Odontoid fractures are the most common injury of the axis – often resulting in atlantoaxial instability. They are especially common in the elderly. The mean age of the patients in our study was 81 years. This correlates well with the results of Ryan and Henderson, who demonstrated that type II fractures according to the Anderson and d’Alonso classification were most common in people aged 70 years and older [18].
In our study, most injuries were attributed to a simple low energy fall.
Contrary to the investigations of Ryan and Taylor, who found a high incidence of concomitant spinal cord injuries especially in the elderly, no neurological deficits could be documented in our study [11]. Other previous investigations are in line with our results showing a low incidence of concomitant neurological deficits [9].
Odontoid fractures – diagnosed as well as missed- pose a difficult clinical problem in elderly patients with an alarming rate of complications, such as respiratory problems, non-union, pain, and death. At the same time, the indication for surgical treatment of odontoid fractures in the elderly as well as choice of procedure remains controversial with no present consensus.
Our clinical study focused on the outcome of patients undergoing external splinting, anterior screw fixation and dorsal C1/C2 fusion. We attempted to elaborate advantages of each treatment modality by evaluating fusion rate, magnitude of fracture angulation and mortality.
In all non-operatively treated patients, the follow up examination showed non-union of the type II fractures (Figure 2). All patients were treated with a cervical orthosis for at least six weeks. One possible cause for the high rate of non-union may be the on-going motion of the cervical spine, even after properly placed external splinting [10]. Residual C1/C2 instability may cause persistent pain, upper extremity paraesthesia and myelopathy [4]. In our opinion, this is therefore an unacceptable result.
Although a posterior displacement of the odontoid could be observed in 62,5% of the non-operatively treated cases, all patients were free of neurologic symptoms. However, there are but a few data in literature describing a relationship between the displacement of the odontoid and myelopathy. For instance, Ryan and Taylor found an incidence of 70% of posterior displacement in patients with myelopathy and Müller et al. mentioned that neurological impairment was associated with posterior displacement of the odontoid in the majority of the cases in their cohort [11, 19]. The absence of neurological symptoms in the present study can be explained through Steele’s rule of thirds [20]: The antero-posterior diameter of the ring of the atlas spans approximately 3 cm. The spinal cord and the odontoid process are each approximately 1 cm in diameter; the remaining centimetre of free space allows for some degree of displacement without any neurological sequelae to be expected.
The mortality rate in the non-operatively treated group in our series was 85%, which is substantially higher compared to reports from other clinical series [9–11, 21]. However, the observation period in our cohort was quite long and our study focused on the elderly only. Our results are further relativised when considering that Müller et al. demonstrated a 35% in-hospital mortality rate following odontoid fractures in patients aged >70 years [19]. Reasons for the increased mortality in many cases are respiratory-related complications due to prolonged periods of immobilisation.
As mentioned, several options are available for treating odontoid fractures, but currently there is no consensus concerning the best treatment. Surgical options include anterior odontoid screw fixation and posterior atlantoaxial arthrodesis, which is most commonly combined with agglomeration of autologous bone. Both treatment modalities allow early and effective patient mobilisation, which decreases complication rates such as respiratory failure, pneumonia and cardiac arrest [22]. Furthermore, the overall reported high rates of pseudarthrosis with this fracture pattern are substantially lower in the operatively treated groups compared to a non-operatively treated cohort [23].
Anterior direct fracture stabilisation has shown good results in young patients (Figure 4). It has become a popular choice of treatment in noncomminuted fractures that can readily be realigned. The advantage of this procedure is the less traumatic approach allowing rapid postoperative mobilisation. Furthermore, a reduction of morbidity and mortality in comparison to external splinting has been previously extensively investigated and was confirmed by several studies [21, 24, 25]. In a study conducted by Chiba et al., the authors concluded that anterior screw fixation was the best therapeutic option, but also mentioned the need for bone of decent quality for adequate screw fixation [26]. In a young patient cohort, fusion rates of up to 95% for single screw fixation were found [27]. Significant differences in union rates when using one- versus two-screw fixation techniques in anterior fixation could not be demonstrated in a number of studies [28, 29] and satisfying clinical results have been recorded with single screw fixation [30, 31]. As mentioned above, anterior fixation was performed by the use of one cannulated screw in our study.
The rate of delayed- or non-union was 77% in our anterior fixation cohort, which is rather high compared to the figures found in literature (Figure 3). However, it is necessary to mention that the patients included in our study were significantly older compared to most studies [27]. Accordingly, the rate of osteoporotic bone was also substantially higher. This is in line with findings in previous investigations, where a high complication rate following anterior fixation could be demonstrated in the elderly [14]. Further causes for these poor results may also be higher rates of comminution at the fracture site or stiffness of the cervical spine preventing ideal positioning of the screw. Furthermore, concomitant thoracic kyphosis or barrel chest deformities make the anatomic reduction difficult in elderly patients. Due to these facts, some authors have come to view this kind of operation as contraindicated in such cases [24].
In contrast to anterior fixation, posterior fusion of C1-C2 results in a high rate of bony union. In a study by Omeis et al., a fusion technique was the treatment of choice and the reported fusion rate was above 90% [13]. Nevertheless, loss of motion at the atlantoaxial joint will follow with this treatment [14] as the high bony fusion rate is achieved at the cost of an almost 50% reduction of cervical rotation and a 10% reduction for flexion and extension [22]. However, the assumption that anterior fixation preserves the atlantoaxial motion is merely theoretical with a reported reduction of up to 50% for C1-2 range of motion following anterior fixation as well [32]. Our results confirm this hypothesis, as the range of motion after posterior fusion and anterior screw fixation was comparable in our study.
Posterior atlanto-axial fusion can be obtained in different ways. We used the screw-rod fixation according to Harms. Correct positioning of the screw-rod system with a complete bony fusion of the posterior structures could be observed in all of the follow-up patients with a posterior C1-C2 fusion. This is in accordance with recent literature demonstrating healing rates of around 93% following posterior atlantoaxial fusion [5, 33]. The advantage of this method is the ability to achieve stability without need for prior anatomical reduction of the atlantoaxial articulation. Therefore, thoracic spine and chest deformities in elderly patients have no further negative impact. Furthermore, posterior fusion can also be performed as a salvage operation when anterior screw fixation has failed.
We acknowledge several limitations of the present study. First, due to the retrospective study design we were dependant on complete and accurate patient medical charts to evaluate the physical condition on admission. However, although data collection was performed in a routine setting by trained personal of the Trauma center, we could not ensure with final certainty the completeness of our data. Second, the study was conducted at a single designated trauma centre without randomisation making comparison of the different groups difficult. However, physical conditions were comparable in all three groups. Finally, the relatively low sample size makes further investigation necessary in order to confirm the clear trend demonstrated in our present study and to optimise the statistical power.