ORIGINAL ARTICLE

Rod fracture after long construct fusion for spinal deformity:clinical and radiographic risk factors

Tsutomu Akazawa • Toshiaki Kotani •

Tsuyoshi Sakuma • Tetsuharu Nemoto •

Shohei Minami

Received: 11 May 2013 / Accepted: 26 August 2013 / Published online: 14 September 2013

� The Japanese Orthopaedic Association 2013

Abstract

Background No reports have been published on detailed

risk factors for rod fracture after spinal deformity correc-

tion and fusion. The purpose of this study was to analyze

clinical and radiographic risk factors of rod fracture after

long construct fusion for spinal deformity.

Methods The survey subjects were 155 cases who were

diagnosed with spinal deformity and underwent correction

and fusion surgery with long construct instrumentation ([3

levels, average 10.3 levels) between July 2004 and June

2010. The subjects comprised 32 males and 123 females

with a mean age of 19.0 (range 8–78) years. The mean

Cobb angle was 61.0 ± 16.1� preoperatively and

25.7 ± 16.9� postoperatively. Univariate analysis and

logistic regression analysis were performed.

Results Rod fracture occurred in 8 of 155 cases (5.2 %).

The mean period from surgery to rod fracture was

18.1 months (range 2–37). The level of fracture ranged

from the thoracolumbar junction to the lumbosacral ver-

tebrae. Six patients had fracture near the fused lower end

and two patients had fracture at the thoracolumbar junc-

tion. Univariate analysis revealed that non-ambulatory

status, preoperative kyphosis, small-diameter rods, multi-

ple surgery, and use of iliac screws were significant risk

factors for rod fracture. Sex, obesity, severity of preoper-

ative scoliosis, and rod material were not significant risk

factors. Logistic regression analysis revealed that use of

iliac screws (odds ratio: 81.9, 95 % confidence interval:

7.2–935.0, p \ 0.001) and small-diameter (\6 mm) rods

(odds ratio: 16.3, 95 % confidence interval: 1.7–152.6,

p = 0.015) were risk factors for rod fracture.

Conclusions The incidence of rod fracture after long

construct fusion for spinal deformity was 5.2 %. Iliac

screw fixation and small-diameter rods were risk factors for

rod fracture.

Introduction

Complications of spinal deformity surgery include nerve

injuries, great vessel injuries, infections, and pulmonary

embolism. In addition to these complications, implant-

related complications have attracted attention. In the mor-

bidity and mortality database of the Scoliosis Research

Society, the incidence of implant-related complications is

1.5 % for pediatric scoliosis patients [1] and 1.6 % for

adult scoliosis patients [2]. The incidence of implant failure

is 4.4 % for patients with adult spinal deformity [3] and

8.8 % for patients with curvature of at least 100� [4]. The

papers from which these data were obtained did not

describe the details of implant breakage, including ‘‘rod

fracture’’ as mechanical rod breakage.

Rod fracture is a serious complication. There have been

several reports on rod fracture after spinal deformity sur-

gery. Yang et al. [5] reported the incidence of rod fracture

as 15 % for growing rods. Because growing rod surgery is

non-fusion surgery, the incidence of rod fracture is high.

Fusion surgery is predicted to have a lower incidence.

Erwin et al. [6] reported the incidence of rod fracture as

2.1 % for Harrington distraction rods. However, no reports

have been published on detailed risk factors for rod fracture

after spinal deformity correction and fusion using third-

generation implants or newer. The purpose of our study

T. Akazawa (&) � T. Kotani � T. Sakuma � T. Nemoto �S. Minami

Department of Orthopedic Surgery, Seirei Sakura Citizen

Hospital, 2-36-2 Ebaradai, Sakura, Chiba 285-8765, Japan

e-mail: cds00350@par.odn.ne.jp

123

J Orthop Sci (2013) 18:926–931

DOI 10.1007/s00776-013-0464-4

was to analyze clinical and radiographic risk factors for rod

fracture after long construct fusion for spinal deformity.

Materials and methods

The subjects were 221 patients whose spinal deformities

were treated by correction and fusion by use of spinal

implants. The institutional review board approved this

study. The procedure was performed at one institution

between July 2004 and June 2010. The patients were

excluded if they had received non-bone graft surgery, for

example those using the growing rod method, fusion of 3 or

fewer vertebral bodies, or surgery for degenerative lumbar

scoliosis with a Cobb angle of less than 30�. A total of 155

patients were reviewed for this study. The patients con-

sisted of 32 men and 123 women, and their mean age was

19.0 ± 12.8 years (range 8–78 years). The mean follow-up

period was 46.1 ± 17.8 months (range 24–94 months).

