675

C OPYRIGHT Ó 2012 BY T HE J OURNAL OF B ONE AND J OINT S URGERY, I NCORPORATED

Operative Versus Nonoperative Care of Displaced

Midshaft Clavicular Fractures: A Meta-Analysis
of Randomized Clinical Trials
Robbin C. McKee, Daniel B. Whelan, MD, FRCS(C), Emil H. Schemitsch, MD, FRCS(C), and Michael D. McKee, MD, FRCS(C)

Investigation performed at St. Michael’s Hospital and the University of Toronto, Toronto, Ontario, Canada

Background: Recent studies have suggested benefits following primary operative fixation of substantially displaced
midshaft fractures of the clavicle. We reviewed randomized clinical trials of operative versus nonoperative treatment of these
fractures, and pooled the functional outcome and complication rates to arrive at summary estimates of these outcomes.
Methods: A systematic review of the literature was performed to identify studies of randomized clinical trials comparing
operative versus nonoperative care for displaced midshaft clavicular fractures.
Results: Six studies (n = 412 patients, mean Detsky score = 15.3) were included. The nonunion rate was higher in the
nonoperatively treated patients (twenty-nine of 200) than it was in patients treated operatively (three of 212) (p = 0.001).
The rate of symptomatic malunion was higher in the nonoperative group (seventeen of 200) than it was in the operative
group (0 of 212) (p < 0.001).
Conclusions: Operative treatment provided a significantly lower rate of nonunion and symptomatic malunion and an
earlier functional return compared with nonoperative treatment. However, there is little evidence at present to show that
the long-term functional outcome of operative intervention is significantly superior to nonoperative care.
Level of Evidence: Therapeutic Level I. See Instructions for Authors for a complete description of levels of evidence.

A
midshaft fracture of the clavicle is a common skeletal
injury that accounts for 2.6% to 10% of all adult frac-
tures1,2. Older studies report that the nonunion rate with
conservative treatment is low (less than 1%)3,4. Recent pro-
spective studies focusing on the nonoperative treatment of
displaced midshaft fractures in the adult population describe
nonunion rates of 15% to 20%, objective shoulder muscle
strength loss of 18% to 33%, poor early functioning of the
injured shoulder, and up to 42% of patients with residual se-
quelae at six months after injury5-7.
Due to the increasing recognition of suboptimal out-
come following nonoperative treatment, primary operative
fixation has become increasingly popular for these injuries.
Operative implants and techniques vary, and include in-
tramedullary osteosynthesis and plate fixation8-11. There are a
number of recently published, randomized prospective clin-
ical trials12-17. Since this injury commonly afflicts a young,
active population, rapid return to function and early union
is a priority in addition to the more traditional outcomes,
such as recovery of muscle strength, range of motion, and the

Disclosure: None of the authors received payments or services, ei- This article was chosen to appear
ther directly or indirectly (i.e., via his or her institution), from a third electronically on March 14, 2012, in
party in support of any aspect of this work. One or more of the advance of publication in a regularly
authors, or his or her institution, has had a financial relationship, in scheduled issue.
the thirty-six months prior to submission of this work, with an entity in
the biomedical arena that could be perceived to influence or have the A commentary by William T. Obremskey,
potential to influence what is written in this work. No author has had MD, MPH, is linked to the online version of
any other relationships, or has engaged in any other activities, that this article at jbjs.org.
could be perceived to influence or have the potential to influence
what is written in this work. The complete Disclosures of Potential
Conflicts of Interest submitted by authors are always provided with
the online version of the article.

