With the development of the spinal corrective techniques and the advancement of the instrumentation, severe and rigid scoliosis which used to be difficult to correct became manageable. At present, the definition of severe scoliosis remains controversial. Greiner et al. determined that AIS patients did not exhibit clinically significant respiratory symptoms until their curves were 60 to 100°, so he defined severe scoliosis as Cobb angle larger than 60°. Lenke et al.  have defined it as Cobb angle ≥ 70°, and Tokunaga  thought that Cobb angle > 80° could be treated as severe scoliosis. As for the rigid scoliosis, its definition was also unclear until recently. According to author's clinical experience, the results of one stage posterior surgery for the scoliosis with a coronal Cobb angle less than 70 degrees and a flexible index on Bending films more than 40% was satisfactory. Therefore, patients with severe and rigid scoliosis were recruited in current study with a coronal Cobb angle larger than 70° and flexible index on bending films less than 40%.
The aim of the anterior spinal release was to increase spinal flexibility and to improve subsequent correction rate at posterior instrumentation [12, 13]. Tokunaga et al.  reported that staged surgery including anterior release was an effective surgical method for patients with severe scoliosis, where a rigid curve or the risk of neurological complications due to acute forceful correction may exist. Mehlman et al.also reported that the spinal release and halo-femoral traction protocol outlined offer a safe, controlled approach to the reduction of severe spine deformities before fusion. In current study all the patients received anterior spinal release first.
Traction as a method of correction of spinal deformity could be dated back to 3500 BC . Perry and Nickel first introduced the halo device in 1959  during which time a jacket or cast was used for caudal support. Then several other count-traction methods were invented: halo-gravity, halo-pelvic and halo-femoral traction [17–19]. In terms of halo-femoral traction, Kane et al.  reported their series of 30 scoliotic patients in 1967. The average original curve measured 112° and reduced to 58° after final correction. Four patients got pin-site irritation and the pins were reinserted. Paresthesia developed in 3 patients, and 1 had abducens nerve palsy; all the symptoms recovered with the reduction of traction forces. Details about the types of curves treated and specific treatment regimens were not provided in this paper. Bonnett et al.  reported that preoperative halo-femoral traction resulted in 57% correction of scoliosis as well as 53% correction of pelvic obliquity in 37 patients with paralytic scoliosis. Arlet et al.  reported on the usage of halo-femoral traction to treat a 17-year-old girl with congenital scoliosis of 145° and cor pulmonale. Correction of the deformity and improvement in pulmonary function were well achieved. Huang et al.  reported on the usage of intra-operative halo-femoral traction to treat severe scoliosis and associated pelvic obliquity in a 14-year-old patient with cerebral palsy. The patient underwent one stage anterior and posterior spinal fusion, the posterior procedure was performed with the patient under halo-femoral traction. The patient responded well to the surgical intervention and had a stable correction of his pelvic obliquity. Mehlman et al.  assessed the effectiveness of spinal release and halo-femoral traction in the management of severe spinal deformity in 2004. Twenty-four patients were treated with halo-femoral traction at the interval between anterior spinal release and posterior surgery. The etiology of the deformity included IS, CS, Scheuermann's kyphosis, Neuromuscular scoliosis, and Osteogenesis imperfecta. The correction obtained after Halo-feromal traction averaged 59% (ranged 14–100%).
In current series, compared with CS with similar curve magnitude, the patients with severe and rigid idiopathic scoliosis were slightly more flexible on side bending film (IS 24.3% correction vs. CS 22.5%). Curve correction obtained after traction has a significant improvement when compared with the correction obtained from side bending film in our study. This statistically significant difference confirms the efficacy of the technique of Halo-femaral traction. We also found that the average correction obtained from posterior fusion was 57.5% in IS group, significantly higher than that in CS group (45.2%, p < 0.001). Current results demonstrated less overall curve correction rate when compared with the reports of Kane et al. , Bonnett et al.  and Mehlman et al. . This may be due in part to lower traction forces used in our study (only 36% of the average body weight) than Mehlman study (54% of the average body weight). Furthermore, the curves in Kane and Bonnett's study were less rigid than current study.
Leatherman  first described a two-stage procedure for the treatment of congenital scoliosis. In his study, the mean curve correction obtained after the second stage was 45.6% and the correction of kyphosis was 44.4%. Author's results demonstrate that after posterior surgery the curve correction obtained averaged 45.2% and the thoracic kyphosis magnitude decreased to 39.0° (ranged 11°–82°) with average correction rate of 43.5%. Although the curve correction rate in two studies were similar, the initial curve angel of CS patients in current study were far more serious than that in Leatherman's study. Therefore we could conclude that Halo-femaral traction had a enormous effectiveness for the correction of patients with severe and rigid congenital scoliosis.
Severe coronal curve usually associated with significant deformity on sagittal plane. In current study, twenty-three patients with IS and twenty-five CS patients had pre-operative thoracic kyphosis (T5–T12 > 40°). Compared with the IS patients, the mean pre-operative thoracic kyphosis for CS was significant higher (70.2° vs. 50.6°). Combined with Halo-femoral traction, modern spinal instrumentation system provided good correction on sagittal plane for severe scoliosis. Thoracic kyphosis of patients in our study corrected well after posterior surgery, especially for CS patients.
Complications related to the halo itself included pin loosening, superficial, and deep pin tract infections. Brain abscess has also been previously described with halo pins . Halo-femoral traction compiled certain neurological complications . Rinella  reported a total of 42 consecutive patients with severe operative scoliosis, kyphoscoliosis, or kyphosis treated with halo-gravity traction. Triceps palsy (2.38%), and brachial plexus palsy (2.38%) occurred during halo traction. Traction-related complications were also encountered in our study. In the present study, 4 cases suffered from brachial plexus palsy (1 CS patients, 3 IS patients). All patients restored their complete neural function at two months follow-up. The most likely cause of the injury was thought to be due to the hyper-abduction of the arm and over-stretched of the brachial plexus. Brachial plexus palsy associated with Halo-femoral traction in severe and rigid scoliosis was a temporary, revertible damage to nerve function. If the symptoms were promptly detected and rehabilitation training and appropriate medication were prescribed timely, complete nerve functional restoration could be achieved.
Spinal cord injury and paralysis were the most serious complications of spinal corrective surgeries. Cotrel  reported that the incidence was 0.8%. Patients with severe and rigid scoliosis were thought to be at greater risk of these complications. Some authors advocated rapid correction via one stage anterior release and posterior surgery for patients with severe scoliosis without an intervening period of traction . Long term follow-up and big sample size were mandatory to support these one-stage or "rapid correction" conception. Our results showed that Halo-femoral traction was a safe, well-tolerated and efficacious method in the treatment of this formidable disease. Combined with anterior spinal release and posterior fusion, it could notably reduce the incidence of severe complication such as spinal cord injury, at the time of good correction of severe spinal deformity. In addition, curve correction obtained after traction was significantly superior than that achieved on side bending film in current study, therefore the pre-operative side bending radiography may not able to accurately predict the correction rate of posterior instrumentation for severe scoliosis.