During the observation period of 10 years, 189 patients (mean age 72 years, SD 19, range 6–97 years) were treated with a fracture of the atlas. The majority of the patients (108/189, 57.1%) were of male gender. Leading trauma mechanism was a low-energy trauma (113/189, 59.8%), in three of which this was associated with a metastatic lesion of the first vertebra.
Of 189 atlas fractures included, only 63 were isolated injuries. One or more concomitant injuries to the head and spine were present in 66.7% (Fig. 2), a concomitant fracture of the cervical spine (C2–C7) in 59.8%, and at least one fracture of the whole spine (C2–L5) in 64.6%. A concomitant head injury was found in 25.9% of the cases, and in 15/189 patients (7.9%) this included intracranial hemorrhage. Fractures of the occipital condyles were seen in 10/189 cases (5.3%). A combined atlanto-axial fracture was present in 107/189 cases (56.6%), and concomitant fractures of the thoraco-lumbar spine were observed in 23/189 patients (15.4%).
When classified by Gehweiler, 23.3% represented type 1 fractures, 22.2% type 2 fractures, 32.8% type 3 fractures, 19.0% type 4 fractures and 1.1% type 5 fractures (Fig. 3). Two fractures could not be classified: one combination of an anterior arch and a comminuted lateral mass fracture and one transverse split of the whole ring. In one patient, only MRT imaging was available that did not allow for a proper classification.
Two of the 189 patients were treated in an outpatient setting only. The mean duration of hospitalization was 12 days (SD 10, range 1–77). The average time between trauma and treatment was 4 days (SD 7, range 0–44). In-hospital complications included thromboembolic events in 2/187 patients (1.1%), surgical-site infection (n = 2) or systemic infection (n = 25) in 27/187 patients (14.4%), and delirium in 18/187 patients (9.6%). All-cause 30-day mortality was 12.2% (23/189). Causes of death could not be attributed to the atlas fractures except for three patients with cardiac arrhythmia and/or respiratory failure on admission that might be related to brainstem compression. Seven patients died before treatment could be applied.
Treatment consisted of non-operative treatment with soft or rigid collars in 82/182 (45.1%), with halo fixation in 13 (7.1%) and with open surgical procedures in 87 patients (47.8%, Table 1). However, in many cases, the indication for surgery was made due to concomitant injuries of the adjacent cervical spine segments. In isolated fractures of the atlas without concomitant injuries to the occipital condyle or C2–C7 that were amenable for treatment (n = 67), non-operative treatment with a soft or rigid collar was performed in 45/67 cases (67.1%), halo fixation in 4/67 (6.0%) and open surgical fixation in 18/67 patients (26.9%, Fig. 4).
Secondary interventions during the whole follow-up were necessary in 14/189 patients (7.4%); this included one patient who had a re-fracture of the atlas after a second fall, three patients with non-operative treatment that required secondary stabilization with halo fixation, and two patients with implant removal after more than a year, three patients with halo fixation that required secondary C1/2 fusion. Revision surgery of open fixation was necessary in 5/87 (5.7%) cases; this included surgical-site infection in two and loss of reduction in three cases.
Patients older than 65 years were more likely to be female (p < 0.001, Table 2), sustain a low-energy trauma (p < 0.001) and to suffer of in-hospital complications including death (p = 0.004). In addition, patients older than 65 years were more likely to present with fractures Gehweiler type 1 and 2 while younger patients more frequently had fractures type 3 and 4 (p = 0.008). Younger patients had more concomitant fractures of the occipital condyle (p = 0.028) and elderly patients had more combined C1/C2 fractures (p < 0.001).
Patients who were treated non-operatively had a shorter hospital stay (p = 0.002, Table 3) but were more likely to die during the hospitalization. The difference in hospital stay was still significant when comparing only patients who did not die during their initial hospitalization (non-operative: 10 days, SD 10; operative: 14 days, SD 11; p < 0.001). The occurrence of in-hospital infections (p = 0.836), thromboembolic events (p = 1.0) and delirium (p = 0.806) was not different between operatively and non-operatively treated patients. Patients with a high-energy trauma (p = 0.028) and patients with a concomitant fracture of C2 (p < 0.001) were more likely to receive operative treatment.