- Email: [email protected]
Influence of Intraoperative Neuromonitoring on the Outcomes of Surgeries for Pediatric Scoliosis in the United States
Spine Deformity 7 (2019) 27e32 www.spine-deformity.org
Case Series
Influence of Intraoperative Neuromonitoring on the Outcomes of Surgeries for Pediatric Scoliosis in the United States Jaiben George, MBBSa,*, Soumabha Das, MBBSb, Anthony C. Egger, MDa, Reid C. Chambers, DOa, Thomas E. Kuivila, MDa, Ryan C. Goodwin, MDa b
a Department of Orthopaedic Surgery, Cleveland Clinic, 9500 Euclid Avenue, Cleveland, OH 44915, USA Department of Physical Medicine and Rehabilitation, Westchester Medical Center, 100 Woods Rd, Valhalla, NY 10595, USA Received 12 December 2017; revised 7 May 2018; accepted 18 May 2018
Abstract Background: Intraoperative neuromonitoring (IONM) is used to detect impending neurologic damage during complex spinal surgeries. Although IONM is increasingly used during pediatric scoliosis surgeries in the United States, the effect of IONM on the outcomes of such surgeries at a national level is unclear. Methods: Using National Inpatient Sample (NIS) from 2009 to 2012, 32,305 spinal fusions performed in children 18 years old or younger of age with scoliosis were identified using ICD-9 procedure and diagnosis codes. IONM was identified using the ICD-9 procedure code 00.94. The effects of IONM use on length of stay (LOS), discharge disposition, hospital charges, and in-hospital complications were assessed using multivariate regression analysis adjusting for patient and hospital characteristics. Results: IONM was used in 5,706 (18%) of the surgeries. IONM was associated with increased home discharge (adjusted odds ratio [AOR] 5 1.25 [95% confidence interval 1.10-1.40], p 5 .001). There was no difference in LOS (p 5 .096) and hospital charges (p 5 .750). Neurologic complications were noted in 52 (0.9%) surgeries using IONM and 368 (1.4%) surgeries without IONM (p 5 .005). Although IONM use trended toward lower risk of neurologic complications in multivariate analysis, it failed to achieve statistical significance (AOR 5 0.77 [0.57-1.04], p 5 .084). Conclusions: Reported use of IONM in this database was significantly less compared with other databases, suggesting that IONM might be underreported in the NIS database. Nevertheless, in this database, IONM was significantly associated with increased home discharge. Hospital charges and LOS were not affected by IONM. There was a trend toward lower risk of neurologic complications with IONM use, though this finding was not statistically significant. Ó 2018 Scoliosis Research Society. All rights reserved. Keywords: Scoliosis; Spinal deformity; Neuromonitoring; Spinal fusion; Complications
Introduction Surgical correction of scoliosis and other pediatric spinal deformities involves substantial manipulation of the vertebral column and the associated neural structures [1-3].
Author disclosures: JG (none), SD (none), ACE (none), RCC (none), TEK (none), RCG (none). The authors declare no conflicts of interest. No funding was obtained for this study. The study was exempt from IRB approval as it used a publicly available database. *Corresponding author. Department of Orthopaedic Surgery, Cleveland Clinic, 9500 Euclid Avenue, A40, Cleveland, OH 44915, USA. Tel.: (216) 972-1965; fax: (216) 445-3585. E-mail address: [email protected] (J. George).
Although a number of complications can occur after such complex spinal deformity surgeries, neurologic complications are among the most feared [1,3]. Neurologic injury can result from excessive stretching of spinal cord or the nerve roots, hypoperfusion, direct injury during implant placement, or from patient positioning [2] Historically, the rates of neurologic complications were reported to be 3% to 4% [4,5]. However, more recent studies report an incidence between 0.5% and 2% [1,3,6,7]. This decline in complication rates over the last few decades is, at least in part, due to the availability of intraoperative neuromonitoring (IONM) [2,8]. IONM can detect impending neurologic injury intraoperatively, allowing the surgeon to take action to prevent long-lasting neurologic deficits [2]. IONM is becoming
2212-134X/$ - see front matter Ó 2018 Scoliosis Research Society. All rights reserved. https://doi.org/10.1016/j.jspd.2018.05.013
28
J. George et al. / Spine Deformity 7 (2019) 27e32
more common in pediatric spinal deformity surgeries and is available in most of the United States [2,3]. Despite increased use, very few studies have compared the outcomes of operations with and without IONM [8-10]. Although IONM is proven to be useful in detecting early neurologic insult, there is little evidence to support that IONM decreases the incidence of neurologic complications [8,11]. Moreover, given the rising health care costs, there are also concerns about the utility of IONM in less complex spinal surgeries [9,10,12]. As the frequency of IONM increases in pediatric spinal deformity surgeries performed in the United States, it is important to study the effects of IONM on the outcomes of such surgeries at a national level. Therefore, the present study used the Nationwide Inpatient Sample (NIS), a nationally representative database, to compare 1) the length of stay (LOS), discharge disposition, and hospital charges and 2) in-hospital complications (including neurologic complications) between pediatric fusion surgeries for scoliosis with and without IONM.
previous codes; including kyphosis, kyphoscoliosis, etc). The surgical approach was divided into two groups based on ICD9 procedure codes: posterior-only approach (only 81.03, 81.05, 81.07, and 81.08) and combined anterior/posterior approach (contains 81.02, 81.04, and 81.06). The number of levels fused were grouped as follows: three or fewer vertebrae (81.62), four to eight vertebrae (81.63), and more than eight vertebrae (81.64). Patient demographics, comorbidities provided by AHRQ, and hospital characteristics were recorded [14]. The baseline characteristics of patients in the monitoring and nonmonitoring groups are provided in Tables 1 and 2.
Methods
Variable
Monitoring group (n 5 5,706)
Nonmonitoring group (n 5 26,599)
Age, mean SD (in years) Sex Male Female Race White Black Hispanic Other/missing Insurance Medicare/Medicaid Private Other/missing Diagnoses Adolescent idiopathic scoliosis Congenital scoliosis Neuromuscular scoliosis Other curvature deformities Levels fused !3 4e8 O8 Unspecified Approach Posterior Anterior posterior Number of comorbidities 0 1 or 2 >3 Year 2009 2010 2011 2012
13.72.6
13.82.6
32.5 62.8
31.7 67.9
51.5 11.8 7.2 7.1
53.1 12.8 1.5 8.4
26 68.4 5.6
31.9 60.2 7.9
68.9
71.2
7 7.4 16.8
6.1 5.1 17.6
5.7 20.9 73.1 0.3
7.4 20 70.8 1.8
92.4 7.6
90.5 9.5
61.5 34.1 4.4
58.3 36.1 5.6
18.9 16.9 29.7 34.5
22.8 28.5 24.2 24.5
Outcomes The effects of IONM on LOS, discharge disposition, hospital charges, and in-hospital complications were Table 1 Baseline patient and surgical characteristics of spinal fusion surgeries performed for pediatric scoliosis in the United States.
