Secondary Logo

Journal Logo


The Use of Traction in the Treatment of Severe Spinal Deformity

Sponseller, Paul D., MD*; Takenaga, Ryan K., MA*; Newton, Peter, MD; Boachie, Oheneba, MD; Flynn, Jack, MD§; Letko, Lynn, MD; Betz, Randal, MD; Bridwell, Keith, MD**; Gupta, Munish, MD††; Marks, Michelle, PT, MA; Bastrom, Tracey, MA

Author Information
doi: 10.1097/BRS.0b013e318184ef79
  • Free

Correction of large, rigid spinal deformities carries an increased risk of neurologic compromise.1 In addition, instrumentation anchor sites may fail with extreme corrective forces.2 Methods of preoperative traction have been devised to more slowly and completely correct severe deformities. A primary benefit of traction is that correction of a severe curve can be done gradually so that less demand is placed on the bone-anchor interface when eventual instrumentation of the curve is accomplished. Also, except for periods of sleep, the patient is awake while in halo-gravity traction, which allows continuous neurologic monitoring. External traction is applied with cranial traction through a halo and counter-traction through the femur, pelvis, or body weight (gravity). The literature contains uncontrolled series of all of these methods (Figures 1 and 2).3–8

Figure 1
Figure 1:
Severe idiopathic scoliosis in a 15 + 4-year-old female patient treated with traction [A, preoperative scoliosis (146°); preoperative kyphosis (51°); B, week 1 in traction; C, last follow-up (8 years 1 month) scoliosis (43°); kyphosis (72°)].
Figure 2
Figure 2:
Severe idiopathic scoliosis in a 12 + 1-year-old female patient treated without traction [A, preoperative scoliosis (111°); preoperative kyphosis (55°); B, first postoperative erect; C, last follow-up (2 years 7 months) scoliosis (38°); kyphosis (49°)].

Shortfalls of the prior studies, however, are the lack of a comparison control group, as well as relatively small number of idiopathic scolioses studied. Because of the limitations of these studies, the member of the Harms Study Group performed a multicenter, retrospective, nonrandomized comparison study. The primary goal of the study was to compare the surgical correction of large scoliotic curves in patients with idiopathic scoliosis (IS) with traction and without traction.

Materials and Methods

Retrospective data were collected by 9 centers of the study group. The study was surgeon randomized. Surgeons who preferred to use traction for their large or rigid curves were compared with surgeons who did not. Thirty consecutive cases treated after 1995 with external traction (Tx group) before definitive fusion from 11 centers were collected. These were cases with large rigid curves greater than 90° or having flexibility less than 25%. Control (C group) cases of scoliosis greater than 100° or kyphosis over 120° having correction of less than 25% on traction or bending films were collected from 9 centers. A slight difference in the inclusion criteria was established to prevent control cases of large curves, which were still flexible but not treated with traction. All curve etiologies were included. Patients were 18 years old or younger at the time of surgery and had a minimum of 2 years of follow-up.

Technique of Halo Traction

The halo was applied at the time of anterior or posterior spinal release (20 patients) or, if none was performed (10 patients), it was applied with sedation and local anesthesia. The pins were tightened to 6- to 8-in pounds in older children or adults, assuming normal cranial bone density. Traction application was standard with halo principles according to age and initial weights of 5 to 10 lb.9 Traction was started immediately with approximately 5 to 10 lb of weight for young children and 10 to 15 lb for those closer to maturity. It was gradually increased 2 to 3 lb per day as tolerated up to a maximum 33% to 50% of body weight. In most cases, the head of the bed was inclined upward. Skin was inspected regularly; this was especially important in patients with kyphosis. In most centers, patients were upright in a halo wheelchair or walker during a portion of the day. The traction was applied continuously throughout the day. Neurologic checks of the upper and lower extremities were performed 3 times per day and cranial nerve function was checked once per day. The duration of preoperative halo traction varied from 2 to 12 weeks depending on the curve, its response to traction, and surgeon discretion. Patients with borderline pulmonary or nutritional reserve had longer periods of traction to optimize their nutrition and their pulmonary restrictive defect.

A total of 53 patients with severe scoliosis or kyphosis were studied including review of hospital records and radiographs. Thirty of these patients were treated with traction and 23 were treated without traction. Patients within each group were analyzed based on demographics (age, height, weight), diagnosis, perioperative data (time, blood loss, complications, hospital stay, vertebral column resection, instrumentation used), and radiographic data (main coronal curve, coronal compensatory curve, sagittal plane curves, flexibility, percent correction, and spine length). Patients within the traction group were also analyzed based on time in traction, whether a spinal release procedure was performed before or after traction, and type of distal counter traction.

