Orthopaedics & Traumatology: Surgery & Research 101 (2015) 197–200

Available online at

ScienceDirect
www.sciencedirect.com

Original article

Intrinsic and extrinsic risk factors for nonunion after nonoperative

treatment of midshaft clavicle fractures
W. Liu a,1 , J. Xiao b,1 , F. Ji a,∗ , Y. Xie a , Y. Hao a
a
Department of orthopedics, Huai’an First People’s Hospital, Nanjing Medical University, 6, Beijing Road West, Huai’an, Jiangsu 223300, PR China
b
Department of orthopedics, The 101st Hospital of Chinese PLA, No. 101 North Xingyuan Road, Wuxi, Jiangsu 214044, PR China

a r t i c l e i n f o a b s t r a c t

Article history: Background: The optimal treatment of midshaft clavicle fractures remains controversial. Nonunion is
Received 19 June 2014 usually considered to be an uncommon complication following a nonoperatively treated clavicle fracture.
Accepted 14 November 2014 Hypothesis: Not every midshaft clavicular fractures shares the same risk of developing nonunion after
nonoperative treatment. The present study was performed to identify the intrinsic and extrinsic indepen-
Keywords: dent factors that are independently predictive of nonunion in patients with midshaft clavicular fractures
Clavicle fracture after nonoperative treatment.
Treatment
Materials and methods: We performed a retrospective study of a series of 804 patients (391 men and 413
Risk factor
Multivariate analysis
women with a median age of 51.3 years) with a radiographically confirmed midshaft clavicle fracture,
Nonunion which was treated nonoperatively. There were 96 patients who underwent nonunion. Putative intrinsic
(patient-related) and extrinsic (injured-related) risk factors associated with nonunion were determined
with the use of bivariate and multivariate statistical analyses.
Results: By bivariate analysis, the risk of nonunion was significantly increased by several intrinsic risk
factors including age, sex, and smoking and extrinsic risk factors including displacement of the frac-
ture and the presence of comminution (P < 0.05 for all). On multivariate analysis, smoking (OR = 4.16,
95% CI: 1.01–14.16), fracture displacement (OR = 7.81, 95% CI: 2.27–25.38) and comminution of fracture
(OR = 3.86, 95% CI: 1.16–13.46) were identified as independent predictive factors.
Conclusion: The risk factors for nonunion after nonoperative treatment of midshaft clavicle fractures are
multifactorial. Smoking, fracture displacement and comminution of fracture are independent predictors
for an individual likelihood of nonunion. Further studies are still required to evaluate these factors in the
future.
Level of evidence: Level III, case-control study.
© 2015 Elsevier Masson SAS. All rights reserved.

1. Introduction of treatment, with early mobilization of the shoulder as the pain

subsides.
Clavicular fractures accounts for 5% of all fractures in adults. A Nonunion is usually considered to be an uncommon compli-
large number of these fractures, about 69–82%, are located in the cation following a clavicular fracture. Between five and 20% of
midshaft of the clavicle [1–5]. Most midshaft clavicular fractures are patients with midshaft clavicular fractures develop nonunion if
caused by a direct axial compression to the shoulder after a sudden treated by nonoperative intervention [7–10]. Patients who undergo
stop or fall during sports, such as cycling and horse riding [5,6]. To primary fixation have a lower rate of nonunion and report better
date, the optimal treatment of midshaft clavicular fractures is still functional outcomes than those treated nonoperatively [11]. Out-
in controversy. Although many forms of nonoperative treatment comes following primary fixation are also better than outcomes
have been described, the most widely accepted treatment involves following secondary fixation in patients who develop nonunion fol-
the provision of a simple sling for support during the initial phase lowing nonoperative treatment [7]. This has resulted in the growing
support for a policy of primary fixation for midshaft clavicular frac-
tures in adult patients [7]. However, a blanket surgical approach
would exert potential complications of surgery on huge numbers
∗ Corresponding author.
of patients who would have healed without surgical intervention.
E-mail address: [email protected] (F. Ji).
Despite the risk of nonunion after these fractures, most are still
1
The first two authors contributed equally to this work. treated nonoperatively. Therefore, it is imperative to identify the

