Currently, there are many methods of classifying injuries related to the thoracolumbar spine. Different scholars use different classifications to report their findings, and it is difficult to compare their results. The reason for this phenomenon is that there is no convincing scheme for the classification of thoracolumbar spine injuries. A perfect injury classification should reflect the full extent of injury, guide treatment, and predict outcomes. This article reviews the research related to the classification and evaluation of thoracolumbar spine injury scoring systems. I. Theories related to the stability of thoracolumbar spine injuries The need for surgical treatment of thoracolumbar spine injuries depends on the stability and instability of the spine after trauma. In the late 1940s, Nicoll [1] believed that the mechanical stability of the spine was determined by four factors, namely the vertebral body, intervertebral disc, small joints and interspinous ligaments, of which the interspinous ligaments were the most important. In 1968, Kelly and Whitesides [3] redefined the two-column theory of the spine and concluded that a burst fracture is an unstable fracture regardless of the integrity of the posterior ligamentous complex [4]. In 1982, Denis [6] proposed the three-column theory of the spine: the anterior column consists of the anterior longitudinal ligament, the anterior half of the fibrous ring and the anterior half of the vertebral body, the middle column consists of the posterior longitudinal ligament, the posterior half of the fibrous ring and the posterior half of the vertebral body, and the posterior column consists of the arch, the interspinous ligament, the supraspinous ligament, the yellow ligament and small joint capsule; he considered all burst fractures as unstable fractures.McAfee et al [7] proposed their own three-column theory of the spine by combining basic biomechanical research, in which the middle column consists of the posterior longitudinal ligament, the posterior 1/3 part of the fibrous ring and the posterior 1/3 part of the vertebral body; they believed that the degree of injury and injury mechanism of the middle column determine the stability of the spine, and the posterior ligament complex is important factor for stability. Some clinical scholars determine whether the fracture is stable or not based on the degree of kyphosis, the occupancy in the spinal canal, the rate of vertebral height loss and the neurological function status, and decide whether surgical treatment is needed accordingly [8-9]. Therefore, there is no consensus on the understanding of spinal stability, especially the role of soft tissues in maintaining spinal stability is highly controversial. We believe that the stability of spinal injury should include three aspects namely vertebral body, soft tissue and nerve condition, one of which is indispensable, but the interrelationship of the three aspects and the magnitude of their role on stability may be different and needs further study. Second, the classification system of thoracolumbar spine injury: 1, classification based on the description of fracture morphology and injury mechanism: the description of fracture morphology is based on radiological diagnosis, with the development of diagnostic radiology, the understanding of thoracolumbar fractures also gradually deepened, as shown in the increasingly detailed classification of fractures. 1938 Watson-Jones [10] divided thoracolumbar fractures into simple wedge fractures, comminuted fractures and In 1949, Nicoll [1] classified thoracolumbar spine injuries into four main types based on morphological analysis of different injury mechanisms, namely, anterior wedge fracture, lateral wedge fracture, fracture-dislocation type, and arch fracture.Holdsworth [2] first considered the role of ligaments in maintaining spinal stability and classified thoracolumbar spine fractures according to whether the posterior ligament complex Denis [6], based on the three-column theory, classified thoracolumbar spine injuries into four major categories: compression fractures, seat belt fractures, burst fractures and fracture-dislocation types, and further subdivided them into 16 subcategories. The three-column theory and its classification used to be widely used in clinical practice, especially the five subtypes of burst fractures in Denis’ classification, which are still in use by scholars today. Due to the different understanding of burst fractures, McAfee et al [7] classified thoracolumbar injuries into the following six types based on their own three-column theory of the spine: wedge compression fractures, stable burst fractures, unstable burst fractures, Chance fractures, flexion-distraction fracture injuries, and shear injuries. With the promotion of the AO/ASIF (Arbeitsge-meinschaft für Osteosynthesefragen/The Association for the Study of Internal Fixation) theory in clinical practice, Magerl et al [11] in a multicenter AO classification was proposed by Magerl et al [11] on the basis of a multicenter radiographic and CT study of 1400 patients with thoracolumbar spine injuries. This classification divided the injury into three basic types of compression, distraction and rotational violence according to Holdsworth’s two-column theory and injury mechanism, and then into different subtypes according to the fracture site, fracture pattern, ligament damage and the direction of displacement, which divided the thoracolumbar fractures into 3 categories, 9 groups and 27 types, as many as 55 types. Some domestic scholars have also done a lot of research on this and put forward their own classification schemes, such as Jin Dadi [12] who classified thoracolumbar fractures into three categories according to the mechanism of injury and its pathological morphology, each category into two types, and each type into three subtypes for a total of 18 types. These classifications are based on imaging data, combined with the patient’s mechanism of injury, and only consider the injury of bony structures without combining soft tissue and nerve injury, much less quantifying the severity of the injury. Therefore, some scholars have studied [13-15] the thoracolumbar spine injury severity scoring system to quantitatively evaluate the injury severity, reflect the degree of spinal stability and guide the treatment and predict the prognosis. 