SPINE Volume 40, Number 13, pp 992-1000 ©2015, Wolters Kluwer Health, Inc. All rights reserved.
CERVICAL SPINE
Analysis of Delays to Surgery for Cervical Spinal Cord Injuries Andre M. Samuel, BBA, Daniel D. Bohl, MPH, Bryce A. Basques, BS, Pablo J. Diaz-Collado, MD, Adam M. Lukasiewicz, MSc, Matthew L. Webb, AB, and Jonathan N. Grauer, MD
Study Design. A retrospective study of surgically treated patients with cervical spinal cord injury (SCI) from the National Trauma Data Bank Research Data Set. Objective. To determine how time to surgery differs between SCI subtypes, where delays before surgery occur, and what factors are associated with delays. Summary of Background Data. Studies have shown that patients with cervical SCI undergoing surgery within 24 hours after injury have superior neurological outcomes to patients undergoing later surgery, with most evidence coming from the incomplete SCI subpopulation. Methods. Surgically treated patients with cervical SCI from 2011 and 2012 were identified in National Trauma Data Bank Research Data Set and divided into subpopulations of complete, central, and other incomplete SCIs. Relationships between surgical timing and patient and injury characteristics were analyzed using multivariate regression. Results. A total of 2636 patients with cervical SCI were identified: 803 with complete SCI, 950 with incomplete SCI, and 883 with central SCI. The average time to surgery was 51.1 hours for patients with complete SCI, 55.3 hours for patients with incomplete SCI, and 83.1 hours for patients with central SCI. Only 44% of patients with SCI underwent surgery within the first 24 hours after injury, including only 49% of patients with incomplete SCI. The vast majority of time between injury and surgery was after admission, rather than in the emergency department or in the field. Upper cervical SCIs and greater Charlson Comorbidity Index were associated with later surgery in all 3 SCI subpopulations. Conclusion. The majority of patients with SCI do not undergo surgery within the first 24 hours after injury, and the majority of From the Department of Orthopaedics and Rehabilitation, Yale School of Medicine, New Haven, CT. Acknowledgment date: October 29, 2014. Revision date: January 21, 2015. Second revision date: February 24, 2015. Acceptance date: March 10, 2015. The manuscript submitted does not contain information about medical device(s)/drug(s). No funds were received in support of this work. Relevant financial activities outside the submitted work: consultancy, expert testimony, grants. Address correspondence and reprint requests to Jonathan N. Grauer, MD, Department of Orthopaedics and Rehabilitation, Yale School of Medicine, 800 Howard Ave, New Haven, CT 06510; E-mail: [email protected] DOI: 10.1097/BRS.0000000000000883
992
delays occur after inpatient admission. Factors associated with these delays highlight areas of focus for expediting care in these patient populations. Key words: spinal cord injury, cervical, complete spinal cord injury, incomplete spinal cord injury, central spinal cord injury, surgical timing, decompression, comorbidities, upper cervical spinal cord injuries, inpatient delay. Level of Evidence: 4 Spine 2015;40:992–1000
T
here is growing interest in the ideal timing of surgery after cervical spinal cord injury (SCI). Early clinical studies of the effect of surgical timing on outcomes showed mixed results,1–4 likely due to inadequate sample sizes and the logistical challenges of substantially shortening the time to surgery. More recently, however, the Surgical Timing in Acute Spinal Cord Injury Study (STASCIS), a large prospective multicenter study of 313 patients published in 2012, was able to demonstrate neurological benefits with early surgical intervention.5 STASCIS showed that, compared with later surgery, surgery within 24 hours after injury was associated with a higher rate of improvement of 2 or more grades on the American Spinal Injury Association Impairment Scale at 6-month follow-up. As STASCIS was a prospective cohort study, and not a randomized controlled trial, there are limits to the conclusions that can be drawn from the results. As an example, significant differences in the baseline characteristics of the experimental groups and heterogeneity regarding SCI types in the study have been highlighted recently.6 However, the difficulties of conducting a randomized prospective study of SCIs are clear,7 and the results of STASCIS seem to confirm prevailing opinions of surgeons regarding optimal treatment of acute SCIs.8 In a survey of more than 900 spine surgeons, more than 80% preferred to decompress acute SCIs within the first 24 hours after injury.9 Recently, the results of STASCIS were also corroborated in a larger retrospective study of 1410 patients.10 In addition, a metanalysis of 1687 patients showed improved outcomes after early surgical decompression compared with both late decompression and nonoperative treatment.11 Cervical SCIs, however, present in a variety of different injury patterns, which each may respond differently to early
