The Functional Outcome of Patients Following Spinal Injury in Asia Metropolitans: Does the Prehospital Spine Immobilization Matter?

Prehospital spinal immobilization is a widely used procedure in the emergency medical service (EMS) system worldwide, while the incidence of patients with spinal injury (SI) is relatively low, and unnecessary prehospital spinal immobilization is associated with patient complications. This study aimed to determine the association between prehospital spine immobilization and favorable functional outcomes at hospital discharge among trauma patients with SI. We conducted a retrospective cohort study using the Pan-Asia Trauma Outcomes Study (PATOS) registry data from January 1, 2016, to November 30, 2018. A total of 759 patients with SI were enrolled from 43,752 trauma patients in the PATOS registry during the study period. The subjects had a median age of 58 years ( Q1−Q3,41−72), and 438 (57.7%) patients had prehospital spine immobilization. Overall, prehospital spinal immobilization was not associated with favorable functional outcomes at discharge in multivariable logistic regression (aOR, 1.06; 95%CI 0.62−1.81, p = 0.826). However, in the subgroup of cervical SI, prehospital spinal immobilization was associated with favorable functional outcomes at discharge (aOR, 3.14; 95%CI 1.04−9.50; p = 0.043). Therefore, we suggest that paramedics should be more careful when determining cervical SI and using prehospital immobilization. in our study were prehospital immobilization, which was dened as the neck collar and/or backboard used. The basic characteristics of the patients in our study included country, age, sex, mechanism of injury (trac, fall, others), location of SI (cervical, thoracic, or lumbar spine), and torso injury. Torso injury was dened as injury involving the chest or abdomen, including fractures of the clavicle, rib, pelvis, traumatic pneumothorax, hemothorax, intra-abdominal bleeding, laceration, or contusion of the spleen, liver, or bowel. Data on prehospital management, including rescue


Introduction
Patients with spinal injuries (SIs) following trauma are at risk of spinal cord injuries (SCIs) with severe neurological consequences and disability in life, which are estimated in approximately 20% of these cases 1 . However, the incidence of traumatic SCI differs from region to region, with a relatively low incidence of 16 per million in Western Europe to 40 per million in the United States 2 . The incidence of SI in the trauma population also varied from 4.58% in China to all trauma patients 3 and 9.6% in Europe in patients following major trauma 4 , while some studies reported only 1−2% of all trauma patients 5 .
Despite the relatively low incidence of SI and SCI, prehospital spinal immobilization has been widely adopted worldwide as well as in trauma courses, such as prehospital trauma life support (PHTLS) and advanced trauma life support (ATLS) 6,7 . It aims to minimize further movement of the spine and reduce the risk of secondary injury, but the procedure lacks high-quality evidence of clear bene ts in decreasing disability 8, 9 and is associated with complications such as respiratory restriction 10,11 , elevated intracranial pressure 12,13 , pressure ulcers 14 and changing the results of physical examination 15 while patients are immobilized. In addition, previous studies also revealed that prehospital immobilization may be associated with higher mortality in patients with penetrating trauma and gun-shot wounds [16][17][18] .
Due to the growing concerns regarding prehospital spinal immobilization, this study aimed to determine the association between prehospital spinal immobilization and favorable functional outcomes at hospital discharge among adult trauma patients with SI.

Study design and setting
We conducted a retrospective cohort study using a prospectively collected database from the Pan-Asia Trauma Outcomes Study (PATOS), which was a cross-national trauma registry network initiated in 2015 and consists of 33 participating sites from 14 Asian countries, including Australia, Hong Kong, India, Indonesia, Japan, Korea, Malaysia, Philippines, Singapore, Sri Lanka, Taiwan, Thailand, the United Arab Emirates, and Vietnam. Urban areas are covered by all countries 19 . The variables of PATOS include demographics, injury epidemiology, prehospital care, emergency department (ED) and hospital care, injury severity, and clinical outcomes 20 . The PATOS Trauma Database was characterized as an emergency medical service (EMS)based registry. Participation in the PATOS registry was voluntary. Patient's data were recorded in the registry if they were sent to the participating hospitals due to trauma, either from the scene or via interhospital transport.