The diagnosis was adolescent idiopathic scoliosis for 95

patients, congenital scoliosis for 15 patients, neuromuscu-

lar scoliosis for 8 patients, scoliosis associated with neu-

rofibromatosis for 7 patients, adult spinal deformity for 13

patients, early onset scoliosis (after final fusion surgery) for

11 patients, and other syndromic scoliosis for 6 patients.

Surgical methods were posterior correction and fusion for

139 patients, anterior correction and fusion for 12 patients,

and combined anterior and posterior procedure for 4

patients. The mean number of fused vertebral bodies was

10.3 ± 2.9. The scoliosis angle was 61.0 ± 16.1� before

surgery and 25.7 ± 16.9� soon after surgery (1 week after

surgery or on first standing erect), indicative of 58.3 %

correction. The spinal implants were made of titanium

alloy or commercially pure (CP) titanium. The bone grafts

were autografts with or without beta-tricalcium phosphate

bone grafts, and no allogeneic bone or bone morphogenetic

protein was used. For patients who underwent lumbosacral

fusion, S1 pedicle screws and iliac screws were used.

Fusion was not performed using S1 pedicle screws alone or

Jackson’s techniques. All patients were examined for rod

fracture in X-rays at a follow-up visit.

Statistical analysis

A Mann–Whitney U test was used to compare age, body

weight, preoperative scoliosis angle, postoperative scolio-

sis angle (1 week after surgery or on first standing erect),

and number of fused vertebral bodies between patients with

rod fracture and those without.

The potential risk factors for rod fracture were sex

(male), obesity (body mass index [25 kg/m2), non-ambu-

latory status, severity of preoperative scoliosis (80� or

more), preoperative kyphosis (40� or more), use of small-

diameter (\6 mm) rods, rod material (CP titanium rods),

multiple surgery, and use of iliac screws. Pearson’s chi-

squared test was used for univariate analysis. Logistic

regression analysis was performed using the aforemen-

tioned factors as dependent variables and rod fracture as

independent variable to examine the risk factors for rod

fracture. SPSS Statistics version 19.0 (International Busi-

ness Machines Corporation, NY, USA) was used for sta-

tistical analysis, and the level of statistical significance was

set at 5 %.

Results

Rod fracture was observed in 8 of 155 patients (5.2 %).

The fractured rods were titanium alloy rods of 4.5-mm

diameter in 1 patient, CP titanium rods of 5.5-mm diameter

in 5 patients, and CP titanium rods of 6.35-mm diameter in

2 patients. Implant breakage other than rod fracture was

disassembly of an axial connector in 1 patient. No screw or

hook breakage was observed in any patient. Among

patients with rod fracture, two had adolescent idiopathic

scoliosis, two had an adult spinal deformity, two had early

onset scoliosis (after final fusion surgery), one had con-

genital scoliosis, and one had scoliosis associated with

neurofibromatosis. Surgical methods were posterior cor-

rection and fusion for 6 patients and anterior correction and

fusion for 2 patients.

With regard to rod material and diameter, incidence of

rod fracture was 33.1 % (1 of 3 patients) with 4.5-mm

diameter titanium alloy rods, 0 % (0 of 17 patients) with

5.5-mm diameter titanium alloy rods, 25 % (5 of 20

patients) with 5.5-mm diameter CP titanium rods, 0 % (0

of 25 patients) with 6.35-mm diameter titanium alloy rods,

and 2.2 (2 of 90 patients) with 6.35-mm diameter CP

titanium rods. With regard to diagnosis, incidence of rod

fracture was 2.1 % (2 of 95 patients) for adolescent idio-

pathic scoliosis, 6.7 % (1 of 15 patients) for congenital

scoliosis, 0 % (0 of 8 patients) for neuromuscular scoliosis,

14.3 % (1 of 7 patients) for scoliosis associated with neu-

rofibromatosis, 15.4 % (2 of 13 patients) for adult spinal

deformity, 18.2 % (2 of 11 patients) for early onset scoli-

osis (after final fusion surgery), 0 % (0 of 6 patients) for

early onset scoliosis (after final fusion surgery), and 18.2 %

for other syndromic scoliosis. With regard to surgical

procedure, incidence of rod fracture was 4.3 % (6 of 139

patients) for posterior correction and fusion, 16.7 % (2 of

12 patients) for anterior correction and fusion, and 0 % (0

of 4 patients) for combined anterior and posterior proce-

dure. For patients who underwent iliac screw fixation, the

incidence of rod fracture was 57.1 % (4 of 7 patients).