www.spinejournal.com
July 2015
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. SPINE141418_LR 992
09/06/15 7:12 AM
CERVICAL SPINE surgery. With incomplete SCIs, there is the greatest evidence for improved neurological function with early surgery.11 Therefore, for these patients, surgery is traditionally thought to be more urgent than that for patients with other SCI. With complete SCIs, on the contrary, there is disagreement among surgeons concerning the benefits of rapid surgery.9 Central SCIs, a subset of incomplete SCIs that often occur in older patients with pre-existing stenosis, are commonly managed with delayed surgical intervention, allowing time for hemodynamic stabilization and reduction of spinal cord edema.9,12–14 At the very least, there seems to be some benefit with early surgical intervention after incomplete SCIs (not including central SCIs). Early surgery also has value for other reasons such as decreased hospital costs and length of stay15 and decreased hospital morbidities.16,17 However, no studies to date have considered where delays may occur prior to surgery for these patients and what factors may be associated with delays. In addition, there has been little research specifically comparing current surgical timing after the various types of SCIs. The current study aims to investigate these 3 questions using data from the American College of Surgeons National Trauma Data Bank Research Data Set (NTDB RDS). This national data set has the large sample size necessary to appropriately power a study of the subtypes of this rare injury pattern. NTDB RDS also includes the preadmission data necessary to analyze the entire time interval from initial injury to surgery.
MATERIALS AND METHODS Data Source A retrospective cohort study was performed using the NTDB RDS, the largest national database of trauma patients containing data from more than 900 centers in the United States. The database contains both administratively coded and chartabstracted data on more than 100 different variables. Data from 2011 and 2012 were used. This study was approved by the local institutional human investigations committee.
Study Population International Classification of Diseases, Ninth Revision (ICD-9) diagnosis and procedure codes were used to identify all patients with cervical SCIs undergoing surgery during their hospitalization. The population was further limited to patients for whom the SCI could be identified as a complete, incomplete (excluding central SCIs), or central SCI using ICD-9 diagnosis codes and patients with complete time interval information (from ambulance dispatch to surgical intervention). Supplemental Digital Content Appendix 1, available at: http://links.lww.com/BRS/A973, lists all ICD-9 codes used to identify the study population.
Data Collection Patient characteristics and injury characteristics were assessed using NTDB RDS variables. Sex, age, insurance status, comorbidities, and presentation to an academic facility were directly Spine
Timing of Surgery for Cervical Spinal Cord Injury • Samuel et al
extracted from the database. Comorbidities are reported in NTDB RDS with specific comorbidity codes. The comorbidities analyzed were alcoholism, cancer, prior cerebrovascular accident, congestive heart failure, coronary artery disease, dementia, diabetes mellitus, functionally dependent status, hypertension requiring medication, obesity, renal failure, and respiratory disease. A modified Charlson Comorbidity Index (CCI) was then calculated.18 Modified CCIs have comparable predictive value to the original CCI19 and have been used previously with national databases studies.20 Computation of the modified CCI is detailed in Supplemental Digital Content Appendix 2, available at: http://links.lww.com/BRS/A973. Injury characteristics assessed included mechanism of injury, hospital-assigned injury severity score (ISS), and the presence of specific associated injury diagnoses based on ICD-9 coding. These codes can be found in Supplemental Digital Content Appendix 1, available at: http://links.lww.com/BRS/A973. Time intervals included in NTDB RDS include prehospital time from emergency medical services (EMS) dispatch to arrival at emergency department (ED), duration in ED, and duration between admission and surgery. Total time to surgery from EMS dispatch was computed using these intervals.