De nition of exposure
The key exposures in our study were prehospital immobilization, which was de ned as the neck collar and/or backboard used. The basic characteristics of the patients in our study included country, age, sex, mechanism of injury (tra c, fall, others), location of SI (cervical, thoracic, or lumbar spine), and torso injury. Torso injury was de ned as injury involving the chest or abdomen, including fractures of the clavicle, rib, pelvis, traumatic pneumothorax, hemothorax, intra-abdominal bleeding, laceration, or contusion of the spleen, liver, or bowel. Data on prehospital management, including spinal immobilization, rescue airway (supraglottic airway or endotracheal tube), the establishment of uid access either by intravenous line (IV) or by intraosseous line (IO), and scene-to-hospital time (S to H time) were collected. Because one of the major contributing countries (Korea) mostly reported the S to H time rather than transport time, we used S to H time as variables to maximize the valid value. We used the RTS and ISS as indices of trauma severity. RTS is a physiological triage score using the Glasgow Coma Scale (GCS), systolic blood pressure (SBP), and respiratory rate (RR). The RTS formula was as follows: RTS = (GCS score coded × 0.9368) + (SBP coded × 0.7326) + (RR coded × 0.2908) 22 . ISS was calculated by summing the square of the three highest Abbreviated Injury Scale scores for injuries to different body regions 23 . We divided the ISS score into three groups: <9, minor injury; 9-15 de ned as moderate injury; and ≥16, severe injury 24,25 . We also collected data from patients who had undergone spine surgery or surgery of other body regions except for the spine. Missing data of the variables were excluded.

Outcome measurements
The primary outcome was the modi ed Rankin Scale (mRS) score at discharge. MRS was used to evaluate the functional outcome in patients with stroke at rst 26 and was then applied to measure the disability caused by TBI and general trauma [27][28][29][30] . No signi cant disability to moderate disability (mRS 0-3) was de ned as favorable functional outcome, and moderately severe disability to death (mRS 4-6) was de ned as poor functional outcomes 31 . Subgroup analyses were performed in different subgroups of age (<65 and ≥65 years), RTS (<7 and ≥7), ISS (<9, ≥9), and the location of SI (cervical, thoracic, and lumbar regions). We used 65 years old as the cut-off value because it represents the most commonly accepted age to consider a patient as an elderly patient and was used according to the guidelines of the Eastern Association for the Surgery of Trauma 32,33 . A score of 7 as the cut-off point in RTS and 9 as the cut-off point in ISS were used in previous studies for major trauma 25,34,35 .

Statistical methods
Continuous variables are reported as mean or median (Q1-Q3), as appropriate. Dichotomous and categorical variables are presented as numbers (percentages). Continuous variables were compared using the Mann-Whitney U test. Categorical and nominal variables were compared using Pearson's chi-square test or Fisher's exact test. We examined the association between prehospital spinal immobilization and favorable functional outcomes using univariate logistic regression and multivariable logistic regression. Variables with p < 0.05 on univariable logistic regression and the major variable (prehospital spinal immobilization) were selected for multivariable logistic regression analysis using the forced entry method. For subgroup analyses, we also conducted multiple logistic regression using the forced entry method for all variables. Statistical analysis was performed using SPSS version 25.0 (IBM, Armonk, NY, USA). All tests were two-sided, and a p-value of less than 0.05 was considered statistically signi cant.

Ethics approval and consent to participate
The PATOS collaboration was approved by the Institutional Review Board of the National Taiwan University Hospital and Far Eastern Memorial Hospital. This study was conducted in accordance with the Declaration of Helsinki (as revised in 2013) and the study proposal and methods were approved by the PATOS Taipei Meeting. The data were anonymized before being released to the authors in 2019. Institutional Review Board of Far Eastern Memorial Hospital waived the need for Informed consent due to the retrospective nature of this study.

