Paediatric spine injuries, even in spine centres, are rare in clinical practice, and their appearance varies from 1 to 4% [1]. The predominant mechanism of injury leading to spinal injuries in younger children is falls from larger heights and bicycle and sports accidents [2,3,4,5].

Currently, thoracic and lumbar spine injuries constitute about 20 to 40% of all paediatric spinal injuries [6,7,8]. In particular, fractures of the thoracolumbar junction have shown to have an incidence of 0.6 to 0.9% and being associated with an age between 14 and 16 years and male gender [9, 10]. The analysis of spinal injury incidences in young patients present with significant regional differences. Transferring other epidemiological data to Europe or even Germany would not be appropriate [11].

The aim of this retrospective multicentre study, initiated by the spine section of the German Association of Orthopedic and Trauma Surgeons (DGOU), was to provide epidemiological data on paediatric patients with thoracic and lumbar spine trauma in Germany to improve the quality of the decision-making in relation to diagnostic pathways and any necessary future therapy.

Materials and methods

This retrospective multicentre study was initiated by the paediatric spinal trauma working group, the spine section of the German Society for Trauma and Orthopaedic Surgeons. It included all paediatric patients up to 16 years of age with spinal column injuries who were admitted to one of six level I spine centres between January 2010 and December 2016. The participating clinics were located in Aachen, Dresden, Karlsbad, Leipzig, Ludwigshafen and Murnau and were thus distributed all over Germany. Patients were eligible for further analysis if they met the following inclusion criteria: age between 0 and 16 years; thoracic and/or lumbar spine injury; a fully available diagnostic workflow, especially for spinal injury classification according to the established AOSpine Classification and the estimation of neurological deficits according to the ASIA Impairment Scale (AIS) [12].

Demographic data, circumstances of the accident, overall injury severity score (ISS), kind of image diagnostic, associated injuries and complications were documented. The measured outcome parameters were length of hospital stay (LOS) and onset of complications during the hospital stay.

The trauma mechanism was recorded out of the medical chart and classified as:

  • Fall < 3 m.

  • Fall > 3 m.

  • Car accident.

  • Bicycle accident.

  • Sports accidents.

  • Equestrian sports.

  • Jump in shallow water.

  • Others.

Patients were divided into three age groups using the classification proposed by Meinig et al. [13]: age group I (0–6 years), age group II (7–9 years) and age group III (10–16 years).

Data were collected from the in-house clinical documentation systems. Counts and frequencies were used to describe the sample. Correlation analysis (Spearman) was used to determine any dependence between the variables. A group analysis was performed using the Student’s t-test. Significance was set at p < 0.05 for all statistical tests. All testing procedures were performed exploratively, so no adjustment for multiple testing was made. Statistical analyses were performed using IBM SPSS software (version 23; IBM Corp., Armonk, New York, NY, USA). The study complies with the principles of the Declaration of Helsinki (2013), and at each centre, the present retrospective multicentre study was approved by the local ethics committee. Approval for the main study centre was granted by the ethics committee of the State Medical Association of Rheinland-Pfalz, Germany (file number: 837.295.17).


A total of 153 paediatric patients with spinal injuries within the thoracic and lumbar spines were identified. The patient cohort consisted of 65 female (42.5%) and 88 male children (57.5%), with a mean age at the time of injury of 12.9 (± 3.1) years. The most common injury patterns were falls from lower heights (n = 48, 31.4%) followed by traffic accidents (motor vehicle, bicycle and pedestrian accidents; n = 45, 29.4%) and falls from heights more than 3 m (n = 26, 17.0%). The thoracic spine was predominately affected (Table 1). Higher age was correlated with high-impact trauma (r = 0.151, p = 0.005).

Table 1 Baseline demographics for spinal fractures within the thoracic and lumbar spine

A total of 9 (5.9%) paediatric patients were included in age group I, 13 (8.5%) in age group II and 131 patients (85.6%) in age group III. Ninety-five patients (62.1%) were treated conservatively, whereas 58 (37.9%) children had to undergo surgery.


