Cervical Spine Trauma in East Africa: Presentation, Treatment, and Mortality

Study Design and Clinical Setting

A retrospective analysis of prospectively collected data from the Muhimbili Orthopaedic Institute (MOI), a tertiary referral hospital in Dar es Salaam, Tanzania, was conducted. MOI houses approximately 120 general ward beds. At the time of data collection, the hospital housed 8 intensive care unit (ICU) beds, but has since grown to 18 ICU beds with 18 high-dependency unit surgical step-down beds. Local institutional review board approved the current study and informed consent was obtained from all patients.

Patient Identification

All patients who presented with CST from October 2016 to June 2019 (33 months) were included. CST was operationally defined as any injury to the spinal column involving the occiput to C7-T1 disc space, with or without evidence of SCI. Exclusion criteria included patients with the following conditions: only thoracic/lumbar injuries, <14 years old, concomitant brain injury, or underwent an operation >4 months from the time of injury. This case series represents an extension of a previously published cohort.^17,21

Clinical and Operative Data

Several demographic and injury-specific data points were collected, including age, sex, and mechanism of injury. Injury levels were categorized according to prior studies and described per cervical level.²⁵ Insurance status was classified as public (required to provide all funds prior to receiving hospital services) or private (no additional funds required to receive hospital services).

Fracture type was defined descriptively using a combination of prior cervical spine fracture classification schemes.^26–29 Unilateral facet dislocations were defined as listhesis of 25% or less, bilateral facet dislocations had listhesis of 25%–75%, and spondyloptosis had listhesis of ≥100%.²¹ Central cord syndrome was defined according to prior studies and by radiographic and/or clinical presence of a cervical SCI without ongoing compression, which most often occurred in the setting of preexisting spondylosis and canal narrowing.^30,31 We acknowledge that the definition of central cord syndrome is controversial, and any spondylosis with active compression due to a fracture, disc prolapse, ligamentum flavum buckling, or facet arthropathy, was not defined as central cord syndrome. Neurologic exams were obtained upon admission and discharge according to the American Spinal Injury Association (ASIA) Impairment Scale.³² In-hospital mortality was recorded for all patients.

One of four operations was performed: (1) anterior cervical discectomy and fusion with tricortical iliac crest autograft and plate, (2) anterior cervical corpectomy with tricortical iliac crest autograft and plate, (3) posterior cervical laminectomy and fusion with lateral mass screws and rods (PCLF), and (4) posterior cervical laminectomy only. Tricortical iliac crest autograft was used for bone graft in anterior cases, whereas local autograft was used for posterior cases. Decompression was performed at the site of active cord compression. Time in days was recorded during the following points: injury to admission, admission to operating room, operating room to discharge, and total length of stay.

Surgical Indication

All patients were retrospectively reviewed to determine whether an indication for surgery was present using a combination of the AO fracture classification system²⁶ and neurologic function, in accordance with prior studies.³⁰ Fracture type was defined using the AO Classification System (A0–4, B1–3, C).²⁶ Patients with traumatic disk herniation or central cord syndrome without a fracture were classified as A0. An injury was considered to have a surgical indication in one of two ways: (1) it was an unstable fracture, classified as type A4, B, or C, which required stabilization regardless of neurologic injury or (2) it was a potentially unstable fracture, classified as A0-A3, with neurologic impairment (ASIA A–D). Given the limited resources, some patients only had x-rays available due to cost, and in these cases, a best estimation of fracture type was made. All surgical indication determinations were performed by 2 individuals: a neurosurgery trainee and fellowship-trained neurosurgeon.

Statistical Methods

All continuous data were presented as mean (SD) and/or median with interquartile range (IQR) and/or full range, whereas all count data were presented as n (%). T-tests and χ² tests or the Fisher exact test were used to assess differences among continuous and categorical variables, respectively. Univariate followed by multivariate logistic regression was used to assess predictors of 3 outcomes: operative treatment, improvement in neurologic function, and mortality. Any variable with P < .10 after univariate regression was included in the multivariate model. The second outcome of improved neurologic function was subdivided into surgical group. Significance was considered at an αof < .05. All statistical analyses were performed in RStudio, version 1.2.5033.