Data source This study used the NIS database from 2009 to 2012. The NIS database is a part of the Healthcare Cost and Utilization Project managed by the Agency for Healthcare Research and Quality (AHRQ) and is the largest all-payer database in the United States containing information about inpatient admissions [13]. The NIS is a stratified sample of 20% of all the hospitalizations in the United States and contains an array of information including demographics, International Classification of Disease, ninth edition (ICD-9), procedure and diagnosis codes, insurance information, admission data, hospital data, LOS, discharge disposition, and total charges. The NIS database provides sampling weights for each discharge that can be used to obtain national-level estimates of these discharges. The NIS data are publically available and deidentified; therefore, the study was deemed exempt from approval by our institutional review board. Study population All discharges in children 18 years old or younger admitted for spinal fusion were identified using the primary clinical classification software code of 158. These fusion procedures were screened to identify spinal fusions performed in children with scoliosis (isolated or associated with other conditions/ deformities) using the ICD-9 diagnosis code 737.x. The use of IONM was identified using the ICD-9 procedure code 00.94. A total of 32,305 (unweighted 5 6,642) spinal fusion surgeries performed for scoliosis were identified, and IONM was used in 5,706 (18%) of the operations (monitoring group). Patients were grouped into the following diagnosis groups: adolescent idiopathic scoliosis (737.30), congenital scoliosis (737.31, 737.32, 754.2), neuromuscular scoliosis (737.40-3), and other curvature deformities (those without any of the
p value
.149 .634
.051
!.001
.01
!.001
.034
.065
!.001
Values are percentages unless otherwise noted.
J. George et al. / Spine Deformity 7 (2019) 27e32 Table 2 Hospital characteristics of spinal fusion surgeries performed for pediatric scoliosis in the United States. Variable
Location Rural Urban nonteaching Urban teaching Missing Region Northeast Midwest South West Size Small Medium Large Missing Control Government Private, nonprofit Private, for profit Missing
Monitoring group, % (n 5 5,706)
Nonmonitoring group, % (n 5 26,599)
1.6 4.6 93.3 0.5
1.9 8.4 88.7 1.0
8.4 32.8 34.3 24.5
13.9 21.6 40.4 24.1
19.1 25.7 54.7 0.5
15.6 27.9 55.6 1.0
8.9 88.8 1.8 0.5
8.0 83.4 7.7 1.0
29
Analysis All analyses were performed using sampling weights provided by NIS to obtain the national estimates. As NIS changed the sampling method in 2012, we used the updated trend weights for all the years used in the study. Descriptive statistics were used to compare the baseline characteristics of monitoring and nonmonitoring groups. Linear regression analysis was used for continuous outcome variables like LOS and hospital charges, whereas logistic regression analysis was used for binary outcome variables like discharge disposition and complications. Multivariate regression analyses were performed to adjust for potential confounding factors. Only the baseline variables that were different between the groups at a p value of .10 were included in the multivariate analysis to avoid overfitting of the model. As IONM might be underreported in the NIS database due to coding inaccuracies, a sensitivity analysis was performed to include only the hospitals that performed at least one surgery with a code for IONM [10]. The results of multivariate logistic regression analyses were reported using adjusted odds ratios (AORs), whereas that of multivariate linear regression analyses were reported using adjusted unstandardized coefficients. The frequency of missing values among different variables, if applicable, are given in Tables 1 and 2. Missing values were excluded from the multivariate regression analyses. Confidence intervals (CIs) of 95% were calculated and the level of significance for all analysis was set at p!.05. Statistical analysis was performed using R software (version 3.1.3, Vienna, Austria) [16].
p value
!.001
!.001
.011
!.001
assessed. The LOS was calculated as the number of days from admission to discharge, and the discharge disposition was categorized as home or a care facility. Hospital charges provided in the NIS were converted to 2012 US dollars using the Consumer Price Index inflation calculator provided by Bureau of Labor Statistics to adjust for inflation [15]. Neurologic complications were defined using the following ICD-9 diagnosis codes: 9970.0-2, 997.09, and 907.0-9. Other in-hospital complications assessed in this study included mortality, respiratory complications (997.31-2,997.39), cardiovascular complications (997.1, 997.7x, 997.8x), wound complications (998.1x, 998.3x, 99883), and surgical site infections (998.51, 998.59).
Results Majority of the patients in both the groups were discharged to home (monitoring group [MG] 5 5,355 [93.8%], nonmonitoring group [NMG] 5 24,309 [91.4%]) (Table 3). MG was more likely to be discharged to home
Table 3 Results of univariate regression analyses showing the effects of intraoperative neuromonitoring on various outcomes after pediatric spinal fusions for scoliosis.* Outcome
Monitoring group (n 5 5,706)
Nonmonitoring group (n 5 26,599)
Length of stay, d, meanSD Discharge to home, % Hospital charges, in 2012 US dollars, meanSD In-hospital complications, % Mortality Neurologic Respiratory Cardiovascular Wound complications Surgical site infections
5.85.1 93.8 165,3639,4404
6.26.1 91.4 171,49710,5988
0.0 0.9 2.0 0.5 1.1 0.5
0.2 1.4 2.9 1.2 3.2 0.6
Unadjusted coefficient (95% CI)
p value
0.42 (e0.80, e0.05) 1.44 (1.28, 1.62) 6134 ( 12764, 496)
d 0.66 0.68 0.44 0.32 0.73
(0.49, (0.56, (0.30, (0.25, (0.49,
0.88) 0.84) 0.64) 0.42) 1.10)
.028 !.001 .070
d .005 !.001 !.001 !.001 .130
CI, confidence interval; SD, standard deviation. Linear regression analyses were used for continuous variables like length-of-stay and hospital charges, and regression coefficients are reported. Logistic regression analyses were used for binary outcome variables like discharge disposition and complications, and odds ratios are reported. *
30
J. George et al. / Spine Deformity 7 (2019) 27e32
Table 4 Results of multivariate regression analyses showing the effects of intraoperative neuromonitoring on various outcomes after pediatric spinal fusions for scoliosis.* Outcome
Adjusted coefficient (95% CI)
p value
Discharge to home Length-of-stay Hospital charges In-hospital complications Neurologic Respiratory Cardiovascular Wound complications Surgical site infections
1.25 (1.10, 1.41) 0.31 ( 0.68, 0.06) 1,037.8 (e5,339.1, e7,414.6)
!.001 .096 .75
0.77 0.73 0.48 0.37 0.81
(0.57, (0.59, (0.33, (0.28, (0.53,
1.04) 0.90) 0.71) 0.48) 1.24)
.084 .003 !.001 !.001 .341
CI, confidence interval. Linear regression analyses were used for continuous variables like length-of stay and hospital charges, and regression coefficients are reported. Logistic regression analyses were used for binary outcome variables like discharge disposition and complications, and odds ratios are reported. *
after adjusting for baseline characteristics (AOR 5 1.25 [95% CI: 1.10e1.41], p!.001) (Table 4). The mean LOS in the MG versus NMG were 5.85.1 and 6.26.1 days, respectively. On multivariate analysis, there was no difference in LOS between the groups (adjusted coefficient 5 e0.31 [e0.68 to 0.06], p 5 .096). The mean hospital charges in the MG versus NMG were 165,363 94,404 and 171,497 105,988 US dollars, respectively. On multivariate analysis, there was no difference in hospital charges between the groups (adjusted coefficient 5 1,038 [e5,339 e7,415], p 5 .096). Neurologic complications were lower in the MG (MG 5 52 [0.9%], NMG 5 368 [1.4%], OR 5 0.66 [0.49e0.88], p 5 .005) (Table 3). Although the incidence of neurologic complications tended to be lower in MG, this failed to statistically significant in the multivariate analysis (AOR 5 0.77 [0.57e1.04], p 5 .084) (Table 4). Among
Table 5 Results of sensitivity analyses showing the effects of intraoperative neuromonitoring on various outcomes after pediatric spinal fusions for scoliosis.* Outcome Discharge to home Length-of-stay Hospital charges In-hospital complications Neurological Respiratory Cardiovascular Wound complications Surgical site infections