In the overall cohort, a total of 215 imaging modes were used to detect a potential spinal injury. Plain X-rays were performed in 83 (54.2%) patients. A total of 112 (73.2%) children and adolescents received additional imaging, including computed tomography (CT; n = 61, 39.9%) and/or magnet resonance imaging (MRI; n = 51; 33.3%). MRI with anaesthesia was necessary in eight patients (5.2%). At this point, it cannot be ascertained whether the anaesthesia was necessary due to MRI-associated sedation or anaesthesia that was already present due to the underlying general trauma. Dynamic fluoroscopic examination was performed in a minority of six (3.9%) paediatric patients. An expectable significant correlation between age distribution and diagnostics was found: older patients demonstrated a higher likelihood of needing a CT scan (r = 0.317; p = 0.001) as part of diagnostic imaging. Conventional radiographs, in turn, were more often used in younger children (r = −0.142; p = 0.122). While MRI was more frequently used in younger children (r = −0.135; p = 0.169), the use of thoracic and lumbar fluoroscopy was only used in a few cases (Fig. 1).

Fig. 1
figure 1

Distribution of the diagnostics used in the patients (data given in absolute numbers)

Additional injuries

A total of 172 concomitant injuries were observed. Both head injuries (n = 53, 30.8%) and injuries of the extremities (n = 52, 30.2%) dominated in the patient cohort, followed by chest injuries (n = 42, 24.4%). Abdominal (n = 6, 3.5%) and pelvic trauma (n = 17, 9.9%) was minor entities. Further spinal fractures located at the cervical spine were found in two cases (1.2%). There was a correlation between age and the number of accompanying injuries (r = 0.138, p = 0.012). No correlation was found between gender (r = −0.072, p = 0.195) and further trauma localizations (Fig. 2).

Fig. 2
figure 2

Absolute count of accompanying injuries

Injury location and spinal injury severity according to the AOSpine classification

A total of 256 injuries (74.2%) were located at the thoracic spine and 89 injuries (25.8%) within the lumbar spine. The detailed distribution is shown in Fig. 3. Most of the injuries involved the thoracolumbar junction (n = 90, T11 to L2). A total of 299 type-A fractures (86.6%) were observed in the cohort. Most of the paediatric patients showed type-A1 endplate fractures (n = 226, 65.5%). Twenty fractures were classified as segmental type B-injuries (5.8%), whereas 12 type C-injuries (3.5%) were found. Eighteen thoracic and 14 lumbar spinal injuries were associated with neurological impairment (ASIA D to ASIA A). Sprain and contusion injuries were seen in 14 cases (4.1%). Thoracic injuries led to significantly more neurological impairment in comparison with lumbar spine injuries (r = 0.121, p = 0.024). With an increasing severity grade of the spinal injury, the degree of a concomitant neurological deficit rises at an equal rate (r = 0.396; p = 0.001). The risk of neurological impairment is also higher with older age (r = 0.145; p = 0.008), and the older patients showed a higher graded severity of spinal injury according to the AOSpine classification (r = 0.188, p < 0.001). As expected, patients with neurological deficits demonstrated a longer overall hospital stay, with a maximum stay of up to 218 days (r = 0.401, p = 0.001).

Fig. 3
figure 3

Injury location (absolute numbers). The 14 cases of sprain and contusion injuries were not considered


In the total cohort, 95 patients (62.1%) were treated conservatively, whereas 58 children (37.9%) were treated surgically. The age distribution was heterogenous. No more than four children were younger than 10 years (range 3–9 years). Thirty-one treated injuries were located at the lumbar spine, while 27 injuries at the thoracic spine were surgically treated (Table 2 and Table 3). The used surgical techniques did range from percutaneous interventions, open instrumentation techniques to open stabilization including decompression. In 19 cases, percutaneous instrumentations were performed, and in 39 cases, an open surgical intervention was used. Out of the mentioned cases, 16 patients initially did present with some kind of neurological deficit (ASIA A to ASIA E). Eight out of them were treated with open stabilization and decompression within the injured spine segment. In one case of a C-type lesion at level T7, a combined posterior/anterior approach was performed. In another case, a secondary anterior cage implantation was necessary (> 3 months).