http://dx.doi.org/10.1016/j.otsr.2014.11.018
1877-0568/© 2015 Elsevier Masson SAS. All rights reserved.
198 W. Liu et al. / Orthopaedics & Traumatology: Surgery & Research 101 (2015) 197–200

patient who is at higher risk of nonunion if they will treated by non- of mobility or pain on stressing the site of the fracture and evi-
operative intervention. In addition, identification of patients with dence of bridging callus on radiographs. On each radiograph, the
relative risk factors of nonunion is desirable at the time of the ini- cortices were evaluated for the amount of bridging. Healing time
tial treatment to improve patient counseling and enable targeted was set as the time when the fracture was bridged, defined by the
surgical treatment. disappearance of the cortical interruption at the fracture site as a
Several studies have attempted to evaluate the risk factors of result of callus formation. Nonunion was judged as a fracture that
nonunion in patients after nonoperative intervention. A wide range remained unhealed according to these above criteria at 24 weeks
of factors have been hypothesized to contribute to the risk of after the injury. Eighty-three patients with nonunion were offered
nonunion after injury. These include intrinsic factors, such as the operative open reduction and plate fixation after 24 weeks unless
age and gender of the patient, and extrinsic factors, such as the loca- they were unfit for surgery. 16 patients who was uncertainty in
tion and extent of displacement of the fracture [12–15]. However, fracture union underwent exploratory operation, and three were
these studies have included participants in children [14], fractures found to be united; the remaining 13 had a definite nonunion,
of the medial and lateral ends of the clavicle [9], or only displaced which was treated with plate fixation and bone-grafting. These 13
midshaft fractures of clavicle [16]. Thus, the models in these previ- patients were considered to be nonunion at 24 weeks. All patients
ous studies are limited in their ability to predict nonunion in adult with nonunion were united following surgery.
patients with midshaft fractures of clavicle [9]. Therefore, we take
a hypothesis that not every midshaft clavicular fractures shares the
2.5. Putative intrinsic (patient-related) and extrinsic
same risk of developing nonunion. In the present study, we intend
(injured-related) risk factors
to identify the intrinsic and extrinsic factors that are independently
predictive of nonunion through a retrospective cohort study.
All demographic and outcome data were gathered by two
authors (W.D.L and Y.D.H). The displacement of fracture was
2. Materials and methods
defined as at least one residual cortical not contact between
2.1. Study design bone ends. The intrinsic (patient-related) information recorded
at age, gender, with or without medical comorbidities (includ-
A database was compiled of patients who were treated nonoper- ing rheumatoid disease, immunocompromise, renal failure and
atively in an academic hospital – a university-based medical center etc), tobacco consumption, alcohol consumption, employment sta-
from 1st February 2008 to 31st January 2013 following a midshaft tus insurance or medicolegal claim pending and mental status.
clavicular fracture. We performed a retrospective analysis of the The extrinsic (injured-related) information included mechanism of
data that included only patients who were 18 years of age or older. injury, displacement of fracture (including translation, angulation
and shortening of the fracture which was recorded from the initial
2.2. Inclusion criteria anteroposterior radiographs made after the injury), comminution
of fracture, presence of neurological deficit.
Patients who were at least 18 years of age were included in the
study if they had:
2.6. Statistical analysis

• a fracture in the middle three-fifths of the clavicle;

