Skip to content
This GLP
← Research
Evidence

Frailty did not change what semaglutide did

Split 17,604 trial participants three ways by a frailty index and the cardiovascular benefit holds across all three. The frailty scores themselves moved too.

Dana Sullivan7 min read
Semaglutide vs placebo, by frailtyno effectnot frail0.84 (0.651.07)more frail0.70 (0.590.82)most frail0.92 (0.761.10)

Frailty is the usual reason to withhold a drug from the people a trial says should get it. A secondary analysis of SELECT tested whether that caution has a basis here, building a 31-item frailty index across 17,604 adults with established cardiovascular disease and no diabetes [1]. It is a different question from the one asked in the heart failure trial, and a much larger sample.

Thirty-one percent scored at or below 0.210 on the index, 47% between 0.211 and 0.310, and 22% at 0.311 or above. Events were more common as frailty rose, which is what a frailty index is for. The treatment effect did not track it: hazard ratios for the primary outcome ran 0.84 (95% CI 0.65–1.07) in the least frail, 0.70 (95% CI 0.59–0.82) in the middle group, and 0.92 (95% CI 0.76–1.10) in the frailest, with the test for interaction at P = .09.

Two results are not subgroup slicing and carry more weight. Benefit on the EQ-5D-5L quality of life score was larger at higher frailty, with the interaction test reaching P = .02. And frailty was not a fixed trait in this trial: between baseline and week 104, participants on semaglutide were more likely to improve a frailty category (OR 2.46, 95% CI 1.80–3.37) and less likely to worsen one (OR 0.47, 95% CI 0.34–0.65).

The safety result runs against the intuition that drives the caution. Adverse events leading to permanent discontinuation were relatively lower, not higher, among participants with greater baseline frailty (P < .001 for interaction). That does not mean frail patients tolerate the drug better in some absolute sense, since the comparison is against placebo within each stratum and frail placebo patients stop for their own reasons. It does mean the trial found no frailty-specific tolerability penalty, which is the thing the caution assumes.

What none of this settles is whether people this sick are being offered the drug at all, which is a separate and worse-documented problem covered in uptake after a heart attack or stroke. SELECT enrolled people without diabetes and with treated cardiovascular disease, so it does not describe the frailest patients in a clinic, and the body composition question that frailty raises most sharply is handled on its own terms in what these drugs do to muscle and in the wider cardiovascular results summarized in the heart failure evidence.

Frequently asked

Did semaglutide work better in some frailty groups than others?
No. The test for interaction was P = .09, meaning the differences between the three subgroup estimates are not distinguishable from chance. Only the middle group's interval excluded 1, which reflects how a trial behaves when split into thirds.
Can a drug change how frail someone is?
In this trial frailty category moved. By week 104, participants on semaglutide were more likely to improve a category (OR 2.46) and less likely to worsen one (OR 0.47) than those on placebo.
Were frail participants more likely to stop because of side effects?
The opposite pattern appeared. Hazard ratios for adverse events leading to permanent discontinuation were lower at higher baseline frailty (P < .001 for interaction), so the trial found no frailty-specific tolerability penalty.

Sources

  1. [1] Ostrominski JW, Plutzky J, Scirica BM, Hovingh GK, et al. (2026). Efficacy and Safety of Semaglutide According to Frailty Status: A Post Hoc Analysis of the SELECT Randomized Clinical Trial JAMA Cardiology. PMID 42747817

Where to get it

Best GLP-1 injections

Every injectable seller we can verify, with the price each one publishes and an honest read of what the trials measured.

Compare providers →

More in Evidence