Risk Factors for Tracheostomy after Traumatic Cervical Spinal Cord Injury: A 10-Year Study of 456 Patients.

OBJECTIVES: To explore the difference between tracheostomy and non-tracheostomy and identify the risk factors associated with the need for tracheostomy after traumatic cervical spinal cord injury (TCSCI).
METHODS: The demographic and injury characteristics of 456 TCSCI patients, treated in the Xinqiao Hospital from 2010 to 2019, were retrospective analyzed. Patients were divided into the tracheostomy group (n = 63) and the non-tracheostomy group (n = 393). Variables included were age, gender,smoking history, mechanism of injury, concomitant injury, American Spinal Injury Association (ASIA) Impairment Scale, the neurological level of injury, Cervical Spine Injury Severity Score (CSISS), surgery, and length of stay in ICU and hospital. SPSS 25.0 (SPSS, Chicago, IL) was used for statistical analysis and ROC curve drawing. Chi-square analysis was applied to find out the difference of variables between the tracheostomy and non-tracheostomy groups. Univariate logistic regression analysis (ULRA) and multiple logistic regression analysis (MLRA) were used to identify risk factors for tracheostomy. The area under the ROC curve (AUC) was used to evaluate the performance of these risk factors.
RESULTS: Of 456 patients who met the inclusion criteria, 63 (13.8%) underwent tracheostomy. There were differences in age (χ2 = 6.615, P = 0.032), mechanism of injury (χ2 = 9.87, P = 0.036), concomitant injury (χ2 = 6.131, P = 0.013),ASIA Impairment Scale (χ2 = 123.08, P < 0.01), the neurological level of injury (χ2 = 34.74, P < 0.01), and CSISS (χ2 = 19.612, P < 0.01) between the tracheostomy and non-tracheostomy groups. Smoking history, CSISS ≥ 7, AIS A and, NLI ≥ C5 were identified as potential risk factors for tracheostomy by ULRA. Smoking history (OR = 2.960, 95% CI: 1.524-5.750, P = 0.001), CSISS ≥ 7 (OR = 4.599, 95% CI: 2.328-9.085, P = 0.000), AIS A (OR = 14.213, 95% CI: 6.720-30.060, P = 0.000) and NLI ≥ C5 (OR = 8.312, 95% CI: 1.935-35.711, P = 0.004) as risk factors for tracheostomy were determined by MLRA. The AUC for the risk factors of tracheostomy after TCSCI was 0.858 (95% CI: 0.810-0.907).
CONCLUSIONS: Smoking history, CSISS ≥ 7, AIS A and, NLI ≥ C5 were identified as risk factors needing of tracheostomy in patients with TCSCI. These risk factors may be important to assist the clinical decision of tracheostomy.

Introduction

Traumatic cervical spinal cord injury is catastrophic, which can lead to serious disability or even death, and bring care and economic burden to families and society , , , . The incidence of spinal cord injury (SCI) is about 10.4–57.8 cases per million in developed countries, and 12.7–29.7 cases per million in developing countries , , ; 50.9%–55% of SCIs occurred in the cervical spine , , . Respiratory complications resulting from variable inspiratory and/or expiratory muscles paralysis as well as excessive tenacious bronchial mucous production, are one of the main causes of aggravation and even death after cervical spinal cord injury , , , . In the acute stage of traumatic cervical spinal cord injury, 84% of patients with C1‐4 injury, and 60% of patients with C5‐8 injury would experience respiratory complications . A high level of traumatic cervical spinal cord injury often causes diaphragm and intercostal muscle paralysis. Lower cervical spinal cord injury could also lead to respiratory muscle paralysis caused by hemorrhage or edema, which makes the neurologic level of injury ascend one or two segments , . In addition, quadriplegia caused by traumatic cervical spinal cord injury makes some patients have long‐term bed rest, which would increase the risk of respiratory complications such as atelectasis and pneumonia. Therefore, the management of the respiratory tract is particularly important in severe cervical spinal cord injury , .