Factors associated with nonunion after nonoperative treatment
• no fracture in other parts of body;
of midshaft clavicular fractures were identified using univariate
• nonoperative treatment (brace or sling) until either confirmed
analysis. The data analysis was performed using SPSS version 19.0
fracture-healing or the development of nonunion; (Chicago, IL, USA). Continuous data were compared between the 2
• adequate documentation of demographic details and clinical and
groups using the Student’s t-test, whereas discontinuous data were
radiographic follow-up until fracture-healing or the development analyzed using the Chi-squared test. Fisher’s exact test was used for
of nonunion. small data subsets (n < 5). All significance tests were 2-tailed, with
P < 0.05 representing statistical significance. In addition, a multi-
2.3. Patients excluded from study
variate logistic regression analysis was performed to identify which
1059 patients identified as having a midshaft clavicular fracture independent factors helped predict the probability of nonunion.
were treated nonoperatively. Of the 1059 patients, 804 satisfied
the above inclusion criteria and were consider further. 34 patients 3. Results
were excluded because of no demographic data could be gained
during the follow-up study. Moreover, 51 patients were excluded 3.1. Demographic data
for incomplete clinical or demographic data. 63 patients were
excluded because they were lost to follow-up before fracture union Of the 804 patients, 96 underwent nonunion, representing a risk
was determined. 80 patients were excluded because they under- of 11.9%. A summary of the demographic data of union group and
went operative treatment after nonoperative treatment (within nonunion group is presented in Table 1. The data about putative
two weeks of injury). The surgery was performed as a result of skin extrinsic risk factors is present in Table 2.
or neurovascular compromise in 13; pathological fracture, float-
ing shoulder, or other multifocal shoulder girdle injury in 20; a
request by the patient in 24; and a decision of the treating surgeon 3.2. Bivariate analysis
in 23. 27 patients were excluded for the patients underwent early
operative treatment from two to 24 weeks after injury before the Age, sex, smoking are the intrinsic risk factors for nonunion
development of definite nonunion. after nonoperative treatment for midshaft clavicular fractures on
bivariate analysis. While, for the extrinsic risk factors, overall frac-
2.4. Assessment of fracture union ture displacement, including presence of complete displacement
of fracture, shortening of an off-ended fracture, translation of frac-
The union of midshaft clavicle fracture was evaluated by two ture, and angulation of fracture and comminution of fracture were
authors (W.D.L and Y.X). Fracture union was judged as the absence associated with increased risk of nonunion on bivariate analysis.
 W. Liu et al. / Orthopaedics & Traumatology: Surgery & Research 101 (2015) 197–200 199

