This was a multicenter, prospective, randomized clinical trial involving
eight centers, including St. Michael's Hospital, Toronto; Sunnybrook and
Women's College Health Sciences Centre, Toronto; McMaster University Medical
Center, Hamilton; Brantford General Hospital, Brantford; London Health
Sciences Centre, London, Ontario; Royal Columbian Hospital, New Westminster,
British Columbia; Montreal General Hospital, Montreal, Quebec; and Foothills
Medical Centre, Calgary, Alberta, Canada. Institutional approval was obtained
from the research ethics board at each participating site prior to the
initiation of the study. Between April 2001 and December 2004, 132 patients
were enrolled in the study from eight participating study centers (seven
university-affiliated and one community hospital). Eligible patients (see
below) were randomized to nonoperative or operative care for completely
displaced (no cortical contact between the proximal and distal fragments)
midshaft fractures of the clavicle. Patients with isolated fractures and those
with concomitant shoulder girdle fractures were included. The primary outcome
measure was the Disability of the Arm, Shoulder and Hand (DASH)
score17, while
secondary outcome measures included the Constant shoulder score, union rate,
and complication rates. The null hypothesis was that there would be no
differences between the operative and nonoperative groups with respect to
surgeon-based and patient-based upper extremity outcome scores.
Inclusion Criteria
Patients were included in the study if they had (1) a completely displaced
midshaft fracture of the clavicle (no cortical contact between the main
proximal and distal fragments), (2) a fracture in the middle third of the
clavicle (a fracture amenable to plate fixation with a minimum of three screws
in each proximal and distal fragment), (3) an age between sixteen and sixty
years, (4) no medical contraindications to general anesthesia, and (5)
provided informed consent.
Exclusion Criteria
Patients were excluded from the study if they had (1) an age of less than
sixteen years or greater than sixty years, (2) a fracture in the proximal or
distal third of the clavicle, (3) a pathological fracture, (4) an open
fracture, (5) a fracture seen more than twenty-eight days after the injury,
(6) an associated neurovascular injury with objective neurological findings on
physical examination, (7) an associated head injury (a Glasgow Coma Scale
score of <12), (8) an upper extremity fracture distal to the shoulder, (9)
an inability to comply with follow-up (a transient or an inability to read or
complete forms), (10) a medical contraindication to surgery and/or anesthesia
(such as heart disease, renal failure, or active chemotherapy), and (11) a
lack of consent.
Sample Size Calculation
Before beginning of the study, a power analysis was performed. The choice
of sample size was made on the basis of the primary outcome of shoulder
function scores. Assuming a beta error of 0.05 and a power of 0.80, it was
anticipated that sixty patients would be required in each group in order to
demonstrate a 15% difference in the shoulder scores between the two
groups.
Randomization
In the fracture clinic or emergency room, the attending surgeon or
orthopaedic resident identified a patient as being eligible for the study and
the study protocol was introduced. The patient was then seen by the research
nurse, the nature of the study was explained, and consent was obtained.
Typically, the patient took a consent form home for perusal and completion.
Once consent was obtained, randomization was made by the research nurse using
a sequentially numbered, opaque, sealed envelope to either nonoperative care
(a sling) or open reduction and plate fixation in a 1:1 ratio.
Nonoperative Care
Patients randomized to nonoperative care received a standard sling for six
weeks, although compliance was variable: most patients discarded the sling
when the pain subsided. There is no convincing clinical evidence that a closed
reduction of a displaced clavicular fracture can be
maintained18. In a
prior randomized clinical trial comparing a sling and a figure-of-eight
bandage for displaced clavicular shaft fractures, Andersen et al. showed no
functional or radiographic difference at the time of final follow-up and the
patients favored the
sling18. Therefore,
no attempt was made at a closed reduction nor was a figure-of-eight bandage
applied. Following healing, a course of physiotherapy for strengthening was
prescribed.
Operative Technique
Patients randomized to plate fixation had the operation within twenty-eight
days after the injury. Prophylactic antibiotics were given. Under a general
anesthetic, the patient was positioned in a beach-chair semi-sitting position.