J Bone Joint Surg Am. 2012;94:675-84 d http://dx.doi.org/10.2106/JBJS.J.01364

 676
TH E JO U R NA L O F B O N E & JO I N T SU RG E RY J B J S . O RG
d
V O LU M E 94-A N U M B E R 8 A P R I L 18, 2 012
d d
avoidance of long-term complications (such as nonunion and
symptomatic malunion).
There are numerous studies on nonoperative and oper-
ative care, but few randomized clinical trials have prospectively
compared the two. In this systematic review, the primary aim
was to use Level-I evidence from recent randomized clinical
trials to arrive at summary estimates of treatment effect, po-
tential harm, and patient outcome.
Materials and Methods
Research Questions
W e posed three questions with regard to the treatment of acute, displaced
midshaft fractures of the clavicle: (1) Is there a significant difference in
patient and surgeon-assessed outcomes of primary operative fixation versus
nonoperative treatment? (2) What are the differences in the rate and severity
of a defined set of complications (nonunion or delayed union, symptomatic
malunion, infection, hardware removal, neurologic compromise, and refrac-
ture) for operative fixation versus nonoperative treatment at one year? and (3)
Is there evidence that, when compared with nonoperative care, primary fixation
of displaced midshaft clavicular fractures minimizes disability and improves
function at twelve months after injury?
Eligibility Criteria
Studies were included in this review if they were randomized prospective trials
of operative versus nonoperative treatment of acute, completely displaced
midshaft fractures of the clavicle. Quasi-randomized studies (nonrandom
treatment allocation) were excluded, as were studies that included delayed
union or nonunion, children (i.e., younger than sixteen years of age), or
pathological fractures.
Literature Search and Review of the Proceedings of
Annual Meetings
An online MEDLINE database was used with the search string ‘‘clavicle’’ [key-
word] and ‘‘fracture’’ [keyword] and randomized clinical trial [publication type:
ptyp]. The two senior authors of this paper (M.D.M. and E.H.S.) reviewed the
titles, and appropriate abstracts were reviewed. Content experts were contacted to
inquire about unpublished trials or trials in progress. Proceedings of the annual
meetings of the American Academy of Orthopaedic Surgeons (AAOS), Ortho-
paedic Trauma Association (OTA), and American Shoulder and Elbow Surgeons
(ASES) for the years 1990 to 2010 (inclusive) were also manually reviewed by one
of the authors (R.C.M.). Abstracts for all eligible citations were reviewed by the
two senior authors (M.D.M. and E.H.S.) for inclusion with use of the above
criteria.
Outcome Scores
A variety of outcome scores were acknowledged in this review, including the
Constant Shoulder Score (CS), the Disabilities of the Arm, Shoulder and Hand
(DASH) score, the L’Insalata Shoulder Rating Questionnaire, and the Single As-
sessment Numeric Evaluation (SANE)
18-21
. Secondary outcome measures of this
review include complications, range of motion, return to work, and clavicular
shortening.
Validity Assessment
Two senior authors (M.D.M. and D.B.W.) independently assessed the
methodological quality of the selected studies, without blinding, with use of
the Detsky score, a fourteen-item index containing questions arranged in five
equally weighted categories of randomization, outcome measures, eligibility
22
criteria, interventions, and statistics . The Detsky scale was chosen as it
has been utilized previously to determine the scientific quality of published
orthopaedic randomized trials and has demonstrated consistency and
23
reliability .
O P E R AT I V E V E R S U S N O N O P E R AT I V E C A R E OF DI SP L ACE D
M I D S H A F T C L AV I C U L A R F R A C T U R E S
Data Extraction and Analysis
Two senior authors (M.D.M. and D.B.W.) independently extracted data from all
included studies with the use of a prespecified data extraction form. A third
investigator (R.C.M.) then reviewed the forms for completeness and agreement
and entered the data (population demographics, mechanism of injury, operative
methods, shoulder range of motion, and patient and surgeon outcome scores,
complications [nonunion or delayed union, symptomatic malunion, infection,
hardware removal, neurologic compromise, and refracture], and return to
activity and/or work) into a custom spreadsheet. Disagreements in data ex-
traction were resolved by discussion and consensus. Data across the six studies
were pooled and summary estimates of treatment effect (relative risks with
associated 95% confidence intervals) were calculated. A random effects model
was used to ensure that these studies represent a random sample of all po-
24
tentially available studies . In the pooled analysis, studies were weighted by the
inverse of the variance for the reported outcome. Homogeneity across the
studies was assessed with a chi-square analysis, with p £ 0.10 being considered
25
significant. The Review Manager (RevMan) software program (The Nordic
Cochrane Centre, Copenhagen, Denmark), provided by The Cochrane Col-
laboration, was used for graphical representation of the pooled data.
Source of Funding
This study was supported by an internal educational grant from the St.
Michael’s Hospital Orthopaedic Research and Education Fund (to R.C.M.).
Research Ethics Board approval is not required for this type of study at our
institution.
Results
Literature Search and Review of the Proceedings of
Annual Meetings
T he keywords ‘‘clavicle’’ and ‘‘fracture’’ produced 988 hits, of
which twelve were randomized clinical trials. Eight trials