Table 2 Data of the 31 surgically treated patients at the lumbar spine (ASIA: American Spinal Injury Association; MIS: minimally invasive surgery)
Table 3 Data of the 27 surgically treated patients at the thoracic spine (ASIA: American Spinal Injury Association; MIS: minimally invasive surgery)


The mean hospital stay was 14.2 (range 1–218) days. A neurological deficit (r = 0.401, p < 0.001), a thoracic injury location (r = 0.141, p = 0.009), a higher spinal injury severity (r = 0.258, p < 0.001), a higher age (r = 0.270, p < 0.001) and necessary surgical therapy (r = 0.630, p < 0.001) led to a significantly longer length of hospital stay.


Regarding the treatment of spinal injuries, most of the patients showed no evidence of complications during the hospital stay (96.7%). Surgical-related complications were seen in five cases (8.6%): two patients (3.4%) presented with early implant failure, and in two cases a superficial wound infection occurred (3.4%). The development of a segmental hyperkyphosis with anterior/posterior revision after a follow-up of three months was necessary in one case (1.7%).


Paediatric spinal injuries can range from minor sprains and strains to severe conditions such as fractures, dislocations, or even neurological deterioration. The symptoms of spinal injury in children vary widely, but usually include pain, weakness and reduced mobility. Missed spinal injuries in children can lead to significantly impaired range of motion and may lead to future deformities. An exclusion of a spinal injury and a reliable diagnosis are therefore essential. In comparison to adults, however, the circumstances are often more difficult, leading to a higher probability for uncertainties during the diagnostic process. The medical history is often aggravated, and third-party case history is impossible. Physical examination can be of reduced value as patients do not tolerate investigations, making any necessary differentiation (e.g. ASIA score) impossible. Plain X-ray diagnostics do not always offer sufficient sensitivity, and further radiological diagnosis is most often more complex to perform [14]. Due to the described rarity of spinal injuries, doctors often lack the necessary knowledge and experience to treat spinal injuries in children. Epidemiological data on this topic can help; however, the literature is limited to data from monocentric case series.