Characteristics of study objects
From January 1, 2016, to November 30, 2018, a total of 1,573 cases with SIs were extracted from the PATOS registry of 43,752 EMS-transported trauma patients (3.5% of all cases) using ICD-9 or ICD-10. We further excluded patients aged <16 years (n = 30), pre-existing disability (GOS 1-3) before the injury (n = 274), TBI (n = 136), and incomplete data (n = 374). The remaining 759 adult patients diagnosed with SI were included in the study, and 438 (57.7%) patients had prehospital spine immobilization. Figure 1 shows a detailed ow diagram of the patients enrolled in the nal analysis.
The demographics of the 759 patients with prehospital spinal immobilization are shown in Table 1. The study cohort consisted of patients from four countries, including Korea, Malaysia, Japan, and Vietnam. Korea and Malaysia accounted for 97.1% of the study group. Injury mainly resulted from tra c accidents and falling accidents, and all patients had blunt injuries. The combination of cervical SI and thoracic or lumbar SI was only 3.1% (n = 24).

Association between prehospital spinal immobilization and favorable functional outcome
For the primary outcome of the association between prehospital spinal immobilization and favorable functional outcome, the demographics and the results of univariable and multivariable logistic regression are shown in

Subgroup analysis
In the subgroup analysis (Figure 2), we found that prehospital spinal immobilization was associated with favorable functional outcomes in the subgroup of cervical SI (aOR, 3.14; 95%CI 1.04-9.50; p = 0.043), but not in the subgroup ISS <9, ISS ≥9, RTS ≥7, thoracic SI, and lumbar SI. Unfortunately, multivariable logistic regression could not be performed due to the limited number of patients in subgroup RTS <7.

Discussion
In this cross-national retrospective cohort study from January 1, 2016, to November 30, 2018, we found that prehospital spinal immobilization was not associated with favorable functional outcomes at discharge in all EMS-transported patients with SI; however, in the subgroup of cervical SI, the procedure still mattered. There was a signi cant association between prehospital spinal immobilization and favorable functional outcomes at discharge. Our study has several strengths. First, this was a cross-national study in Asia with different EMS systems. Hence, the results may be applicable to other Asian populations. Second, we excluded preexisting disability before injury and TBI to minimize the effect on functional outcome at discharge and to focus more on the impact of SI.
The bene t of spinal immobilization came from indirect and low evidence that numerous studies found that delayed diagnosis of SI without immobilization was correlated with worse disability 36-38 . Toscana et al. 38 conducted a case series of 123 patients with blunt SI and found that 32 (26%) patients had major neurological deterioration between the time of injury and the time of admission. Of these 32 patients, 19 (59%) had no immobilization, suggesting that neurological deterioration resulted from not being immobilized. Meanwhile, Hauswald et al. 39 retrospectively reviewed patients with traumatic SI from the USA and Malaysia, where the former had prehospital spinal immobilization, but the latter did not have prehospital spinal immobilization in all patients. The results revealed that non-immobilized Malaysian patients had a less neurologic disability (OR, 2.03; 95%CI 1.03-3.99; p = 0.04). However, the study was criticized by some points, such as patients who died in the scene or transportation were excluded, no matching of the patient's injury severity, and relatively small sample size. In our study, we did not exclude patients with traumatic out-of-hospital cardiac arrest, and we performed a subgroup analysis to match the injury severity of patients.
In the subgroup analysis of our study, there was a statistically signi cant association between prehospital spinal immobilization and favorable functional outcome when limited to patients with cervical SI. Excluding patients with TBI may be a cause. The prevalence of concomitant TBI in patients with an SI was 32.5% (95%CI 10.8-59.3%) and 40.4% (95%CI 33.0-48.0%) in patients with cervical SI 21 . Severe TBI can cause severe functional disability or death; approximately 38% of patients die 40 and, as estimated, 43% are discharged with long-term disability 41 . Moreover, spinal immobilization could also increase intracranial cerebral pressure in immobilized patients, which could worsen the situation of TBI 12 . Excluding patients with concomitant TBI decreased the in uence of brain injury and helped us focus on the effects of SCI and SI.
The incidence of SI in our study was 3.5% in all EMS-transported trauma patients, which was similar to a previous study from China with an incidence of 4.58% in the trauma population 3 . The percentage of patients with cervical SI and other locations of SI were also similar to those reported in previous studies 3,42 , indicating the generalizability of our ndings. The incidence data for Europe and Canada were relatively high, at 9.6% and 23.2%, respectively 4,42 , which may be due to the study population with polytrauma and higher severity in their studies. The leading causes of SI were falls (47.3%) and tra c accidents (41.9%), which is consistent with previous studies 2, 43 . The rate of cervical SI combined with thoracic or lumbar SI in our study was 3.1%, which was lower than previous studies (approximately 10%) 4,6 . Patients with lower trauma severity in our study may have been the cause.
In patients in the immobilized group, statistically signi cant higher trauma severity (ISS, RTS), higher percentage of combined torso injury, prehospital uid management, and SI were noted than non-immobilized groups. It is reasonable that severely injured patients would receive more treatment at the scene. Subgroup analysis using ISS and RTS was performed to reduce the impact of injury severity, and the association between favorable functional outcome and prehospital spinal immobilization was not signi cant in the subgroup with ISS<9, ISS≥9, and RTS≥7. The scene-to-hospital time was not signi cantly different between the immobilized and non-immobilized groups, although the immobilized group received more treatment out of the hospital. This might be due to the familiarity of the procedure from paramedics, and only 42% (N = 184) of patients received full spine immobilization in the immobilized group.

