Cost-effectiveness of intraoperative neurophysiological monitoring for spinal surgeries: Beginning steps

Abstract

highlights Cost-effectiveness modeling of multimodal intraoperative neurophysiologic monitoring (IONM) for spinal surgery relies on assumptions based on pooled results of uncontrolled observational studies, with uncertainty evaluated through probabilistic sensitivity analysis of Monte Carlo simulation results. Multimodal IONM reduces the relative risk of post-operative neurological complications by an estimated 49.4% (p < 0.001) at a cost of $63,387 (95%CI $61,939–$64,836) per neurological deficit averted. 2012 Published by Elsevier Ireland Ltd. on behalf of International Federation of Clinical Neurophysiology. The use of evoked potentials (EPs) and electromyograms (EMGs) in the operating room has increased in popularity over the past several decades, beginning with brainstem auditory EPs for acoustic neuroma resections to the most common usage today, neck and back surgeries. Intraoperative neurophysiologic monitoring (IONM) holds the promise of prevention of neurological complications by detecting evolving abnormalities and alerting the operating team to take actions to normalize the electrodiagnostic abnormality, including repositioning the patient, taking a different surgical approach, adjusting anesthesia, giving blood pressure support or waking the patient. By monitoring motor and somatosensory EPs with free run and triggered EMG, multimodal IONM may have greater diagnostic sensitivity than any single modality. However, the evidence for efficacy of IONM in spinal surgeries is scant, as seen in a recent systematic review (Fehlings et al., 2010), which concluded that IONM may be more helpful for complicated surgeries and recommended usage based on the discretion of the operating surgeon. Furthermore, multimodal IONM adds cost to already expensive procedures (average US cost of spinal fusion in 2007 was $24,600 (Elixhauser and Andrews, 2010)). The value of IONM in spinal procedures is in the avoidance of neurological deficits, based on IONM diagnostic characteristics, likelihood of preventing a post-operative injury following an IONM alert, and the baseline (a priori) rate of neurological complications from a given spinal operation. We created an economic decision model which incorporates the available evidence for these parameters. To account for uncertainty inherent in observational, nonrandomized data, the diagnostic characteristics of IONM and neurological complication rates for the eleven studies using multimodal IONM in spinal surgeries identified by Fehlings et al. (n = 2162) were pooled in a random effects meta-analysis to give estimates of average effect size and standard errors. This yielded a 5.0% baseline neurological complication rate for spinal surgeries (95% CI, 3.0– 7.0%), 94.3% sensitivity (95% CI, 92.3–96.3%) and 95.6% specificity (95% CI, 93.3–97.4%). The rate of prevention of post-operative deficits given an IONM alert has only been evaluated in a single study (Wiedemayer et al., 2002), which reported a 52.4% prevention rate (95%CI 37.3–67.3%) when actions were taken subsequent to IONM alerts compared to no actions taken. The per-operation cost of multimodal IONM for four-limb EMG, upper and lower extremity motor and somatosensory EPs and continuous IONM was calculated from 2009 Medicare reimbursement rates (national rating, global fee) for corresponding Current Procedure Terminology codes totaled $1535. The cost-effectiveness outcome was the cost per undesirable consequence avoided, calculated as the mean expense of IONM divided by the difference in post-operative neurological deficit rates in spinal surgeries using IONM compared with those not using IONM. In the IONM usage arm of the model, the likelihood of preventing a neurological deficit is the baseline risk of neurological deficit for the surgery X diagnostic sensitivity of IONM X probability of prevention of neurological deficit given an IONM alert, where the post-operative neurological deficit rate in the non-IONM arm is equal to the baseline risk of neurological complications for the surgery. To incorporate uncertainty, we conducted a probabilistic sensitivity analysis (PSA) with Monte Carlo simulation. Probability