A subgroup of 23 adolescent idiopathic scoliosis (AIS) patients was studied separately. Fifteen of these were treated with preoperative traction and 8 were not. Patients within each group were analyzed based on the same data mentioned above.


Summary of All Patients

A total of 30 patients were treated with traction and 23 without traction. Patient demographics were similar with the mean age at surgery for the Tx and C groups being 14 and 13, respectively (Table 1). Mean weights were not statistically different at 40 kg (Tx) and 32 kg (C). The average pretreatment curve magnitudes for the Tx versus C groups were as follows: main curve 93° versus 105°, compensatory curve 67° versus 56°, and kyphosis 63° versus 56°. The traction and control groups had mean flexibility of 22% and 21%, respectively, and spinal lengths of 34 cm and 32 cm, respectively. Those who underwent traction had almost twice the average hospital stay as the control group: 36 days versus 14 days (P = 0.011) and there was a trend toward a difference in the total complication rate (27% Tx vs. 52% C) (P = 0.058). The operative times were not significantly different (Tx: 380 minutes vs. C: 406 minutes), nor was the blood loss at the time of definitive fusion (Tx: 1957 mL vs. C: 2021 mL). See Tables 1 and 2 for complete statistical data.

Table 1
Table 1:
Patient Characteristics (n = 53)
Table 2
Table 2:
Operative, Postoperative Data, Instrumentation (n = 53)

At 2-year postoperative, the mean percent correction of the main coronal curve was 62% in the halo traction group, and 59% in the nontraction (C) group. The mean percent correction of the compensatory curve was 55% in the halo traction group and 51% in the C group. The kyphosis decreased 10° in the Tx group and 21° in the C group. None of the differences in these results was statistically significant (Table 2).

Traction was used for a mean of 4.9 weeks before surgery (range, 1–19). Femoral pins were used as distal counter traction in 6 patients; in all others it was gravity in bed, with or without additional time in a wheelchair (14 patients) or walker (8 patients). There was a transient (3 days) leg paresis in 1 patient with halo-gravity traction but no other serious complications of the traction itself. Both groups had 1 instance of an intraoperative pulmonary and neurologic complication. The neurologic complication in the traction group was the loss of spinal cord monitoring during operation; however, no motor deficit resulted. The neurologic complication in the control group was abnormal somatosensory-evoked potentials. However, no neurologic deficit resulted from the surgery. The control group had 2 cases of intraoperative instrumentation complications, whereas the traction group only had one such case. The control also had 3 “other” intraoperative complications, and the traction group had 2 other complications. After surgery, the non-AIS patients in the control group had 5 other complications, which included complications such as back and neck pain, chest pain, problems with brace fitting, poor feeding tolerance, which required reintubation, and adding on. On the other hand, the non-AIS patients in the traction group only suffered one other complication, which was a curve progression requiring extension of fusion. During the deformity correcting surgery, vertebral column resection was more commonly performed in the C (30%) than Tx (3%) group (P = 0.015). The types of instrumentation were similarly distributed across the Tx and C groups, with the majority being hybrid instrumentation in both (80% Tx, 70% C) (Table 2).

Summary of Only AIS Patients

Of the 53 total patients with severe scoliosis or kyphoscoliosis included in the study, the most common diagnosis was IS (18 Tx and 11 C). Of the 29 patients with IS, 2 had infantile onset, whereas 4 had juvenile onset. Thus, there were 23 patients with AIS included in our study. Fifteen received halo-gravity traction and 8 did not. Patient demographics were similar. The mean age at surgery for the Tx and C groups was 14 and 15, respectively (Table 3). Mean weight and height for the Tx group were 43 kg and 154 cm, respectively, although they were 41 kg and 152 cm, respectively, for the control group. The average pretreatment curve magnitudes for the Tx versus C groups were as follows: main curve 97° versus 93°, compensatory curve 64° versus 61°, and kyphosis curve 50° versus 42°. The traction and control groups had mean flexibility of 19% and 20%, respectively. Both groups also had similar spinal lengths (37 vs. 38 cm). However, those who underwent traction had almost twice the average hospital stay as the control group: 41 versus 21 days (P = 0.212). As in the study population as a whole, the operative times (405 vs. 496 minutes) and blood loss (2057 vs. 2975 mL) were also not statistically significantly different. However, during the deformity correcting surgery, vertebral column resection was more commonly performed in the C (25%) than Tx (0%) group, and this difference was statistically significant (P = 0.043). See Tables 3 and 4 for complete statistical data.