The mean length of time between surgery and rod

fracture was 18.1 ± 10.5 months (range 2–37 months).

Rod fracture in spinal deformity 927

123

The clinical symptoms at the time of fracture were pain for

2 patients and the breaking sound of a crack for two

patients. There was no symptom for 4 patients and the

fracture was discovered in a radiograph at a follow-up visit

(Table 1). The level of fracture ranged from the thoraco-

lumbar junction to the lumbosacral vertebrae (Fig. 1). Six

patients had fracture near the fused lower end and two

patients had it at the thoracolumbar junction (Fig. 2). Sal-

vage surgery was performed for 6 patients, and replace-

ment of fractured rods and bone regrafting were performed.

Mean age at the time of surgery was 28.8 ± 24.7 years

in the group with fracture and 18.5 ± 11.8 years in the

group with no fracture, indicating no significant difference

in age. Mean body weight was 42.6 ± 12.2 and

46.0 ± 11.2 kg, respectively, indicating no significant

difference. Mean preoperative scoliosis angle was

55.6 ± 26.5� and 61.3 ± 15.5�, respectively, indicating no

significant difference. Mean postoperative scoliosis angle

(1 week after surgery or on first standing erect) was

40.9 ± 25.3� and 24.9 ± 16.1�, respectively, indicating a

significant difference (p = 0.029). Mean number of fused

vertebral bodies was 10.8 ± 5.5 and 10.3 ± 2.8, respec-

tively, indicating no significant difference (Table 2).

Analysis of risk factors

Univariate analysis revealed that sex, obesity, severity of

preoperative scoliosis (80� or more), and rod material were

not significant risk factors for rod fracture. The group with

fracture contained significantly more non-ambulatory

patients than the group with no fracture (12.5 vs 0.7 %,

p = 0.004). The incidence of preoperative kyphosis (40� or

more) was significantly higher in the group with fracture

than in the group with no fracture (50 vs 10.2 %,

Table 1 Details of patients with rod fracture

Case no Age Sex Diagnosis Operative

method

Rod

diameter

(mm)

Rod

material

Length of time

between surgery

and rod fracture

(month)

Fusion

level

Level of

fracture

Clinical

symptom

1 17 F AIS ASF 4.5 Ti alloy 37.3 T12-L3 L2-3 Breaking sound

2 18 F AIS ASF 5.5 CP Ti 24.1 T12-L3 L2-3 None

3 17 F NF PSF 5.5 CP Ti 13.8 T9-S1, Iliac S1 Pain

4 16 F EOS PSF (final fusion) 5.5 CP Ti 20.8 T2-L4 T12-L1 Pain

5 11 F EOS PSF (final fusion) 5.5 CP Ti 15.2 C6-S1, Iliac L3-4 None

6 14 F Congenital PSF 6.35 CP Ti 6.3 T8-L7 L1 None

7 66 F ASD PSF 5.5 CP Ti 2.2 T9-S1, Iliac L4-5 None

8 71 F ASD PSF 6.35 CP Ti 25.5 T6-S1, Iliac L3-4 Breaking sound

AIS adolescent idiopathic scoliosis, NF neurofibromatosis scoliosis, EOS early-onset scoliosis, ASD adult spinal deformity, ASF anterior spinal

fusion, PSF posterior spinal fusion, CP commercially pure, Ti titanium

Fig. 1 Examples of rod

fracture. a Rod fracture at the

fused lower end after anterior

spinal fusion. b Rod fracture

near the fused lower end after

posterior spinal fusion using

iliac screws. c Rod fracture at

the thoracolumbar junction after

posterior spinal fusion

928 T. Akazawa et al.

123

p = 0.001). The incidence of small-diameter (\6 mm)

rods was significantly higher in the group with fracture

than in the group with no fracture (75 vs 23.1 %,

p = 0.001). The incidence of multiple surgery was sig-

nificantly higher in the group with fracture than in the

group with no fracture (62.5 vs 14.3 %, p \ 0.001). The

incidence of the use of iliac screws was significantly higher

in the group with fracture than in the group with no fracture

(50 vs 2.0 %, p \ 0.001) (Table 3).

Logistic regression analysis revealed that use of iliac

screws (odds ratio: 81.9, 95 % confidence interval:

7.2–935.0, p \ 0.001) and small-diameter (\6 mm) rods

(odds ratio: 16.3, 95 % confidence interval: 1.7–152.6,

p = 0.015) were risk factors for rod fracture (Table 4).

