Utilization trends and outcomes of computer-assisted navigation in spine fusion in the United States.

BACKGROUND CONTEXT: Computer-assisted navigation (CAN) has emerged in spine surgery as an approach to improve patient outcomes. While there is substantial evidence demonstrating improved pedicle screw accuracy in CAN as compared to conventional spinal fusion (CONV), there is limited data regarding clinical outcomes and utilization trends in the United States. PURPOSE: The purpose of this study was to determine the utilization rates of CAN in the United States, identify patient and hospital trends associated with both techniques, and to compare their results. STUDY DESIGN: Retrospective review of national database. PATIENT SAMPLE: Nationwide Inpatient Sample (NIS), United States national database. OUTCOME MEASURES: CAN utilization, mortality, medical complications, neurologic complications, discharge destination, length of hospital stay, cost of hospital stay. METHODS: The NIS database was queried to identify patients undergoing spinal fusion with CAN or CONV. CAN and CONV utilization were tracked by year and anatomic location (cervical, thoracic, lumbar/lumbosacral). Patient demographics, hospital characteristics, index length of stay (LOS), and cost of stay (COS) were compared between the cohorts. After multivariate adjustment, index hospitalization clinical outcomes were compared. RESULTS: A total of 4,275,413 patients underwent spinal fusion surgery during the study period (2004 to 2014). CONV was performed in 98.4% (4,208,068) of cases and CAN was performed in 1.6% (67,345) of cases. The utilization rate of CAN increased from 0.04% in 2004 to 3.3% in 2014. Overall, CAN was performed most commonly in the lumbar/lumbosacral region (70.4%) compared to the cervical (20.4%) or thoracic (9.2%) regions. When normalized to region-specific rates of fusion with any technique, the proportional utilization of CAN was highest in the thoracic spine (2.7%), followed by the lumbar/lumbosacral (2.2%) and cervical (0.9%) regions. CAN utilization was positively correlated with patient factors including increasing age and number of medical comorbidities. Multivariate adjusted clinical outcomes demonstrated that compared to CONV, CAN was associated with a statistically significant decreased risk of mortality (0.28% vs 0.31%, OR=0.67, 95% CI: 0.46-0.97, p=.035) and increased risk of blood transfusions (9.1% vs 6.7%, OR=1.19, 95% CI: 1.02-1.39, p=.032). However, there was no difference in risk of neurologic complications. CAN patients had an increased average LOS (4.44 days vs. 3.97 days, p<.0001) and average COS ($34,669.49 vs $26,784.62, p<.0001) compared to CONV patients. CONCLUSIONS: CAN utilization increased in the United States from 2004-2014. Use of CAN was proportionately higher in the thoracic and lumbar/lumbosacral regions and in older patients with more comorbidities. Given the continued trend towards increased CAN utilization, large-scale studies are needed to determine the impact of this technology on long-term clinical outcomes.