Analysis Three sets of analyses were conducted in this study. First, the 3 subpopulations of patients with cervical SCIs were described on the basis of patient characteristics and injury characteristics. To account for nonlinear relationships, continuous variables were converted to categorical variables, and differences between subpopulations were compared using Pearson χ2 test. Second, the total time to surgery was broken down into EMS time, ED time, and inpatient time to determine the percentage of time spent in each setting prior to surgery. The total time to surgery was then compared between the 3 subpopulations using multivariate analysis of variance with Tukey post hoc pairwise comparisons. Because the distribution of total time to surgery had substantial positive skew, this variable was log transformed to approximate a normal distribution in the analysis of variance. Third, patient and injury characteristics that are associated with increased time to surgery were determined using multivariate linear regression. Time to surgery was again log transformed as before. This analysis was performed for each of the 3 subpopulations separately. The variables included in these analyses were sex, CCI, mechanism of injury, presentation to academic trauma center, level of cervical SCI, and presence of 9 individual associated injuries. Age was not included in the analysis because it is already included in calculation of CCI and likely collinear. ISS was not included in the analysis as it is likely collinear with the individual-associated injuries of specific interest in the study (i.e., upper cervical SCIs, injuries, head injury, pelvic fracture, etc.). All statistical tests were 2-tailed and controlled for institutional clustering using linearized standard errors. Because of multiple hypothesis testing in this study, false discovery rate control21 was used to avoid type 1 error. A false discovery rate www.spinejournal.com
993
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. SPINE141418_LR 993
09/06/15 7:12 AM
CERVICAL SPINE of d = 0.05 was set and only P value of less than 0.017 was taken as significant.
RESULTS NTDB RDS contained 2636 patients with cervical SCIs who met the inclusion criteria in admission years 2011 and 2012. Of these, 803 (30.5%) had complete SCIs, 950 (36.0%) had incomplete SCIs (excluding central SCIs), and 883 (33.5%) had central SCIs.
Comparison of SCI Subpopulations Table 1 summarizes patient characteristics of the 3 SCI subpopulations. In all 3 subpopulations, the majority of patients were male. The central SCI subpopulation was older than the complete and incomplete SCI subpopulations and had higher rates of all pre-existing comorbidities, except for obesity, which was highest in the complete SCI subpopulation. Table 2 summarizes injury characteristics of the cervical SCI subpopulations. The patients with central SCI primarily experienced falls, whereas the most common mechanism of injury in the complete and incomplete SCI subpopulations was automobile accidents. ISS, which is predictive of mortality in trauma patients,22 was highest in the complete SCI subpopulation compared with both the incomplete and central SCI subpopulations (P < 0.001). Head injuries (intracranial injuries and skull fractures) were the most commonly associated injuries in all 3 subpopulations, occurring in 28.0% to 29.8% of patients. The rates of all other associated injuries were greatest in the complete SCI subpopulation, except for pelvic and lower extremity fractures, which were greatest in the incomplete SCI subpopulation. In the complete SCI subpopulation, SCIs were primarily in the lower cervical spine (levels C5–C7), with 72.2% of patients having injuries in this region only. In the incomplete and central SCI subpopulations, isolated lower cervical spinal cord injuries occurred in only 48.5% and 46.0% of patients, respectively.