were excluded because of repeated citations (n = 2), lack of a
nonoperative control group (n = 4), lack of an operative group
(n = 1), or because the study was a cadaver study (n = 1). The
remaining four articles were included in this meta-analysis.
Content expert knowledge and meeting abstracts yielded four
additional studies: two are in progress (no data were avail-
able), one had been accepted for publication (full manuscript
was available for review), and one had been published in
abstract form (unpublished article and data were forwarded
by the author) (Fig. 1). These latter two are included in this
review.
Six studies (four published, one accepted for publication,
and one abstract with unpublished data available) were in-
cluded (n = 412 patients, 212 treated operatively, 200 treated
nonoperatively); all were randomized clinical trials that en-
rolled patients with completely displaced midshaft clavicular
fractures. The studies were performed in various countries,
including Finland13, the United States14,16, Austria17, Canada12
and Germany15 (Table I). There were no significant differences
in baseline demographics between the operative and non-
operative groups (Appendix), and individuals enrolled in all
six studies were homogeneous, consisting of young (mean age
range, 25.0 to 41.3 years), predominantly male patients. Two
studies examined specific subgroups (Judd et al. studied active
duty military personnel16, and Witzel studied athletes15). Op-
erative techniques included plate fixation in three studies and
intramedullary pin fixation in three studies. Nonoperative
treatment consisted of a standard sling in all six studies.
 677
TH E JO U R NA L O F B O N E & JO I N T SU RG E RY J B J S . O RG
d
V O LU M E 94-A N U M B E R 8 A P R I L 18, 2 012
d d
Fig. 1
Graph showing flow of studies retrieved for systematic review of random-
ized clinical trials for operative versus nonoperative treatment of clavicular
fractures.
Validity Assessment
Of the six studies, only one study reported its level of evidence
(Therapeutic Level I) in the publication itself 12. Study quality
was independently assessed through the use of the Detsky score
(five articles). One study was in abstract form and an accurate
score could not be assigned13. The mean Detsky score was 15.3
O P E R AT I V E V E R S U S N O N O P E R AT I V E C A R E OF DI SP L ACE D
M I D S H A F T C L AV I C U L A R F R A C T U R E S
(range, 14 to 16) (Table II). All five studies that reported the
method of randomization used a sealed envelope technique, a
method considered suboptimal by modern trial standards22.
Treatment
Fixation was accomplished with a variety of small fragment
plates as the operative technique in three studies12-14 and with
pin osteosynthesis in three studies15-17 (i.e., small-diameter
[2.0, 2.5, or 3.0 mm] intramedullary titanium pins in two
studies15,17 and a modified Hagie pin in one study16). The
postoperative rehabilitation was similar in all four studies in
which it was described: a sling for comfort for ten days, fol-
lowed by early active range-of-motion exercises. At six weeks
postoperatively, strengthening and a gradual return to sport
and/or heavy lifting, overhead, or repetitive activities was al-
lowed12,14,16,17. All studies used a standard sling for nonoperative
care12-17, with institution of range-of-motion exercises, as pain
allowed, followed by physiotherapy12,14,16,17.
Study Synopses (Table III)
Virtanen et al. (2010)
Virtanen et al. randomized sixty patients to sling (n = 32), or
plate (n = 28) treatment13. At one year, twenty-six operatively
treated patients and twenty-five nonoperatively treated patients
had been followed and Constant Shoulder Scores (a mean of
86.1 for the nonoperative group and 86.5 for the operative
group, p = 0.90), DASH scores (a mean of 7.1 for the nonop-
erative group and 4.3 for the operative group, p = 0.81), and
visual analog scale scores for pain (on a scale from 1 to 100
[with 1 representing no pain and 100 representing the worst
pain imaginable], a mean of 7 for the nonoperative group and
3 for the operative group, p = 0.88) were recorded. There were
six nonunions in the nonoperative group versus none in the
operative group (six [24%] of twenty-five versus 0 [0%] of
twenty-six, p = 0.01). These six patients with nonunion had
greater initial displacement of the fracture (p = 0.002) and
worse DASH scores (mean, sixteen points lower, p = 0.05).
Smekal et al. (2009)
Smekal et al. randomized sixty-eight patients to sling or elastic
titanium intramedullary pin fixation17. Sixty patients (thirty in
each group) were assessed two years after the injury occurred.
Time to union was shorter in the operative group (12.1 weeks)
compared with the nonoperative group (17.6 weeks) (p =
0.04). The DASH scores remained significantly lower (p < 0.05)
in the operative group for the first eighteen weeks after injury.
The CS was significantly higher (p < 0.05) at both the six
month and two-year follow-up in favor of the operative group.
Delayed union (no evidence of healing at twenty-four weeks
after injury) developed in six patients in the nonoperative
group (six of thirty, 20%) versus one of the operative group
(one of thirty, 3%) (p = 0.02).
Judd et al. (2009)
In the study by Judd et al.16, fifty-seven military personnel were
randomized to Hagie pin fixation (twenty-nine) or a standard
 678
TH E JO U R NA L O F B O N E & JO I N T SU RG E RY J B J S . O RG
d
O P E R AT I V E V E R S U S N O N O P E R AT I V E C A R E OF DI SP L ACE D
V O LU M E 94-A N U M B E R 8 A P R I L 18, 2 012
d d
M I D S H A F T C L AV I C U L A R F R A C T U R E S

