Nonoperatively treated type A spinal fractures: mid-term versus long-term functional outcome

Approximately 66% of spinal fractures can be classified as being type A fractures. Of those, 86% present without neurological deficits [13]. Often, these fractures are treated nonoperatively. Consequently, nonoperatively treated type A spinal fractures form the large majority of spinal fractures. This retrospective, cross-sectional study was conducted to assess long-term functional outcome after nonoperatively treated type A spinal fractures.

Our data show good results at both four years as well as ten years after the fracture. Functional outcome as measured by the VAS and RMDQ appears good at both measurements. A VAS score of 73–75 can be interpreted as an admirable figure. Furthermore, an average RMDQ score of 5 reflects almost no disability; compared to a healthy reference group (VAS score 92–93, RMDQ score of 0.5), patients are only slightly impaired [10, 17]. When considering these numbers, however, it should be kept in mind that patients with primary neurological deficit were not included in this study. It is well known that neurological injury negatively affects functional outcome in patients with a spinal fracture [14]. Our results using the RMDQ are similar to that of Siebenga et al. who investigated functional outcome seven years after nonoperatively treated type A spinal fractures; an average RMDQ score of 4.6 was found [21]. A study concerning functional outcome 16 years after nonoperatively treated type A spinal fractures reported a mean VAS score of 58 points [19]. Our results concerning VAS scores are more favourable. An explanation might be that in the aforementioned paper the study group comprised some neurologically injured patients. Recently a VAS score of 67 was found five years after nonoperative treatment for type A fractures [5]. Our mid-term VAS score is to some extent comparable. Tezer et al. as well as Butler et al. [6, 23] also found satisfactory results six years after nonoperative treatment of compression and “burst” fractures, similar to our results. Weinstein et al., as one of the first to study of functional outcome after spinal fractures, found a mean RMDQ score of 13, measured 20 years after nonoperatively treated “burst” fractures [26]. Our results seem better; a clarification might be found in the fact that Weinstein’s paper included patients with neurological deficit and that the fracture classification and distribution was dissimilar to our data.

No difference was found between functional outcome at the mid- and long-term assessments. Furthermore, the time between the two measurements did not show a correlation with VAS and RMDQ difference scores. This indicates that a “steady state” on a group level exists from (at least) four to ten years post injury. Previous studies mention a status quo in functional outcome ranging from two to four years after a spinal fracture [1, 7]. Data on the course of outcome after two to four years, however, are not available. To our knowledge, there is no previously published paper available studying the course of functional outcome after a spinal fracture in the same cohort.

As mentioned before, pain and neurological deficit can arise long after a spinal fracture (mostly after nonoperative treatment, but rarely after operative treatment) [25]. In these cases, operative treatment might be necessary [4, 25]. Late onset pain requiring operative interference has also been reported for type A fractures [3]. In our series however, none of the patients required surgery for late neurological symptoms or pain. Three patients had a higher RMDQ score at the long-term assessment (difference scores −13 and −8), indicating more impairment. Since the minimal clinically important change for the RMDQ is 3.5 points, the alteration in RMDQ scores in these patients indicates a clinically important change [16]. When looking closer at these patients, none of them had a significantly different gender, age, follow-up time, or fracture type compared to the rest of the study group.

Patient one, a 44-year-old woman, sustained a type A1.2 fracture of T12 due to a car accident in 1999. She was treated by direct mobilisation without brace. At the mid-term measurement (three years post injury) she had a RMDQ score of 7 and a VAS score of 23. At the long-term follow-up (8.1 years post injury), a RMDQ score of 20 was found and a VAS score of 14. After the regular clinical control visits, she never contacted us concerning severe pain. Her general practitioner had requested an X-ray in 2002 since the patient suffered from back pain (see Fig. 2). This X-ray showed osteoarthritis in the facet joints, which could explain her complaints. On the other hand, her VAS score had deteriorated 9 points, which is less than 20 points, the minimal clinically important change when using a VAS [16].

The second patient was a 44-year-old man, who had sustained a T12 fracture in 1997, type A1.2, due to a fall. He was treated by two weeks of bed rest followed by a brace. At the mid-term measurement (5.4 years post injury) he had a RMDQ score of 8 and a VAS score of 23. At the long-term follow-up (10.3 years post injury), a RMDQ score of 21 and a VAS score of 27 were found. After the regular follow-up visits, the patient was seen by a neurologist in 2004 because of low back pain and a strange feeling in both legs. No pathological neurological conditions were found. An X-ray showed spondylosis at L2-L3 (see Fig. 3). This could explain the patient’s discomfort.

The third patient was a 38-year-old male, who sustained a type A1.1 fracture of L3 in 2000 following a sports accident. In addition, he sustained a femoral fracture and an acetabulum fracture. The femoral fracture was treated by intramedullary nailing, the acetabulum fracture was treated nonoperatively. At the mid-term measurement (2.1 years post injury) he had a RMDQ score of 0 and a VAS score of 89. At the long-term follow-up (7.3 years post injury), a RMDQ score of 8 and a VAS score of 72 were found. Why this patient deteriorated is unknown.

Given the favourable outcomes found in our series, nonoperative treatment is an appropriate approach for type A spinal fractures without primary neurological deficit. Three patients however, had poor outcomes, although their fractures were classified as being the rather “simple” type A1.1 and A1.2. Why these patients had poor outcomes is unknown. As we made no MRI scans at that time, potential posterior ligamentous complex (PLC) injuries cannot be excluded. The PLC is important in maintaining stability in the spinal column, and rupture may result in instability and severe back pain [15, 23, 25]. As reported by Leferink et al., 30% of type A fractures (classified on CT images) appeared to be type B fractures (PLC lesions present) during operation [12].

We could not demonstrate a relationship between VAS and RMDQ difference scores compared to age, gender, fracture type, or time between measurements. Since no correlation was found between outcome and sub-classification, it is open to discussion whether such an extensive classification as the CC is required in daily practice. Perhaps there is need for a new, less extensive classification system, which gives more direction to treatment and uses MRI for detecting PLC injuries. The thoracolumbar injury classification and severity score (TLICS), as recently developed by the Spine Trauma Study Group, recognises the aforementioned criteria; possibly it will replace the CC in the future [11].

This study is subject to certain restrictions and limitations that are worth mentioning. When assessing outcome ten years after a fracture, scores on the questionnaires might be influenced by back pain unrelated to the spinal fracture. For example, pain may arise solely due to the normal process of ageing and osteoarthritis, not per se at the level of the injured vertebra. Furthermore, it is known that a variety of other factors, which we did not consider (e.g. chronic illness, lower education level) might influence back pain [9]. Finally, the small sample size might have introduced a type 2 statistical error [2].