Comparison of Time Interval to Surgical Intervention Between Subpopulations Figure 1 shows Kaplan-Meier curves of time to surgical intervention in the 3 cervical SCI subpopulations. The dashed line indicates 24 hours after injury; the STASCIS recommended time interval for early surgical intervention.5 At 24 hours, only 44.9% of all patients had undergone surgery, or 57.0% of the patients with complete SCI , 49.4% of the patients with incomplete SCI, and 28.9% of the patients with central SCI. Figure 2 shows the differences in total time to surgery between the SCI subpopulations. The vast majority of time spent between EMS dispatch and surgical intervention occurred during the inpatient stay, rather than with EMS or in the ED. Inpatient time accounted for 90.0%, 89.7%, and 91.9% of the total time to surgery in the complete, incomplete, and central SCI subpopulations, respectively. 994
Timing of Surgery for Cervical Spinal Cord Injury • Samuel et al
Mean time to surgery was longest for patients with central SCI (83.1 hr) compared with patients with complete (51.1 hr) and incomplete (55.5 hr) SCIs. After controlling for patient characteristics (age, sex, comorbidities) and injury characteristics (mechanism of injury, ISS, associated injuries) using multivariate analysis with Tukey post hoc pairwise comparisons, there were statistically significant differences in time to surgery between the patients with central SCI and the patients with both complete and incomplete SCIs (P < 0.001, P < 0.001). There was no significant difference in time to surgery between the patients with complete and incomplete SCIs (P = 0.956), largely due to similar total inpatient time prior to surgery (P = 0.744).
Analysis of Factors Associated With Time to Surgery Table 3 summarizes the statistically significant results of multivariate analyses of characteristics associated with time to surgery after cervical SCIs. Upper cervical SCI (levels C1–C4) and elevated CCI were found to be associated with increased time to surgery in all 3 subpopulations. Complete regression results from the 3 multivariate analyses can be found in Supplemental Digital Content Appendices 3 to 5, available at: http://links.lww.com/BRS/A973.
DISCUSSION In 2012, STACSIS provided evidence that patients with cervical SCI surgically treated within the first 24 hours after injury had superior neurological outcomes to those treated later.5 Although there are inherent limitations to this prospective cohort study, the results do support other large retrospective studies10,11 and the current practices of the majority of spine surgeons.8,9 STASCIS does not, however, distinguish between subtypes of SCIs. It has been suggested that the greatest potential benefit of early surgery may be in the incomplete SCI subpopulation.11 However, no data exist identifying where delays to early surgery occur and what factors may be associated with greater time to surgery in the different SCI subpopulations. In the context of this knowledge gap, the current study examined 2636 patients with cervical SCI from the NTDB RDS. The results indicate that the majority of patients do not undergo surgery within 24 hours of injury. With the most evidence and surgeon agreement supporting early surgery after incomplete SCI,9,11 only 49.4% of these patients had undergone surgery within 24 hours. The longer time to surgery seen in the central SCI subpopulation is consistent with the lack of evidence supporting early surgery in this subpopulation.13,23 In all 3 subpopulations of patients with cervical SCI, the vast majority of time between injury and surgery occurs during the inpatient hospitalization, rather than in the ED or during EMS response and transport. Therefore, it is unlikely that delays in ED or EMS management substantially contribute to delayed surgical management. Instead, specific delays after inpatient admission, which have yet to be specifically identified, contribute to later surgery. Possible delays may include imaging, need for medical stabilization, and management of other more emergent injuries.
www.spinejournal.com
July 2015
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. SPINE141418_LR 994
09/06/15 7:12 AM
CERVICAL SPINE
Timing of Surgery for Cervical Spinal Cord Injury • Samuel et al
TABLE 1. Patient Demographics and Comorbidities by Spinal Cord Injury Type Complete Spinal Cord Injuries N = 2636 Total
Incomplete Spinal Cord Injuries
Central Spinal Cord Injuries
Frequency
Percent
Frequency
Percent
Frequency
Percent
803
30.5%
950
36.0%
883
33.5%
P 80
21
2.6%
54
5.7%
59
6.7%
Female
162
20.2%
228
24.0%
194
22.0%
Male
641
79.8%
722
76.0%
689
78.0%
Sex
0.156
5
47
5.9%
78
8.2%
120
13.6%
134
16.7%
167
17.6%
190
21.5%
0.023
Cancer
4
0.5%
3
0.3%
11
1.3%
0.040
CVA with deficit
10
1.3%
17
