Motor evoked potential recovery with surgeon interventions and neurologic outcomes: A meta-analysis and structural causal model for spine deformity surgeries.

OBJECTIVE: To improve estimates of motor evoked potential (MEP) performance during spine deformity surgeries by accounting for potential confounders. METHODS: A meta-analysis of MEPs for spine deformity surgeries determined the probability of a MEP deterioration which recovered by the end of surgery, P(RSC), and the conditional probability of no new post-operative deficit given an RSC, P(NND|RSC), stratified by category of intraoperative adverse event associated with the MEP deterioration. A structural causal model (SCM) and propensity score matching accounted for intraoperative adverse events and patient diagnosis as potential confounders. RESULTS: MEPs changes (either reversible, RSC or irreversible, IRREV) were reported for 295 of 5055 cases (6%) in 21 studies. The probability of no new motor deficit, P(NND), plotted against the probability of a RSC, P(RSC), for studies in the meta-analysis was highly significant (r = 0.71, p < 0.001). P(RSC) was 0.76 for an alert associated with correction, less for osteotomies (0.48, p = 0.0008), and tended to be higher for hypotension (0.92, p = 0.06). P(NND|RSC) was 0.94 for correction, less for positioning (0.82), and osteotomies (0.86), and greater for hypotension (1.0). In the SCM, a RSC after an alert was a highly significant and independent predictor of no new motor deficits (odds 25.2, p < 0.001). CONCLUSION: There are significant differences in P(RSC) for hypotension and osteotomies, and in P(NND) for osteotomies and instrumentation, compared to correction. P(RSC) is a significant and independent predictor of outcomes. SIGNIFICANCE: When MEPs are used for spine deformity surgeries, accounting for adverse events associated with an alert and patient diagnosis as potential confounders is expected to improve RSC prediction of post-operative outcomes and estimates of RSC efficacy in improving outcomes.