Spine Journal - 2026-09-01 - Journal Article
Frequency and risk factors for symptomatic pulmonary embolism following spinal fusion surgery exceeding 6 hours: dynamic risk stratification using a 2-stage predictive model.
Heinig SA, Corr F, Stieger A, Luedi MM, Bozinov O, Martens B, Stienen MN
Topics
Key Takeaway
Symptomatic PE occurred in 10.8% of adults undergoing instrumented spinal fusion ≥6 hours, with fusion ≥8 segments (OR 3.81) and osseous bone-anchoring complications (OR 5.07) as the strongest independent predictors.
Summary Depth
Choose how much analysis to show on this article page.
Summary
This single-center retrospective cohort study asked whether a 2-stage clinical prediction model could identify PE risk factors in adults undergoing instrumented spinal fusion lasting ≥6 hours. Multivariable logistic regression identified fusion ≥8 segments (OR 3.81), L2-L5 lordosis-kyphosis inflection (OR 2.87), osseous bone-anchoring complications (OR 5.07), and fibrinogen administration per gram (OR 1.55) as independent predictors; the model achieved bootstrap-corrected AUC 0.787. PE patients had 13.3-fold higher discharge mortality and 84.2% ICU admission rate versus 56.2% in non-PE patients.
Key Limitation
Absence of external validation in an independent cohort means the model's discriminative performance (AUC 0.787) cannot be assumed to generalize beyond this single institution's case mix, prophylaxis protocols, and imaging thresholds.
Original Abstract
BACKGROUND CONTEXT
Major spinal fusion involving extensive surgery time carries substantial thromboembolic risk, yet the incidence and determinants of pulmonary embolism (PE) in complex procedures remain insufficiently defined.
PURPOSE
To estimate the frequency of symptomatic PE after instrumented spinal fusion lasting ≥6 hours and to develop, as a hypothesis-generating model derivation study, a 2-stage clinical prediction score integrating pre- and intraoperative risk factors.
STUDY DESIGN/SETTING
Retrospective single-center cohort study.
PATIENT SAMPLE
A total of 351 adults underwent instrumented fusion ≥6 hours for degenerative disease, deformity or trauma (2016-2023).
OUTCOME MEASURES
Primary outcome: PE within 90 days confirmed by CT-angiography.
SECONDARY OUTCOMES
PE severity according to American Heart Association (AHA) and European Society of Cardiology (ESC), perioperative risk factors, functional outcome (modified MacNab criteria), complications (Therapy-Disability-Neurology, TDN classification), and mortality at discharge, 90 days, and 12 months.
METHODS
Risk factors were evaluated through multivariable logistic regression with multiple imputation. A hierarchical prediction model was constructed (Stage 1: preoperative variables; Stage 2: intraoperative modifiers). Model performance was assessed using area under the receiver operating characteristic curve (AUC), calibration metrics, and bootstrap validation.
RESULTS
Symptomatic PE occurred in 10.8% (n=38), diagnosed at median postoperative day 2. Independent preoperative predictors were fusion ≥8 segments (odds ratio [OR] 3.81, 95% confidence interval [CI] 1.91-7.61) and a lumbar lordosis-thoracic kyphosis inflection at L2-L5 (OR: 2.87, 95% CI: 1.21-6.77). Intraoperative risk modifiers included osseous bone-anchoring complications (OR: 5.07, 95% CI: 1.59-16.20) and fibrinogen administration per gram (OR: 1.55, 95% CI: 1.09-2.21). Apparent discrimination in the derivation cohort was good (AUC 0.818; optimism-corrected 0.787 after bootstrap internal validation); external validation is lacking. PE patients had higher intensive care unit (ICU) admission (84.2% vs 56.2%, p=.001) and markedly increased mortality at discharge (OR: 13.3, p=.005), at 90 days (OR: 5.7, p=.029), and at 12 months (OR: 4.52, p=.039).
CONCLUSIONS
In adults undergoing extensive instrumented spinal fusion exceeding 6 hours, symptomatic PE was observed in 10.8% of cases, occurred early, and was associated with substantial morbidity and mortality despite institutional prophylaxis. Findings should be interpreted in the context of symptom-triggered imaging without routine postoperative VTE screening, likely underestimating the total PE burden. This hypothesis-generating, internally validated 2-stage model illustrates the feasibility of dynamic perioperative risk stratification in this specific subgroup but cannot be recommended for clinical implementation before prospective multicenter external validation.