1.8%
21
2.4%
0.220
Congestive heart failure
12
1.5%
21
2.2%
20
2.3%
0.456
Coronary artery disease
7
0.9%
10
1.1%
24
2.7%
0.003
Dementia
2
0.3%
5
0.5%
5
0.6%
0.577
Diabetes mellitus
74
9.2%
107
11.3%
154
17.4%
CERVICAL SPINE
Analysis of Delays to Surgery for Cervical Spinal Cord Injuries Andre M. Samuel, BBA, Daniel D. Bohl, MPH, Bryce A. Basques, BS, Pablo J. Diaz-Collado, MD, Adam M. Lukasiewicz, MSc, Matthew L. Webb, AB, and Jonathan N. Grauer, MD
Study Design. A retrospective study of surgically treated patients with cervical spinal cord injury (SCI) from the National Trauma Data Bank Research Data Set. Objective. To determine how time to surgery differs between SCI subtypes, where delays before surgery occur, and what factors are associated with delays. Summary of Background Data. Studies have shown that patients with cervical SCI undergoing surgery within 24 hours after injury have superior neurological outcomes to patients undergoing later surgery, with most evidence coming from the incomplete SCI subpopulation. Methods. Surgically treated patients with cervical SCI from 2011 and 2012 were identified in National Trauma Data Bank Research Data Set and divided into subpopulations of complete, central, and other incomplete SCIs. Relationships between surgical timing and patient and injury characteristics were analyzed using multivariate regression. Results. A total of 2636 patients with cervical SCI were identified: 803 with complete SCI, 950 with incomplete SCI, and 883 with central SCI. The average time to surgery was 51.1 hours for patients with complete SCI, 55.3 hours for patients with incomplete SCI, and 83.1 hours for patients with central SCI. Only 44% of patients with SCI underwent surgery within the first 24 hours after injury, including only 49% of patients with incomplete SCI. The vast majority of time between injury and surgery was after admission, rather than in the emergency department or in the field. Upper cervical SCIs and greater Charlson Comorbidity Index were associated with later surgery in all 3 SCI subpopulations. Conclusion. The majority of patients with SCI do not undergo surgery within the first 24 hours after injury, and the majority of From the Department of Orthopaedics and Rehabilitation, Yale School of Medicine, New Haven, CT. Acknowledgment date: October 29, 2014. Revision date: January 21, 2015. Second revision date: February 24, 2015. Acceptance date: March 10, 2015. The manuscript submitted does not contain information about medical device(s)/drug(s). No funds were received in support of this work. Relevant financial activities outside the submitted work: consultancy, expert testimony, grants. Address correspondence and reprint requests to Jonathan N. Grauer, MD, Department of Orthopaedics and Rehabilitation, Yale School of Medicine, 800 Howard Ave, New Haven, CT 06510; E-mail: [email protected] DOI: 10.1097/BRS.0000000000000883
992
delays occur after inpatient admission. Factors associated with these delays highlight areas of focus for expediting care in these patient populations. Key words: spinal cord injury, cervical, complete spinal cord injury, incomplete spinal cord injury, central spinal cord injury, surgical timing, decompression, comorbidities, upper cervical spinal cord injuries, inpatient delay. Level of Evidence: 4 Spine 2015;40:992–1000
T
here is growing interest in the ideal timing of surgery after cervical spinal cord injury (SCI). Early clinical studies of the effect of surgical timing on outcomes showed mixed results,1–4 likely due to inadequate sample sizes and the logistical challenges of substantially shortening the time to surgery. More recently, however, the Surgical Timing in Acute Spinal Cord Injury Study (STASCIS), a large prospective multicenter study of 313 patients published in 2012, was able to demonstrate neurological benefits with early surgical intervention.5 STASCIS showed that, compared with later surgery, surgery within 24 hours after injury was associated with a higher rate of improvement of 2 or more grades on the American Spinal Injury Association Impairment Scale at 6-month follow-up. As STASCIS was a prospective cohort study, and not a randomized controlled trial, there are limits to the conclusions that can be drawn from the results. As an example, significant differences in the baseline characteristics of the experimental groups and heterogeneity regarding SCI types in the study have been highlighted recently.6 However, the difficulties of conducting a randomized prospective study of SCIs are clear,7 and the results of STASCIS seem to confirm prevailing opinions of surgeons regarding optimal treatment of acute SCIs.8 In a survey of more than 900 spine surgeons, more than 80% preferred to decompress acute SCIs within the first 24 hours after injury.9 Recently, the results of STASCIS were also corroborated in a larger retrospective study of 1410 patients.10 In addition, a metanalysis of 1687 patients showed improved outcomes after early surgical decompression compared with both late decompression and nonoperative treatment.11 Cervical SCIs, however, present in a variety of different injury patterns, which each may respond differently to early