2, classification system based on injury severity score: McCormack et al [13] proposed load sharing classification (LSC) based on the study of the causes of short segmental pedicle nail fixation failure. The sum of the three scores is the total LSC score, and the higher the score, the more unstable the fracture. This classification was the first to quantify the severity of injury, but it also only considered vertebral fractures, ignoring the role of soft tissue and nerve injury on spinal stability. In 2005, The Spine Trauma Study Group (STSG) in the United States proposed the thoracolumbar injury severity score (TLISS), which combines radiographs, CT and MRI examinations, and the patient’s nerve injury [14]. The TLISS was based on three main aspects: (1) understanding the mechanism of injury of the fracture based on imaging data; (2) the integrity of the posterior ligamentous complex structure of the vertebral body; and (3) the neurological functional status of the patient. The STSG later improved the TLISS by replacing the subjective mechanism of injury with a more objective description of fracture morphology, and called the thoracolumbar injury classification and severity score system (TLICS) [15]. The specific criteria were (1) radiological morphology of the fracture: compression fracture 1; burst fracture 2; rotational fracture 3; and distraction fracture 4. In case of duplication, the highest score was taken. (2) Structural integrity of the posterior ligament complex: 0 points for intact; 3 points for complete rupture; 2 points for incomplete rupture or suspected rupture. (3) The neurological functional status of the patient: 0 points for no neurological damage; 2 points for complete spinal cord injury; 3 points for incomplete injury or cauda equina syndrome. The sum of the scores is the total TLICS score, with higher scores representing more severe injury. This classification is the first time to consider the role of vertebral body, ligament and nerve function on the stability of spinal injury, and to quantify the severity of injury, which is a significant progress. Burstein [16], writing an editorial review in The Journal of Bone and Joint Surgery (JBJS), pointed out that the essence of the fracture classification system is the application of a tool, and the evaluation of the usefulness of this tool in clinical practice is to see whether it can be used by different physicians at different According to Audige et al [17], the evaluation of a fracture classification should consist of three stages: the first stage is the evaluation of the validity and reliability of the classification by its authors; the second stage is the multicenter evaluation of the validity and reliability of the classification; and the third stage is a prospective clinical trial to confirm its usefulness for clinical treatment and prognosis. The third stage is to conduct prospective clinical trials to confirm its validity for clinical treatment and prognosis. 1. Evaluation of classification reliability and validity: Reliability reflects the ability to reach the same conclusion after reading the imaging data of the same group of patients using the same fracture classification system, which is divided into reliability between different observers and consistency of the judgment of the same observer at different times. Validity refers to the degree of accuracy of the classification system’s response to the true status of the fracture. The percent compliance and Kappa value are commonly used statistically to evaluate the reliability and validity of the classification [18], with a Kappa value >0.75 indicating a very good level of compliance and <0.50 indicating a poor level of compliance. Xia Liangzheng et al [19] evaluated two classifications, Denis and Gertzbein, and the results of the interobserver mean Kappa index: 0.588 for Denis four types and 0.134 for 16 subtypes; 0.603 for Gertzbein three types and 0.420 for nine subtypes. mean Kappa index before and after observers: 0.706 for Denis four type was 0.706 and 0.432 for 16 subtypes; Gertzbein type III was 0.746 and 0.511 for nine subtypes. wood et al [20] evaluated the Magerl and Denis classifications and the mean Kappa for Magerl classification of types A, B and C and nine subtypes between observers was 0.475 and 0.537. The mean Kappa values for the Denis classification of four types and 16 subtypes were 0.606 and 0.173, and the pre- and post-observer compliance rates were 0.82 and 0.67 for Magerl types and subtypes, respectively, and 0.79 and 0.56 for Denis classification. the Magerl, Denis and Gertzbein classifications all had only moderate reliability and validity, with varying degree of deficiencies. Dai et al [21] performed a kappa analysis of the load-sharing classification, where x-rays and CT films of 45 cases were evaluated by five observers yielding a mean interobserver kappa of 0.79 for the first time and 0.84 after three months, and a mean pre- and post-observer kappa of 0.78. For each of the three components of this classification, the kappa means were 0.90 for the first time between observers, 0.92 for the second time, and 0.89 for before and after observers.The authors concluded that the load sharing classification had good reliability and validity. In 48 cases of thoracolumbar segment fractures, rated according to the TLICS system and re-evaluated after 3 months, Zhang et al [22] analyzed interobserver agreement and repeatability using Cohen-weighted Kappa coefficients and calculated that the TLICS subcategory Kappa coefficients were located between moderate and higher reliability (0.43 to 0.72), and the TLICS system repeatability The Kappa coefficient of the TLICS system is between moderate and high reproducibility (0.55-0.76), and the diagnostic accuracy of the TLICS system is 95.1%, sensitivity 87.5%, and specificity 96.8%, and the TLICS classification system is considered to be highly reliable and reproducible. 