The involved shoulder was prepared and draped, and an oblique incision was
made over the fracture site. Larger branches of the identifiable
supraclavicular nerves were identified and protected throughout the procedure;
smaller branches were sacrificed at the surgeon's discretion. The fracture
site was identified, and the fracture was reduced and fixed with a
small-fragment plate on the superior surface of the bone, with the goal being
a minimum of three screws in the main proximal and distal fragments
(forty-four patients were managed with limited contact dynamic compression
plates; fifteen, with 3.5-mm reconstruction plates; four, with precontoured
plates; and four, with other plates) (Figs.
1-A, 1-B, and 1-C). Reconstruction plates were used for physically
smaller individuals (<70 kg). Comminuted fragments were secured with lag
screws if possible, with care being taken to preserve soft-tissue attachments,
and a longer plate was selected to maintain a minimum of three screws in the
primary proximal and distal fragments. If the fragments were too small to
accept fixation, they were loosely sutured into place with number-1 absorbable
suture and positioned under the plate. Bone-grafting was not performed. The
deltotrapezial fascia was closed with interrupted number-1 absorbable sutures
as a distinct layer, followed by skin closure. No drains were used.
A sling was used for comfort for seven to ten days, and then a
physiotherapist instructed the patient in active range-of-motion exercises
that were performed at home. When fracture union (defined as radiographic
union [see below] with no pain or motion with manual stressing of the
fracture) was evident, typically at six weeks, strengthening was allowed, with
a return to full activities (including sports) at three months. However,
compliance with this regimen was variable as the patients were predominantly
young men, and many returned to more aggressive recreational and occupational
activities earlier than recommended.
Assessment
Following enrollment in the study, the patients were seen at six weeks and
at three, six, and twelve months. Assessment included standardized clinical
evaluation and completion of the Constant shoulder score and the Disability of
the Arm, Shoulder and Hand (DASH) score. Both an anteroposterior and a 20°
cephalad radiograph were made for each patient. Radiographic union was defined
as complete cortical bridging between proximal and distal fragments on both
radiographs as determined by the treating surgeon.
Adverse Events and/or Complications
An adverse event or complication was defined as any event that necessitated
another operative procedure or additional medical treatment. Nonunion was
defined as the lack of radiographic healing with clinical evidence of pain and
motion at the fracture site at one year. Radiographic malunion, defined as
loss of anatomic contour of the clavicle, was universal in the nonoperative
group. Symptomatic malunion was defined as union of the fracture in a
shortened, angulated, or displaced position with weakness, easy fatigability,
pain with overhead activity, neurologic symptoms, and shoulder asymmetry with
a completed or planned corrective osteotomy. Complex regional pain syndrome
was diagnosed by the presence of dysesthetic pain and hyperesthesia extending
into the hand of the involved limb, vasomotor changes, skin atrophy, and
diffuse
osteopenia19.
Statistical Analysis
Statistical analysis was performed with SPSS software (version 13.0; SPSS,
Chicago, Illinois). All scale variables were tested for normality with the
Kolmogorov-Smirnov test. The main effect of treatment on the DASH and Constant
scores was analyzed with use of a two-way analysis of variance with treatment
(operative or nonoperative) and time (six, twelve, twenty-four, fifty-two, and
104 weeks) as independent factors and the Tukey post hoc method for the
comparison of means. The Student t test was used for the comparison of means
for parametric scale variables in independent groups. Nominal variables were
tested by the chi-square or Fisher exact test. The Pearson correlation
coefficient was used for comparison of the DASH scores at one year with the
Injury Severity Scores (ISS) and total vertical and horizontal displacement
(total xy). All tests were two-sided. The results were considered to be
significant at p < 0.05.
One hundred and thirty-two patients with 132 fractures were entered in the
study between April 2001 and December 2004. Sixty-seven patients were
randomized to the operative group and sixty-five to the nonoperative group.
