Minimizing the risk of neurological deficits in spinal deformity surgery is a top priority for every spine surgeon in the operating room. But despite the advancements in technology and intraoperative neuromonitoring, we are still seeing a prevalence of intraoperative neurological events as high as 23%. There is a plethora of studies in the literature that discuss a variety of different risk factors to try an anticipate these problems and potentially adjust the surgical plan accordingly, such as the Cobb measurements, deformity angular ratio (DAR), 3DCT DAR, and the spinal cord shape classification system (SCSCS) among others. However, no preoperative risk stratification tool of IONM loss exists to help delineate important preoperative factors that should be considered in the decision making process prior to deformity correction. In our newly published study in The Journal of Bone and Joint Surgery, Inc., we leveraged a machine learning approach to develop the first preoperative prediction tool for spinal cord-level IONM data loss during adult and pediatric spinal deformity surgery.
From a total of 1,106 patients, we were able to delineate 8 critical non-modifiable preoperative factors with the following scores: conus level below L2 (2 points), type 3 spinal cord (2 points), cervical UIV (2 points), preop thoracic cobb angle ≥75 degrees (2 points), sagittal-DAR ≥15 (2 points), preop lower extremity deficit (2 points), preop TK ≥ 80 degrees (1 point), and the total-DAR (1 point). Scores between 0 and 2, 3 and 6, and 7 and 12 resulted in low, moderate, and high risk of cord-level IONM data loss, respectively. The performance of our model had an AUC of 0.921 and 0.898 on our training and testing sets with over 90% accuracy! By employing predictive modeling for proper risk stratification, clinicians can engage in more informed discussions with patients and potentially adjust operative plans to optimize neurologic safety, particularly those with elevated risk factors for cord-level IONM data loss and resultant neurologic deficit.
Link to the full study:
https://lnkd.in/g3vz2_U3
Nathan J. Lee, MD Varun Arvind Ted Shi Alexandra Dionne Chidebelum Nnake Anastasia Ferraro Matthew Cooney Erik Lewerenz Justin L. Reyes Steven Roth Justin K. Scheer Tom Zervos Earl Thuet, CNIM Joseph Lombardi Zeeshan Sardar Ron Lehman, MD Benjamin Roye MD-MPH Michael Vitale Fthi M. Hassan, MPH