Defining the Pathway to Definitive Care and Surgical Decompression after Traumatic Spinal Cord Injury: Results of a Canadian Population-Based Cohort Study

Introduction

Over the past few decades, there has been increased recognition that timely treatment of patients with acute traumatic spinal cord injury (SCI) may lead to improved clinical outcomes. ^{1 –3} Pathologically, this is related to the fact that after the primary cord injury, secondary mechanisms expand the zone of neural tissue damage in a time-dependent fashion. ^4,5 Mitigating these secondary mechanisms after SCI has the potential to limit the volume of tissue injury and improve clinical outcomes. ^{6 –8} Examples of studied acute therapies shown effective at improving outcomes by mitigating secondary injury events include methylprednisolone sodium succinate when administered in the first 8 h post-injury and decompressive surgery performed within the first 24 h post-injury. ^{9 –11} Other promising acute phase therapies currently in the early phases of clinical study include the sodium/glutamate blocking drug riluzole, the anti-inflammatory drug minocycline, and systemic hypothermia. ^{12 –14} Lastly, early optimization of cardiorespiratory and hemodynamic parameters, as well as management within an acute spinal cord injury unit, have been associated with diminished morbidity and mortality, in addition to improved neurological outcomes at follow-up. ^{15 –18}

Importantly, delays in the transfer of patients with SCI to health centers specialized in the management of spinal trauma and SCI represent a critical barrier to the delivery of optimal acute phase medical and surgical management. Previous studies from both North America and Europe have estimated that only 24–51% of SCI patients are transferred to such centers and are eligible for surgical decompression within 24 h of injury. ^19,20 This implies that a substantial proportion of injured patients arrive at a specialized center outside the critical acute phase therapeutic window and hence are deprived of timely therapy that could potentially facilitate enhanced outcomes. From a research perspective, such delays also have implications for future clinical trials studying acute phase therapeutics. If patients are arriving at the hospital after the time-window for the investigational intervention has elapsed, study enrollment may be limited and the external validity of eventual study findings may be compromised. In order to prevent such occurrences, the pre-eminent goal would be to study and modify the existing logistics to allow a streamlined path to a specialized acute care center for those patients suffering an acute SCI. However, in order to accomplish this goal, it is first necessary to understand at a population level the current pathway of patients in the immediate post-injury period and in the process, to identify obstacles preventing expedient acute medical and surgical care.

Based on this background, our primary objective was to complete a population-based cohort study in order to: 1) characterize and quantify patients' pathway to definitive care and to decompressive surgery post-SCI; and 2) to identify patient-specific and system factors that may pose barriers to expeditious care along the course of this pathway.

Methods

A population-based retrospective cohort study for the province of Ontario was undertaken using administrative health data accessed through the Institute for Clinical Evaluative Sciences at the University of Toronto. Specific datasets utilized included: 1) the Discharge Abstract Database, which contains hospital discharge records from across Canada, including diagnostic codes under the International Classification of Diseases and Related Health Problems, 10th Revision, Canada (ICD-10-CA); 2) the National Ambulatory Care Resource System database, which contains data for hospital-based and community-based emergency and ambulatory care; and 3) The Ontario Registered Persons Database, which contains information regarding age, sex, residence postal code, and vital statistics, including deaths, for Ontario residents.

The use of these datasets allows for capture of all SCI events within the province of Ontario, providing a comprehensive population-based evaluation of the study question. Using de-identified patient information, these data sets were linked to serve as the basis for subsequent cohort creation and analysis. University and institutional research ethics approval was received prior to study commencement.

The cohort of interest was defined as all adult patients with incident traumatic SCI treated with surgery and presenting for the first time to an Ontario health care center between 2002 and 2011. Patients with SCI were identified through the use of SCI-specific ICD-10-CA codes (Appendix). Patients younger than 16 years of age, those with non-traumatic SCI etiology (including a vascular, degenerative, or neoplastic cause of SCI), those with injuries to the cauda equina or penetrating SCI (including those with stab or missile related SCI), and those with missing information on gender were excluded.