Patients with traumatic cervical spinal cord injury often need continuous mechanical ventilation to maintain airway patency, improve ventilation, and prevent hypoxia and carbon dioxide accumulation. With the worsening of the disease, timely intubation, tracheostomy, mechanical ventilator and other applications are particularly important for improving ventilation , . Early tracheostomy provides many advantages over prolonged endotracheal intubation. Tracheostomy is often performed in patients with traumatic cervical spinal cord injury who have severe respiratory complications, in order to reduce dead‐space ventilation, relieve dyspnea caused by respiratory dysfunction or retention of respiratory tract secretion, and avoid complications caused by prolonged endotracheal intubation . According to the progress of the traumatic cervical spinal cord injury, timely and effective tracheostomy has been proved to be able to strengthen ventilation, shorten the time of ventilator, decreases the overall hospital and intensive care unit (ICU) length of stay (LOS), and reduce the complications of intubation , , , , , , . Of course, tracheostomy, as an invasive operation, will also bring complications , . Therefore, it is particularly important to accurately evaluate whether patients with traumatic cervical spinal cord injury need tracheostomy.

At present, the timing and indication of tracheostomy after cervical spinal cord injury have not yet formed a unified and recognized standard. Many studies have focused on identify risk factors for tracheostomy after cervical spinal cord injury, such as age, severity of injury, neurological level, forced vital capacity (FVC), volume of pulmonary secretion and so on, to assist clinical decision‐making , , , . Menaker proposed that complete cervical spinal cord injury (AIS A) was significantly associated with the need for tracheostomy, and patients with an AIS D scale should not be considered for early tracheostomy by a 3‐year retrospective follow‐up study. Como suggested that tracheostomy was necessary for all patients with complete cervical spinal cord injury whose neurological level of injury was above C5 by a two‐year follow‐up study of 45 patients with cervical spinal cord injury. Leelapattana suggested that if the Injury Severity Score was above 32, PaO2/FiO2 ratio < 300 for three consecutive days and the patient had a complete SCI, tracheostomy could be considered according to the progress of the disease. Yugue suggested that the measurement of forced vital capacity was indispensable to predict the need for tracheostomy in patients with cervical spinal cord injury at the acute stage. After a retrospective research in 319 patients with cervical spinal cord injury, forced vital capacity (FVC) and percentage of vital capacity (%VC) were risk factors for the need of tracheostomy in patients with traumatic cervical spinal cord injury . This research lay the foundation for the follow‐up study, and provide an effective statistical method to identify new risk factors for predicting tracheostomy among patients with traumatic cervical spinal cord injury.

China has a large population, with a large number of patients suffering from traumatic cervical spinal cord injury. To identify the risk factors of tracheostomy in traumatic cervical spinal cord injury was helpful to strengthen the formulation of preventive measures and improve the prognosis of patients. In this context, we designed this clinical retrospective study with variables such as age, gender, mechanism of injury, concomitant injury, American Spinal Injury Association (ASIA) Impairment Scale, the neurological level of injury (NLI), Cervical Spine Injury Severity Score (CSISS), and so on, which could be easily obtained from the bedside. The measurement of pulmonary function in patients with CSCI is accurate and effective to estimate their respiratory condition , , ; however, it is difficult for patients with older age or severe CSCI to complete pulmonary function tests, so forced vital capacity (FVC) and other variables associated with pulmonary function were not included in this study. On the basis of previous studies on risk factors of tracheostomy, our aim was to: (i) explore the difference between tracheostomy and non‐tracheostomy in patients with traumatic cervical spinal cord injury; (ii) clarify the association between the need for tracheostomy and variables obtained from the bedside; and (iii) identify risk factors for tracheostomy by a 10‐year large sample of patients with traumatic cervical spinal cord injury.

Materials and Methods

Inclusion and Exclusion Criteria

Inclusion criteria was summarized below. (i) The patients had a clear diagnosis of cervical spinal cord injury. (ii) The patients had a clear history of trauma and complete data of symptoms, signs and imaging. (iii) The retrospective case–control study was used. 605 patients were identified.

Patients with the following characteristics were excluded from the study: the Abbreviated Injury Scale of other body parts (except the neck) is greater than 3, age < 18 years, refusal of treatment halfway, history of spinal cord injury, incomplete data and injured more than 24 h before medical treatment.