www.spinejournal.com
July 2015
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. SPINE141418_LR 992
09/06/15 7:12 AM
CERVICAL SPINE surgery. With incomplete SCIs, there is the greatest evidence for improved neurological function with early surgery.11 Therefore, for these patients, surgery is traditionally thought to be more urgent than that for patients with other SCI. With complete SCIs, on the contrary, there is disagreement among surgeons concerning the benefits of rapid surgery.9 Central SCIs, a subset of incomplete SCIs that often occur in older patients with pre-existing stenosis, are commonly managed with delayed surgical intervention, allowing time for hemodynamic stabilization and reduction of spinal cord edema.9,12–14 At the very least, there seems to be some benefit with early surgical intervention after incomplete SCIs (not including central SCIs). Early surgery also has value for other reasons such as decreased hospital costs and length of stay15 and decreased hospital morbidities.16,17 However, no studies to date have considered where delays may occur prior to surgery for these patients and what factors may be associated with delays. In addition, there has been little research specifically comparing current surgical timing after the various types of SCIs. The current study aims to investigate these 3 questions using data from the American College of Surgeons National Trauma Data Bank Research Data Set (NTDB RDS). This national data set has the large sample size necessary to appropriately power a study of the subtypes of this rare injury pattern. NTDB RDS also includes the preadmission data necessary to analyze the entire time interval from initial injury to surgery.
MATERIALS AND METHODS Data Source A retrospective cohort study was performed using the NTDB RDS, the largest national database of trauma patients containing data from more than 900 centers in the United States. The database contains both administratively coded and chartabstracted data on more than 100 different variables. Data from 2011 and 2012 were used. This study was approved by the local institutional human investigations committee.
Study Population International Classification of Diseases, Ninth Revision (ICD-9) diagnosis and procedure codes were used to identify all patients with cervical SCIs undergoing surgery during their hospitalization. The population was further limited to patients for whom the SCI could be identified as a complete, incomplete (excluding central SCIs), or central SCI using ICD-9 diagnosis codes and patients with complete time interval information (from ambulance dispatch to surgical intervention). Supplemental Digital Content Appendix 1, available at: http://links.lww.com/BRS/A973, lists all ICD-9 codes used to identify the study population.
Data Collection Patient characteristics and injury characteristics were assessed using NTDB RDS variables. Sex, age, insurance status, comorbidities, and presentation to an academic facility were directly Spine
Timing of Surgery for Cervical Spinal Cord Injury • Samuel et al
extracted from the database. Comorbidities are reported in NTDB RDS with specific comorbidity codes. The comorbidities analyzed were alcoholism, cancer, prior cerebrovascular accident, congestive heart failure, coronary artery disease, dementia, diabetes mellitus, functionally dependent status, hypertension requiring medication, obesity, renal failure, and respiratory disease. A modified Charlson Comorbidity Index (CCI) was then calculated.18 Modified CCIs have comparable predictive value to the original CCI19 and have been used previously with national databases studies.20 Computation of the modified CCI is detailed in Supplemental Digital Content Appendix 2, available at: http://links.lww.com/BRS/A973. Injury characteristics assessed included mechanism of injury, hospital-assigned injury severity score (ISS), and the presence of specific associated injury diagnoses based on ICD-9 coding. These codes can be found in Supplemental Digital Content Appendix 1, available at: http://links.lww.com/BRS/A973. Time intervals included in NTDB RDS include prehospital time from emergency medical services (EMS) dispatch to arrival at emergency department (ED), duration in ED, and duration between admission and surgery. Total time to surgery from EMS dispatch was computed using these intervals.
