Diagnostic Accuracy of Somatosensory Evoked Potentials in Evaluating New Neurological Deficits After Posterior Cervical Fusions

作者:Thirumala Parthasarathy D*; Melachuri Samyuktha R; Kaur Jaspreet; Ninaci David; Melachuri Manasa K; Habeych Miguel E; Crammond Donald J; Balzer Jeffrey R
来源:Spine, 2017, 42(7): 490-496.
DOI:10.1097/BRS.0000000000001882

摘要

Study Design. This study examined the diagnostic accuracy of significant changes of somatosensory evoked potentials (SSEPs) to evaluate and predict postoperative neurological deficits after posterior cervical fusions (PCF). Eight hundred forty six eligible patients underwent PCF at the University of Pittsburgh Medical Center (UPMC), from 2010 to 2012. Objective. To assess the specificity and sensitivity of intraoperative monitoring in predicting postoperative neurological deficits during PCF. Summary of Background Data. We calculated the predictive value, including sensitivity and specificity, of changes in SSEPs to identify neurological deficits postoperatively. We used a receiver operating characteristic (ROC) curve with SSEP categories as cutoff values to further evaluate the diagnostic accuracy of change in SSEPs and postoperative neurological deficit. Methods. All patients had preposition baselines and continuous SSEP monitoring throughout the surgery. Statistical analysis was completed using SPSS version 22 (IBM Corp., Armonk, NY). Results. Age and sex did not influence outcomes. Obesity affected patient outcome. The SSEP categories of significant changes and loss of responses resulted in a sensitivity/specificity of 0.30/0.96 and 0.16/0.98, respectively. The receiver operating characteristic curve has an area under the curve for significant change in/loss of SSEPs of 0.62/0.65 with a 95% confidence interval of 0.525 to 0.714/0.509 to 0.797. Conclusion. Significant SSEP changes during PCF are a very specific but poorly sensitive indicator of postoperative neurological deficits. The odds ratio for significant changes in SSEPs and loss of waveforms was 9.80 and 11.82, respectively, with a 95% confidence interval of 4.695 to 20.46 and 4.45 to 31.41, respectively.

  • 出版日期2017-4-1