Surgeons have spent the past two years wary of the GLP-1 drugs. The medicines that slow the stomach to blunt appetite, semaglutide ↗ and tirzepatide ↗ among them, keep slowing it under anesthesia, which raised a fear that food left in the gut could be inhaled into the lungs mid-operation. The American Society of Anesthesiologists took the worry seriously enough to issue guidance on holding the drugs before surgery ↗. But aspiration is only the acute risk. The slower question is what these drugs do to healing, especially operations that depend on bone knitting itself back into one piece.
Cervical fusion is exactly that kind of case. It is the neck operation that locks two vertebrae into a single block of bone, and it only counts as a success if the bone actually fuses. If a GLP-1 drug interfered with that, the graft could fail to join, the hardware could loosen, or the patient could end up back in the operating room.
A team publishing in the North American Spine Society Journal ↗ pooled six retrospective studies covering 17,310 patients to ask the question head-on. All six were propensity-score matched, a statistical trick that pairs people on the drug against otherwise similar people who were not, to imitate the balance a real trial would build in. The authors ran a random-effects meta-analysis and rated the strength of the evidence with GRADE, a formal scoring system for how much to trust a pooled result.
The result was nothing, four times over
None of the four main outcomes moved. For pseudarthrosis, the technical name for bone that never fuses, the odds ratio was 0.98, meaning the drug group and the no-drug group had almost identical odds of a failed fusion. Trouble swallowing after the operation, called dysphagia, came in at 0.89. Hospital readmission was 0.67. Hardware failure was 0.83. An odds ratio near 1.0 means no difference between the groups, and everything here sat close to 1.0.
That reads like reassurance, and it may turn out to be. But the confidence intervals tell a harder story. The pseudarthrosis interval ran from 0.33 to 2.89, which means the true effect could be anything from cutting the odds of a failed fusion by two-thirds to nearly tripling them. Dysphagia ran from 0.38 to 2.05. Both intervals comfortably straddle the line where the drug does nothing, and both are wide enough to hide a real effect in either direction. Readmission and hardware failure were tighter but still crossed that line.
What the null actually means
The authors were careful about this. They graded the certainty of the evidence as low and called for prospective studies with clean definitions of who took what drug, at what dose, and for how long before surgery. That last point is the weak joint in every study of this kind. Six retrospective chart reviews can tell you that a patient had a GLP-1 prescription on file. They cannot tell you whether the patient took it the morning of surgery or quit it three months earlier, and that difference is the whole ballgame for a drug whose surgical risk depends on timing.
So the honest read is not "GLP-1 drugs are safe for cervical fusion." It is "six databases were too small and too blurry to find a signal if one exists." Absence of evidence, not evidence of absence. This piece sits alongside a growing pile of GLP-1 surgical-outcome studies, including one where the drugs tracked with fewer knee-implant infections but no change in mechanical failures ↗, and they all share the same limit. The patient counts are large, the exposure data is thin, and the answer keeps coming back as a shrug.
For a surgeon deciding whether to make a patient stop a GLP-1 receptor agonist ↗ before a neck fusion, that shrug is not permission. It is a reason to keep asking.