Adjusted coefficient (95% CI) 1.21 (1.06, 1.38) 0.29 ( 0.66, 0.08) 2,291 ( 8,474, 3,874) 0.88 0.75 0.46 0.42 0.77
(0.64, (0.61, (0.31, (0.31, (0.49,
1.22) 0.94) 0.70) 0.55) 1.20)
p value .006 .128 .465 .471 .012 !.001 !.001 0.251
CI, confidence interval. Multivariate linear regression analyses were used for continuous variables like length-of stay and hospital charges, and regression coefficients are reported. Multivariate logistic regression analyses were used for binary outcome variables like discharge disposition and complications, and odds ratios are reported. *
those who did not develop neurologic complications, 92% (29,346/31,885) were discharged to home, whereas only 76% (318/420) of those who developed complications were discharged to home (p !.001). The incidence of neurologic complications were higher in MG, had a lower risk of respiratory (AOR 5 0.73 [0.59e0.90], p 5 .003), cardiovascular (AOR 5 0.48 [0.33e0.71], p!.001), and wound complications (AOR 5 0.37 [0.28e0.48], p !.001). Surgical site infections were similar between the two groups (AOR 5 0.81 [0.53e1.24], p 5 .341). There were no mortalities in the MG, whereas 49 (0.2%) patients died in the NMG. For the sensitivity analysis, 194 of 493 hospitals had at least one surgery with a code for IONM. Within these hospitals, a total of 20,978 (unweighted 5 4,331) surgeries were performed and IONM was used in 5,706 (27%) surgeries. The effect of IONM on various outcomes did not show a considerable variation with the sensitivity analysis (Table 5). Discussion Neuromonitoring is frequently used during pediatric spinal surgeries to detect impending neurologic damage. Although many studies have validated the effectiveness of IONM in assessing neurologic integrity, the effect of IONM on outcomes of pediatric spinal deformity surgeries has not been studied at a national level. Using the NIS database, this study found that IONM use was reported in 18% of the surgeries, which is lower than expected. This study also found that IONM was associated with increased home discharge, and did not result in increased hospital charges. Additionally, IONM appeared to be associated with lower risk of neurologic complications, though this finding was not statistically significant. Although the use of the NIS allowed for evaluation of the effects of IONM at a national level, it was not without limitations. Being an administrative database, NIS uses ICD-9 procedure and diagnosis codes to identify various diseases and procedures, and is therefore prone to coding inaccuracies [17,18]. Although the extent of coding errors with respect to IONM use is uncertain, the low prevalence of IONM suggests that there is underreporting of IONM in the nation’s largest all-payer database. However, a systematic bias in underreporting is unlikely, and therefore, the coding inaccuracies may not have affected the results related to the effects of IONM on various outcomes. This is further supported by the sensitivity analysis, which showed comparable results [10]. There are different methods for neuromonitoring such as somatosensory evoked potentials (SSEPs), motor evoked potentials (MEPs), and electromyography, and a combination of these are often used in spinal surgeries. However, as only one ICD-9 diagnosis code is available to represent all types of IONM, the different methods of neuromonitoring could not be separately studied. Also, this study is limited by the clinical details about
J. George et al. / Spine Deformity 7 (2019) 27e32
the complications. It is possible that some of the neurologic complications identified using codes might have been transient, whereas some might not have been iatrogenic. Moreover, this study does not provide information about the signal changes observed during IONM, and the subsequent actions taken. Nevertheless, this study provides a national overview on the outcomes of pediatric scoliosis surgeries using IONM in the United States. The prevalence of IONM reported in the NIS database is lower than that reported in many single- or multi-institutional studies [19-22]. In a study by Hamilton et al. [19] using the Scoliosis Research Society (SRS) database, the authors found the prevalence of IONM to be 65% among 108,419 adult and pediatric spinal surgeries. As most pediatric spinal fusion surgeries are performed by members of SRS, the low prevalence of IONM in the NIS database highlights the shortcomings of NIS database. Other studies using administrative databases similar to this study have reported comparable rates of IONM use, suggesting that lower prevalence in this study could be related to under-reporting [6,10,12]. As increasing number of studies are being performed using NIS and similar administrative databases, it is important to ensure procedures are accurately coded in these databases. The use of IONM did not increase hospital charges in the present study. However, Cole et al. [6] reported that the use of IONM was associated with increased payments for single-level spinal procedures in adults. Similarly, Ney et al. [10] also reported that hospital charges were higher in adult noncomplex spinal surgeries using IONM. The surgeries included by Cole et al. [6] and Ney et al. [10] were less complicated spinal surgeries, the overall costs of which are lower, and hence, might have been more likely to be influenced by costs of IONM. It is also possible that the lack of significance might have been because many patients in the NMG group were not coded for IONM. In this study, IONM was associated with increased home discharges. This was similar to James et al. [12], who found that IONM increased discharge to home in patents with thoracic fusion; however, they did not find any difference for other types of spinal procedures. Both the increased rate of discharge to home and the lack of higher hospital charges in surgeries using IONM might be due to the lower complications observed in the monitoring group, suggested by the increased rate of home discharge among those who did not have complications. In this study, there was a trend toward lower rate of neurologic complications in surgeries using IONM, though this finding was not statistically significant. This is similar to the findings of Fu et al. [7], who found that IONM did not reduce neurologic complications in 23,918 pediatric spinal surgeries using the SRS database. In fact, their study found that IONM was associated with higher rates of neurologic complications [7]. Cole et al. [6] evaluated single-level spinal procedures in adults undergoing lumbar laminectomies with IONM and demonstrated decreased odds of neurologic complications. However, they did not find any benefits with IONM with respect to other spinal
31
procedures. Similarly, a number of other studies evaluating various types of spinal procedures have failed to demonstrate a decrease in neurologic complications with IONM [10,12,19,23]. The lack of reduction in neurologic complications with IONM in this study and others could be due to the presence of unadjusted confounding factors and/or due to the low incidence of complications. Because of the retrospective nature of many studies including that of the present study, it is possible that IONM was only used in high-risk cases indicating a selection bias. Interestingly, a number of other complications were lower in the monitoring group. Although the reasons for these findings are unclear, it could be related to the overall expertise of the surgical and anesthesia team at institutions using IONM. In summary, there appears to be underreporting of IONM in the NIS database. Our study results highlights that researchers using these databases should be aware of such coding inaccuracies which needs to be acknowledged, and the results of database studies should be interpreted with caution. Despite the limitations of this database, this study demonstrates that IONM was significantly associated with increased home discharge, and we did not find significant increase in hospital charges with use of IONM. There was a trend toward lower risk of neurologic complications with IONM use, though this finding failed to achieve statistical significance. Given the rising health care costs and the concerns about benefits of widespread use of IONM, the results of this study suggest that IONM might improve the outcomes of pediatric spinal fusion surgeries for scoliosis without increasing the hospital charges.