Fig. 2 Case no. 6, 14-year-old

female, congenital

kyphoscoliosis. a Preoperative

radiograph. b Posterior

correction and fusion with

vertebral column resection were

performed. c Rod fracture and

loss of correction occurred

6 months after initial surgery.

Replacement of fractured rod

and bone regrafting were

performed

Table 2 Comparison of rod

fracture and no fracture groupsGroup with rod fracture Group with no fracture P value

Age (years), mean 28.8 ± 24.7 18.5 ± 11.8 0.241

Body weight (kg), mean 42.6 ± 12.2 46.0 ± 11.2 0.674

Preoperative scoliosis angle (�), mean 55.6 ± 26.5 61.3 ± 15.5 0.457

Postoperative scoliosis angle (�), mean 40.9 ± 25.3 24.9 ± 16.1 0.029

Number of fused vertebral bodies, mean 10.8 ± 5.5 10.3 ± 2.8 0.832

Table 3 Univariate analysis of

risk factors for rod fracture

BMI body mass index,

CP commercially pure

Rod fracture (%) No fracture (%) P value

Sex (male) 0 21.8 0.138

Obesity (BMI [25) 0 4.8 0.528

Non-ambulatory status 12.5 0.7 0.004

Severity of preoperative scoliosis (80� or more) 25 12.9 0.331

Preoperative kyphosis (40� or more) 50 10.2 0.001

Use of small-diameter (\6 mm) rods 75 23.1 0.001

Rod material (CP titanium rods) 87.5 70.1 0.290

Multiple surgery 62.5 14.3 \0.001

Use of iliac screws 50 2.0 \0.001

Rod fracture in spinal deformity 929

123

Discussion

Several reports have been published on spinal implant

breakage. Okamoto et al. [7] reported that implant break-

age occurred in 4.2 % of patients who underwent cervical

vertebral fusion but that no rod fracture occurred. Suda

et al. reported that implant breakage occurred in 5.9 % of

patients with lumbar isthmic spondylolisthesis in whom the

pedicle screw system and posterolateral fusion were used.

All cases of implant breakage were screw breakage and

there was no rod fracture [8]. Neo et al. [9] reported that

implant breakage occurred in 3.2 % of patients who

underwent thoracic or lumbar spinal instrumentation and

that rod fracture occurred in patients who underwent multi-

segmental thoracolumbar surgery. In our study, all cases of

implant breakage except one (disassembly of an axial

connector) were rod fracture and the incidence was 5.2 %.

No screw or hook breakage was observed in any patient.

Titanium and titanium alloys are notch sensitive. This

property of titanium has come to be known as notch sen-

sitivity. Use of the French bender and connecting bolts on a

titanium rod creates notches. Fatigue life was markedly

shorter for CP titanium and titanium alloy rods notched by

the French bender [10]. Rod contouring using the French

bender is believed to be related to the fracture of CP tita-

nium and titanium alloy rod.

Use of iliac screws was the most significant risk factor in

this study. Iliac screws have been shown to increase the

stiffness of lumbosacral constructs, but disadvantages

include difficulty in connecting the iliac screw to adjacent

S1 pedicle screws. Excessive stress of rod contouring is

necessary to connect iliac screws and S1 pedicle screws.

Excessive stress of rod contouring is believed to be related

to rod fracture. We currently use S2 alar iliac screws

instead of iliac screws. The S2 alar iliac screws were all in-

line and connected easily to the S1 pedicle screws [11].

There remains debate about whether it is better to stop a

long fusion distally at L5 or S1 with pelvic fixation. Fusion

to S1 with pelvic fixation may have a potential effect on a

rod fracture. We currently perform S2 alar iliac screw

fixation in cases of fusion to S1. In a future study we will

examine whether use of S2 alar iliac screws prevents rod

fracture.

Use of small-diameter rods was the second risk factor in

this study. Rod stiffness for a given material correlates with

the 4th power of the diameter; for example, the stiffness of

a 6-mm rod is 2.07 times that of a 5.5-mm rod, and the

stiffness of a 6.35-mm rod is 2.6 times that of a 5.5-mm rod

[12]. Larger-diameter rods are recommended.

Our study has some limitations. The odds ratios might

have been relatively high because the number of patients

with rod fracture was small (8 cases). Because the range

of the 95 % confidence interval was wide, the odds

ratios could change with an increase in the number of

cases. Although there was no patient with screw break-

age, it is a problem that should be noted because there

are a sufficient number of reported cases with screw

breakage in the literature [7, 8, 13]. In our study,

examination of pseudarthrosis was not performed for

patients without rod fracture. We did not check CT

routinely to evaluate bone union because we should

reduce radiation exposure, especially in pediatric cases.