Table 3
Table 3:
AIS Patient Characteristics (n = 23)
Table 4
Table 4:
AIS Patient Operative, Postoperative Data (n = 23)

At 2-year postoperative, there were statistically insignificant differences in the mean curve corrections. The mean percent correction of the main coronal curve was 64% in the Tx group, and 61% in the C group. The mean percent correction of the compensatory curve was 56% in both the Tx and C groups. The kyphosis decreased 3° in the Tx group and 22° in the C group (Table 4).

Complications were 33% in the traction group and 25% the control group, and the difference was not statistically significant (P = 0.68) (Table 5). Each group had 1 instance of an intraoperative respiratory complication. The halo traction group also had 1 intraoperative instrumentation complication and 1 instance of an ischial pressure sore. In addition, there was a dural tear in the traction group and dilutional coagulopathy in the control group. By the 2 years follow-up, the halo traction group had 1 reoperation that required 4 rib excisions and revision of a hypertrophic chest scar. There were no neurologic complications associated with the correction of IS in either group.

Table 5
Table 5:
Complications in AIS Patients (n = 23)


Axial traction attempts to achieve safe and effective correction of severe spinal deformities. The gradual increase in traction over a period of weeks allows for a partial correction of the large curvature and associated extraspinal deformity so that when surgical correction is performed it is done on a less pronounced spinal deformity. Theoretically, this should allow for a better overall correction without the complications that are associated with pure surgical correction of large curves.

Mehlman et al,3 studied 24 patients who had halo-femoral traction and spinal release. Only 1 patient suffered neurologic complication, a transient bilateral lower extremity sensory deficit. Another complication associated with halo-femoral traction is temporary brachial plexus palsy, which was noted by Qian et al.4 Another form of halo traction is halo-pelvic traction. This has the benefit of allowing more direct tension to be applied to the spine, but the high level of traction is associated with several consequences. One of these consequences is cranial nerve palsy, which was found in 6 of 70 patients treated with halo-femoral or halo-pelvic traction studied by Wilkins and MacEwen.5 The sixth cranial nerve was the most commonly affected, with involvement of the ninth, tenth, and twelfth nerves less common. Other complications of halo-femoral and halo-pelvic traction are pin site infections, peritoneal penetration or intestinal perforation, and hip dislocation.2

In contrast to halo-femoral and halo-pelvic traction, halo-gravity traction, which was popularized by Stagnara,6 uses the weight of the patient’s body as the counterforce. Unlike halo-femoral traction, which requires prolonged bed rest, halo-gravity traction can be applied while a patient is in either a bed, a wheelchair, or a walking frame. Contraindications to halo traction include cervical kyphosis or stenosis, or significant instability or ligamentous laxity.

A few uncontrolled case series have investigated the success of halo-gravity traction in correcting severe spinal deformities. Rinella et al7 performed a retrospective analysis of 33 patients with severe scoliosis, kyphoscoliosis, or kyphosis. The main coronal curve ranged from 22° to 158° with a mean of 84°. Four of the 33 patients had IS, and in these patients the main coronal curve ranged from 84° to 131° with a mean of 101°. For all 33 patients, the main coronal curve reduced 38° or 46% after posterior spinal fusion compared with pretreatment radiographs. Periopoerative complications included temporary respiratory distress, malignant hyperthermia, coagulopathy, supraventricular tachycardia, triceps palsy, brachial plexus palsy, and anterior strut graft dislodgement requiring revision anterior fusion. Long-term complications included superficial wound infections, curve progression, and rod migration. Sink et al8 conducted a retrospective review of 19 children with severe scoliosis who underwent spinal fusion surgery following 6 to 21 weeks of preoperative halo-gravity traction. Four of the 19 patients had IS. For all 19 patients, the Cobb angle ranged from 63° to 100° with a mean of 84°. After traction and immediately preceding fusion, the Cobb angle ranged from 22° to 76° with a mean of 55°. The average improvement was 35%. Complications that occurred during traction included pin loosening and superficial pin site infections. Complications after posterior spinal fusion included wound infection and loss of fixation of the inferior laminar hooks.