The present multicentre study represents the largest patient cohort from German-speaking countries, including 153 children with injuries within the thoracic and lumbar spine. While the thoracolumbar junction was the most affected region in all patients, older children and adolescents suffered from injuries within the middle thoracic spine. Almost 1/3 of the children presenting with thoracic and/or lumbar spine injuries were treated surgically. The surgical procedures varied from percutaneous instrumentation to open instrumentation and decompression. Single anterior approaches were of minor relevance in the present cohort. The average age of the patients included in the current study was 12.9 (± 3.1) years and so comparable to the results of Kraus et al. [5]. In a patient cohort of 86 children with thoracolumbar injuries, a gender ratio of 1.5:1 (male/female) was found. In contrast, the present study showed a more homogeneous gender distribution of 1:1.3. Statements in other studies that almost exclusively boys do suffer from spinal injuries could not be verified for the described cohort [11]. Several studies have shown that boys are likelier than girls to sustain spinal injuries due to traumatic events, such as sports injuries, motor vehicle accidents and falls [15]. The main trauma mechanisms that cause spinal injuries have been shown to be age related. In the international literature, traffic accidents are described as the main causes of spinal injuries at the thoracolumbar spine [16]. Knox et al. found road traffic accidents to be the most common cause of spinal injuries in children aged 0 to 3 years (49%) and children aged 4 to 9 years (58%). This is in contrast with the results of the present study, in which car accidents are only the second most common cause (29.4%) of paediatric spinal injuries in all age groups. In the presented cohort, falls from a height < 3 m were the most common cause of injury (31.4%) independent from patients age. The reason for that remarkable difference might be the varying definitions used for traffic accidents in different studies. In the present study, car accidents were differentiated from pedestrian and two-wheeler accidents. If all of them were summarized under the term "traffic accident", the numbers would undergo a significant rise to 34.0%. Accident events that were described to end up in spinal injuries in adult patients do not differ to the ones found in children [17, 18]. As high-energy impacts are responsible for causing paediatric spinal trauma, they are usually accompanied by relevant concomitant injuries. According to the literature, craniocerebral and thoracic trauma, along with injuries of the extremities, are the most frequent concomitant injuries in adults and children [18,19,20]. Both head injuries and injuries of the extremities dominated in this patients’ cohort. Depending on the underlying trauma mechanisms, the probability of further spinal injuries in other segments and spinal regions does rise significantly, with a combination of cervical and thoracic spine injuries being the most common. The risk of concomitant non-contiguous spine injuries has been well characterized in adults and varies between 4 and 11% [21, 22]. This finding could not be fully confirmed by the present study. No more than 1.2% of the patients had an additional injury in another spinal segment. In children, on the other hand, neighbouring vertebrae are more frequently affected [23]. As not all patients undergo MRI investigations, low-grade injuries in particular might not be diagnosed. Although there are no national guidelines, plain radiographs are used as an initial imaging technique. In general, plain radiographs can be used to detect most osseous injuries [24]. Analysis of the diagnostics showed that more than half of the children in all age groups were exposed to radiation exposure induced by X-ray (54.2%). In suspicion of an undetected spine trauma, Srinivasan et al. recommended a CT as a further diagnostic tool [15]. In more than two-thirds of the present cohort, further diagnostical imaging was necessary. Clark et al. showed that CT and MRI imaging are recommended in the case of inadequate conventional imaging [25]. MRI is especially useful in ruling out lesions that may be missed on CT, such as epidural hematoma, disco-ligamenteous injuries, or traumatic disk herniation. In our study, 39.9% of the patients received a CT scan, which was also for the purpose of preoperative planning. The main concern with regard to patient safety with both, radiographs and CT, is radiation dose, especially as the infantile thyroid gland is very sensitive to radiation [26]. An MRI was performed in one-third of patients in the present cohort. This comparatively high rate of MRI diagnostics can be explained by their increasing availability, and not at least because of the lack of radiation exposure associated with the technique. Whenever feasible, an MRI is considered as the initial imaging procedure in cases of suspected spinal injuries in children [27]. In contrast with the current literature, the need for sedation was unexpectedly low (5.2%; n = 8) in our cohort [28] and might be avoided in some indications.

In paediatric patients, the most commonly affected levels are L2 and L3 in contrast with the thoracolumbar junction in adults [29]. Compression fractures are the most prevalent ones in the paediatric spine, with most located near the thoracolumbar junction [16, 30]. In contrast with the literature, the thoracolumbar junction was the most affected region in the present cohort and the levels of L2 and L3 were affected less frequently. The most common fracture type in this cohort with 87% is type-A according to the AOS classification. Although a high proportion of B-injuries (especially type B2 lesions) is described in the literature, the proportion in this cohort is very low (5.6%) [31]. A reason might be the lower overall number of accidents that are often associated with B-type injuries. The proportion of children suffering from spinal trauma and concomitant neurological deterioration ranges from 2.5 to 35% throughout the literature [5, 30]. For adults, higher numbers of neurological deteriorations ranging from 20 to 31% have been published [18, 20]. Overall, more severe neurological deficits in children are rare and represent only 2 to 8% of the overall number of paraplegics [8]. Neurological deficits were also rare in the present patients’ cohort (9%). As expected, with a higher grade of spinal injury the severity of neurological deficit also rises (r = 0.396; p = 0.001). Higher patient age is also associated with more severe neurological deficits. The main indications for surgical treatment in the literature are malalignment, instability and spinal cord injury with neurological deficits. In the present cohort, a high number of children underwent surgical treatment (37.9%). This number is higher compared to previous studies, which have found that between 7.5% and 30% of patients with thoracolumbar fractures require surgical intervention [5, 16, 32, 33]. One reason may be selection bias, as solely highly specialized spine centres were involved in this multicentre study.