One patient randomized to operative repair declined surgery, and one patient
randomized to nonoperative treatment insisted on operative repair. Both were
followed in their original groups with use of the intention-to-treat
principle20. One
patient in the nonoperative group died in a subsequent motor-vehicle accident,
and fifteen were lost to follow-up by one year. Five patients in the operative
group were lost to follow-up at one year. Significantly more patients in the
nonoperative group were lost to follow-up (p = 0.008; see Discussion). Thus,
sixty-two patients in the operative group and forty-nine in the non-operative
group completed the one-year assessment. There were no demographic differences
between the operative and nonoperative groups, and there were no differences
with regard to mechanism of injury, associated fractures and/or injuries, or
ISS between the groups (Table
I).
Constant Shoulder Scores
The operative group had significantly superior Constant shoulder scores at
all time-points (p < 0.01) (Fig.
2). This difference persisted at one year (p = 0.001), and the
magnitude of the difference was approximately 10 points, which is considered a
clinically measurable
amount21.
DASH Scores
The operative group had significantly superior (i.e., lower) DASH scores at
all time-points (Fig. 3),
persisting to one year (p < 0.01). The magnitude of this difference was
approximately 10 points, which is considered a clinically relevant
amount17.
Patient Satisfaction
At each assessment, patients were asked "Are you satisfied with your
shoulder?" At each assessment, patients in the operative group were more
likely to reply "yes" to this question (p = 0.02, odds ratio = 3.8
at six weeks; p = 0.001, odds ratio = 4.4 at twelve weeks; p = 0.03, odds
ratio = 3.2 at twenty-four weeks; and p = 0.002, odds ratio = 3.5 at fifty-two
weeks).
Range of Motion
Range of motion was well maintained, and there were no significant
differences in measured range of motion between the two groups. No patient
lost >10° of motion in any plane.
Fracture Union
The mean time to union was 16.4 weeks in the operative group and 28.4 weeks
in the nonoperative group (p = 0.001). Nonunion occurred in two patients in
the operative group and in seven in the nonoperative group
(Table II). One non-union in
the operative group was in the patient who had been randomized to operative
fixation but declined surgery and had subsequent development of the nonunion.
He was followed with the intention-to-treat
principle20,
although technically this was not a failure of operative intervention.
Adverse Events and/or Complications
Complications, including nonunion and symptomatic mal-union, were more
frequent in the nonoperative group. Complications in the operative group
tended to be hardware-related (plate irritation and removal, and wound
problems). Wound infection and dehiscence following plate fixation of the
clavicle has been a feared complication. We had three patients with such
complications, and all were managed with antibiotics and local wound care.
Once fracture union had occurred, each patient underwent hardware removal and
irrigation and/or débridement with successful resolution of the
infection (Table II). One
patient in the operative group experienced premature hardware failure in an
all-terrain vehicle accident six weeks after fixation and required repeat
fixation.
Appearance of the Shoulder
Patients were specifically questioned about their satisfaction or
dissatisfaction regarding the appearance of the shoulder (and incision, if
applicable) at one year following the injury
(Table III). Patients in the
operative group were more likely to be satisfied with the appearance of the
shoulder (p = 0.001).
Radiographic Outcome
Anatomic reduction was obtained and maintained in all sixty-two patients in
the operative group except for one in whom early mechanical failure of the
plate occurred at six weeks. Correlating displacement and outcome in the
operative group was not possible since anatomic reduction was obtained and
maintained in all patients, but an association was found between total
displacement at the fracture site (vertical displacement and shortening
combined) and DASH scores at one year in the nonoperative group (r = 0.326, p
= 0.05); that is, greater displacement correlated with a higher DASH score or
more patient-related disability. With the numbers available, patients with
multiple shoulder girdle injuries did not demonstrate significantly worse
scores than those with isolated injuries (p = 0.24).
Traditionally, clavicular fractures have been treated nonoperatively. In
the 1960s, Neer and Rowe reported on the nonoperative treatment of clavicular
fractures3,4.
Neer reported nonunion in only three of 2235 patients with middle-third
fractures treated by closed
methods3, while Rowe
reported nonunion in four of 566 clavicular
fractures4. This
information dominated the clinical approach to displaced clavicular fractures.