Analysis Three sets of analyses were conducted in this study. First, the 3 subpopulations of patients with cervical SCIs were described on the basis of patient characteristics and injury characteristics. To account for nonlinear relationships, continuous variables were converted to categorical variables, and differences between subpopulations were compared using Pearson χ2 test. Second, the total time to surgery was broken down into EMS time, ED time, and inpatient time to determine the percentage of time spent in each setting prior to surgery. The total time to surgery was then compared between the 3 subpopulations using multivariate analysis of variance with Tukey post hoc pairwise comparisons. Because the distribution of total time to surgery had substantial positive skew, this variable was log transformed to approximate a normal distribution in the analysis of variance. Third, patient and injury characteristics that are associated with increased time to surgery were determined using multivariate linear regression. Time to surgery was again log transformed as before. This analysis was performed for each of the 3 subpopulations separately. The variables included in these analyses were sex, CCI, mechanism of injury, presentation to academic trauma center, level of cervical SCI, and presence of 9 individual associated injuries. Age was not included in the analysis because it is already included in calculation of CCI and likely collinear. ISS was not included in the analysis as it is likely collinear with the individual-associated injuries of specific interest in the study (i.e., upper cervical SCIs, injuries, head injury, pelvic fracture, etc.). All statistical tests were 2-tailed and controlled for institutional clustering using linearized standard errors. Because of multiple hypothesis testing in this study, false discovery rate control21 was used to avoid type 1 error. A false discovery rate www.spinejournal.com
993
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. SPINE141418_LR 993
09/06/15 7:12 AM
CERVICAL SPINE of d = 0.05 was set and only P value of less than 0.017 was taken as significant.
RESULTS NTDB RDS contained 2636 patients with cervical SCIs who met the inclusion criteria in admission years 2011 and 2012. Of these, 803 (30.5%) had complete SCIs, 950 (36.0%) had incomplete SCIs (excluding central SCIs), and 883 (33.5%) had central SCIs.
Comparison of SCI Subpopulations Table 1 summarizes patient characteristics of the 3 SCI subpopulations. In all 3 subpopulations, the majority of patients were male. The central SCI subpopulation was older than the complete and incomplete SCI subpopulations and had higher rates of all pre-existing comorbidities, except for obesity, which was highest in the complete SCI subpopulation. Table 2 summarizes injury characteristics of the cervical SCI subpopulations. The patients with central SCI primarily experienced falls, whereas the most common mechanism of injury in the complete and incomplete SCI subpopulations was automobile accidents. ISS, which is predictive of mortality in trauma patients,22 was highest in the complete SCI subpopulation compared with both the incomplete and central SCI subpopulations (P < 0.001). Head injuries (intracranial injuries and skull fractures) were the most commonly associated injuries in all 3 subpopulations, occurring in 28.0% to 29.8% of patients. The rates of all other associated injuries were greatest in the complete SCI subpopulation, except for pelvic and lower extremity fractures, which were greatest in the incomplete SCI subpopulation. In the complete SCI subpopulation, SCIs were primarily in the lower cervical spine (levels C5–C7), with 72.2% of patients having injuries in this region only. In the incomplete and central SCI subpopulations, isolated lower cervical spinal cord injuries occurred in only 48.5% and 46.0% of patients, respectively.
Comparison of Time Interval to Surgical Intervention Between Subpopulations Figure 1 shows Kaplan-Meier curves of time to surgical intervention in the 3 cervical SCI subpopulations. The dashed line indicates 24 hours after injury; the STASCIS recommended time interval for early surgical intervention.5 At 24 hours, only 44.9% of all patients had undergone surgery, or 57.0% of the patients with complete SCI , 49.4% of the patients with incomplete SCI, and 28.9% of the patients with central SCI. Figure 2 shows the differences in total time to surgery between the SCI subpopulations. The vast majority of time spent between EMS dispatch and surgical intervention occurred during the inpatient stay, rather than with EMS or in the ED. Inpatient time accounted for 90.0%, 89.7%, and 91.9% of the total time to surgery in the complete, incomplete, and central SCI subpopulations, respectively. 994
Timing of Surgery for Cervical Spinal Cord Injury • Samuel et al
Mean time to surgery was longest for patients with central SCI (83.1 hr) compared with patients with complete (51.1 hr) and incomplete (55.5 hr) SCIs. After controlling for patient characteristics (age, sex, comorbidities) and injury characteristics (mechanism of injury, ISS, associated injuries) using multivariate analysis with Tukey post hoc pairwise comparisons, there were statistically significant differences in time to surgery between the patients with central SCI and the patients with both complete and incomplete SCIs (P < 0.001, P < 0.001). There was no significant difference in time to surgery between the patients with complete and incomplete SCIs (P = 0.956), largely due to similar total inpatient time prior to surgery (P = 0.744).