Table 1 study did not report the treatment of midshaft clavicular fracture.
Baseline intrinsic (patient-related) characteristics of the two groups with or without
The differences in nonunion frequency might be partly related to
nonunion.
the chosen definition of a nonunion and different treatment.
Union Nonunion P Although age and sex were identified as the risk factors in the
Characteristics n = 708 n = 96 bivariate analysis, age and sex no longer represented an inde-
Age (y) 55.3 ± 7.3 49.1 ± 6.4 0.049 pendently significant predictor of nonunion in multiple logistic
Male, n (%) 335 (47.3%) 56 (58.3%) 0.043 regression analysis. Only three factors, namely smoking, com-
Height (cm) 173.1 ± 13.1 169.8 ± 15.7 0.818
minution of fracture and overall fracture displacement, were
Weight (kg) 72.5 ± 11.7 77.4 ± 15.3 0.798
BMI (kg/m2 ) 19.5 ± 2.5 20.7 ± 3.1 0.799 independently predictive of nonunion. Using a multivariate regres-
Smoker, n (%) 125 (17.7%) 30 (31.3%) 0.002 sion model that takes the three identified risk factors into account,
Alcohol, n (%) 212 (30.0%) 21 (21.9%) 0.102 estimates of the risk of nonunion after nonoperative management
Comorbidity, n (%) 25 (3.5%) 4 (4.2%) 0.754
can be produced.
Employment, n (%) 493 (69.6%) 60 (62.5%) 0.157
Mental disorder, n (%) 6 (0.8%) 1 (1.0%) 0.848 Several studies have included children and the lateral and
medial ends of the clavicle fracture [9,18], which place an influ-
ence on the stability of the results. We, therefore, only investigate
Table 2 midshaft fractures in adults in the present study, aiming to reduce
Baseline extrinsic (injured-related) characteristics of the two groups with or without the confounding effects of age and anatomical location.
nonunion. A young patient predominantly male has a higher risk to have
Union Nonunion P a clavicle nonunion than an old one in the bivariate analysis.
The reason is that these fractures occurred in young patient and
Characteristics n = 708 n = 96
male population may be caused by high-energy injury such as a
Mechanism of injury sports injury or a traffic accident. The severity of the fracture may
Simple fall 220 (31.1%) 25 (26.0%) 0.315
influence the healing progress of fracture. Smoking is the only
Fall from a height 165 (23.3%) 23 (24.0%) 0.887
Sports 133 (18.8%) 16 (16.7%) 0.616 one intrinsic risk factor for nonunion after nonoperative treat-
Traffic accident 79 (11.2%) 13 (13.5%) 0.491 ment for midshaft clavicular fracture. Previously, several clinical
Direct violence 76 (10.7%) 11 (11.5%) 0.830 and experimental studies have confirmed the association between
Other 35 (4.9%) 8 (8.3%) 0.166 the fracture union and smoking [19,20]. However, several studies
Displacement of fracture 186 (26.3%) 55 (57.3%) 0.031 reported that smoking was not a risk factor for clavicular nonunion
Complete displacement of fracture 35 (5.9%) 11 (11.5%) 0.019 [8,18]. Although smoking was identified as an independent risk fac-
Translation of fracture 58 (8.2%) 15 (15.6%) 0.038
tor for nonunion in the present study, the strength of association
Angulation of fracture 32 (4.5%) 10 (10.4%) 0.021
Shortening of fracture 61 (8.6%) 19 (19.8%) 0.033 cannot be conduced to establish whether heavier smokers were at
great risk.
Comminution of fracture 170 (24.0%) 38 (39.6%) 0.039
As extrinsic risk factors, overall displacement and comminu-
Presence of neurological deficit 35 (4.9%) 6 (6.3%) 0.585 tion of fractures are identified as the two independent risk factors
for nonunion. The above two factors place more attentions to the
reduction and fracture morphology on healing. Comminution and
Table 3 displacement of fractures may be associated with higher-energy
Independent risk factors for nonunion in patients after nonoperative treatment for
trauma and, therefore, add to the severity of underlying osseous
midshaft clavicular fracture.
and soft tissue injuries. Several studies also argued that fracture
95% confidence interval comminution was associated with poorer outcome [8–10]. Previ-
Characteristics Odds ratio Lower Upper P ous studies demonstrated that high nonunion rates (up to 29%)
value limit limit have been observed in displaced fractures [8,9,21,22]. Bernstein
Smoking 4.16 1.01 14.16 0.031 [7] reported that a direct relationship existed between increased
Fracture displacement 7.81 2.27 25.38 0.001 displacement and worse functional outcome scores. Hill et al. [8]
Comminution of fracture 3.86 1.16 13.46 0.035 argued that there was a significant association between initial
shortening and the development of nonunion. It has also been
reported that displaced midshaft fractures were 18.5 times more
3.3. Logistic regression analysis likely to result in delayed union or nonunion compared with
nondisplaced fractures [9].
These above parameters were entered into the logistic regres- The present confirm the need for consideration of all three
sion model. By multivariate logistic regression analysis, several variables when identifying patients at greatest risk of nonunion.
independent factors were identified to be related to a higher risk Although three independent risk factors are associated with
of nonunion. Smoking, fracture displacement and comminution of nonunion in patients with midshaft clavicular, the ability to accu-
fracture are independently predictive. The results were presented rately predict nonunion in individual patients may be poor, because
in Table 3. of the relatively low prevalence of nonunion. Although it is possible
to determine whether patients have the above-mentioned risk fac-
4. Discussion tors, it is less certain that patients with one or more independent
risk factors will develop nonunion. Many patients at high-predicted
The present study confirms that the nonunion after nonopera- risk of nonunion will heal without nonunion, and a number of
tive treatment for midshaft clavicular is an uncommon occurrence. patients with few risk factors will nevertheless develop nonunion.
The prevalence is higher than the previously reported in other The limitations of the present study include the following:
retrospective studies [4,17]. The present findings support the although X-ray allows a qualitative assessment of callus forma-
increased prevalence of nonunion reported in contemporary stud- tion and cortical bridging, doubt has been cast over its reliability
ies [7]. In the present study, nonunion occurred in 96 (11.9%) of for the assessment of fracture-healing [23]. The complex three-
the 804 patients who were at least 18 years of age. Although this dimensional configurations of fractures are not fully appreciated
prevalence is twice that observed by Nowak et al. [6], the previous on radiographs, limiting the accuracy of judgments of shortening,
200 W. Liu et al. / Orthopaedics & Traumatology: Surgery & Research 101 (2015) 197–200