TABLE I Study Characteristics (Patients at the Time of Final Follow-up)*

No. of Nonunions per

Total No. of Patients
No. of Followed Outcome Scores
Journal or Patients Operative
Year Author Meeting Followed Type Operative Nonoperative Operative Nonoperative
13
2010 Virtanen et al. Annual Meeting 51 Plate 0/26 6/25 CS = 86.5, CS = 86.1,
of the American DASH = 4.3 DASH = 7.1
Academy of
Orthopaedic
Surgeons
17
2009 Smekal et al. Journal of 60 Pin 0/30 3/30 CS = 97.9 CS = 93.7
Orthopaedic DASH scores
Trauma were significantly
better for the first
18 weeks in the
operative group
16
2009 Judd et al. The American 57 Pin 1/29 1/28 3-weeks 3-weeks
Journal of SANE = 49.8, SANE = 36.4,
Orthopedics 1-year 1-year
SANE = 93.5 SANE = 97
2007 Canadian The Journal of 111 Plate 2/62 7/49 CS = 96.1, CS = 90.8,
Orthopaedic Bone DASH = 5.2 DASH = 13.0
Trauma and Joint Surgery
Society (American)
12
(COTS)
15
2007 Witzel Zeitschrift für 68 Pin NA/35 NA/33 Postoperative
Orthopädie und mobility and
Unfallchirurgie strength measures
were significantly
better in the
operative group
2000 Smith Annual Meeting 65 Plate 0/30 12/35 NA NA
14
et al. of the American
Shoulder and
Elbow Surgeons
Totals 412 3/212 29/200 Weighted Weighted
average average
of CS, 94.3 of CS, 90.2

*CS = Constant Score, DASH = Disabilities of the Arm, Shoulder and Hand score, SANE = Single Assessment Numeric Evaluation score, NA = not
available.