References [1] Coe JD, Arlet V, Donaldson W, et al. Complications in spinal fusion for adolescent idiopathic scoliosis in the new millennium. A report of the Scoliosis Research Society Morbidity and Mortality Committee. Spine (Phila Pa 1976) 2006;31:345e9. [2] Devlin VJ, Schwartz DM. Intraoperative neurophysiologic monitoring during spinal surgery. J Am Acad Orthop Surg 2007;15:549e60. [3] Reames DL, Smith JS, Fu K-MG, et al. Complications in the surgical treatment of 19,360 cases of pediatric scoliosis: a review of the Scoliosis Research Society Morbidity and Mortality database. Spine (Phila Pa 1976) 2011;36:1484e91. [4] Thompson GH, Wilber RG, Shaffer JW, et al. Segmental spinal instrumentation in idiopathic scoliosis. A preliminary report. Spine (Phila Pa 1976) 1985;10:623e30. [5] Wilber RG, Thompson GH, Shaffer JW, et al. Postoperative neurological deficits in segmental spinal instrumentation. A study using spinal cord monitoring. J Bone Joint Surg Am 1984;66:1178e87. [6] Cole T, Veeravagu A, Zhang M, et al. Intraoperative neuromonitoring in single-level spinal procedures: a retrospective propensity scorematched analysis in a national longitudinal database. Spine (Phila Pa 1976) 2014;39:1950e9. [7] Fu K-MG, Smith JS, Polly DW, et al. Morbidity and mortality associated with spinal surgery in children: a review of the Scoliosis Research Society morbidity and mortality database. J Neurosurg Pediatr 2011;7:37e41. [8] Schwartz DM, Auerbach JD, Dormans JP, et al. Neurophysiological detection of impending spinal cord injury during scoliosis surgery. J Bone Joint Surg Am 2007;89:2440.
32
J. George et al. / Spine Deformity 7 (2019) 27e32
[9] Sankar WN, Skaggs DL, Emans JB, et al. Neurologic risk in growing rod spine surgery in early onset scoliosis: is neuromonitoring necessary for all cases? Spine (Phila Pa 1976) 2009;34:1952e5. [10] Ney JP, van der Goes DN, Nuwer MR. Does intraoperative neurophysiologic monitoring matter in noncomplex spine surgeries? Neurology 2015;85:2151e8. [11] Gunnarsson T, Krassioukov AV, Sarjeant R, et al. Real-time continuous intraoperative electromyographic and somatosensory evoked potential recordings in spinal surgery: correlation of clinical and electrophysiologic findings in a prospective, consecutive series of 213 cases. Spine (Phila Pa 1976) 2004;29:677e84. [12] James WS, Rughani AI, Dumont TM. A socioeconomic analysis of intraoperative neurophysiological monitoring during spine surgery: national use, regional variation, and patient outcomes. Neurosurg Focus 2014;37:E10. [13] HCUP-Nationwide Inpatient Sample (NIS). Healthcare Cost and Utilization Project (HCUP). Rockville, MD: Agency for Healthcare Research and Quality; 2015. Available at: www.hcup-us.ahrq.gov/ nisoverview.jsp. Accessed January 1, 2015. [14] Agency for Healthcare Research and Quality. HCUP Comorbidity software. Healthcare Cost and Utilization Project (HCUP) 2014. Available at: https://www.hcup-us.ahrq.gov/toolssoftware/ comorbidity/comorbidity.jsp. Accessed September 15, 2017. [15] US Department of Labor. Bureau of Labor Statistics. Consumer Price Index 2017. Available at: http://www.bls.gov/data/inflation_ calculator.htm. Accessed September 15, 2017.
[16] R Core Team. R: a language and environment for statistical computing. [17] George J, Newman JM, Ramanathan D, et al. Administrative databases can yield false conclusionsdan example of obesity in total joint arthroplasty. J Arthroplasty 2017;32(9 suppl):S86e90. [18] Lawson EH, Louie R, Zingmond DS, et al. A comparison of clinical registry versus administrative claims data for reporting of 30-day surgical complications. Ann Surg 2012;256:973e81. [19] Hamilton DK, Smith JS, Sansur CA, et al. Rates of new neurological deficit associated with spine surgery based on 108,419 procedures: a report of the Scoliosis Research Society morbidity and mortality committee. Spine (Phila Pa 1976) 2011;36:1218e28. [20] McGrory BJ, Klassen RA. Arthrodesis of the cervical spine for fractures and dislocations in children and adolescents. A long-term follow-up study. J Bone Joint Surg Am 1994;76:1606e16. [21] Molinari RW, Bridwell KH, Lenke LG, et al. Complications in the surgical treatment of pediatric high-grade, isthmic dysplastic spondylolisthesis. A comparison of three surgical approaches. Spine (Phila Pa 1976) 1999;24:1701e11. [22] Qiu Y, Wang S, Wang B, et al. Incidence and risk factors of neurological deficits of surgical correction for scoliosis: analysis of 1373 cases at one Chinese institution. Spine (Phila Pa 1976) 2008;33: 519e26. [23] Ajiboye RM, Zoller SD, D’Oro A, et al. Utility of intraoperative neuromonitoring for lumbar pedicle screw placement is questionable. Spine (Phila Pa 1976) 2017;42:1006e10.