translation, and angulation. Therefore, the potential false-positive number of patients undergoing unnecessary surgery by providing
rate may affect the stability of the results of the present study. current estimate of risk factors of nonunion. High-quality, random-
Three-dimensional computed tomography would improve the ized, controlled trials are still required to evaluate these factors in
interpretation of fracture morphology and provide a more accurate the future.
assessment of healing in future studies [24], but its use was limited
in the present study by its cost. Patients who underwent operative Disclosure of interest
treatment before the last follow-up for reasons including skin infec-
tion, patient request, or a decision of the surgeon were excluded. The authors declare that they have no conflicts of interest con-
It is almost certain that some of these patients were at high risk cerning this article.
of nonunion, which will affect the external validity of the present
study. The present study has not recorded the specific smoking References
consumption for individual patients. Thus, we cannot determine
the threshold of risk level of smoking for nonunion. According to [1] Allman FJ. Fractures and ligamentous injuries of the clavicle and its articulation.
J Bone Joint Surg Am 1967;49:774–84.
the above limitation, it is significant that the results of the present [2] Khan LA, Bradnock TJ, Scott C, Robinson CM. Fractures of the clavicle. J Bone
study will be validated in independent samples. Moreover, the spe- Joint Surg Am 2009;91:447–60.
cific nonoperative treatments, including brace and sling, were not [3] Postacchini F, Gumina S, De Santis P, Albo F. Epidemiology of clavicle fractures.
J Shoulder Elbow Surg 2002;11:452–6.
evaluated, which may influence the stability of the results. The
[4] Robinson CM. Fractures of the clavicle in the adult. Epidemiology and classifi-
proportion of patients were lost to follow-up before the final assess- cation. J Bone Joint Surg Br 1998;80:476–84.
ment. We were unable to confirm whether these fractures had [5] Stanley D, Trowbridge EA, Norris SH. The mechanism of clavicular fracture. A
united, and it could be argued that many of these patients probably clinical and biomechanical analysis. J Bone Joint Surg Br 1988;70:461–4.
[6] Nowak J, Mallmin H, Larsson S. The aetiology and epidemiology of clavicular
failed to return for the follow-up examination because they were fractures. A prospective study during a two-year period in Uppsala, Sweden.
asymptomatic, with a healed fracture. This limitation, therefore, Injury 2000;31:353–8.
could overestimate the risk of nonunion in the present study. [7] Bernstein J. Nonoperative treatment compared with plate fixation of displaced
midshaft clavicular fractures. J Bone Joint Surg Am 2007;89:1866–7.
Although several putative intrinsic and extrinsic risk factors [8] Hill JM, McGuire MH, Crosby LA. Closed treatment of displaced middle-third
for nonunion were evaluated in the present study, other risk fac- fractures of the clavicle gives poor results. J Bone Joint Surg Br 1997;79:537–9.
tors such as interposition of soft tissue between fragments and [9] Robinson CM, Court-Brown CM, McQueen MM, Wakefield AE. Estimating the
risk of nonunion following nonoperative treatment of a clavicular fracture. J
genetic predisposition were not identified. Moreover, risk factors Bone Joint Surg Am 2004;86-A:1359–65.
of low prevalence may influence the development of nonunion in [10] Zlowodzki M, Zelle BA, Cole PA, Jeray K, McKee MD. Treatment of acute mid-
individual cases. In particular, comorbidities, including rheuma- shaft clavicle fractures: systematic review of 2144 fractures: on behalf of
the Evidence-Based Orthopaedic Trauma Working Group. J Orthop Trauma