Patient Cohort

Of 101 patients presenting to MOI with CST, 25 (24.8%) were treated operatively on a median (IQR) of postadmission day 16.0 (7.0–25.0) (Table 1). Motor vehicle and motorcycle collisions were the most common injury mechanism in both groups, and the majority of patients had public insurance (88.1%). Almost all patients were admitted to an outside hospital before MOI (96.0%), leading to a median (IQR) time from injury to MOI admission of 2.0 (1.0–5.0) days. Nonoperative patients received x-rays more often, whereas operative patients more commonly received computed tomography (CT) and magnetic resonance imaging scans. Of the nonoperative cohort, 93.4% had an indication for surgery, signifying the severity of injury in this population. Twenty-six patients (25.7%) died, with 3 (12.0%) in the operative cohort and 23 (30.3%) in the nonoperative cohort. The most common fracture pattern was bilateral facet dislocation (n = 27, 26.7% of patients) followed by unilateral facet dislocation (n = 22, 21.8% of patients) and burst/teardrop fracture (n = 20, 19.8% of patients) (Figure 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Fracture types among 101 cervical spine trauma patients (each fracture type counted separately when more than 1 injury was present).

Operative Treatment

Though 95.0% of all CST patients had an indication for surgery, 25 patients (24.8% of total sample; 26.0% of those with an operative indication) went to surgery, which means 71 patients (74.0% of those with an operative indication) did not undergo surgery. Of the 25 operative patients, PCLF was the most common operation (48%), followed by anterior cervical corpectomy (28%) (Table 2). A representative case of a C5 burst fracture and anterior cervical corpectomy with iliac crest autograft is shown Figure 2A–E). Median time from admission to operating room was 16.0 days (0–74). Postoperative imaging was obtained in only 56% of patients, likely due to cost, and was almost exclusively x-rays. Two independent predictors of undergoing operative treatment emerged after multivariate logistic regression: (1) having an injury at C4–C7 compared to higher occiput–C3 levels (odds ratio [OR] 6.36, 95% confidence interval [CI] 1.71–32.28, P = .011) and (2) having an incomplete injury (ASIA B–D) compared to complete or intact (ASIA A/E) (OR 3.64; 95% CI 1.19–12.25; P = .029) (Table 3). Thus, it appears those most likely to undergo surgery were patients with lower cervical injuries (C4–C7) and incomplete SCI injuries (Figure 3).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2A-E

(A) Patient with C5 burst fracture. (B) Incision over right anterior iliac crest in preparation for (C) tricortical iliac crest autograft. (D, E) postoperative x-rays.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Rates of operative treatment by cervical level and neurologic exam.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Mortality rates based on neurologic exam and treatment.

Neurologic Outcomes

In the 76 patients with an exam recorded at discharge (75.2%), 12 patients (15.8%) improved neurologically by at least 1 ASIA grade (Tables 4 and 5). After univariate logistic regression in all patients (operative and nonoperative), the only factor associated with neurological improvement was having an incomplete SCI (ASIA B–D) (OR 4.38, 95% CI 1.18–21.20, P = .038); however, this lost statistical significance after multivariate analysis (Table 6). In the smaller subset of 22 patients who underwent surgery, univariate logistic regression showed a trend toward faster time to surgery being associated with an increased odds of neurologic improvement (OR 1.08, 95% CI 1.02–1.21, P = .061), but this too lost significance after multivariate analysis. Importantly, operative treatment was not associated with neurologic improvement on univariate testing (OR 0.79, 95% CI 0.16–2.99, P = .743).

Mortality

A total of 26 patients died (25.7%), with 3 (12.0%) in the operative group and 23 (30.3%) in the nonoperative group. After multivariate testing, 2 factors were significantly associated with increased odds of mortality (Table 7). Counterintuitively, a longer time from injury to MOI admission was associated with a 34% decreased odds of mortality (OR 0.66, 95% CI 0.48–0.85, P = .006). Said in other words, a faster time from injury to MOI admission was associated with increased odds of death. Additionally, having a complete injury (ASIA A) was associated with over 11 times the odds of mortality compared to incomplete or intact patient status (ASIA B–E) (OR 11.75, 95% CI 3.29–54.72, P < .001). Though a trend for nonoperative treatment associated with increased odds of mortality was seen, statistical significance was not achieved (OR 36.7, 95% CI 0.87–20.35, P = .097).