Case Series
Influence of Intraoperative Neuromonitoring on the Outcomes of Surgeries for Pediatric Scoliosis in the United States Jaiben George, MBBSa,*, Soumabha Das, MBBSb, Anthony C. Egger, MDa, Reid C. Chambers, DOa, Thomas E. Kuivila, MDa, Ryan C. Goodwin, MDa b
a Department of Orthopaedic Surgery, Cleveland Clinic, 9500 Euclid Avenue, Cleveland, OH 44915, USA Department of Physical Medicine and Rehabilitation, Westchester Medical Center, 100 Woods Rd, Valhalla, NY 10595, USA Received 12 December 2017; revised 7 May 2018; accepted 18 May 2018
Abstract Background: Intraoperative neuromonitoring (IONM) is used to detect impending neurologic damage during complex spinal surgeries. Although IONM is increasingly used during pediatric scoliosis surgeries in the United States, the effect of IONM on the outcomes of such surgeries at a national level is unclear. Methods: Using National Inpatient Sample (NIS) from 2009 to 2012, 32,305 spinal fusions performed in children 18 years old or younger of age with scoliosis were identified using ICD-9 procedure and diagnosis codes. IONM was identified using the ICD-9 procedure code 00.94. The effects of IONM use on length of stay (LOS), discharge disposition, hospital charges, and in-hospital complications were assessed using multivariate regression analysis adjusting for patient and hospital characteristics. Results: IONM was used in 5,706 (18%) of the surgeries. IONM was associated with increased home discharge (adjusted odds ratio [AOR] 5 1.25 [95% confidence interval 1.10-1.40], p 5 .001). There was no difference in LOS (p 5 .096) and hospital charges (p 5 .750). Neurologic complications were noted in 52 (0.9%) surgeries using IONM and 368 (1.4%) surgeries without IONM (p 5 .005). Although IONM use trended toward lower risk of neurologic complications in multivariate analysis, it failed to achieve statistical significance (AOR 5 0.77 [0.57-1.04], p 5 .084). Conclusions: Reported use of IONM in this database was significantly less compared with other databases, suggesting that IONM might be underreported in the NIS database. Nevertheless, in this database, IONM was significantly associated with increased home discharge. Hospital charges and LOS were not affected by IONM. There was a trend toward lower risk of neurologic complications with IONM use, though this finding was not statistically significant. Ó 2018 Scoliosis Research Society. All rights reserved. Keywords: Scoliosis; Spinal deformity; Neuromonitoring; Spinal fusion; Complications
Introduction Surgical correction of scoliosis and other pediatric spinal deformities involves substantial manipulation of the vertebral column and the associated neural structures [1-3].
Author disclosures: JG (none), SD (none), ACE (none), RCC (none), TEK (none), RCG (none). The authors declare no conflicts of interest. No funding was obtained for this study. The study was exempt from IRB approval as it used a publicly available database. *Corresponding author. Department of Orthopaedic Surgery, Cleveland Clinic, 9500 Euclid Avenue, A40, Cleveland, OH 44915, USA. Tel.: (216) 972-1965; fax: (216) 445-3585. E-mail address: [email protected] (J. George).
Although a number of complications can occur after such complex spinal deformity surgeries, neurologic complications are among the most feared [1,3]. Neurologic injury can result from excessive stretching of spinal cord or the nerve roots, hypoperfusion, direct injury during implant placement, or from patient positioning [2] Historically, the rates of neurologic complications were reported to be 3% to 4% [4,5]. However, more recent studies report an incidence between 0.5% and 2% [1,3,6,7]. This decline in complication rates over the last few decades is, at least in part, due to the availability of intraoperative neuromonitoring (IONM) [2,8]. IONM can detect impending neurologic injury intraoperatively, allowing the surgeon to take action to prevent long-lasting neurologic deficits [2]. IONM is becoming
2212-134X/$ - see front matter Ó 2018 Scoliosis Research Society. All rights reserved. https://doi.org/10.1016/j.jspd.2018.05.013
28
J. George et al. / Spine Deformity 7 (2019) 27e32
more common in pediatric spinal deformity surgeries and is available in most of the United States [2,3]. Despite increased use, very few studies have compared the outcomes of operations with and without IONM [8-10]. Although IONM is proven to be useful in detecting early neurologic insult, there is little evidence to support that IONM decreases the incidence of neurologic complications [8,11]. Moreover, given the rising health care costs, there are also concerns about the utility of IONM in less complex spinal surgeries [9,10,12]. As the frequency of IONM increases in pediatric spinal deformity surgeries performed in the United States, it is important to study the effects of IONM on the outcomes of such surgeries at a national level. Therefore, the present study used the Nationwide Inpatient Sample (NIS), a nationally representative database, to compare 1) the length of stay (LOS), discharge disposition, and hospital charges and 2) in-hospital complications (including neurologic complications) between pediatric fusion surgeries for scoliosis with and without IONM.
previous codes; including kyphosis, kyphoscoliosis, etc). The surgical approach was divided into two groups based on ICD9 procedure codes: posterior-only approach (only 81.03, 81.05, 81.07, and 81.08) and combined anterior/posterior approach (contains 81.02, 81.04, and 81.06). The number of levels fused were grouped as follows: three or fewer vertebrae (81.62), four to eight vertebrae (81.63), and more than eight vertebrae (81.64). Patient demographics, comorbidities provided by AHRQ, and hospital characteristics were recorded [14]. The baseline characteristics of patients in the monitoring and nonmonitoring groups are provided in Tables 1 and 2.
Methods
Variable
Monitoring group (n 5 5,706)
Nonmonitoring group (n 5 26,599)
Age, mean SD (in years) Sex Male Female Race White Black Hispanic Other/missing Insurance Medicare/Medicaid Private Other/missing Diagnoses Adolescent idiopathic scoliosis Congenital scoliosis Neuromuscular scoliosis Other curvature deformities Levels fused !3 4e8 O8 Unspecified Approach Posterior Anterior posterior Number of comorbidities 0 1 or 2 >3 Year 2009 2010 2011 2012
13.72.6
13.82.6
32.5 62.8
31.7 67.9
51.5 11.8 7.2 7.1
53.1 12.8 1.5 8.4
26 68.4 5.6
31.9 60.2 7.9
68.9
71.2
7 7.4 16.8
6.1 5.1 17.6
5.7 20.9 73.1 0.3
7.4 20 70.8 1.8
92.4 7.6
90.5 9.5
61.5 34.1 4.4
58.3 36.1 5.6
18.9 16.9 29.7 34.5
22.8 28.5 24.2 24.5
Outcomes The effects of IONM on LOS, discharge disposition, hospital charges, and in-hospital complications were Table 1 Baseline patient and surgical characteristics of spinal fusion surgeries performed for pediatric scoliosis in the United States.