For purposes of standardizing terminology for analyses, the site of definitive care was defined as a hospital admission to a tertiary or quaternary health care center with: 1) an intensive care unit familiar with the monitoring and medical management of patients with acute SCI according to the most recent American Association of Neurological Surgeons/Congress of Neurological Surgeons guidelines for the management of acute SCI, ^{21 –24} and 2) a spine surgery service familiar with the surgical management of acute spine trauma and SCI. Based on these criteria, a total of 12 definitive care sites were identified in the province of Ontario. Following initial injury, the number of stops en route to one of these definitive care sites fell into one of two general patterns: 1) patient transported directly from site of injury to the site of definitive care (includes time spent at an on-site emergency department) or 2) patient transported from site of injury to one or more intermediate health centers for diagnosis and/or medical stabilization prior to transfer to definitive care. Since little prehospital data was available, patients' first presentation to a health center defined time zero when determining time to definitive care and time to surgery. Further, patients with time to definitive hospital or time to surgery exceeding 336 h were considered outliers and were deleted from the final analysis. Final analysis was conducted on a total cohort of 1111 patients with SCI who were treated with spinal surgery from 2002 to 2011. Information on timing of surgery was available only in administrative data from 2009, thus analyses examining surgical time included 369 patients with traumatic SCI who were admitted for spinal surgery from 2009 to 2011.

Results

Over the 10-year study period (2002–2011), a total of 1111 patients met eligibility criteria and were included in the cohort. Given that surgical timing data was collected from 2009 onwards, analyses related to this outcome are based on a smaller cohort of patients (n = 369) who sustained injury in the 3-year period between 2009 and 2011. The overall characteristics of the cohort, as well as description of the number of patients included from each year, are summarized in Table 1, with the later cohort of patients with surgical timing data summarized in Table 2. Of patients included, 64 (5.8%) died during acute hospital admission, with the remaining patients transferred either to rehabilitation/long-term care or to home with home-care services.

During the study period, the mean and median (interquartile range) times to arrival at the site of definitive care were 8.1 ± 25.5 and 3 (1–6) h, respectively. A total of 921 patients (82.9%) were transported directly from the site of injury to the site of definitive care, while 190 patients (17.1%) were transferred to at least one intermediate health center before transfer to the site of definitive care. The mean and median times to surgery were 49.4 ± 65.0 and 21 (9–57) h, respectively. Table 3 displays the frequency and percentage of patients arriving at definitive care for the entire cohort (Table 3A) and to surgery (Table 3B) by time interval in subgroups where data were available (2009–2011). With respect to surgery, while 197 patients (53.3%) underwent surgery prior to 24 h, 172 patients (46.6%) underwent surgery beyond this cut-off point. Further, only 34.2% reached surgery within 12 h of arrival at the site of definitive care, despite the vast majority of patients (88.4%) presenting within 6 h.

Bivariate negative binomial regression analyses demonstrated that a total of 10 predictor variables were associated with increased time to arrival at definitive care (p ≤ 0.20), namely increased age, male sex, small urban versus large urban center of presentation, lower ISS, increased number of imaging investigations, increased number of stops at intermediate health care centers, complete versus incomplete injuries, increased comorbidity index, motor vehicle accident etiology versus fall etiology, and pedestrian struck etiology vs motor vehicle accident etiology. When included in a multi-variate model, four of the covariates were statistically significant—specifically, older age (incidence rate ratio [IRR] = 1.01; 95% CI: 1.01, 1.02), increased number of stops at intermediate health care centers (IRR = 7.70; 95% CI: 7.54, 7.86), higher comorbidity index (IRR = 1.43; 95% CI: 1.14, 1.72), and fall-related SCI etiology (IRR = 1.16; 95% CI: 1.02, 1.29) were associated with increased time to arrival at definitive care (Table 4).

