Both ways, depending on which operation. Take anterior cervical fusion first. One large database put failed fusion lower among GLP-1 users, at OR 0.52 (95% CI 0.40 to 0.67)[1]. A semaglutide-specific posterior fusion cohort found the opposite, at OR 4.79 (95% CI 3.11 to 7.37). Pooling everything gives an interval from 0.35 to 4.70, which is not a finding. The question is adjacent to whether to pause them around an operation and separate from it.
What was gathered
Twelve retrospective cohorts, all published in 2025 or 2026. All cover GLP-1 exposure and outcomes after cervical fusion. Five databases were searched to June 2026. The primary outcome was pseudarthrosis, which is the fusion failing to knit.
Take anterior cervical discectomy and fusion first. One large claims network put the odds of failed fusion lower among GLP-1 users, at OR 0.52, 95% CI 0.40 to 0.67. An independent estimate from a different network was neutral, at OR 0.86, 95% CI 0.56 to 1.32. The anterior subgroup as a whole came to 0.65, 95% CI 0.40 to 1.06, with heterogeneity of 74.5%.
A semaglutide-specific posterior cervical fusion cohort went the other way entirely. Failed fusion at two years came in at OR 4.79, 95% CI 3.11 to 7.37. A dysphagia signal ran alongside it.
Twelve studies, fewer than twelve samples
This is the part worth carrying away. Several of the included cohorts query the same commercial claims networks. The same patient can appear in more than one apparently independent study. A reader counting twelve papers may be looking at far fewer distinct groups of people.
The authors prespecified how to handle double counting. They used non-overlapping estimates for the primary synthesis. That is careful and it is rare, because most reviews in this literature simply count the papers. Large shared claims networks make the problem worse each year, and it affects every study built on the same data.
Why the direction might genuinely differ
Anterior and posterior cervical fusion are different operations with different biology and different failure modes. A drug may affect one and not the other, or affect them oppositely. The authors take that position explicitly.
The posterior finding may also reflect who gets a posterior fusion. That is generally more complex disease, more levels, more comorbidity. One cohort in one procedure is thin ground for a fourfold estimate. The review classes all of this as hypothesis-generating rather than actionable. The distance between a pooled figure and a patient is rarely this visible.
What the dose evidence adds
One study sorted patients by prescription strength rather than by exposure [2]. It set high-dose GLP-1 use against standard-dose use before neck fusion surgery. After matching, the two dose groups held together on readmission, emergency visits, complications and failed fusion.
Both dose groups had less failed fusion than matched non-users. The higher dose added nothing on top. That is an absence of an obvious dose penalty, not a demonstration that no difference exists. This site reads it in twice the dose, the same recovery. The wider surgical picture is in what a GLP-1 does to surgery outcomes.
What it means before an operation
That the question is live and unsettled, and that your surgeon should be told you are taking one of these drugs. Whether to pause it around surgery is a separate question, and the evidence there is mostly about anesthesia rather than about bone.
No seller on this roster asks whether you have surgery scheduled. None would know what to do with the answer. the census of what goes unasked counts how much a prescription arrives without. Anybody facing a fusion should raise it with the surgeon who will do it.