Drug safety methods, worked through with real cases.
Drug Safety Matters, from Uppsala Monitoring Centre, builds its conversations around named guests and practical questions: disproportionality analysis, medical device incident reporting, artificial intelligence and the WHO Global Smart Pharmacovigilance Strategy. The talk runs on worked cases, such as the amyotrophic lateral sclerosis drug edaravone, whose huge statistical signal reflects why patients take it rather than any reaction to it.
PRESS PLAY
Drug Safety Matters is the pharmacovigilance podcast of Uppsala Monitoring Centre. Its subject is the machinery of medicine safety: how harms get reported, counted, analysed and acted on. Episodes centre on named guests. The titles give the range: 'How to perform better disproportionality analyses' with Michele Fusaroli and Eugene van Puijenbroek, two parts of 'Rethinking medical device safety' with Omar Aimer, an episode with Shanthi Pal on the WHO Global Smart Pharmacovigilance Strategy, and two parts of 'How to use artificial intelligence in pharmacovigilance' with Niklas Norén. That run leans toward method and practice. The disproportionality conversation opens on a method it calls easy and fast, then adds a caveat. Its clearest example is edaravone, a drug prescribed for amyotrophic lateral sclerosis, which shows a huge disproportionality with the disease it treats. The explanation offered is reverse causality: the event is the reason for taking the drug, and spotting that takes clinical understanding. Confounding comes next as a second trap. The medical device episodes widen the frame beyond pills, to everything from bandages to MRI machines. Devices can malfunction, degrade and interact with users in ways drugs cannot, and a software update can shift a device's risk profile. Cybersecurity and usability become safety questions, and health apps that measure signs on phones and smartwatches count as devices too. The fixes proposed are practical. Reporting should be simpler and built into everyday tools such as prescription software. It should also come with follow-up loops that show healthcare professionals what their reports led to. In Shanthi Pal's episode the theme is reliance: countries pooling data to strengthen a study, or one country building on another's results instead of repeating the work. Repeated active surveillance of malaria and HIV medicines is raised as spending that limited resources cannot support. The artificial intelligence conversation, with Niklas Norén, head of research at the centre, is split into two parts. It covers the uneven cost of mistakes. Missing a pair of duplicate reports wastes time, while wrongly removing an important report could delay a signal. Explainability gets the same task-by-task treatment. It is seen as essential for ambiguous work like signal detection and less needed for de-identification, where measured performance may be enough. Several voices present the show, including Federica Santoro and Alexandra Coutinho. Sign-offs credit listeners by first name for the questions they sent in, and a Fan Mail feature lets anyone text the show anonymously through a link in the show notes. The disproportionality episode is a good first listen, because the edaravone case shows in one example why a statistical signal is not yet a verdict.
Made for: Suits a listener who works with adverse event reports or safety data and wants methods explained through examples. It also suits anyone curious about how a spike in reports is told apart from a genuine drug reaction.
What sets it apart: It comes from Uppsala Monitoring Centre itself, and the centre's own head of research is among its guests. It keeps checking methods against specific cases, like the edaravone signal that turns out to be reverse causality or the duplicate-detection error that could hide a real signal.
Drug Safety Matters brings you the best stories from the world of pharmacovigilance. Through in-depth interviews with our guests, we cover new research and trends, and explore the most pressing issues in medicines safety today. Produced by Uppsala Monitoring Centre, the WHO Collaborating Centre for International Drug Monitoring. The views and opinions expressed in the podcast are those of the hosts and guests respectively and, unless otherwise stated, do not represent the position of any institution to which they are affiliated.
The denominator can change a vaccine safety signal; you'll appreciate the second look.
Press play if you want
Short reviews from the PlayNext desk, based on the episodes we processed.
A focused guide to why pharmacovigilance signals can look alarming when bias, confounding, or reverse causality is doing the work.
Disproportionality analysis is pharmacovigilance's quick, familiar way of spotting possible safety signals, and this episode is most useful on what can go wrong next: treating a statistical alert too quickly as evidence of a drug reaction.
A grounded review of explainability, privacy and human judgment as AI moves into pharmacovigilance.
AI arrives in pharmacovigilance here not as a spectacle, but as a question of professional judgment: what does someone need to know before trusting a system's output?
Oncology evidence translated into clinical decisions.
Critical care evidence, carefully talked through.
Clinical research, reduced to its practical signal.
["Drug Safety Matters asks how to perform better disproportionality analyses with Michele Fusaroli of Uppsala Monitoring Center and Eugene van Puyenbroek of the Netherlands Pharmacovigilance Center, LAROV.", "The episode centers on their paper, \"Charting and Sidestepping the Pitfalls of Disproportionality Analysis,\" and uses the fire-alarm analogy to show why a signal is not proof.", "They give concrete examples of reverse causality with edaravone and amyotrophic lateral sclerosis, confounding in child vaccines and growth retardation, and amplified media attention after the late-1990s vaccine-autism wave.", "The discussion ends with training, case-by-case assessment, triangulation with literature and pharmacology, and stricter standards for journal papers and tools like FAERS and VigiBase."]
Medical device incidents may capture only 1 to 5% of actual events, below the 5 to 10% rate for drugs.
Drug Safety Matters examines why medical-device vigilance extends from bandages and thermometers to MRI machines and robotic surgical systems.
This episode covers the WHO's new Global Smart Pharmacovigilance Strategy, discussed at an international meeting.
["Drug Safety Matters #41 continues a two-part interview with Niklas Nor\u00e9n, the head of research at Uppsala Monitoring Centre, on artificial intelligence in pharmacovigilance.", "Nor\u00e9n argues that AI review has to separate model transparency from performance transparency, because a system can be understandable on paper and still fail on the data that matter.", "He uses duplicate detection and low-prevalence targets to show how accuracy can be misleading, and he recommends pilot studies in the real use case before trusting a score.", "A listener question from Harika leads into practical guidance on SIOMS, data privacy in prompts, model freezing, and the possibility that AI could co-pilot adverse-event reporting."]
Niklas Norén, head of research at Uppsala Monitoring Centre, says that using disproportionality analysis to triage signals is a basic form of AI, and he cites the chess engine Deep Blue as an example of AI without machine learning.
This episode introduces a new listener feedback feature and discusses risk communication in medicine with expert Alexandra Freeman.