These studies also suggested a higher non-union rate with operative care.
However, more recent studies have shown that the union rate for displaced
midshaft fractures of the clavicle may not be as favorable as once thought. In
a prospective, observational cohort study, Robinson et al. described a
consecutive series of 868 patients with clavicular fractures, 581 of whom had
a midshaft diaphyseal
fracture22. They
found a significantly higher nonunion rate (21%) for the displaced, comminuted
midshaft fractures (p < 0.05). In a letter to the editor, Brinker et al.
analyzed the data in that study and suggested a nonunion rate ranging between
20% and 33% for displaced, comminuted fractures in
males23. Similarly,
in a study of fifty-two displaced midshaft clavicular fractures, Hill et al.
reported that eight patients had a nonunion and sixteen patients had an
unsatisfactory outcome on the basis of patient-oriented
measures6. They
concluded that displacement of the fracture fragments by >2 cm was
associated with an unsatisfactory result. A meta-analysis of recent studies
revealed that the rate of nonunion for displaced midshaft clavicular fractures
was 2.2% (ten of 460 patients) after plate fixation compared with 15.1%
(twenty-four of 159 patients) after nonoperative care, a relative risk
reduction for nonunion of
86%14. That
meta-analysis also showed that primary plate fixation was, contrary to
prevailing opinion, a safe and reliable
procedure14.
Previously, malunion of the clavicle (which is typical with displaced
fractures) was thought to be of radiographic interest only and required no
treatment. However, it is becoming increasingly apparent that clavicular
malunion is a distinct clinical entity with radiographic, orthopaedic,
neurologic, and cosmetic features. Nowak et al. examined the late sequelae in
208 adult patients with clavicular fractures and found that, at ten years
after the injury, ninety-six patients (46%) still had symptoms despite the
fact that only fifteen (7%) had a
non-union24. McKee
et al. described the typical inferior, shortened, and anteriorly rotated
position of the distal fragment in clavicular malunion and the symptoms that
resulted from it8.
Corrective osteotomy and plate fixation improved the DASH score from 32 to 11,
with fourteen of fifteen patients who were satisfied with the
procedure25.
Similar results were found with corrective osteotomy for clavicular malunion
in studies by Basa-mania, Bosch et al., and Chan et
al.9-11.
In the forty-nine patients in our study who were treated nonoperatively and
had a healed fracture, many (nine; 18%) had the typical symptoms of malunion
develop and they elected corrective
osteotomy26-29.
Most of the malunions were associated with substantial clavicular displacement
and shortening, although the effect of shortening on function remains
controversial30,31.
Our study found (in the nonoperative group) a direct relationship between
increased displacement and a worse DASH score.
While it is unclear why there is such a dramatic difference between the
outcome of clavicular fractures in previous reports and those in contemporary
studies, there are several possibilities. The initial reports often included
data on clavicular fractures in children, who have inherent healing abilities
and remodeling potential, and their data may have artificially improved the
overall
results1-4.
Second, the use of patient-oriented outcome measures, as in the studies by
Hill et al. and McKee et al., has been shown to reveal functional deficits in
the upper extremity that are not detected by traditional surgeon-based scores;
it is unlikely that such patient dissatisfaction would be detected in a 1960
study that focused on radiographic
outcome6,8.
A related issue is changing patient expectations: most active clinicians are
acutely aware that a patient today is more likely to expect a rapid return to
pain-free function following a fracture (and be more vocal when this does not
occur) than his or her 1960 counterpart. Last, it may be that injury patterns
are changing. In one study of clavicular shaft fractures in patients with
polytrauma, the presence of a clavicular fracture was found to be associated
with a mortality rate of 32% (thirty-four of 105 patients) (mainly due to
concomitant chest and head
injuries)5.
Survivors displayed a substantial level of residual disability in the involved
shoulder. Most studies have revealed a correlation between fracture
comminution (and displacement) and worse outcome, and these fracture features
are associated with higher-energy
trauma5,6,14,22.