Data source This study used the NIS database from 2009 to 2012. The NIS database is a part of the Healthcare Cost and Utilization Project managed by the Agency for Healthcare Research and Quality (AHRQ) and is the largest all-payer database in the United States containing information about inpatient admissions [13]. The NIS is a stratified sample of 20% of all the hospitalizations in the United States and contains an array of information including demographics, International Classification of Disease, ninth edition (ICD-9), procedure and diagnosis codes, insurance information, admission data, hospital data, LOS, discharge disposition, and total charges. The NIS database provides sampling weights for each discharge that can be used to obtain national-level estimates of these discharges. The NIS data are publically available and deidentified; therefore, the study was deemed exempt from approval by our institutional review board. Study population All discharges in children 18 years old or younger admitted for spinal fusion were identified using the primary clinical classification software code of 158. These fusion procedures were screened to identify spinal fusions performed in children with scoliosis (isolated or associated with other conditions/ deformities) using the ICD-9 diagnosis code 737.x. The use of IONM was identified using the ICD-9 procedure code 00.94. A total of 32,305 (unweighted 5 6,642) spinal fusion surgeries performed for scoliosis were identified, and IONM was used in 5,706 (18%) of the operations (monitoring group). Patients were grouped into the following diagnosis groups: adolescent idiopathic scoliosis (737.30), congenital scoliosis (737.31, 737.32, 754.2), neuromuscular scoliosis (737.40-3), and other curvature deformities (those without any of the
p value
.149 .634
.051
!.001
.01
!.001
.034
.065
!.001
Values are percentages unless otherwise noted.
J. George et al. / Spine Deformity 7 (2019) 27e32 Table 2 Hospital characteristics of spinal fusion surgeries performed for pediatric scoliosis in the United States. Variable
Location Rural Urban nonteaching Urban teaching Missing Region Northeast Midwest South West Size Small Medium Large Missing Control Government Private, nonprofit Private, for profit Missing
Monitoring group, % (n 5 5,706)
Nonmonitoring group, % (n 5 26,599)
1.6 4.6 93.3 0.5
1.9 8.4 88.7 1.0
8.4 32.8 34.3 24.5
13.9 21.6 40.4 24.1
19.1 25.7 54.7 0.5
15.6 27.9 55.6 1.0
8.9 88.8 1.8 0.5
8.0 83.4 7.7 1.0
29
Analysis All analyses were performed using sampling weights provided by NIS to obtain the national estimates. As NIS changed the sampling method in 2012, we used the updated trend weights for all the years used in the study. Descriptive statistics were used to compare the baseline characteristics of monitoring and nonmonitoring groups. Linear regression analysis was used for continuous outcome variables like LOS and hospital charges, whereas logistic regression analysis was used for binary outcome variables like discharge disposition and complications. Multivariate regression analyses were performed to adjust for potential confounding factors. Only the baseline variables that were different between the groups at a p value of .10 were included in the multivariate analysis to avoid overfitting of the model. As IONM might be underreported in the NIS database due to coding inaccuracies, a sensitivity analysis was performed to include only the hospitals that performed at least one surgery with a code for IONM [10]. The results of multivariate logistic regression analyses were reported using adjusted odds ratios (AORs), whereas that of multivariate linear regression analyses were reported using adjusted unstandardized coefficients. The frequency of missing values among different variables, if applicable, are given in Tables 1 and 2. Missing values were excluded from the multivariate regression analyses. Confidence intervals (CIs) of 95% were calculated and the level of significance for all analysis was set at p!.05. Statistical analysis was performed using R software (version 3.1.3, Vienna, Austria) [16].
p value
!.001
!.001
.011
!.001
assessed. The LOS was calculated as the number of days from admission to discharge, and the discharge disposition was categorized as home or a care facility. Hospital charges provided in the NIS were converted to 2012 US dollars using the Consumer Price Index inflation calculator provided by Bureau of Labor Statistics to adjust for inflation [15]. Neurologic complications were defined using the following ICD-9 diagnosis codes: 9970.0-2, 997.09, and 907.0-9. Other in-hospital complications assessed in this study included mortality, respiratory complications (997.31-2,997.39), cardiovascular complications (997.1, 997.7x, 997.8x), wound complications (998.1x, 998.3x, 99883), and surgical site infections (998.51, 998.59).
Results Majority of the patients in both the groups were discharged to home (monitoring group [MG] 5 5,355 [93.8%], nonmonitoring group [NMG] 5 24,309 [91.4%]) (Table 3). MG was more likely to be discharged to home
Table 3 Results of univariate regression analyses showing the effects of intraoperative neuromonitoring on various outcomes after pediatric spinal fusions for scoliosis.* Outcome
Monitoring group (n 5 5,706)
Nonmonitoring group (n 5 26,599)
Length of stay, d, meanSD Discharge to home, % Hospital charges, in 2012 US dollars, meanSD In-hospital complications, % Mortality Neurologic Respiratory Cardiovascular Wound complications Surgical site infections
5.85.1 93.8 165,3639,4404
6.26.1 91.4 171,49710,5988
0.0 0.9 2.0 0.5 1.1 0.5
0.2 1.4 2.9 1.2 3.2 0.6
Unadjusted coefficient (95% CI)
p value
0.42 (e0.80, e0.05) 1.44 (1.28, 1.62) 6134 ( 12764, 496)
d 0.66 0.68 0.44 0.32 0.73
(0.49, (0.56, (0.30, (0.25, (0.49,
0.88) 0.84) 0.64) 0.42) 1.10)
.028 !.001 .070
d .005 !.001 !.001 !.001 .130
CI, confidence interval; SD, standard deviation. Linear regression analyses were used for continuous variables like length-of-stay and hospital charges, and regression coefficients are reported. Logistic regression analyses were used for binary outcome variables like discharge disposition and complications, and odds ratios are reported. *
30
J. George et al. / Spine Deformity 7 (2019) 27e32
Table 4 Results of multivariate regression analyses showing the effects of intraoperative neuromonitoring on various outcomes after pediatric spinal fusions for scoliosis.* Outcome
Adjusted coefficient (95% CI)
p value
Discharge to home Length-of-stay Hospital charges In-hospital complications Neurologic Respiratory Cardiovascular Wound complications Surgical site infections
1.25 (1.10, 1.41) 0.31 ( 0.68, 0.06) 1,037.8 (e5,339.1, e7,414.6)
!.001 .096 .75
0.77 0.73 0.48 0.37 0.81
(0.57, (0.59, (0.33, (0.28, (0.53,
1.04) 0.90) 0.71) 0.48) 1.24)
.084 .003 !.001 !.001 .341
CI, confidence interval. Linear regression analyses were used for continuous variables like length-of stay and hospital charges, and regression coefficients are reported. Logistic regression analyses were used for binary outcome variables like discharge disposition and complications, and odds ratios are reported. *
after adjusting for baseline characteristics (AOR 5 1.25 [95% CI: 1.10e1.41], p!.001) (Table 4). The mean LOS in the MG versus NMG were 5.85.1 and 6.26.1 days, respectively. On multivariate analysis, there was no difference in LOS between the groups (adjusted coefficient 5 e0.31 [e0.68 to 0.06], p 5 .096). The mean hospital charges in the MG versus NMG were 165,363 94,404 and 171,497 105,988 US dollars, respectively. On multivariate analysis, there was no difference in hospital charges between the groups (adjusted coefficient 5 1,038 [e5,339 e7,415], p 5 .096). Neurologic complications were lower in the MG (MG 5 52 [0.9%], NMG 5 368 [1.4%], OR 5 0.66 [0.49e0.88], p 5 .005) (Table 3). Although the incidence of neurologic complications tended to be lower in MG, this failed to statistically significant in the multivariate analysis (AOR 5 0.77 [0.57e1.04], p 5 .084) (Table 4). Among
Table 5 Results of sensitivity analyses showing the effects of intraoperative neuromonitoring on various outcomes after pediatric spinal fusions for scoliosis.* Outcome Discharge to home Length-of-stay Hospital charges In-hospital complications Neurological Respiratory Cardiovascular Wound complications Surgical site infections