toid disease, immunocompromise, renal failure, epilepsy, and use 2005;19:504–7.
of drugs (corticosteroids and those interfering with vitamin-D [11] Altamimi SA, McKee MD. Nonoperative treatment compared with plate fixation
metabolism), may increase the risk of nonunion. Because of insuf- of displaced midshaft clavicular fractures. Surgical technique. J Bone Joint Surg
Am 2008;90(Suppl. 2 Pt 1):1–8.
ficient frequency of these comorbidities, they cannot investigate
[12] Edwards DJ, Kavanagh TG, Flannery MC. Fractures of the distal clavicle: a case
their identification as statistically significant risk factors. for fixation. Injury 1992;23:44–6.
It is better that these independent risk factors should be used [13] Neer CN. Fractures of the distal third of the clavicle. Clin Orthop Relat Res
to guide clinicians in counseling patients. In addition, patients with 1968;58:43–50.
[14] Neer CN. Nonunion of the clavicle. J Am Med Assoc 1960;172:1006–11.
these risk factors does not imply that their outcome would always [15] Nordqvist A, Petersson C, Redlund-Johnell I. The natural course of lateral
be improved by primary operative intervention. Primary fixation clavicle fracture. 15 (11–21) year follow-up of 110 cases. Acta Orthop Scand
of midshaft clavicular fractures has been advocated in recognition 1993;64:87–91.
[16] Murray IR, Foster CJ, Eros A, Robinson CM. Risk factors for nonunion after non-
of the increased rate of nonunion and inferior functional outcomes operative treatment of displaced midshaft fractures of the clavicle. J Bone Joint
associated with nonoperative treatment of fractures of this type Surg Am 2013;95:1153–8.
[7]. The present study aims to improve awareness of which patients [17] Rowe CR. An atlas of anatomy and treatment of midclavicular fractures. Clin
Orthop Relat Res 1968;58:29–42.
are at greatest risk of nonunion while minimizing the number of [18] Nowak J, Holgersson M, Larsson S. Can we predict long-term sequelae after
patients undergoing unnecessary surgery. fractures of the clavicle based on initial findings? A prospective study with
nine to ten years of follow-up. J Shoulder Elbow Surg 2004;13:479–86.
[19] Gullihorn L, Karpman R, Lippiello L. Differential effects of nicotine and smoke
5. Conclusion condensate on bone cell metabolic activity. J Orthop Trauma 2005;19:17–22.
[20] Moghaddam A, Zimmermann G, Hammer K, Bruckner T, Grutzner PA, von
The risk factors for nonunion after nonoperative treatment Recum J. Cigarette smoking influences the clinical and occupational outcome
of patients with tibial shaft fractures. Injury 2011;42:1435–42.
in midshaft clavicular fractures are multifactorial. Multivari- [21] Kulshrestha V, Roy T, Audige L. Operative versus nonoperative management
ate logistic regression analysis suggests that smoking, fracture of displaced midshaft clavicle fractures: a prospective cohort study. J Orthop
displacement and comminution of fracture are independently pre- Trauma 2011;25:31–8.
[22] Wick M, Muller EJ, Kollig E, Muhr G. Midshaft fractures of the clavicle with a
dictions for an individual likelihood of nonunion. We believe that
shortening of more than 2 cm predispose to nonunion. Arch Orthop Trauma
the predicted risk factors of nonunion should be used to guide clini- Surg 2001;121:207–11.
cians in counseling patients. Primary fixation of midshaft clavicular [23] Morshed S, Corrales L, Genant H, Miclau TR. Outcome assessment in clinical
fractures may be advocated in patients with above-mentioned risk trials of fracture-healing. J Bone Joint Surg Am 2008;90(Suppl. 1):62–7.
[24] Braunstein EM, Goldstein SA, Ku J, Smith P, Matthews LS. Computed tomogra-
factors. In summary, we hope to improve awareness of which phy and plain radiography in experimental fracture-healing. Skeletal Radiol
patients are at greatest risk of nonunion while minimizing the 1986;15:27–31.