X and 0.321 for Bluesky. The six-day X window gave users fewer opportunities to post repeatedly, making this comparison sensitive to the sampling periods. No formal bot-detection analysis was conducted, leaving the influence of automated accounts unresolved.

Why Negative Words Can Mislead

An automated sentiment analysis initially found more negatively classified posts than positively classified ones on Bluesky, including over 3,000 negative and approximately 2,600 positive classifications. Medical terms such as cancer, death, pain and risk helped produce that pattern, even though these words can appear in straightforward research updates. A post explaining a cancer study does not necessarily express pessimism, distrust or opposition to clinical research.

Removing selected common medical terms changed Bluesky's results to an approximately balanced positive–negative distribution. Positive classifications remained more common on X, where words such as "exciting" and "groundbreaking" contributed to a more promotional tone. A separate emotion dictionary found relatively more fear- and trust-associated vocabulary on Bluesky and more joy- and anticipation-associated vocabulary on X. These results describe word choices, rather than directly measuring the feelings or beliefs of the people posting.

A focused comparison of the cancer treatment pembrolizumab, sold as Keytruda, brought these differences into view. Bluesky discussions linked the treatment with survival, patient groups, registries, effectiveness and pharmaceutical companies. X combined clinical terms about treatment, surgery and combination therapies with informal or critical language, including references to recovery, failure and distrust. The word associations suggest different ways of discussing the same therapy, but cannot establish whether individual claims were accurate or reveal the intentions behind them.

What This Could Mean for Patients

The researchers suggest that X could help trial teams reach people who might take part in clinical trials, and Bluesky could help them connect with doctors, researchers and patient groups that refer participants. These platforms could help teams explain studies and understand people's concerns about treatment risks, access and trust. The study did not test whether using them this way would attract more participants or people from different backgrounds.

Clear, everyday explanations of medical research could help people understand clinical trials and the concerns they see online. Looking at conversations across several platforms might help researchers spot discussions about possible medicine side effects. These are ideas for future research, and the study does not show that they improve drug safety, build trust or help patients communicate with doctors.