A network meta-analysis published in Endocrinology, Diabetes & Metabolism pooled 43 studies covering 932,380 patients and compared bariatric surgery against GLP-1 receptor agonists on hard cardiovascular outcomes. Surgery was associated with reductions in major adverse cardiovascular events, heart failure and myocardial infarction. The drugs were associated with a reduction in major adverse cardiovascular events but not, at conventional significance, in the other two. The authors are explicit that heterogeneity across the included studies was high and that the comparison is largely indirect.
The numbers
Against controls, bariatric surgery returned a hazard ratio of 0.66 for major adverse cardiovascular events (95% CI 0.60 to 0.74), 0.45 for heart failure (0.38 to 0.53) and 0.53 for myocardial infarction (0.45 to 0.63). GLP-1 receptor agonists returned a hazard ratio of 0.85 for major adverse cardiovascular events (0.77 to 0.94), with effects on heart failure and myocardial infarction whose intervals crossed 1.
Read as plain arithmetic, a hazard ratio of 0.66 means roughly a third fewer events over the follow-up period, and 0.85 means roughly 15% fewer. The surgical intervals are also narrower, which reflects both larger accumulated cohorts and a longer observational history. The authors searched PubMed, Embase, the Cochrane Library and Google Scholar for studies published between January 2000 and September 2024, assessed heterogeneity with I², and tested consistency by node-splitting. I² ran from 66% to 93%.
Why the comparison is harder than it looks
An I² of 93% means the included studies disagree with each other far more than chance would explain, and pooling them yields a number that no single study supports. More importantly, almost none of this evidence is randomised head to head. People who have bariatric surgery are selected, for fitness for an operation, for the ability to complete a pre-surgical programme, for insurance approval. People who take a GLP-1 drug are selected differently. An observational comparison inherits every one of those differences, and no propensity adjustment fully removes them.
The search cut-off matters too. It ends in September 2024, which excludes the more recent cardiovascular readouts for this drug class and the tirzepatide cardiovascular data that has become available since. The authors say plainly that future studies on newer agents are required and that their result should inform individualised treatment selection rather than settle it. For someone weighing the two options, the honest summary is that surgery has the longer and more consistent cardiovascular record, and that this analysis cannot tell you what would happen to you specifically.
What it cannot tell you is whether surgery would beat a modern GLP-1 drug in a fair fight. The agents in most of the pooled data are earlier ones studied at earlier doses, and the analysis is dominated by observational cohorts rather than trials. The non-significant heart failure and myocardial infarction results in the drug arm are not evidence of no effect; they are wide intervals around a plausible effect, which is a different thing. No randomised trial has assigned people to surgery or to a GLP-1 drug and followed them for cardiovascular events, and until one does, part of the gap reported here remains a gap in study design rather than a gap in biology.
Sources: Shirmohammadi E, Ghasemloo N, Ebrahimi N, Mohammadzadeh N, Ehtemami A, Babaei M, Fardoost S. Comparative Cardiovascular Outcomes of GLP-1 Receptor Agonists Versus Bariatric Surgery: A Systematic Review and Network Meta-Analysis. Endocrinology, Diabetes & Metabolism, September 2026;9(5):e70311. doi:10.1002/edm2.70311.