2. The significance of the classification in guiding clinical treatment and prognosis: according to Denis' three-column theory and its classification [3], all burst fractures are unstable injuries, which can lead to further impairment of nerve function and require surgical treatment. However, many studies have found that conservative treatment of thoracolumbar burst fractures without neurological injury rarely or not show neurological impairment, indicating that the classification is of little significance in guiding treatment.Weninger [23] analyzed the imaging data and functional assessment of 136 cases of conservatively treated thoracolumbar compression and burst fractures and found that conservative treatment was safe and effective for thoracolumbar burst fractures.In 2003 Wood [24] published a prospective randomized controlled article on the treatment of thoracolumbar burst fractures, in which 47 patients with thoracolumbar burst fractures without neurological injury were randomized to surgical and conservative treatment, and found no statistically significant differences in either radiological or functional evaluation (including SF-36 evaluation) between the surgical and conservative treatment groups. The difference between the surgical and conservative treatment groups was not statistically significant for either radiological or functional evaluation (including SF-36 evaluation). When McCormack [6] proposed LSC, he noted that this classification is only used to guide the choice of surgical fixation, and that posterior short-segment fixation can be performed for fractures with <7 points, and for severe fractures with ≥7 points, posterior short-segment fixation alone has a high rate of internal fixation failure.Dai [25] retrospectively analyzed 127 cases of burst fractures treated conservatively after a mean follow-up of 7.2 years to analyze the patients' posterior Korovessis et al [26] conducted a prospective randomized study of 40 patients with L2-4 vertebral fractures with LSC scores >6 and found that short segment pedicle screw fixation was not effective for burst fractures with LSC scores >6. In a 5-7 year prospective randomized controlled clinical trial, Dai et al [27] found that short-segment pedicle screw fixation was safe and effective for thoracolumbar burst fractures with a load-sharing score ≤6. The LSC classification is useful in guiding clinical work, especially for selecting surgical fixation segments to prevent loosening, fracture, and re-loss of fracture reduction. The LSC classification is particularly useful for guiding clinical work, especially for selecting surgical fixation segments, preventing internal fixation loosening, fracture and fracture loss. TLICS [15] recommends that surgical treatment should be considered for ≥5 points, non-surgical treatment for ≤3 points, and surgical or non-surgical treatment for 4 points, and suggests that the surgical access and surgical approach should be decided according to the neurological status and the damage to the posterior ligamentous complex. The classification system recommends: (1) anterior decompression and internal fixation for incomplete neurological impairment from anterior compression; (2) posterior surgery for posterior ligamentous complex injury; (3) combined anterior and posterior approach for both incomplete neurological impairment and posterior ligamentous complex injury. No article has been retrieved to evaluate the significance of this classification for clinical guidance. IV. Summary and outlook In the management of thoracolumbar spine injuries, in addition to injuries to bone structures, attention should be paid to soft tissue injuries, including ligamentous structures and intervertebral discs. Neglecting the presence of soft tissue injuries often affects the outcome of treatment [28]. The early injury classification is only based on the fracture morphology of the radiographs to classify the fracture simply, and whether the posterior ligamentous complex is injured can only be judged from the physical examination and the separation of the spinous process seen on the radiographs, which is easy to miss the diagnosis of the posterior ligamentous complex injury and has limited clinical guidance and cannot determine the prognosis. the Denis and Magerl classification is too complicated, its validity and reliability are low, and its clinical guidance is limited. The LSC classification only considers the stability of the spine in terms of mechanical force transmission, without considering the nerve injury and the important role of soft tissues on the stability of the spine, and cannot guide the need for surgical treatment. the TLICS classification combined with MRI examination, especially the T2-weighted sagittal images of the lipid suppression sequence, has significantly improved the sensitivity of the diagnosis of posterior ligamentous complex injury, and can consider the spine comprehensively. The classification has good reliability and validity, and can be used to guide clinical work based on the classification results, which is a classification method worth promoting and applying in clinical practice. However, the classification still needs to be improved, especially whether the quantitative criteria truly reflect the degree of spinal stability, and clinical trials are needed to verify the significance of the classification scoring system for guiding clinical treatment and predicting prognostic ability.