The basic goals of treating injuries to the thoracic and lumbar spine in children are comparable to the ones in adults—restoration of stability and alignment, as well as protection of neurological function [34]. Depending on the patients’ age and the type of malalignment, enormous correction potential is described in paediatric patients [35, 36]. The decision for non-operative treatment has to include radiological follow-up investigations to detect posttraumatic imbalance.

However, coronal spinal imbalance is less able to be corrected than alignment issues in the sagittal plane. Therefore, early surgical treatment is preferred [37]. In general, the main indications for surgical treatment were defined for malalignment in the frontal and sagittal planes, complete burst fractures (type A4-fractures), C-type fractures and the presence of neurological impairment [34].

Similar to the surgical treatment in adults, there are different surgical approaches available for the treatment of paediatric spinal fractures at the thoracolumbar spine. Besides the well-known open instrumentation with or without decompression, minimally invasive percutaneous pedicle screw placement (MIS) has shown promising results in adults as an alternative for fixation of burst fractures. This technique has the advantages of less time being needed for surgery, less blood loss than with open fixation, less pain and shorter length of hospital stay, and a comparable number of complications [38, 39].

Despite numerous reports on MIS fixation in the adult population and slightly less scientific reports on younger patients/adolescents, outcomes of MIS instrumentation in children have not been reported [40]. Although type C injuries are highly unstable injuries, MIS procedures were used in two cases in our cohort. Cui et al. also showed that temporary fusionless instrumentation can provide successful management of thoracolumbar spine injuries (also for B- and C-type injuries) in paediatric trauma patients [41]. In general, the principles of posterior instrumentation in the surgical treatment of adults are transferable to a younger patient cohort. Even in younger age groups I (0–5 years) and II (6–9 years), surgical stabilization is possible when indicated. However, smaller proportions and anchorage behaviour of bone or cartilage make procedures in smaller children challenging and no general advice can be given by our results. Based on our results, decision-making in younger patients is often driven on case-by-case basis.

Most of the patients showed no complications during their hospital stay (96.7%). Four patients showed implant failure or complications related to surgical intervention. However, assuming the number of 58 surgically treated children, the percentage of complications remains relatively low (n = 4/58, 6,9%). Although international standards on diagnostics and treatment are still missing, the risk of surgical interventions, especially in experienced centres, seems to be acceptable.

The present study has several known limitations. In this multicentre study, only highly experienced spine centres were included and a selection bias could not be avoided. The retrospective study design naturally limits the analysis, as an evaluation of the treatment results is not possible. Due to the rarity of the injuries, it was impossible to generate a sufficiently large number of cases to carry out reasonable statistical subgroup analyses of the individual injury levels and types of injury in relation to patient age. Therefore, only a limited number of conclusions could be drawn about different diagnostic strategies, surgical indications, approaches and techniques, or the outcome and follow-up, all of which influenced the recent published recommendation of the study group of this work [34]. There is a lack of evidence-based standardized diagnostics and treatment algorithms, both nationally and internationally. The work that has been presented in this paper can significantly contribute to the current literature. Prospective data acquisition should be used to overcome the described difficulties and develop consistent algorithms.


Thoracolumbar spine injuries in childhood affect female and male children almost equally. In all age groups, falls and traffic accidents were the most common causes. CT scan and MRI play the most important role in the diagnostic work-up aside from conventional X-ray as the initial diagnostical step. However, fluoroscopy of the thoracolumbar spine plays a minor role in the diagnostic workup. Typical spinal levels affected are located near the thoracolumbar junction (T11 to L2). Older children and adolescents showed significantly higher spinal injury severity in comparison to younger children. Accompanying neurological deficits are rare and related to higher-grade spinal trauma, reflecting the general severity of the injury. Most of the children were treated conservatively. Surgical concepts for adults are mostly transferable to older children, while younger children often need a case-by-case decision. Nevertheless, 37% of the hospitalized children in the present multicentre study required surgical treatment with an acceptable number of complications being associated with said treatment.