| Audience | Patients using GLP-1 therapies, spine clinicians, metabolic-care clinicians, and perioperative teams |
| Primary Topic | GLP-1 | Cervical Fusion | Pseudarthrosis | Surgical Outcomes |
| Source | Read the peer-reviewed PubMed record |
GLP-1 Drugs and Cervical Fusion: What Conflicting Real-World Evidence Shows
A systematic review of 12 retrospective cohorts found no reliable class-wide relationship between GLP-1 receptor agonist exposure and cervical fusion outcomes. Results differed by procedure and dataset, with substantial heterogeneity.
| Study Type | Systematic review, meta-analysis, and evidence map |
| Evidence Base | 12 retrospective cohorts published in 2025 and 2026 |
| Search Through | June 1, 2026 |
| Population | Patients undergoing cervical fusion or decompression-and-fusion |
| ACDF Subgroup | OR 0.65; 95% CI 0.40-1.06; I2 74.5% |
| Posterior Signal | Semaglutide-specific PCF cohort OR 4.79; 95% CI 3.11-7.37 for two-year pseudarthrosis |
| Overall Model | OR 1.28; 95% CI 0.35-4.70; I2 97.3% |
| Main Interpretation | Procedure-specific, hypothesis-generating associations rather than a class-wide effect |
| Journal | European Spine Journal |
| Published | August 24, 2026 |
| PMID / DOI | 42635669 / 10.1007/s00586-026-10293-9 |
The authors conducted a PRISMA-guided systematic review, meta-analysis, and evidence map of GLP-1 receptor agonist or incretin-based therapy exposure in cervical fusion and cervical decompression-and-fusion cohorts.
Twelve retrospective cohorts published in 2025 and 2026 were included. Searches covered PubMed/MEDLINE, Embase, Web of Science Core Collection, CENTRAL, and Scopus from inception through June 1, 2026.
A combined TriNetX estimate for anterior cervical discectomy and fusion was associated with lower odds of pseudarthrosis among exposed patients: odds ratio 0.52, with a 95% confidence interval from 0.40 to 0.67.
An independent MarketScan ACDF estimate was neutral: odds ratio 0.86, with a 95% confidence interval from 0.56 to 1.32. The ACDF subgroup estimate also crossed the null at 0.65, with substantial heterogeneity of 74.5 percent.
A semaglutide-specific posterior cervical fusion cohort showed higher two-year pseudarthrosis, with an odds ratio of 4.79 and a 95% confidence interval from 3.11 to 7.37. The abstract also reports a dysphagia signal in this setting.
This is a procedure-specific observational signal. It does not show that semaglutide caused either outcome, and it should not be generalized to anterior cervical surgery or to every GLP-1 receptor agonist.
The overall model produced an odds ratio of 1.28 with a very wide 95% confidence interval from 0.35 to 4.70. Heterogeneity was 97.3 percent, indicating that the included estimates differed substantially.
The authors therefore favored procedure-specific interpretation over a class-wide conclusion. A pooled number this unstable should not be converted into a simple claim that GLP-1 therapy improves or worsens fusion outcomes.
GLP-1 receptor agonists are increasingly used for type 2 diabetes and obesity, so more surgical patients will arrive taking these therapies.
Cervical fusion outcomes depend on procedure, patient health, surgical factors, and follow-up. Observational database comparisons can identify signals but cannot isolate medication effects as reliably as randomized evidence.
The conflicting estimates make coordination more important than a one-size-fits-all conclusion. Medication decisions should account for established metabolic indications and the specific perioperative plan.
I read this paper as a warning against class-wide shortcuts. One database suggested lower pseudarthrosis after ACDF, another was neutral, and a posterior-fusion cohort showed a strong adverse association.
Those differences may reflect procedure, population, exposure definition, residual confounding, or other study-level factors. The abstract does not let us determine which explanation is correct.
Clinically, this evidence supports careful documentation and multidisciplinary planning. It does not justify prescribing a GLP-1 drug to improve fusion, and it does not justify unilateral discontinuation of an indicated therapy.
How to Read Conflicting Surgical Associations
A meta-analysis can summarize available estimates, but pooling does not erase differences among procedures, databases, patients, or exposure definitions.
Here, inconsistency is the central result, not a statistical inconvenience.
Four Checks for a Careful Reading
Start with study design
All 12 included cohorts were retrospective. Associations may remain confounded even after statistical adjustment.
Separate procedures
ACDF and posterior cervical fusion are not interchangeable. The review’s estimates pointed in different directions across procedure-specific analyses.
Read confidence intervals and heterogeneity
The overall interval was wide and heterogeneity reached 97.3 percent, weakening any class-wide interpretation.
Do not convert association into a medication rule
The paper did not test whether perioperative continuation or discontinuation changes outcomes.
Eight Ways to Interpret Conflicting GLP-1 Fusion Evidence
Procedure, design, patient care, and research implications from an unusually heterogeneous evidence base
A Signal Is Not a Personal Medication Instruction
For a patient preparing for cervical fusion, the most important finding is that the review did not identify a dependable class-wide effect. Results differed across anterior and posterior procedures, and the overall estimate was highly heterogeneous. That makes the evidence unsuitable for a simple message that GLP-1 therapy either protects fusion or causes failure.
The practical implication is to bring the exact drug, indication, dose schedule, and planned operation into a coordinated perioperative conversation. The boundary is equally important: the review did not test whether continuing or stopping treatment changes outcomes, so it cannot determine an individual medication plan.
Procedure Type May Matter More Than a Class Average
The review separates anterior cervical discectomy and fusion from posterior cervical fusion because their estimates did not align. ACDF evidence ranged from an apparent lower-risk association to a neutral association, while one semaglutide-specific posterior cohort reported higher pseudarthrosis and dysphagia signals. A single pooled class statement would conceal those differences.