sling (twenty-eight). The SANE and L’Insalata scores were sig-
nificantly higher at week three for the operative group (p < 0.044
and p < 0.015). There were no significant differences between
these two groups at any other time. One patient in each group
developed a nonunion (one [3%] of twenty-nine in the operative
group versus one [4%] of twenty-eight in the nonoperative
group). This study is notable for the high complication rate in
the operative group (48%) compared with the nonoperative
group (7%). The high complication rate for the operative group
was due primarily to minor complications from pin prominence
posterolaterally (causing pin-track infections and/or require-
ment for early pin removal).
Canadian Orthopaedic Trauma Society (COTS) (2007)
The COTS study randomized 132 patients to a standard sling
(sixty-five) or small-fragment plate fixation (sixty-seven)12. Forty-
nine nonoperatively treated patients and sixty-two operatively
treated patients completed the one year follow-up. The CS and
DASH scores were significantly better at all time points for the
operative group of this study (p < 0.01). There were two
nonunions in the operative group (3%) and seven nonunions in
the nonoperative group (14%) (p = 0.042).
Witzel (2007)
The study by Witzel randomized thirty-three patients to non-
operative care (sling) and thirty-five to operative fixation with
an elastic intramedullary pin, and it concentrated on dis-
tinguishing differences in early return to function15. Early
(one-month) pain scores were significantly better for the
operative group (p = 0.05), as was strength (p = 0.01). It was
also noted that, on the sixtieth day, 80% of the operatively
treated patients resumed athletic activity, while only 55% of
nonoperatively treated patients did.
Smith et al. (2000)
In a randomized clinical trial, Smith and colleagues random-
ized one hundred patients to a sling or small-fragment plate
fixation; only thirty-five nonoperatively treated and thirty
operatively treated patients completed follow-up (mean
follow-up, 18.5 months)14. Of the thirty operatively treated
 679
TH E JO U R NA L O F B O N E & JO I N T SU RG E RY J B J S . O RG
d
O P E R AT I V E V E R S U S N O N O P E R AT I V E C A R E OF DI SP L ACE D
V O LU M E 94-A N U M B E R 8 A P R I L 18, 2 012
d d
M I D S H A F T C L AV I C U L A R F R A C T U R E S

TABLE II Study Quality*

Randomization Lost to Results and Detsky

Technique Follow-up Follow-up Conclusions Comments Score†

Virtanen Blocked randomization, 1 year 15% Equal functional results at No null hypothesis 16/20
13
et al. sized randomly from one year, but nonoperative
(2010) 4-10, sealed envelope group had higher risk of
nonunion
Smekal Balanced block 2 years 13% CS scores were higher and Randomization block 16/20
17
et al. randomization with DASH scores were lower in size known by investigators,
(2009) a block size of 4, and the operative group, and so partially predictable, no
sealed envelope this group also had faster null hypothesis or sample-size
mobilization and less pain calculation stated, take-home
DASH scores not monitored
Judd Sealed envelope 1 year 0% Higher complication rate Inconsistent follow-up at all 16/21
16
et al. in the operative group, Hagie time points (lowest showing
(2009) pin is not recommended is 46%), no null hypothesis
or sample-size calculation
stated
12
COTS Permutation blocks of 1 year 16% Operative care improved Multicenter trial (8 centers), 16/20
(2007) 4 per investigational patient-oriented outcomes number of surgeons not stated
site, sealed envelope and surgeon-oriented
outcomes, patients had
earlier return to function
and decreased rates of
malunion and nonunion
15
Witzel Not stated 120 days Not stated Operative care group had Randomization technique not 14/20
(2007) improved mobility, less pain, described, no null hypothesis or
and were significantly stronger sample-size calculation stated
than the patients who received
nonoperative care
Smith Sealed envelope Until radiographic 35% Results suggest that all Unpublished article, attrition 14/20
14
et al. union or nonunion midshaft fractures should bias—only 65% for follow-up, no
(2000) be operatively fixed if null hypothesis or sample-size
displaced >2 cm, to calculation stated
prevent nonunions

*CS = Constant Score, DASH = Disabilities of the Arm, Shoulder and Hand score, COTS = Canadian Orthopaedic Trauma Society. †Detsky score is
designed to evaluate the quality of a randomized clinical trial (maximum score is 20 for surgical trial).