Thus, further studies are necessary.

In our study, the incidence of rod fracture was 5.2 % for

patients who underwent spinal correction and fusion for

spinal deformities. Implant breakage was mostly rod frac-

ture, and use of iliac screws and small-diameter rods were

risk factors for rod fracture.

Conflict of interest The authors declare that they have no conflict

of interest.

References

1. Reames DL, Smith JS, Fu KM, Polly DW Jr, Ames CP, Berven

SH, Perra JH, Glassman SD, McCarthy RE, Knapp RD Jr, Heary

R, Shaffrey CI; Scoliosis Research Society Morbidity and Mor-

tality Committee. Complications in the surgical treatment of

19,360 cases of pediatric scoliosis—a review of the Scoliosis

Research Society Morbidity and Mortality database. Spine.

2011;36:1484–91.

2. Sansur CA, Smith JS, Coe JD, Glassman SD, Berven SH, Polly

DW Jr, Perra JH, Boachie-Adjei O, Shaffrey CI. Scoliosis

Research Society Morbidity and Mortality of adult scoliosis

surgery. Spine. 2011;36:E593–7.

3. Bridwell KH, Baldus C, Berven S, Edwards C 2nd, Glassman S,

Hamill C, Horton W, Lenke LG, Ondra S, Schwab F, Shaffrey C,

Wootten D. Changes in radiographic and clinical outcomes with

primary treatment adult spinal deformity surgeries from two years

to three- to five-years follow-up. Spine. 2010;35:1849–54.

4. Watanabe K, Lenke LG, Bridwell KH, Kim YJ, Watanabe K,

Kim YW, Kim YB, Hensley M, Stobbs G. Comparison of

radiographic outcomes for the treatment of scoliotic curves

greater than 100 degrees—wires versus hooks versus screws.

Spine. 2008;33:1084–91.

5. Yang JS, Sponseller PD, Thompson GH, Akbarnia BA, Emans

JB, Yazici M, Skaggs DL, Shah SA, Salari P, Poe-Kochert C;

Growing Spine Study Group. Growing rod fractures -risk factors

and opportunities for prevention. Spine. 2011;36:1639–44.

6. Erwin WD, Dickson JH, Harrington PR. Clinical review of

patients with broken Harrington rods. J Bone Joint Surg Am.

1980;62:1302–7.

Table 4 Logistic regression analysis of risk factors for rod fracture

Odds

ratio

95 % Confidence

interval

P value

Use of iliac screws 81.9 7.2–935.0 \0.001

Small-diameter

(\6 mm) rods

16.3 1.7–152.6 0.015

930 T. Akazawa et al.

123

7. Okamoto T, Neo M, Fujibayashi S, Ito H, Takemoto M, Na-

kamura T. Mechanical implant failure in posterior cervical spine

fusion. Eur Spine J. 2012;21:328–34.

8. Suda K, Ito M, Abumi K, Haba H, Taneichi H, Kaneda K.

Radiological risk factors of pseudoarthrosis and/or instrument

breakage after PLF with the pedicle screw system in isthmic

spondylolisthesis. J Spinal Disord Tech. 2006;19:541–6.

9. Neo M, Fujibayashi S, Nakamura T. Mechanical failure of spinal

instrumentation. Seikeigeka. 2009;60:886–93 (in Japanese).

10. Dick JC, Bourgeault CA. Notch sensitivity of titanium alloy,

commercially pure titanium, and stainless steel spinal implants.

Spine. 2001;26:1168–72.

11. O’Brien JR, Matteini L, Yu WD, Kebaish KM. Feasibility of

minimally invasive sacropelvic fixation: percutaneous S2 alar

iliac fixation. Spine. 2010;35:460–4.

12. Serhan H, Mhatre D, Newton P, Giorgio P, Sturm P. Would CoCr

rods provide better correctional forces than stainless steel or

titanium for rigid scoliosis curves? J Spinal Disord Tech.

2013;26:E70–4.

13. Modi HN, Suh SW, Hong JY, Yang JH. Posterior multilevel

vertebral osteotomy for severe and rigid idiopathic and nonidio-

pathic kyphoscoliosis: a further experience with minimum two-

year follow-up. Spine. 2011;36:1146–53.

Rod fracture in spinal deformity 931

123