Limitation
There are some limitations to this study. First, this retrospective study had to address the problem of missing data. Some countries did not record some confounding variables, and we had to exclude them from the analysis, which may have caused selection bias. Although we included many variables in the logistic regression analysis, other unknown factors could in uence the functional outcomes, such as limited prehospital information from EMS, the quality of each EMS team, bystander management of the patients, the quality of in-hospital care, and subsequent rehabilitation programs. These data cannot be included in our analysis because of incomplete data or non-recording of the variables in the registry. Second, the registry data of PATOS were voluntary and could not be considered a representative sample of the included countries. Third, the sample size was relatively small. However, considering the relatively low incidence rate of SI and SCI, we believe that these ndings are still informative to the knowledge gap.

Conclusion
Prehospital spinal immobilization was not associated with favorable functional outcomes in traumatic patients with SI; however, subgroup analysis revealed that it may be bene cial for patients with cervical SI without TBI. Based on our ndings, we suggest that paramedics should be more judicious when determining the cervical SI and use prehospital immobilization, and prospective trials may be needed in the future.

Competing interests
There are no competing interests to declare.

Data availability
The data that support the ndings of this study are available from Pan-Asia Trauma Outcomes Study (PATOS) but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of PATOS. For more detailed information of PATOS, following website is available, http://lems.re.kr/eng/patos-research/.   #Others: hit by person or objects, choking or hanging, drowning, physical overexertion, another mechanism of injury !Other operations: One patient could have undergone several operations during the same hospital stay. Figure 1 Flow diagram of patients included in our study GOS, Glasgow outcome scale; mRS, modi ed Rankin Scale; RTS, revised trauma score; StoH time, Scene-to-hospital time *Patients with spinal injury were de ned as meeting the diagnosis of ICD-9 or 10 described in the methods.

Figure 2
Subgroup analysis for favorable neurologic outcome, ISS<9, ISS≥9, age≥65, age<65, adjusted multiple logistic regression aOR, adjusted odds ratio; SI, spinal injury. *All subgroups except RTS <7 were adjusted by country, age, sex, scene-to-hospital time, mechanism, prehospital uid management, cervical spinal injury, thoracic spinal injury, lumbar spinal injury, torso injury, revised trauma score, ISS, received spine operation, and received other operations. **Due to the small subgroup size, multivariable logistic regression was not applicable.