The perioperative worry about these drugs is specific: delayed stomach emptying, aspiration risk, and possible wound healing problems. This desk has covered the fasting side of it in the pre-procedure hold study. A claims analysis took a different angle, asking what actually happened to nondiabetic patients on a GLP-1 who went through posterior cervical spine surgery [1].
On complications, nothing. No significant difference in surgical or medical complications at 90 days, and none at two years in mortality, revision surgery, surgical site infection or deep infection. For a drug class carrying an active perioperative safety question, a null across that range in 1,719 matched pairs is the useful result.
What did differ was hospital use. Emergency department visits within 90 days came in at an odds ratio of 0.71 (95% CI 0.58–0.87), and hospital admission at 0.54 (95% CI 0.41–0.70) — roughly half as many readmissions. Hospital use as an outcome, and what it does and does not capture, comes up again in the cohort past eighty.
That distinction has a particular edge in this cohort, because everyone in it was nondiabetic. A nondiabetic patient on a GLP-1 drug is, in most cases, paying for it as an elective prescription for weight — which selects for people who are insured, resourced and have a clinician they can reach. Emergency department use falls when there is somewhere else to call. Matching on demographics and comorbidities does not capture that, and it is a plausible explanation for a halving of admissions that produced no change in complications.
For a reader scheduling surgery, the practical content is the null. There is no signal here that taking one of these drugs makes posterior cervical spine surgery go worse, in a sample large enough to have found a moderate effect — which is worth knowing given how much perioperative caution has circulated. The holding question, which is separate and still unsettled, is in still food in there after twelve hours, and the broader pattern of these drugs quietly changing surgical statistics is in the trigger finger data.