Adjusted coefficient (95% CI) 1.21 (1.06, 1.38) 0.29 ( 0.66, 0.08) 2,291 ( 8,474, 3,874) 0.88 0.75 0.46 0.42 0.77
(0.64, (0.61, (0.31, (0.31, (0.49,
1.22) 0.94) 0.70) 0.55) 1.20)
p value .006 .128 .465 .471 .012 !.001 !.001 0.251
CI, confidence interval. Multivariate linear regression analyses were used for continuous variables like length-of stay and hospital charges, and regression coefficients are reported. Multivariate logistic regression analyses were used for binary outcome variables like discharge disposition and complications, and odds ratios are reported. *
those who did not develop neurologic complications, 92% (29,346/31,885) were discharged to home, whereas only 76% (318/420) of those who developed complications were discharged to home (p !.001). The incidence of neurologic complications were higher in MG, had a lower risk of respiratory (AOR 5 0.73 [0.59e0.90], p 5 .003), cardiovascular (AOR 5 0.48 [0.33e0.71], p!.001), and wound complications (AOR 5 0.37 [0.28e0.48], p !.001). Surgical site infections were similar between the two groups (AOR 5 0.81 [0.53e1.24], p 5 .341). There were no mortalities in the MG, whereas 49 (0.2%) patients died in the NMG. For the sensitivity analysis, 194 of 493 hospitals had at least one surgery with a code for IONM. Within these hospitals, a total of 20,978 (unweighted 5 4,331) surgeries were performed and IONM was used in 5,706 (27%) surgeries. The effect of IONM on various outcomes did not show a considerable variation with the sensitivity analysis (Table 5). Discussion Neuromonitoring is frequently used during pediatric spinal surgeries to detect impending neurologic damage. Although many studies have validated the effectiveness of IONM in assessing neurologic integrity, the effect of IONM on outcomes of pediatric spinal deformity surgeries has not been studied at a national level. Using the NIS database, this study found that IONM use was reported in 18% of the surgeries, which is lower than expected. This study also found that IONM was associated with increased home discharge, and did not result in increased hospital charges. Additionally, IONM appeared to be associated with lower risk of neurologic complications, though this finding was not statistically significant. Although the use of the NIS allowed for evaluation of the effects of IONM at a national level, it was not without limitations. Being an administrative database, NIS uses ICD-9 procedure and diagnosis codes to identify various diseases and procedures, and is therefore prone to coding inaccuracies [17,18]. Although the extent of coding errors with respect to IONM use is uncertain, the low prevalence of IONM suggests that there is underreporting of IONM in the nation’s largest all-payer database. However, a systematic bias in underreporting is unlikely, and therefore, the coding inaccuracies may not have affected the results related to the effects of IONM on various outcomes. This is further supported by the sensitivity analysis, which showed comparable results [10]. There are different methods for neuromonitoring such as somatosensory evoked potentials (SSEPs), motor evoked potentials (MEPs), and electromyography, and a combination of these are often used in spinal surgeries. However, as only one ICD-9 diagnosis code is available to represent all types of IONM, the different methods of neuromonitoring could not be separately studied. Also, this study is limited by the clinical details about
J. George et al. / Spine Deformity 7 (2019) 27e32
the complications. It is possible that some of the neurologic complications identified using codes might have been transient, whereas some might not have been iatrogenic. Moreover, this study does not provide information about the signal changes observed during IONM, and the subsequent actions taken. Nevertheless, this study provides a national overview on the outcomes of pediatric scoliosis surgeries using IONM in the United States. The prevalence of IONM reported in the NIS database is lower than that reported in many single- or multi-institutional studies [19-22]. In a study by Hamilton et al. [19] using the Scoliosis Research Society (SRS) database, the authors found the prevalence of IONM to be 65% among 108,419 adult and pediatric spinal surgeries. As most pediatric spinal fusion surgeries are performed by members of SRS, the low prevalence of IONM in the NIS database highlights the shortcomings of NIS database. Other studies using administrative databases similar to this study have reported comparable rates of IONM use, suggesting that lower prevalence in this study could be related to under-reporting [6,10,12]. As increasing number of studies are being performed using NIS and similar administrative databases, it is important to ensure procedures are accurately coded in these databases. The use of IONM did not increase hospital charges in the present study. However, Cole et al. [6] reported that the use of IONM was associated with increased payments for single-level spinal procedures in adults. Similarly, Ney et al. [10] also reported that hospital charges were higher in adult noncomplex spinal surgeries using IONM. The surgeries included by Cole et al. [6] and Ney et al. [10] were less complicated spinal surgeries, the overall costs of which are lower, and hence, might have been more likely to be influenced by costs of IONM. It is also possible that the lack of significance might have been because many patients in the NMG group were not coded for IONM. In this study, IONM was associated with increased home discharges. This was similar to James et al. [12], who found that IONM increased discharge to home in patents with thoracic fusion; however, they did not find any difference for other types of spinal procedures. Both the increased rate of discharge to home and the lack of higher hospital charges in surgeries using IONM might be due to the lower complications observed in the monitoring group, suggested by the increased rate of home discharge among those who did not have complications. In this study, there was a trend toward lower rate of neurologic complications in surgeries using IONM, though this finding was not statistically significant. This is similar to the findings of Fu et al. [7], who found that IONM did not reduce neurologic complications in 23,918 pediatric spinal surgeries using the SRS database. In fact, their study found that IONM was associated with higher rates of neurologic complications [7]. Cole et al. [6] evaluated single-level spinal procedures in adults undergoing lumbar laminectomies with IONM and demonstrated decreased odds of neurologic complications. However, they did not find any benefits with IONM with respect to other spinal
31
procedures. Similarly, a number of other studies evaluating various types of spinal procedures have failed to demonstrate a decrease in neurologic complications with IONM [10,12,19,23]. The lack of reduction in neurologic complications with IONM in this study and others could be due to the presence of unadjusted confounding factors and/or due to the low incidence of complications. Because of the retrospective nature of many studies including that of the present study, it is possible that IONM was only used in high-risk cases indicating a selection bias. Interestingly, a number of other complications were lower in the monitoring group. Although the reasons for these findings are unclear, it could be related to the overall expertise of the surgical and anesthesia team at institutions using IONM. In summary, there appears to be underreporting of IONM in the NIS database. Our study results highlights that researchers using these databases should be aware of such coding inaccuracies which needs to be acknowledged, and the results of database studies should be interpreted with caution. Despite the limitations of this database, this study demonstrates that IONM was significantly associated with increased home discharge, and we did not find significant increase in hospital charges with use of IONM. There was a trend toward lower risk of neurologic complications with IONM use, though this finding failed to achieve statistical significance. Given the rising health care costs and the concerns about benefits of widespread use of IONM, the results of this study suggest that IONM might improve the outcomes of pediatric spinal fusion surgeries for scoliosis without increasing the hospital charges.
