Predictors of Outcome in Patients with Spinal Cord Ischemia after Open Aortic Repair

Introduction Spinal cord ischemia is a potentially devastating complication of thoracic aortic surgery. However, predictors of outcome have not been well characterized. The study objective was to generate a prognostic score that accurately stratifies patient outcomes, aiding in future management and planning. Methods A retrospective review of 224 consecutive open thoracic aortic surgeries identified patients with spinal cord ischemia, defined as changes on intraoperative somatosensory evoked potentials (SSEP) and/or paraparesis/paraplegia postoperatively. The outcome of interest was poor outcome, defined as death or discharge with a lower extremity motor score ≤40, indicating impaired ambulation. Demographic and clinical characteristics were tested in univariate analyses and significant factors were incorporated in multivariate modeling to determine independent predictors of poor outcome. Results Seventy-five patients were identified with spinal cord ischemia, of which 43(57 %) had poor outcomes including 28(37 %) that died prior to discharge. Factors associated with poor outcome in univariate analysis included absent lumbar CSF drain ( p = 0.03), surgical repair that crossed the diaphragm ( p = 0.002), permanent intraoperative SSEP change ( p = 0.02), postoperative renal failure ( p = 0.004), paraplegia ( p = 0.001), and concomitant stroke ( p = 0.04). In multivariable analysis, surgical repair crossing the diaphragm (OR 4.8, CI 1.4–16.7, p = 0.02), paraplegia (OR 4.5, CI 1.4–14.0, p = 0.01), and renal failure (OR 6.1, CI 1.7–21.2, p = 0.005) were independently associated with poor outcome. Patients with transient intraoperative neurophysiologic changes were least likely to have poor outcome when compared to patients with no or permanent SSEP changes, and those not monitored ( p = 0.03). Conclusion Development of spinal cord ischemia with thoracic aortic repair often leads to death or disability. Characteristics known at the time of event can accurately predict the likelihood of poor outcome.