Thus, there are surviving patients with clavicular fractures who have an
intrinsically worse prognosis and in whom long-term sequelae may be more
common.
In contradistinction to earlier case series, modern studies on primary
plate fixation of acute midshaft clavicular fractures have described high
rates of successful results with rates of union ranging from 94% to 100% and
low rates of infection and surgical complications: a recent meta-analysis of
plate fixation for 460 displaced fractures revealed a nonunion rate of only
2.2%14-16.
With improved implants, prophylactic antibiotics, and better soft-tissue
handling, plate fixation has been a reliable and reproducible technique.
Our study examined 111 patients with completely displaced midshaft
clavicular fractures randomized to either traditional sling treatment or open
reduction and internal fixation with a plate. There was a clear superiority of
operative fixation, with significantly superior surgeon-based (Constant
shoulder score) and patient-based (DASH) outcome measures at every time-point
in the study. The improvement in scores (approximately 10 points for each) was
clinically relevant as well as significantly superior. Patients who underwent
operative fixation also had an earlier return to normal function. In addition,
there was a significant reduction in the risk of non-union in the operative
group (two of sixty-two patients had a nonunion) compared with the
nonoperative group (seven of forty-nine patients had a nonunion) (p = 0.001).
Complications in the operative group were typically hardware-related (plate
irritation and wound complications) and were corrected by plate removal in all
cases. Refracture was not seen, despite the fact that many patients returned
to heavy contact and so-called extreme sports (fifty-five of 111 patients in
the study sustained the fracture during sports, bicycling, or skiing and/or
snowboarding). Most of the plates used in our study were straight plates
contoured to fit the clavicle. More recently, we changed to anatomically
designed s-shaped contoured
plates25. Our
preliminary experience with these plates suggests a dramatically reduced
prevalence (and severity) of soft-tissue irritation, and it is possible that
this may decrease the need for plate removal in the operative group.
Appearance is important to patients, and an unsightly scar has been a
traditional deterrent to operative treatment of clavicular fractures. We
specifically investigated this component of patient satisfaction in our study
(see Table III). Despite the
prevalence of hardware prominence and incisional complications (numbness and
sensitivity) in the operative group, more patients in this group (fifty-two of
sixty-two patients) answered "yes" to the question "Are you
satisfied with the appearance of your shoulder?" than in the
nonoperative group (twenty-six of forty-nine; p = 0.001). In this group of
predominantly young male patients, a droopy shoulder (nonoperative group)
seemed to be of greater cosmetic concern than a scar (operative group).
One of the weaknesses of our study is that we used only plate fixation in
the operative group: intramedullary fixation is also an
option32-34.
A direct comparison between the two techniques in a prospective trial is
required. Another weakness of our study is the number of patients who did not
complete the assessment period. However, in a group of fracture patients who
were predominantly young men, the rate of patients lost to follow-up in our
study is comparable with that in other studies and we do not believe that it
jeopardizes our results. Specifically, the greatest concern in a study such as
ours is that a number of complications in the (experimental) operative group
would be missed because of lack of follow-up. However, we followed sixty-two
of sixty-seven operative patients to definitive outcome. We believe that the
patients who did not undergo surgery were less likely to feel committed to the
study, did not return because of a lack of a requirement for postoperative
care, or were potentially unhappy with their allocated treatment. We know of
at least two such individuals who obtained operative treatment for a non-union
elsewhere. Lastly, with time, our reintervention rate may increase, especially
in the operative group (i.e., for hardware removal).
In conclusion, our study shows that early primary plate fixation of
completely displaced midshaft clavicular fractures results in improved
patient-oriented outcomes, improved surgeon-oriented outcomes, earlier return
to function, and decreased rates of nonunion and malunion. There were no
catastrophic complications in the operative group such as brachial plexus
palsy, vascular injury, or pneumothorax; hardware removal was the most common
reason for reintervention. Patients were more satisfied with the shoulder (and
its appearance) following operative intervention. While we stress that our
findings are applicable only to a specific subset of clavicular injuries, our
data support primary plate fixation of completely displaced midshaft
clavicular fractures in active adults. ?