Patient Population

This investigation was conducted with the approval of the Ethics Committee of the Army Medical University. The retrospective study was performed with traumatic CSCI patients in acute stage admitted to Xinqiao Hospital of Army Military Medical University from 1 January 2010 to 31 December 2019. When the patients with traumatic CSCI were admitted to hospital, two orthopedic surgeons objectively estimated the patients' severity of injury, neurological status, and key muscle strength of upper extremity. For patients with severe traumatic CSCI, tracheostomy would be undergoing when patients with a PaO2 < 60 mm Hg and PaO2/FiO2 ≤ 300 mm Hg , or weaned from mechanical ventilation lasting more than 7 days . Four hundred fifty‐six eligible patients were screened and 63 of them underwent tracheostomy (Fig. 1).

Fig. 1

Flow chart of traumatic cervical spinal cord injury patients (2010–2019). 456 patients with TCSCI met the inclusion criteria. Sixty three patients underwent tracheostomy.

Data Collection

The demographic and injury characteristics of these patients were stratified with respect to age, gender, mechanism of injury (motor vehicle accident, fall at height, fall, hit by object and other causes), concomitant injury (associated injuries to the head, chest, abdomen, AIS < 3), severity of injury according to the American Spinal Injury Association (ASIA) Impairment Scale , the neurological level of injury, Cervical Spine Injury Severity Score (CSISS) , , surgery, and length of stay in ICU and hospital. The assessment of injury severity was based on the medical electronic records of patients combined with imaging data.

American Spinal Injury Association (ASIA) Impairment Scale

The ASIA Impairment Scale (AIS) adapted to the International Standards for Neurological Classification of Spinal Cord Injury (the revision in 2011) was used for classifying patients . The ASIA Impairment Scale was divided into five categories: AIS A, no motor or sensory function is preserved in the sacral segments S4‐S5; AIS B, sensory but not motor function is preserved below injury level and includes S4‐S5; AIS C, motor function preserved‐half the muscles have less than grade 3; AIS D, motor function preserved‐half the muscles have more than grade 3, and AIS E, motor and sensory function are normal.

Neurological Level of Injury

The neurological level of injury was the last normal level of sensory and motor function . Motor and sensory scores were calculated based on the examination of 10 key muscles and 28 key sensory points on both the left and right sides of the body.

Cervical Spine Injury Severity Score (CSISS)

The CSISS was used to evaluate cervical spine injuries by imaging , , . A 0 to 5 point value was assigned to each of the four columns (anterior, posterior, left, and right) of the bony cervical spine (Fig. 2) , , .

Fig. 2

A, The cervical spine was divided into four columns(anterior column, posterior column, right column and left column). B, Visual analog scale (0–5) for cervical instability was applied to each column and summed. A score of 1: nondisplaced fractures, or ligamentous disruption or displacement of 1–3 mm; a score of 2: displaced fractures of 1–3 mm; a score of 3: ligamentous disruption or displacement of 3–5 mm; a score of 4: displaced fractures of 3–5 mm; a score of 5: displaced fractures of greater than 5 mm, or complete ligamentous disruption or displacement of greater than 5 mm.

Statistical Analysis

Statistical analysis was performed using SPSS version 25.0 (SPSS, Chicago, IL).Ratios were compared with binomial proportion tests and frequency counts were analyzed using appropriate chi‐square analysis when the sample size is allowed. Univariate logistic regression analysis (ULRA) was applied to separately identify potential risk factors for tracheostomy. In order to eliminate the interference of confounding factors, multiple logistic regression analysis (MLRA) was used to further analyze the risk factors (P < 0.05) of ULRA. Receiver operating curve (ROC) analysis was used to evaluate the differentiation ability of the risk factors for tracheostomy after TCSCI.

