SELECT frailty analysis of semaglutide

A new post hoc analysis of the SELECT randomised trial asks whether baseline frailty changed the balance of benefit and adverse events seen with semaglutide. The short answer is that the investigators did not detect a meaningful difference in the cardiovascular effect across their frailty categories. That is a finding within a specific trial population, not evidence that the same result applies to every older or frail person considering a GLP-1 medicine^[1].

Published in JAMA Cardiology on 16 September 2026, the analysis reused data from SELECT, a completed placebo controlled cardiovascular outcomes trial rather than enrolling a new frailty cohort. Frailty was measured at baseline with a 31 item cumulative deficit index. Participants were then grouped as not frail, more frail, or most frail. The semaglutide evidence overview provides context while this paper evaluates whether treatment effects differed by those starting categories^[1].

Who was included in SELECT

SELECT enrolled 17,604 adults aged 45 years or older with established cardiovascular disease and a body mass index of at least 27. Participants did not have diabetes at entry. They were randomly assigned to semaglutide or placebo alongside standard care, and the trial was designed to assess cardiovascular outcomes. The original report found fewer first cardiovascular death, nonfatal myocardial infarction, or nonfatal stroke events in the semaglutide group during follow up. The semaglutide profile gives readers a separate overview of the peptide^[2].

That eligibility frame sets the boundary for this frailty analysis. It is about people with prior cardiovascular disease plus overweight or obesity, without diabetes, who met SELECT's entry criteria. The scope is narrower than a general trial of frailty treatment. It does not cover every setting where people use semaglutide. For background on the molecule and the evidence categories covered on this site, see the semaglutide overview.

What the frailty analysis found

At baseline, 31% of participants were in the not frail category, 47% in the more frail category, and 22% in the most frail category. The primary cardiovascular outcome became more common as baseline frailty increased. Estimated hazard ratios for semaglutide versus placebo were 0.84 in the not frail group, 0.70 in the more frail group, and 0.92 in the most frail group. The interaction test was not statistically significant, so the analysis did not show that baseline frailty modified the primary cardiovascular effect. See the semaglutide reference page for a separate clinical evidence summary^[1].

The authors also examined a composite heart failure outcome, all cause hospitalisation, all cause mortality, health related quality of life, and adverse events leading to permanent discontinuation. Their reported interaction tests did not indicate frailty related differences for the first three outcomes. Quality of life scores appeared to improve more with semaglutide at higher frailty levels. The study also found that a participant's frailty category was more likely to improve and less likely to worsen by week 104 in the semaglutide group. More context is available in the semaglutide research overview^[1].

These estimates are useful for describing the observed trial pattern, but they are not a personal prediction tool. Each frailty category includes people with different medical histories, functional ability, concurrent treatments, and reasons for stopping medication. The semaglutide regulation information is separate from the clinical question addressed in SELECT.

Why post hoc matters

A post hoc analysis is conducted after a trial has produced its main results. It can test a clinically relevant question, yet it usually has less certainty than a result specified as a primary analysis before the trial began. In this case, the authors report interaction tests rather than proving equivalence among frailty groups. A lack of detected heterogeneity means this dataset did not establish a clear difference in effect by baseline frailty; it does not prove that all such groups respond identically. That distinction matters when comparing subgroup estimates. Readers can also consult the semaglutide evidence page for broader context^[1].

The frailty measure also deserves care in interpretation. SELECT used a cumulative deficit index built from 31 items, which is one operational way to classify frailty. It should not be treated as interchangeable with bedside assessment, gait testing, clinical judgement, or a diagnosis. The original SELECT report documents the cardiovascular disease and no diabetes entry criteria^[2].

Safety, discontinuation, and clinical decisions

The new analysis reports that adverse events leading to permanent discontinuation of semaglutide versus placebo appeared lower among participants with higher baseline frailty, with evidence of interaction. That result should be read alongside the original SELECT report, where discontinuation events were more frequent overall in the semaglutide group than in the placebo group. Context matters when two safety comparisons describe different denominators. The two statements address different comparisons and neither replaces an individual safety assessment. The semaglutide profile is a separate resource, not a substitute for clinical review^[1][2].

Frailty can make medication decisions more complicated because nutrition, muscle function, symptom burden, other medicines, and cardiovascular history may all matter. This paper does not supply a dosing plan or a rule for starting or stopping treatment. Anyone weighing semaglutide in the setting of frailty should review their goals, current symptoms, and medical history with the clinician responsible for their care. The semaglutide page can help readers distinguish the evidence summary from individual medical advice.

What this study adds and leaves open

The added value is a detailed look at frailty within a large randomised cardiovascular outcomes trial. It suggests that the cardiovascular benefit observed in SELECT was not detectably different across the study's baseline frailty categories, while quality of life effects appeared greater at higher frailty levels. It also provides subgroup safety information that can guide future research questions. That is the contribution of this secondary analysis. The semaglutide evidence summary covers the wider topic^[1].

The remaining questions are substantial. The paper cannot establish effects in people excluded from SELECT, including people with diabetes, people without established cardiovascular disease, or people whose frailty is assessed differently. It cannot establish that the subgroup estimates apply to every community setting. It also cannot turn a trial average into a recommendation for one person. Those limits are central to the result, not a footnote to it.

This article is for educational purposes. It reports a secondary analysis of a randomised trial and does not recommend a dose, treatment change, or medication plan.