Analysis of Factors Associated With Time to Surgery Table 3 summarizes the statistically significant results of multivariate analyses of characteristics associated with time to surgery after cervical SCIs. Upper cervical SCI (levels C1–C4) and elevated CCI were found to be associated with increased time to surgery in all 3 subpopulations. Complete regression results from the 3 multivariate analyses can be found in Supplemental Digital Content Appendices 3 to 5, available at: http://links.lww.com/BRS/A973.
DISCUSSION In 2012, STACSIS provided evidence that patients with cervical SCI surgically treated within the first 24 hours after injury had superior neurological outcomes to those treated later.5 Although there are inherent limitations to this prospective cohort study, the results do support other large retrospective studies10,11 and the current practices of the majority of spine surgeons.8,9 STASCIS does not, however, distinguish between subtypes of SCIs. It has been suggested that the greatest potential benefit of early surgery may be in the incomplete SCI subpopulation.11 However, no data exist identifying where delays to early surgery occur and what factors may be associated with greater time to surgery in the different SCI subpopulations. In the context of this knowledge gap, the current study examined 2636 patients with cervical SCI from the NTDB RDS. The results indicate that the majority of patients do not undergo surgery within 24 hours of injury. With the most evidence and surgeon agreement supporting early surgery after incomplete SCI,9,11 only 49.4% of these patients had undergone surgery within 24 hours. The longer time to surgery seen in the central SCI subpopulation is consistent with the lack of evidence supporting early surgery in this subpopulation.13,23 In all 3 subpopulations of patients with cervical SCI, the vast majority of time between injury and surgery occurs during the inpatient hospitalization, rather than in the ED or during EMS response and transport. Therefore, it is unlikely that delays in ED or EMS management substantially contribute to delayed surgical management. Instead, specific delays after inpatient admission, which have yet to be specifically identified, contribute to later surgery. Possible delays may include imaging, need for medical stabilization, and management of other more emergent injuries.
www.spinejournal.com
July 2015
Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. SPINE141418_LR 994
09/06/15 7:12 AM
CERVICAL SPINE
Timing of Surgery for Cervical Spinal Cord Injury • Samuel et al
TABLE 1. Patient Demographics and Comorbidities by Spinal Cord Injury Type Complete Spinal Cord Injuries N = 2636 Total
Incomplete Spinal Cord Injuries
Central Spinal Cord Injuries
Frequency
Percent
Frequency
Percent
Frequency
Percent
803
30.5%
950
36.0%
883
33.5%
P 80
21
2.6%
54
5.7%
59
6.7%
Female
162
20.2%
228
24.0%
194
22.0%
Male
641
79.8%
722
76.0%
689
78.0%
Sex
0.156
5
47
5.9%
78
8.2%
120
13.6%
134
16.7%
167
17.6%
190
21.5%
0.023
Cancer
4
0.5%
3
0.3%
11
1.3%
0.040
CVA with deficit
10
1.3%
17
1.8%
21
2.4%
0.220
Congestive heart failure
12
1.5%
21
2.2%
20
2.3%
0.456
Coronary artery disease
7
0.9%
10
1.1%
24
2.7%
0.003
Dementia
2
0.3%
5
0.5%
5
0.6%
0.577
Diabetes mellitus
74
9.2%
107
11.3%
154
17.4%