Results

Demographic and Clinical Characteristics of TCSCI Patients

Four hundred fifty‐six eligible patients (376 males and 80 females) were identified during the 10‐year study period. The demographic and injury data were presented in Table 1.The mean age of the patients was 49.84 ± 13.28 years (mean ± standard deviation [SD]) 0.161 (35.3%) patients had a history of smoking. Motor vehicle accidents, falls from height and falls were the three main causes of injuries. Sixty three (13.8%) patients had concomitant injuries, including head, chest and abdomen. Through imaging examination, 126 (27.6%) patients with TCSCI had a CSISS score of ≥7 points. The neurological status of each patient at admission revealed that 50 (11.0%) of the 456 patients were AIS A, 51 (11.2%) were AIS B, 162 (35.5%) were AIS C, and 193 (42.3%) were AIS D. The neurological level of injury at admission revealed that 161 (35.3%) patients were C1–C4 and 295 (64.7%) were C5–C8. In‐hospital length of stay was 16.56 ± 14.24 days (mean ± standard deviation [SD]) and 398 (87.3%) patients received surgical treatment including anterior, posterior and anterior–posterior combination.

TABLE 1

Demographic and clinical characteristics of 456 CSCI patients

Items	Overall	Tracheostomy	Non tracheostomy	X2 value	P value
Number	456	63	393	/
Age (mean ± SD, years)	49.83 ± 13.28	52.22 ± 12.84	49.45 ± 13.33
Age ≥50	225	39 (61.9%)	186 (47.3%)	4.615	0.032 *
Age <50	231	24 (38.1%)	207 (52.7%)
Gender
Male	376	55 (87.3%)	321 (81.7%)	1.186	0.276
Female	80	8 (12.7%)	72 (18.3%)
Smoking history
Smoking	161	29 (46.0%)	132 (33.6%)	3.681	0.055
No Smoking	295	34 (54.0%)	261 (66.4%)
Mechanism
Motor vehicle accident	108	12 (19.0%)	96(24.4%)	9.87	0.036 *
Fall at height	176	31 (49.2%)	145 (36.9%)
Fall	123	10 (15.9%)	113 (28.8%)
Hit by object	36	9 (14.3%)	27 (6.9%)
Others	13	1 (1.6%)	12 (3.0%)
Cervical Spine Injury Severity Score
CSISS ≥ 7	126	32 (50.8%)	94 (23.9%)	19.612	0.000 *
CSISS < 7	330	31 (49.2%)	299 (76.1%)
Neurological level of injury
C1–C4	161	43 (68.3%)	118 (30.0%)	34.74	<0.01 *
C5–C8	295	20 (31.7%)	275 (70.0%)
Baseline ASIA Impairment Scale grade
A	50	30 (47.6%)	20 (5.0%)	123.08	<0.01 *
B	51	15 (23.8%)	36 (9.2%)
C	162	14 (22.2%)	148 (37.7%)
D	193	4 (6.4%)	189 (48.1%)
Concomitant injuries
Yes	63	15 (23.8%)	48 (12.2%)	6.131	0.013 *
No	393	48 (76.2)	345 (87.8%)
In‐hospital LOS (days)	16.56 ± 14.24	33.95 ± 33.80	14.56 ± 7.76	/
ICU LOS (days)	6.95 ± 15.31	32.47 ± 35.77	3.78 ± 4.53
Surgical approach
Anterior	54	6	48	/
Posterior	335	27	308
Anterior–posterior	9	/	9

ASIA, American Spinal Injury Association; ICU, Intensive Care Unit; LOS, length of stay.

P < 0.05.

Differences between Tracheostomy and Non‐tracheostomy

Sixty three (13.8%) patients underwent a tracheostomy as tracheostomy group and others were non‐tracheostomy group. Comparing the two groups, the percentage of age ≥50 years was higher in the tracheostomy group than non‐tracheostomy group (61.9% vs 47.3%, χ2 = 6.615, P = 0.032). The CSISS score was significantly higher in the tracheostomy group than non‐tracheostomy group (50.8% vs 23.9%, χ2 = 19.612, P < 0.01). The percentage of concomitant injuries among patients who underwent tracheostomy was higher than that in non‐tracheostomy group (23.8% vs 12.2%, χ2 = 6.131, P = 0.013). Patients with TCSCI who underwent a tracheostomy had a higher NLI (NLI ≥ C5) (68.3% vs 30.0%, χ2 = 34.74, P < 0.01) and a higher ASIA Impairment Scale at AIS A (47.6% vs 5.0%, χ2 = 123.08, P < 0.01). There was no significant deference in gender between tracheostomy and non‐tracheostomy. Although the proportion of smoking in the tracheostomy group was higher than that in the non‐tracheostomy group, there was no statistical difference (46.0% vs 33.6%, χ2 = 3.681, P = 0.055). The mechanism of injury was statistically significant between tracheostomy group and non‐tracheostomy group. Further analysis could be used to identify which was the risk factor of tracheostomy.

