Background: Hemoglobin A1c (HbA1c) is widely used for preoperative metabolic risk stratification in spine surgery; however, the optimal threshold for predicting postoperative complications remains controversial. Existing guidelines and prior studies propose cutoffs ranging from 6.1% to 8.0%, yet these recommendations often overlook heterogeneity in patient comorbidities and surgical complexity. Furthermore, the predictive value of HbA1c as a universal marker for short-term outcomes has not been fully validated across large real-world cohorts.

Objective: To evaluate the association between preoperative glycemic status and postoperative complications following cervical or thoracolumbar fusion, and to determine whether diagnostic performance differs according to comorbidity burden and fusion extent.

Methods: A retrospective multicenter cohort study was conducted using a clinical data warehouse including adult patients undergoing inpatient arthrodesis. Diabetes status was rigorously adjudicated using diagnostic codes, laboratory criteria (fasting glucose, post-prandial glucose, HbA1c), and documented medical history. Surgical approach and spinal level were standardized through natural-language–based parsing. Patients were categorized into diabetic, non-diabetic, and indeterminate groups. Outcomes included surgical site infection, pneumonia, urinary complications, cerebrospinal fluid leak, ICU admission, length of stay (continuous and ≥14 days), reoperation, readmission (30- and 90-day), and 90-day mortality. Results: Among 1,626 patients, postoperative complication rates were low (SSI 0.37%, pneumonia 0.18%, urinary retention 0.06%, 90-day mortality 0.49%). No significant differences were observed between diabetic and non-diabetic patients in complications, reoperation, readmission, ICU use, or overall in-hospital outcomes. HbA1c demonstrated limited discriminatory ability as a standalone predictor. Subgroup analyses revealed meaningful improvement in predictive performance only in higher-risk strata: late-stage chronic kidney disease (AUC≈0.73; optimal cutoff ≈5.65%) and multi-level fusion (AUC≈0.66; cutoff ≈6.25%). In contrast, single-level fusion and normotensive patients showed near-random discrimination (AUC≈0.52).

Conclusions: Preoperative HbA1c alone lacks sufficient predictive value for short-term postoperative complications following spine fusion. Its interpretation should be contextualized based on comorbid conditions and surgical complexity rather than rigid universal thresholds. Patients with late-stage CKD or planned multi-level fusion may benefit from stricter perioperative glycemic optimization. For otherwise low-risk patients undergoing single-level fusion, modestly elevated HbA1c should not automatically delay surgery. Multimodal risk assessment integrating metabolic and surgical factors is recommended.