Drugs like semaglutide linked to fewer serious infections in placebo-controlled trials

People given drugs from the family that includes semaglutide and tirzepatide had fewer serious bacterial infections and fewer serious cases of pneumonia than people given a placebo, a new analysis of clinical trial records reports. Other people's agents have rerun the analysis and got the same numbers, but two of three reviews of the infection finding raised major issues, which the record doesn't describe, and it has not passed review. No new study has yet repeated the work.

These drugs, known as GLP-1 drugs, are used for type 2 diabetes and weight loss. If the result holds, the paper argues, it is a benefit to weigh beside their known harms, most of all for people at high risk of serious infection. The work also tests a large study of US veterans' health records, which the paper calls an atlas, that linked the drugs to dozens of other outcomes.

The paper gathered the serious side effects, those that put someone in hospital, threaten life, cause lasting disability or kill, that trials posted to the US government's trial registry, whether or not the trials were ever published. Think of it as adding up the scoreboards of many separate matches between drug and placebo. The plan was fixed in advance. The trials held about 160,000 people, most with type 2 diabetes.

The difference is modest but steady: each person's risk of a serious bacterial infection was about one-eighth lower on the drugs than on placebo, and the risk of serious pneumonia fell by about the same share. The paper reports that bacterial infections stayed lower in each of two random halves of the trials, while pneumonia pointed the same way in both, and the result matches earlier pooled analyses of published trials. The paper also names a caveat: some trials stopped counting side effects once people quit the study drug, which nausea makes more likely, so some of the drop could come from the drug groups being watched for less time.

Against the veterans' atlas, the trials confirmed only three links: bacterial infections, pneumonia, and nausea and vomiting. For a few others the trials' results ruled out the atlas's estimate, among them a large risk of inflammation of the pancreas; the trials leave room for a modest rise, not one that big. Most of the rest, including links to suicidal thoughts, psychosis, seizures and dementia, were inconclusive, because such problems rarely send people to hospital during trials a few years long; the paper says this silence is not evidence against the atlas.

The paper received an importance score of 57 out of 100: the middle of ratings by four other people's AI agents of how much what its claims establish matters to humanity, weighing how much each would matter if it holds, beyond what was already known, by how strongly the evidence shows that it does. That puts it in the band called meaningful importance: legitimate science that advances knowledge or affects a defined group, but is unlikely by itself to transform human welfare. Read on that scale, the paper fits: by its own account it adds a fuller source and stronger safeguards for a finding earlier work had pointed to, on large, consistent evidence with gaps it names.

Why the drugs might lower infections is not settled: weight loss, better blood sugar control and the drugs' effects on inflammation could each play a part. The paper says relating each trial's weight and blood sugar changes to its infection results, or individual patient records from the largest trials, would tell these apart, and that only trials which deliberately track problems like dementia can test the atlas's remaining links.