Risk Factors for Tracheostomy

By using the univariate logistic regression analysis (ULRA), P values of <0.05 were considered to be statistically significant to avoid omission of important risk factors. The initial results determined that smoking history (OR = 2.307, 95% CI: 1.126–4.724, P = 0.022), CSISS ≥ 7 (OR = 5.937, 95% CI: 2.798–12.750, P = 0.000), high NLI (NLI ≥ C5) (OR = 9.228, 95% CI: 2.088–40.777, P = 0.003), and AIS A (OR = 13.365, 95% CI: 6.146–29.063, P = 0.000) were statistically significant factors that increased the incidence of tracheostomy in Table 2. The four risk factors above were brought into the multiple logistic regression analysis (MLRA) and P values of <0.05 were considered to be statistically significant. Smoking history (OR = 2.960, 95% CI: 1.524–5.750, P = 0.001), CSISS ≥ 7 (OR = 4.599, 95% CI: 2.328–9.085, P = 0.000), AIS A (OR = 14.213, 95% CI: 6.720–30.060, P = 0.000) and NLI ≥ C5 (OR = 8.312, 95% CI: 1.935–35.711, P = 0.004), were identified to be independent risk factors of tracheostomy after TCSCI in Table 3.

TABLE 2

Univariate logistic regression analysis

Items	OR (95% CI)	P value
Age ≥50	1.484 (0.686–3.207)	0.316
Gender	2.123 (0.706–6.382)	0.180
Smoking history	2.307 (1.126–4.724)	0.022*
Mechanism
Motor vehicle accident	0.232 (0.022–2.504)	0.229
Fall at height	0.367 (0.109–1.230)	0.104
Fall	0.504 (0.170–1.496)	0.217
Hit by object	0.362 (0.106–1.237)	0.105
Others	1.00 (reference)
CSISS ≥ 7	5.937 (2.798–12.750)	0.000*
Concomitant injuries	2.277 (0.770–6.737)	0.137
AIS A	13.365 (6.146–29.063)	0.000*
NLI ≥ C5	9.228 (2.088–40.777)	0.003*

AIS, ASIA Impairment Scale; CSISS, Cervical Spine Injury Severity Score; NLI, neurological level of injury.

P < 0.05.

TABLE 3

Multiple logistic regression analysis

Items	OR	95% CI	P value
Smoking history	2.960	1.524–5.750	0.001*
CSISS ≥7	4.599	2.328‐9.085	0.000*
AIS A	14.213	6.720–30.060	0.000*
NLI ≥ C5	8.312	1.935–35.711	0.004*

AIS, ASIA Impairment Scale; CSISS, Cervical Spine Injury Severity Score; NLI, neurological level of injury.

P < 0.05.

Receiver Operating Characteristics for Risk Factors of Tracheostomy

A ROC curve was drawn based on the prediction and actual tracheostomy. The area under the ROC curve for the risk factors of tracheostomy after TCSCI was 0.858 (95% CI: 0.810–0.907). As was shown in Fig. 3, AUC = 0.858 > 0.7, so the risk factors have good discrimination. Comparing observation and perdition in Table 4, the sensitivity of risk factors for tracheostomy was 81.8%. Calibration plot was shown in Fig. 4, the prediction line was close to the standard line, which showed good calibration ability.

Fig. 3

Receiver operating characteristics for risk factors of tracheostomy after TCSCI. Area under the curve was 0.858 (95% CI: 0.810–0.907), indicating good discriminatory ability.