For surgical counseling, procedure-specific risk assessment is therefore more defensible than extrapolation from the overall model. The uncertainty is that retrospective datasets may define exposure, outcomes, and follow-up differently. The abstract does not identify which study-level differences explain the conflicting results.
Established Benefits Still Belong in the Decision
Many patients receive GLP-1 receptor agonists for type 2 diabetes or obesity, conditions that themselves influence perioperative planning. An observational association with a surgical outcome does not erase the established reason a medication was prescribed. Any decision must consider glycemic control, weight management, nutrition, and the broader medical plan.
The review does not compare a coordinated continuation strategy with a coordinated discontinuation strategy, and it does not report the metabolic consequences of changing therapy. Its boundary is therefore clear: it can raise a surgical evidence question, but it cannot determine the net benefit or harm of altering an indicated treatment.
Overlapping Databases Complicate Apparent Sample Size
The authors recognized that several included studies used overlapping databases and prespecified rules to avoid double counting in the primary pseudarthrosis synthesis. That methodological step matters because repeated use of the same underlying patients can make an evidence base look larger and more consistent than it is.
Even with that protection, the estimates remained discordant and the overall heterogeneity reached 97.3 percent. The review’s evidence map is useful for showing where data exist, but the abstract does not provide every cohort’s bias assessment or adjustment set. Residual overlap and residual confounding therefore remain important interpretive boundaries.
The Null-Crossing ACDF Estimate Deserves Emphasis
One combined TriNetX ACDF estimate favored GLP-1 exposure, but an independent MarketScan estimate was neutral. When ACDF estimates were summarized, the odds ratio was 0.65 and the 95% confidence interval extended from 0.40 to 1.06, crossing the null. Heterogeneity within that subgroup was 74.5 percent.
This means the ACDF evidence is collectively inconclusive, not proof of improved fusion. The practical implication is to resist highlighting only the most favorable database. The boundary is that the abstract does not provide enough detail to adjudicate whether differences arose from populations, coding, exposure definitions, or analytical choices.
The Posterior-Fusion Signal Needs Verification, Not Dismissal
A semaglutide-specific posterior cervical fusion cohort reported substantially higher two-year pseudarthrosis, and the abstract also identifies a dysphagia signal. These findings are clinically important enough to investigate because they concern recovery and complications after surgery. They should not be diluted by a broad class average.
At the same time, a single retrospective cohort cannot establish causation or determine whether semaglutide itself produced the association. The abstract does not present a complete adverse-event profile or explain all confounder controls. The responsible response is procedure-specific validation and careful counseling, not a universal warning or automatic discontinuation rule.
Guidance Should Not Outrun the Evidence
Health systems may want standardized perioperative instructions as GLP-1 use grows, but this review does not supply a fusion-specific continuation or discontinuation protocol. Its contribution is an evidence map showing conflicting observational signals across cervical procedures. That can identify questions for pathways and quality review.
The policy boundary is that inconsistent associations cannot establish a universal medication rule. Any institutional guidance should distinguish anesthesia considerations, metabolic management, nutritional assessment, and fusion outcomes rather than treating them as one question. It should also remain updateable as procedure-specific prospective evidence becomes available.
Prospective Procedure-Specific Studies Are the Next Step
The strongest research need is not another undifferentiated class-wide database average. Future studies should prespecify anterior and posterior procedures, define medication exposure and timing clearly, measure pseudarthrosis consistently, and capture relevant surgical and medical outcomes. Independent datasets would help test whether the current signals reproduce.
Prospective designs could better characterize confounding factors and medication changes around surgery, although randomized discontinuation trials may be difficult or inappropriate in some settings. The present review is valuable for hypothesis generation, but its retrospective evidence and extreme heterogeneity limit causal inference. Better design, rather than stronger language, is what can reduce that uncertainty.
Join the Conversation
Have a question about how this applies to your situation? Ask Dr. Caplan
Want to discuss this topic with other patients and caregivers? Join the forum discussion
Frequently Asked Questions
What kind of study was this?
A PRISMA-guided systematic review, meta-analysis, and evidence map of 12 retrospective human cohorts.
Which operations were included?
The review included cervical fusion and cervical decompression-and-fusion cohorts, including ACDF and posterior cervical fusion analyses.
Did GLP-1 drugs reduce pseudarthrosis after ACDF?
The evidence was inconclusive. One combined TriNetX estimate favored exposure, an independent MarketScan estimate was neutral, and the ACDF subgroup confidence interval crossed the null.
What did the posterior cervical fusion cohort find?
A semaglutide-specific retrospective cohort reported higher two-year pseudarthrosis and a dysphagia signal. This association does not prove causation.
What did the overall meta-analysis find?
The overall odds ratio was 1.28 with a wide 95% confidence interval from 0.35 to 4.70 and heterogeneity of 97.3 percent.
Does this prove GLP-1 drugs harm spinal fusion?
No. Results were conflicting, all included cohorts were retrospective, and the review supports procedure-specific rather than class-wide interpretation.
Does this prove GLP-1 drugs protect spinal fusion?
No. The favorable ACDF estimate was not consistent across datasets, and the pooled ACDF interval crossed the null.
Should patients stop GLP-1 treatment before cervical fusion?
This review did not test continuation versus discontinuation and cannot determine an individual medication plan. Patients should coordinate with their prescribing, surgical, and anesthesia teams.
Were adverse events fully characterized?
The abstract mentions a dysphagia signal in a posterior-fusion cohort but does not provide a complete adverse-event profile.
What research is needed next?
Prospective, procedure-specific studies with clear exposure timing, consistent outcome definitions, and independent validation are needed.