References [1] Coe JD, Arlet V, Donaldson W, et al. Complications in spinal fusion for adolescent idiopathic scoliosis in the new millennium. A report of the Scoliosis Research Society Morbidity and Mortality Committee. Spine (Phila Pa 1976) 2006;31:345e9. [2] Devlin VJ, Schwartz DM. Intraoperative neurophysiologic monitoring during spinal surgery. J Am Acad Orthop Surg 2007;15:549e60. [3] Reames DL, Smith JS, Fu K-MG, et al. Complications in the surgical treatment of 19,360 cases of pediatric scoliosis: a review of the Scoliosis Research Society Morbidity and Mortality database. Spine (Phila Pa 1976) 2011;36:1484e91. [4] Thompson GH, Wilber RG, Shaffer JW, et al. Segmental spinal instrumentation in idiopathic scoliosis. A preliminary report. Spine (Phila Pa 1976) 1985;10:623e30. [5] Wilber RG, Thompson GH, Shaffer JW, et al. Postoperative neurological deficits in segmental spinal instrumentation. A study using spinal cord monitoring. J Bone Joint Surg Am 1984;66:1178e87. [6] Cole T, Veeravagu A, Zhang M, et al. Intraoperative neuromonitoring in single-level spinal procedures: a retrospective propensity scorematched analysis in a national longitudinal database. Spine (Phila Pa 1976) 2014;39:1950e9. [7] Fu K-MG, Smith JS, Polly DW, et al. Morbidity and mortality associated with spinal surgery in children: a review of the Scoliosis Research Society morbidity and mortality database. J Neurosurg Pediatr 2011;7:37e41. [8] Schwartz DM, Auerbach JD, Dormans JP, et al. Neurophysiological detection of impending spinal cord injury during scoliosis surgery. J Bone Joint Surg Am 2007;89:2440.
32
J. George et al. / Spine Deformity 7 (2019) 27e32
[9] Sankar WN, Skaggs DL, Emans JB, et al. Neurologic risk in growing rod spine surgery in early onset scoliosis: is neuromonitoring necessary for all cases? Spine (Phila Pa 1976) 2009;34:1952e5. [10] Ney JP, van der Goes DN, Nuwer MR. Does intraoperative neurophysiologic monitoring matter in noncomplex spine surgeries? Neurology 2015;85:2151e8. [11] Gunnarsson T, Krassioukov AV, Sarjeant R, et al. Real-time continuous intraoperative electromyographic and somatosensory evoked potential recordings in spinal surgery: correlation of clinical and electrophysiologic findings in a prospective, consecutive series of 213 cases. Spine (Phila Pa 1976) 2004;29:677e84. [12] James WS, Rughani AI, Dumont TM. A socioeconomic analysis of intraoperative neurophysiological monitoring during spine surgery: national use, regional variation, and patient outcomes. Neurosurg Focus 2014;37:E10. [13] HCUP-Nationwide Inpatient Sample (NIS). Healthcare Cost and Utilization Project (HCUP). Rockville, MD: Agency for Healthcare Research and Quality; 2015. Available at: www.hcup-us.ahrq.gov/ nisoverview.jsp. Accessed January 1, 2015. [14] Agency for Healthcare Research and Quality. HCUP Comorbidity software. Healthcare Cost and Utilization Project (HCUP) 2014. Available at: https://www.hcup-us.ahrq.gov/toolssoftware/ comorbidity/comorbidity.jsp. Accessed September 15, 2017. [15] US Department of Labor. Bureau of Labor Statistics. Consumer Price Index 2017. Available at: http://www.bls.gov/data/inflation_ calculator.htm. Accessed September 15, 2017.
[16] R Core Team. R: a language and environment for statistical computing. [17] George J, Newman JM, Ramanathan D, et al. Administrative databases can yield false conclusionsdan example of obesity in total joint arthroplasty. J Arthroplasty 2017;32(9 suppl):S86e90. [18] Lawson EH, Louie R, Zingmond DS, et al. A comparison of clinical registry versus administrative claims data for reporting of 30-day surgical complications. Ann Surg 2012;256:973e81. [19] Hamilton DK, Smith JS, Sansur CA, et al. Rates of new neurological deficit associated with spine surgery based on 108,419 procedures: a report of the Scoliosis Research Society morbidity and mortality committee. Spine (Phila Pa 1976) 2011;36:1218e28. [20] McGrory BJ, Klassen RA. Arthrodesis of the cervical spine for fractures and dislocations in children and adolescents. A long-term follow-up study. J Bone Joint Surg Am 1994;76:1606e16. [21] Molinari RW, Bridwell KH, Lenke LG, et al. Complications in the surgical treatment of pediatric high-grade, isthmic dysplastic spondylolisthesis. A comparison of three surgical approaches. Spine (Phila Pa 1976) 1999;24:1701e11. [22] Qiu Y, Wang S, Wang B, et al. Incidence and risk factors of neurological deficits of surgical correction for scoliosis: analysis of 1373 cases at one Chinese institution. Spine (Phila Pa 1976) 2008;33: 519e26. [23] Ajiboye RM, Zoller SD, D’Oro A, et al. Utility of intraoperative neuromonitoring for lumbar pedicle screw placement is questionable. Spine (Phila Pa 1976) 2017;42:1006e10.