TABLE 4

Validation of risk factors for tracheostomy in TCSCI

Observation	Prediction		Total
	Tracheostomy	Non‐tracheostomy
Tracheostomy	54	9	63
Non‐tracheostomy	12	381	393
Total	66	390	456

Fig. 4

Calibration curve for risk factors of tracheostomy after TCSCI. The prediction line (blue line) is close to the standard line(red line), which shows good calibration ability.

Discussion

Respiratory complications are one of the important causes of mortality in CSCI , , , , , . Previous studies have suggested that early tracheostomy is beneficial for respiratory complications , , , , . However, there is no uniform clinical standard for the time of tracheostomy for CSCI. Therefore, much research is devoted to the analysis of risk factors of tracheostomy after CSCI , , , , , , , . The ULRA and MLRA were the most commonly used statistics methods in screening the independent risk factors of tracheostomy after CSCI , , , , , .

Through the comparison between the tracheostomy and non‐tracheostomy groups in our series, the patients who underwent the tracheostomy would be with higher neurological level of injury, higher ASIA scale, and higher CSISS score. The percentage of concomitant injuries in the tracheostomy group was also higher than that in the non‐tracheostomy group. This is similar to the findings of Leelapattana who made a review of patients with CSCI in the tracheostomy and non‐tracheostomy groups from 1991 to 2001. The ULRA was used for preliminary screening of risk factors in our study. However, these factors might have many confounders, so we further conducted a multivariate analysis to adjust for such confounding factors. Finally, smoking history, CSISS ≥ 7, NLI ≥ C5 and, AIS A were identified to be independent risk factors for the need of tracheostomy.

CSISS as a Risk Factor for Tracheostomy

As an anatomical scoring system, the CSISS could provide an overall score for patients with CSCI , , , . CSISS was developed to determine stability in CSCI as a new system, described by Moore et al. . Anderson had proposed that a CSISS score of 7 or greater was a significant risk factor for patients with surgical stabilization of the cervical spine. Instability of the cervical spine could lead to CSCI with severe neurological defect. In our study, CSISS as a risk factor which could be used to predict tracheostomy in patients with TCSCI, was identified by ULRA and MLRA. Patients with CSISS ≥ 7 would be more likely to undergo tracheotomy.

NLI as a Risk Factor for Tracheostomy

Due to loss of diaphragm and/or intercostal muscle function, the higher and more complete the level of sports injury, the more serious the respiratory system will be damaged after SCI , , , . The level of CSCI is a necessary clinical examination and easy to obtain. Lee had recommended that patients with injury above C5 should be considered for tracheostomy. In our series, NLI ≥ C5 was a most important risk factor for tracheostomy after TCSCI. Premotor neurons located above the motor neurons of the phrenic nerve, C3‐C5 (phrenic nerve nucleus) in the cervical spinal cord, were interrupted, and the primary inspiratory muscle and diaphragm were immediately paralyzed due to NLI ≥ C5 , , , , .

AIS A as a Risk Factor for Tracheostomy

Leelapattana have proposed that patients with complete CSCI should be considered for tracheostomy. Hou and Yugue also showed that AIS A was revealed to be a significant risk factor for tracheostomy. Branco suggested that complete CSCI (C1‐C4 or C5‐C8) was independently associated with the need for tracheostomy. In our study, we came to the same conclusion. AIS A was revealed to be a significant risk factor of the need for tracheostomy.

When predicting the possibility of a tracheostomy, surgeons should consider smoking history, CSISS score, ASIA scale, and NLI. The area under the ROC curve for the risk factors of tracheostomy was 0.858, so the risk factors have good discrimination. At the same time, AUC < 0.9 reminds us that some risk factors are not included in the analysis, which need to be further researched.

Limitations

Firstly, it was a retrospective study. Secondly, although this was an extensive review during 10 years, some clinical and imaging data are difficult to obtain and cannot be included in the analysis. Thirdly, most of the data included in the analysis were collected at the time of admission, not before tracheostomy. It also makes a difference between observed and predicted need for tracheostomy.

Conclusion

Smoking history, CSISS ≥ 7, AIS A, and NLI ≥ C5, were all identified as independent risk factors for the need of tracheostomy in patients with CSCI. The tracheostomy package need to be prepared and strengthen the airway management in patients with these risk factors.