Examining time to surgery, bivariate negative binomial regression models demonstrated that increased age, female sex, cervical level of injury, increased number of stops at intermediate health centers, and fall injury etiology versus motor vehicle accident etiology predicted increased time to surgery at p ≤ 0.20. In the multi-variate analyses, only increased number of stops at intermediate health centers remained as a significant predictor of time to surgery (IRR = 1.55; 95% CI: 1.23, 1.87; Table 5), suggesting that each additional stop is associated with increased time to surgery. Logistic regression analyses examining time to surgery represented according to the 24 h cut-off showed that three variables (older age, increased number of stops and cervical level of injury) were significantly associated with late surgery in the bivariate analysis; however, only increased age (odds ratio [OR] 1.02; 95% CI: 1.01, 1.03) and increased number of stops at intermediate health centers (OR = 2.48; 95% CI: 1.35, 4.56) remained significantly associated with increased odds of late surgery in the multi-variate analysis (Table 6).

Discussion

The main objectives of this large population-based study were to characterize and quantify patients' pathway to definitive care and to decompressive surgery post-SCI, as well as to identify patient-specific and system factors that may pose barriers to expeditious care along the course of this pathway. The findings presented here indicate that in the province of Ontario (13 million inhabitants), the majority of patients with SCI (82.9%) present directly to the site of definitive care after injury. In addition to stops at intermediate health centers, the multi-variate analysis demonstrated that patient-specific factors, including older age, increased presence of comorbid illness and fall-related SCI etiology, were independently associated with delays in transport to definitive care. When considering time to surgery, substantial delays were seen, with nearly 47% of the study population undergoing surgery more than 24 h after initial presentation to a health center. When considering time to surgery as a binary variable (according to the 24 h cut-off to define early vs. late), increased stops at intermediate health centers and increased age were associated with delays to surgery.

With respect to patient disposition within this 10-year series, it is worth noting that the observed in-patient mortality of 5.8% is consistent with previous literature, across which acute mortality has ranged from 4 to 16%. ^29,30 For those patients who survived acute hospital admission, the majority (76.2%) were transferred to long-term care or rehabilitation facilities. This reflects the fact that in Ontario, many patients undergo a period of SCI inpatient rehabilitation prior to community reintegration. We speculate that the approximately 15% of patients discharged home represent milder forms of SCI or central cord syndrome who received outpatient physiotherapy.

Across the literature, although a number of studies have evaluated the impact of early treatment and surgery on clinical outcomes, few have specifically addressed time to definitive care and surgery, with an overarching aim of identifying barriers to expedient care. In the surgical timing in Surgical Timing in Acute Spinal Cord Injury Study published by Fehlings and colleagues in 2012, 182 of 313 subjects enrolled (58%) underwent surgery prior to 24 h with the remaining 131 subjects (42%) undergoing late surgery. ⁹ In the univariate analysis comparing characteristics between those undergoing late and early surgery, the authors found that patients in the late group had a significantly older mean age and had significantly higher proportion of less severe injuries (American Spinal Injury Association Impairment Scale grade C and D), compared with the early surgery group. In a separate Canadian-based prospective cohort study investigating the impact of surgical timing on clinical outcomes post-SCI in 84 patients, a univariate analysis also demonstrated a tendency towards older subjects with less severe degree of injury in the late surgery group (> 24 h), compared with the early (> 24 h), surgery group. ²⁶

