Association of cannabis use disorder with clinical outcomes following lumbar spinal fusion: a propensity-matched cohort study.

Abstract

Purpose: Cannabis use disorder (CUD) has become increasingly prevalent, yet its association with postoperative outcomes after spine surgery remains poorly characterized. Given the physiologic effects of cannabinoids on pain modulation, inflammatory signaling, and bone metabolism, CUD may be associated with differences in perioperative recovery and spinal fusion outcomes. The purpose of this study was to evaluate the association between a coded preoperative diagnosis of CUD and clinical outcomes following lumbar spinal fusion.

Methods: We conducted a retrospective, propensity-matched cohort study using a multi-institutional electronic medical record database. Patients undergoing single- and multi-level lumbar spinal fusion during the ICD-10-CM era were identified and stratified by a documented preoperative diagnosis of CUD (ICD-10-CM F12) within a 1-year preoperative look-back window (primary analysis). Controls without CUD were matched 1:1 on 37 demographic and clinical covariates. Matched-pair methods were used for outcome comparison, and unadjusted P values were interpreted after Holm-Bonferroni sequential correction.

Results: After propensity matching in the primary (1-year look-back) analysis, the single-level cohort comprised 1266 patients with CUD and 1266 matched controls per arm (1251 per arm at the 2-year horizon), and the multi-level cohort comprised 1086 patients with CUD and 1086 matched controls per arm (1071 per arm at 30 days). At 30 and 90 days, patients with CUD did not demonstrate a consistent increase in coded medical complications after correction for multiple comparisons. Patients with CUD had significantly higher rates of documented opioid prescription orders and inpatient readmission in both cohorts. Emergency department (ED) utilization was significantly higher only in the single-level cohort, and behavioral/psychological service utilization was significantly higher only in the multi-level cohort at 90 days. At 2- and 5-year follow-up, patients with CUD demonstrated similar rates of coded pseudarthrosis, subsequent lumbar procedures, and reoperation components compared with matched controls. No long-term outcome remained significant after correction.

Conclusion: A coded preoperative diagnosis of CUD was not associated with a consistent increase in short-term perioperative morbidity following lumbar spinal fusion but was associated with higher postoperative healthcare utilization, including documented opioid prescription orders and inpatient readmission. These associations may reflect underlying psychosocial complexity and healthcare-access factors rather than a direct pharmacologic effect of cannabis. Prospective studies incorporating detailed exposure metrics are warranted.