patients, there were no nonunions or symptomatic malunions Complications

(0%), and twenty-five patients were satisfied with their out-
come (five were dissatisfied). Of the thirty-five nonoperatively
treated patients, there were twelve nonunions (34%) and four
symptomatic malunions (11%): twenty-six patients were sat-
isfied with their outcome, and nine were dissatisfied.
Outcome Scores
Three studies used Constant Shoulder Scores12,13,17. From these
three studies, the weighted average CS for the operative group
was 94.3, and the weighted average CS for the nonoperative
group was 90.2 (n = 222 total patients followed) at one year.
DASH scores were also used in those three studies, but actual
numbers were only found for two of the three studies: one study
represented the DASH in graphical form only. The weighted
average DASH favors the operative treatment (4.9) at one year in
comparison with conservative care (10.7) (n = 162). The study
that reported DASH scores graphically noted a statistically better
(lower) DASH score in the operative group for the first eighteen
weeks.
We defined a specific set of complications that were reported in
all of the trials: nonunion (usually defined as no evidence of
healing at fifty-two weeks after injury)/delayed union (no evi-
dence of healing at twenty-four weeks after injury), symptomatic
malunion, infection, hardware removal, neurologic symptoms,
and refracture. The complication rate was found to be sixty-two
complications (29%) in 212 operatively treated patients and
eighty-four complications (42%) in 200 nonoperatively treated
patients (n = 412 total patients followed, p = 0.0075) (Fig. 2).
Thus, patients treated nonoperatively were more likely to de-
velop a complication than those treated with primary operative
repair. When complications are summated across studies, an
unweighted chi-square analysis shows a lower complication rate
in the operative group. However, in the weighted test for overall
effect, this difference does not reach significance (Fig. 2) The
predominant complications in the nonoperative group were
nonunion, neurologic (including brachial plexus irritation and
compression), and symptomatic malunion. The commonest
operative complications consisted of local hardware irritation or
 680
TH E JO U R NA L O F B O N E & JO I N T SU RG E RY J B J S . O RG
d
O P E R AT I V E V E R S U S N O N O P E R AT I V E C A R E OF DI SP L ACE D
V O LU M E 94-A N U M B E R 8 A P R I L 18, 2 012
d d
M I D S H A F T C L AV I C U L A R F R A C T U R E S

TABLE III Complications

Operative Group (N = 212 Patients) Nonoperative Group (N = 200 Patients)

13
Virtanen et al. (2010) 1 refracture 1 brachial plexus irritation
3 delayed unions 2 refractures
2 symptomatic malunions
1 delayed union
6 nonunions
17
Smekal et al. (2009) 1 delayed union 6 delayed unions
2 implant failures 2 symptomatic malunions
5 medial nail protrusions 3 transient neurogenic compromises
3 nonunions
16
Judd et al. (2009) 6 pin track infections 1 refracture
3 pin removals 1 nonunion
1 transient radial nerve palsy
1 delayed union
1 pin fracture
1 refracture
1 nonunion
Canadian Orthopaedic Trauma 3 wound infections 9 symptomatic malunions
12
Society (COTS) (2007) 5 hardware irritations (removal required) 7 transient brachial plexuses
8 transient brachial plexuses 7 nonunions
2 nonunions
15
Witzel (2007) NR* NR*
14
Smith et al. (2000) 3 mild nerve compressions 17 mild nerve compressions
15 hardware removals for discomfort 4 symptomatic malunions
12 nonunions
Total Complications 62 84

*NR = not reported.

pin protrusion (treated with removal of hardware), and wound primarily of hardware-related complications from prominence
infection (Table III). One study was a statistical outlier in that the of the pin posterolaterally), compared with 7% in the nonop-
complication rate in the operative group was 48% (consisting erative group16. This was a significant source of heterogeneity in

Fig. 2
Graph showing complication rates comparing operative (experimental) versus nonoperative (control) groups (p = 0.18). The size of each square is
proportional to the weight of the study. Cots = Canadian Orthopaedic Trauma Society. Z = p value of weighted test for overall effect, CI = confidence interval,
df = degrees of freedom, I2 = test statistic.
 681
TH E JO U R NA L O F B O N E & JO I N T SU RG E RY J B J S . O RG
d
O P E R AT I V E V E R S U S N O N O P E R AT I V E C A R E OF DI SP L ACE D
V O LU M E 94-A N U M B E R 8 A P R I L 18, 2 012
d d
M I D S H A F T C L AV I C U L A R F R A C T U R E S

