The Drugs Nobody Writes About

A pharmacovigilance study found muscle atrophy reported disproportionately often with semaglutide and tirzepatide.^[1] The same table shows exenatide and liraglutide — older drugs acting on the same receptor — reported significantly less often than average. That pattern is the most informative thing in the paper, and it does not point at the drugs.

What was measured

The researchers searched the FDA Adverse Event Reporting System — the database anyone can submit a suspected drug side effect to — from October 2003 to March 2024, and found 142 reports of muscle atrophy involving GLP-1 medicines. They then calculated reporting odds ratios: how often muscle atrophy appears in reports about these drugs compared with reports about everything else.

Read all four rows together

Every drug in that table is a GLP-1 receptor agonist. If activating that receptor caused muscle wasting, the effect should appear across the class — strongest where the weight loss is largest, certainly, but present throughout.

Instead the two older agents come in at 0.26 and 0.27, meaning muscle atrophy is reported roughly a quarter as often in connection with them as with drugs generally. Both intervals sit well below 1. This is not an absence of signal; it is a signal pointing the other way.

The two drugs with the signal are the two drugs in the headlines. The two without it are the two nobody writes about.

The authors are clear about this in their own conclusion: the findings should be read as signal detection rather than evidence of causality, and they call for studies using objective measures of muscle mass and function. That is the right conclusion from this method, and it is not what a headline will do with a reporting odds ratio of 2.39.

What this is not saying

The trial evidence on what actually happens to lean tissue is in lean mass on a drug versus dieting, and the distinction between losing muscle bulk and losing strength — which is not the same thing — is in muscle volume versus muscle function.

Two further limits on the numbers above. One hundred and forty-two reports, spread across more than twenty years and many millions of prescriptions, is a very small base for any ratio. And 57% of the reports concerned men, in a population of GLP-1 users that skews substantially female — an oddity the study does not explain and neither will we.

How to read any FAERS study

• It counts reports, not cases. There is no denominator, so no rate can be calculated from it.
• Attention drives the numerator. Media coverage, litigation and label changes all increase reporting without changing what is happening in bodies.
• Compare within the drug class. If a mechanism-based effect were real, related drugs should show it. When they show the opposite, suspect the method.
• Treat it as a question, not an answer. Signal detection exists to decide what to study properly next.