GLP-1 death headlines: what the MHRA’s Yellow Card figures actually show

What the source reports

The Independent reported on 6 August 2026 that more than 150 deaths had been linked to weight-loss injections in UK safety data. Its article quoted 98 fatal outcomes for tirzepatide, 37 for semaglutide and 37 for liraglutide—a total of 172 across the three active-substance profiles.

The same report said patients had filed about 150,000 reports of adverse reactions and serious side effects. It described more than 100,000 complaints for tirzepatide, just under 50,000 for semaglutide and almost 4,000 for liraglutide.

Those are striking numbers, but the article does not clearly separate a Yellow Card report from an adverse reaction recorded within that report. It also does not make clear in its headline that these are suspected events reported after use, rather than deaths proved to have been caused by the medicines.

Reports are not the same as reactions

A Yellow Card report relates to a patient and can contain more than one suspected reaction. Headache, nausea and abdominal pain reported together, for example, would be one report containing three reactions—not three patients.

When OurGLP1 accessed the live MHRA profiles on 9 August 2026, they showed 68,259 reports and 106,309 reactions for tirzepatide; 29,061 reports and 48,148 reactions for semaglutide; and 2,091 reports and 4,528 reactions for liraglutide.

Added across those profiles, that is 99,411 reports and 158,985 reactions. The newspaper’s figure of about 150,000 therefore closely resembles the reaction total, not the number of patient reports. Even the 99,411 profile total should not automatically be treated as 99,411 unique people across all three medicines, because a report can name more than one suspected drug.

What ‘fatal outcome’ means

The live profiles we accessed showed 120 reports with a fatal outcome for tirzepatide, 59 for semaglutide and 37 for liraglutide. That produces a simple profile sum of 216, which is different from the 172 obtained from the figures quoted by The Independent.

That sum is not necessarily 216 unique deaths caused by GLP-1 treatment. A fatal-outcome report means the person was reported to have died after a suspected reaction was submitted. The Yellow Card system requires suspicion, not proof of causality, and an individual report may include several suspected medicines.

A death can occur during treatment because of the condition being treated, another illness, another medicine or a combination of factors. Regulators assess the clinical details, timing, alternative explanations, patterns across reports and evidence from other sources before deciding whether a medicine may have contributed to a safety signal.

Why the live totals differ from the article

The MHRA says its interactive Drug Analysis Profiles are updated regularly and warns that recently submitted reports may take around a month to appear. A figure quoted on one date can therefore differ from a later profile.

However, the article does not identify the extraction date or explain why its quoted fatal-outcome figures differ from the profiles OurGLP1 accessed. The current dashboards display data extracted on 7 July 2026, so routine updating alone does not fully explain an article published on 6 August using lower figures.

There may be a different snapshot, filter or data source behind the newspaper’s numbers, but the available article does not provide enough detail to reproduce it beyond adding the three figures it quotes. The accurate response is to flag the difference, not invent an explanation.

The profiles are not limited to weight-loss treatment

The active-substance dashboards combine Yellow Cards across products and authorised uses. The semaglutide profile, for example, lists Ozempic and Rybelsus as well as Wegovy. Tirzepatide has been used for type 2 diabetes and weight management, while liraglutide products have also covered both areas.

Calling every entry a report from a ‘weight-loss drug’ user therefore goes beyond what these substance-level totals can establish. The profiles do not provide a clean breakdown in their headline totals by indication, brand, dose or whether a medicine was obtained through the NHS, privately or outside an authorised route.

The distinction matters because people taking these medicines can have different underlying health risks. It also means the totals cannot answer a narrower question such as how many deaths were caused by prescribed weight-management treatment in otherwise comparable patients.

Why raw totals cannot measure an individual’s risk

A spontaneous-reporting system does not record every event, and it does not supply a reliable denominator showing exactly how many people used each medicine for how long. Publicity, awareness, time on the market and the extent of use can all alter reporting.

A larger total can reflect wider use, longer availability, closer monitoring or more reporting as well as a genuine safety problem. Conversely, a small total cannot prove that a risk is absent because suspected events may never be reported.

This is why the MHRA explicitly says the profiles cannot be used alone to compare the safety of medicines or calculate how often a reaction occurs. The data are valuable for detecting possible signals, but a personal probability requires exposure data, appropriate comparison groups and clinical assessment of the reported events.

Different sources can tell different parts of the truth

A newspaper focuses on a number that attracts attention. A regulator publishes suspected reports so unusual patterns can be detected. A clinical trial estimates benefits and harms in a defined population, while a manufacturer discusses an authorised product’s evidence and safety information.

These sources are not necessarily describing the same question. The problem arises when a suspected-report count is presented as though it were a confirmed count of medicine-caused deaths, or when the number of recorded reactions is described as the number of patients.

Calling that difference a lie would require evidence of deliberate deception that the public material does not provide. The defensible conclusion is narrower: the headline and wording omit context that is essential for interpreting the numbers accurately.

What remains uncertain

The public profiles do not show how many fatal reports were ultimately judged likely, possibly or unlikely to have been caused by a GLP-1-based medicine. They also do not establish how many profile entries represent the same person or event across suspected drugs.

The figures alone cannot show whether fatal-outcome reporting is higher than expected among comparable people who did not use these medicines. Nor can they reliably separate risk by indication, product, dose, duration, age, underlying disease or route of access.

More detailed regulatory assessments, linked exposure data and carefully designed studies are needed to answer those questions. The existence of uncertainty is not a reason to dismiss reports; it is the reason regulators collect and investigate them.

What this means for the individual

A fatal-outcome headline can be frightening, particularly for someone currently taking one of these medicines. The numbers do not show that a person’s medicine caused 172—or 216—deaths, and they do not provide an individual risk estimate.

They also should not be treated as proof that no risk exists. New or severe symptoms deserve appropriate medical assessment, and suspected side effects can be reported through the Yellow Card scheme even when causation is uncertain.

A useful personal record includes the exact medicine and brand, dose dates, other medicines, symptoms and relevant medical history. That information can support a more informed conversation with a prescriber or urgent-care professional without turning a population-level headline into a personal diagnosis.

Bottom line

The headline is based on real Yellow Card data, but its wording compresses several different concepts. A report is not a reaction, a fatal outcome is not proof that a medicine caused a death, and an active-substance profile is not limited to weight-loss use.

OurGLP1’s check also found that the newspaper’s quoted fatal figures do not match the live profiles and that its approximately 150,000 figure is much closer to the number of reactions than the number of reports. Those distinctions do not make the safety reports unimportant; they make careful language essential.