In addition to the complications noted in the studies above, temporary hypoglossal nerve injury has been reported in the literature.10 This complication may manifest as difficulty swallowing, difficulty speaking, or protrusion of the tongue. Qian et al4 also noted that halo-gravity traction may result in temporary brachial plexus palsies. They studied 300 cases of halo-gravity or halo-femoral traction applied before posterior correction of severe scoliosis. In 3 cases of halo-gravity traction a brachial plexus palsy lasting up to 3 months was discovered. Although there are a few studies that have investigated the safety and efficacy of various methods of halo traction, there were few studies that included a comparison control group. Seller et al1 in one such study analyzed 2 groups of patients with severe neuromuscular spinal deformity: 1 group with and 1 group without preoperative halo traction. After surgery, the average main Cobb angle decreased 57% (77°–33°) and 61% (85°–33°) in the nonhalo traction and halo traction group, respectively. The difference was not significant (P = 0.19). Therefore, they concluded that unless there are specific indications for halo traction, it is not needed as a standard procedure in neuromuscular deformities.

In our study, there was no statistically significant difference between traction and control groups in coronal or sagittal curve correction, spinal length gain, blood loss, operating time, or complications. This was true of AIS alone as well as all diagnoses taken together. This study is limited by the small sample size and retrospective nature. Yet, given the rarity of these severe scoliotic curves, it is uncertain whether a prospective study can be accomplished. Power analysis of the data in this article suggest that it would require a study size of over 2000 patients to determine a possible difference in curve correction. In light of these limitations, we conclude that satisfactory and similar results are obtainable with or without halo gravity in the treatment of severe scoliotic curves. In our study, the patients who had halo traction less frequently had a vertebral body resection, but achieved comparable deformity correction. The surgeon should use his or her own judgment about what will produce the best result for these severe curves given flexibility, medical, and technical factors.

In summary, there are several different options for employing mechanical measures to assist in correcting severe curves. Concerns include minimizing neurologic risk and blood loss while optimizing curve correction. With greater use of operative release and bony resection, there seems to be less need for traction, but with potentially more technically demanding surgeries. Precise knowledge of the patient’s preoperative baseline and quality intraoperative monitoring are essential. The surgeon should consider all of these factors in making a decision on whether to use traction.

Key Points

  • Halo-gravity traction is a safe, well-tolerated method of applying gradual, sustained traction to maximize postoperative correction in patients with severe scoliosis and kyphosis.
  • Prior studies of halo-gravity traction lack a comparison control group and study only a relatively small number of idiopathic scoliosis patients.
  • We created a multicenter, retrospective, comparison group study to compare the treatment of patients with severe scoliosis with and without halo-gravity traction.
  • Our study demonstrated that patients with halo traction less frequently had a vertebral body resection, but achieved comparable deformity correction.


1. Seller K, Haas S, Raab P, et al. Preoperative halo-traction in severe paralytic scoliosis. Z Orthop Ihre Grenzgeb 2005;143:539–43.
2. Toledo LC, Toledo CH, MacEwen GD. Halo traction with the circolectric bed in the treatment of severe spinal deformities: a preliminary report. J Pediatr Orthop 1982;2:554–9.
3. Mehlman CT, Al-Sayyad MJ, Crawford AH. Effectiveness of spinal release and halo-femoral traction in the management of severe spinal deformity. J Pediatr Orthop 2004;24:667–73.
4. Qian BP, Qiu Y, Wang B. Brachial plexus palsy associated with halo traction before posterior correction in severe scoliosis. Stud Health Technol Inform 2006;123:538–42.
5. Wilkins C, MacEwen GD. Cranial nerve injury from halo traction. Clin Orthop Relat Res 1977;126:106–10.
6. Stagnara P. Cranial traction using the “Halo” of rancho los amigos. Rev Chir Orthop Reparatrice Appar Mot 1971;57:287–300.
7. Rinella A, Lenke L, Whitaker C, et al. Perioperative halo-gravity traction in the treatment of severe scoliosis and kyphosis. Spine 2005;30:475–82.
8. Sink EL, Karol LA, Sanders J, et al. Efficacy of perioperative halo-gravity traction in the treatment of severe scoliosis in children. J Pediatr Orthop 2001;21:519–24.
9. Dormans JP, Criscitiello AA, Drummond DS, et al. Complications in children managed with immobilization in a halo vest. J Bone Joint Surg Am 1995;77:1370–3.
10. Ginsburg GM, Bassett GS. Hypoglossal nerve injury caused by halo-suspension sraction: a case report. Spine 1998;23:1490–3.

halo traction; scoliosis; kyphosis; adolescent idiopathic scoliosis

© 2008 Lippincott Williams & Wilkins, Inc.