Current Patterns of Operative Treatment

The most common fracture pattern was bilateral facet dislocation, followed by unilateral facet dislocation and burst fracture, which fits exactly with the 3 most common injuries in a study of CST patients from South Africa.²⁰ We also reported a 24.8% operative rate, which is slightly less than a previous published rate of 38.4% in 125 CST patients from Ghana.³³ While nearly all patients (95%) had an indication for surgery, unfortunately, our low operative rate is common in LMICs, where many patients don't undergo surgery due to insufficient funds or lack of equipment, including spinal implants, operating beds, and surgical tools. PCLF was the most common approach in our study, followed by anterior cervical corpectomy with tricortical iliac crest autograft. In Cambodia, Choi et al²⁵ operated on 20 cervical trauma patients and performed anterior cervical discectomy and fusion most commonly, followed by posterior fusion with wiring, yet only a single corpectomy was required. Our high rate of corpectomies may be attributed to burst fractures being the third most common fracture pattern seen, where the entire vertebral body must be removed to adequately decompress the spinal canal. Tricortical iliac crest autograft was used in all cases, an affordable and convenient source of bone graft.³⁴ In addition, lower-cervical injuries were more commonly operated on than higher-cervical injuries, a trend found in earlier studies.³⁰ Explanations for this finding are likely multifactorial and may be due to high cervical injuries leading to quadriplegia and a poor prognosis. Another reason is that craniocervical injuries require more complex operations to stabilize and treat. Instrumentation of the occiput, C1, and C2 are higher risk, and surgeons may lack either the experience or necessary equipment to safely perform these operations. Lastly, incomplete injuries were taken to surgery more often than complete injuries, a testament to a resource limited setting, which likely represents allotment of resources to those most likely to improve.

Neurologic Improvement

The neurologic improvement rate was low at 15.8%, likely due to long delays from admission to surgery. An earlier CST series from India³⁴ reported a mean neurologic recovery by 1.1 ASIA grades, and their high rate of neurologic improvement may be due to a mean operative day of 7 days postadmission, compared to our median of 16 days postadmission. In high-income countries, where hospital resources and adequate ICU care are ubiquitous, it is well established that faster time to decompression is associated with improved neurologic outcomes.^35–37 However, in an LMIC setting, it appears that faster time to surgery may not offer any benefit without adequate perioperative and ICU services. CST patients require complex postoperative ICU care, including hourly airway and blood pressure monitoring, inotropes to keep the spinal cord perfused, and experienced nursing care. The importance of adequate ICU care cannot be overstated.

Mortality

Mortality rates after CST range from 27.2% in northeast Tanzania⁴ to 30% in rural India,³⁸ which are similar to our rate of 25.7%. Additionally, one Nigerian series reported 72% of all spine trauma deaths occurred from cervical injuries.⁵ We found that having a complete SCI was associated with over an 11-fold increase in odds of death compared to incomplete SCI. In addition, nonoperative treatment was not associated with an increased risk of mortality, which conflicts with prior literature that nonoperative treatment after any spine trauma was associated with a 7.4-fold increase in odds of death.³⁰ This finding may speak to the smaller sample studied, the severity of CST patients, or the higher risk of cervical spine operations.

Perhaps the most counterintuitive finding in our study was that a longer time from injury to admission was associated with decreased odds of mortality, which conflicts with the aforementioned notion that earlier treatment leads to better outcomes. Previous reports of LMIC spine trauma patients have noted a median time from injury to hospital admission of 48 hours.²¹ Thus, it is possible that in the critical 24–48 hours after CST, the more severely injured patients not yet at the hospital may quickly lose vital respiratory or autonomic function and die. Thus, those that survive the early time period may represent a self-selecting, stable group of patients that have passed the critical window of vulnerability. Of note, one complicating factor is that almost all patients (96%) were admitted from some form of an outside healthcare facility, and no information can be gleaned as to the type of care provided in this triage setting.

Equally important aspects of treating CST patients are ICU care and rehabilitation services. Many LMICs lack critical care beds and appropriately trained personnel to manage complex CST patients before and after surgery. It is possible that improvements in critical care provisions may result in reduced mortality but may increase the survival of those with greater neurological disability. It is therefore important to consider the development of postdischarge care and rehabilitation services, such as locally produced wheelchairs and bladder training for newly incontinent patients. Overall, these data reinforce an important conclusion, that CST treatment may need to be adjusted to the resources of a given hospital setting. While earlier treatment may lead to improved neurologic outcomes in well-resourced settings with adequate ICU expertise, up-to-date equipment, and experienced nursing care, the same benefit from earlier surgery may not occur in a hospital without an ICU, reliable access to inotropes, or experienced intensivists and ICU nurses.

Limitations

There are several limitations to the current study. Follow-up information was not collected, so it is unknown how nonoperative treatment impacted long-term neurologic status and radiographic healing. Also, the results may be difficult to generalize to other neurosurgery and orthopedic departments in LMICs, because each institution has a unique set of available resources that dictate operative decision-making. Mortality was not accounted for postdischarge, which may lead to underreporting of true mortality rates. Finally, despite use of a prospective registry, the data were analyzed retrospectively. Determination of surgical indication is a complex process, and doing this retrospectively may have led to misclassifications. Of note, due to new data collection instruments, significant data were missing for systolic blood pressure low and days in ICU.