The current study supports the association of increased age with delays to surgical treatment; however, no significant association was seen between severity of neurological injury and surgical timing. It is worth noting, however, that we relied on a crude measure of injury severity in the current analysis (complete vs. incomplete lesion); if detailed neurological examination information according to the ISNCSCI recommendations were available and used in the current analysis, perhaps results may have differed. In a 2013 publication, Furlan and colleagues performed a process bench-marking appraisal of spinal decompression for cervical SCI, with an overall goal of identifying opportunities to reduce delays in surgical management. ³¹ In contrast to the current analysis, this study, involving 63 patients, reported no clinical or statistically significant differences in patient characteristics (intrinsic variables) between those who underwent early versus later surgery. However, it was noted that patients undergoing late surgery had a longer duration of wait time at intermediate general hospitals and a longer wait time to see a surgeon, as well as for a surgical decision, once having arrived at the site of definitive care. These factors were all considered as extrinsic or system related variables contributing to delays in care. Although we did not have data points related to hospital-specific wait times, we did find that one extrinsic variable, additional stops at intermediate health centers, did delay time to definitive care and to surgical intervention.

When considering the current findings from a policy perspective, several observations are relevant. First, when considering both time to the site of definitive care and time to surgery, it is apparent that stops at intermediate health centers are responsible for delays. When searching for opportunities to expedite and streamline the existing pathway for patients with SCI to definitive care and surgery, this represents an obvious system-related factor potentially amenable to modification. In the context of ischemic stroke, given that early thrombolysis has shown to improve clinical outcomes, many health regions have instituted a prehospital policy wherein individuals demonstrating symptoms/signs of stroke are routed to a designated stroke center, bypassing other intervening non-stroke specialized hospitals. ³² In the realm of traumatic SCI, given that the preponderance of existing and emerging evidence supports the beneficial impact of early medical and surgical therapy on clinical outcomes, perhaps a similar prehospital policy of routing injured patients directly to the site of definitive care should be considered for implementation.

Second, it also is apparent through these analyses that increased age is associated with delays to definitive care and time to surgery. This observation may be explained at least in part by the changing epidemiology of SCI. With the overall aging of the population, a greater proportion of injuries are being recognized among the elderly, typically seen as cervical cord injuries in the setting of low energy mechanisms, such as a falls, in patients with pre-existing cervical stenosis. ^33,34 This SCI phenotype is substantially different than the young SCI patient with multi-system trauma secondary to a high energy mechanism, such as a motor vehicle collision, and therefore may be less likely to raise suspicion of a SCI. As a result, it is possible that elderly patients with SCI are not triggering the same trauma networks as younger patients with SCI, leading to delays.

As evidence of this, Middleton and colleagues have previously reported in an Australian study that patients who sustained their SCI as the result of a low fall were older and less likely to have their SCI identified and treated early, with less than half of this group reaching a specialist SCI center within 24 h in comparison to other patients with SCI. ³⁵ In addition, among elderly patients, comorbid illness and the possibility of other competing diagnoses may cloud the diagnostic picture resulting in delayed diagnosis and work-up. These possibilities speak to the need for enhanced education of primary care doctors, paramedic and emergency medicine personnel surrounding SCI among the elderly, reinforcing the importance of early diagnosis and triage to a site of definitive care. Third, at present, while 12 sites of definitive care were identified in the province, no clear unifying algorithm directing the surgical care of patients with SCI currently exists. In order to streamline the care of these patients, with an overall goal of improving patient flow and clinical outcomes, it is evident that efforts are needed to develop best practice guidelines for the surgical management of SCI, including timelines of trauma/neurological assessment and radiological investigation to facilitate surgical decision-making, that can be adopted into policy development not only in Ontario, but also internationally.

Conclusion

After first presentation to a health center in the province of Ontario post-SCI, this study has shown that while the majority of patients arrive at the site of definitive care in a timely fashion, only approximately half of patients undergo surgical decompression of the spinal cord within 24 h. While older age, increased presence of comorbid illness, stops at intermediate health centers, and fall-related SCI etiology were independently associated with delays in transport to definitive care, only increased age and stops at intermediate health centers were associated with delays to surgery. These results should help to inform policy, to allow for the creation of a streamlined path to specialized acute care, and to provide urgent surgical treatment for those patients suffering an acute SCI, leading to improved neurological outcomes and reduced secondary complications and mortality.