Fig. 3
Graph showing comparison of nonunion and symptomatic malunion rates between operative (experimental) and nonoperative (control) groups (p <0.0001).
The size of each square is proportional to the weight of the study. Cots = Canadian Orthopaedic Trauma Society. Z = p value of weighted test for overall effect,
CI = confidence interval, df = degrees of freedom, I2 = test statistic.

the assessment of complication rates (p = 0.01). The use of this
intramedullary implant (Hagie pin) in this fashion is associated
with a high rate of requirement for removal8,16.
Nonunion and Symptomatic Malunion Rates
We identified nonunion and symptomatic malunion and an-
alyzed them independently. Three of the studies classified a
nonunion as no evidence of osseous union radiographically
at one year after injury12,13,16. Two studies used a similar defi-
nition at an earlier time point (five months in one study and
six months in the other), and one study reported no non-
unions14,15,17. Radiographic malunion is ubiquitous following
nonoperative treatment of a displaced clavicular fracture: this
finding was not included as a complication in these trials.
Symptomatic malunion was defined as a patient with symp-
toms severe enough to warrant corrective osteotomy. Non-
union or symptomatic malunion were significantly more
common in the nonoperative group (forty-six [23%] of 200
nonoperatively treated patients versus three [1%] of 212 op-
eratively treated patients, p < 0.0001) (Fig. 3). The absolute risk
reduction for nonunion and symptomatic malunion (decrease
in risk after primary fixation as compared with nonoperative
treatment) for operative versus nonoperative treatment was
22% (ranging from a 23% risk with nonoperative treatment to
a 1% risk with operative care). The number needed to treat for
one patient to benefit from surgical intervention (in terms of
the prevention of nonunion or symptomatic malunion) was 4.6
(n = 412 total patients). Of the 212 patients in the operative
group, three (1%) developed nonunion. Of the 200 patients in
the nonoperative group, twenty-nine (15%) developed non-
union (p = 0.001) (Fig. 4). The absolute risk reduction for
nonunion for operative versus nonoperative treatment was
14% (ranging from a 15% risk with nonoperative treatment to
a 1% risk with operative care). The number needed to treat for
one patient to benefit from surgical intervention in nonunion
rates was 7.6 (n = 412 total patients). There were no symptom-
atic malunions (0%) reported in the 212 patients treated
operatively and seventeen malunions (9%) in the 200 patients
in the nonoperative group (p < 0.001). The absolute risk re-
duction for symptomatic malunion for operative versus
nonoperative treatment was 9% (ranging from a 9% risk with
nonoperative treatment to a 0% risk with operative care). The

Fig. 4
Graph showing comparison of nonunion rate between operative (experimental) and nonoperative (control) groups (p = 0.001). The size of each square is
proportional to the weight of the study. Cots = Canadian Orthopaedic Trauma Society. Z = p value of weighted test for overall effect, CI = confidence interval,
df = degrees of freedom, I2 = test statistic.
 682
TH E JO U R NA L O F B O N E & JO I N T SU RG E RY J B J S . O RG
d
V O LU M E 94-A N U M B E R 8 A P R I L 18, 2 012
d d
number needed to treat for one patient to benefit from surgical
intervention for symptomatic malunion was 11.7.
Early Return to Function
There were multiple indicators in these six studies that primary
fixation resulted in an earlier return to function12,17. Pain scores
at five, nineteen, and thirty-three days after injury were
significantly better in the operative group in the study by
Witzel15. A return to moderate activity on the sixtieth day after
injury was possible in 80% of the operatively treated patients
but only in 55% of the nonoperatively treated patients in that
study15. Multiple studies demonstrated significantly superior
functional scores early in the period after injury in the opera-
tive group; for example, there was a 15 to 20-point improve-
ment in the DASH scores in the operative group in one study12.
Discussion
T he study populations in the clinical trials reviewed were
predominantly young, active and male. Our meta-analysis
revealed that primary operative fixation does provide more
rapid return of function and minimizes early residual disability
following fracture. In addition, the prevalence of symptomatic
malunion and nonunion is significantly lower (forty-six of 200
in the nonoperative group versus three of 212 in the operative
group, p < 0.0001). However, overall functional improvement
seen at the one-year follow-up was modest, with approximately
a 5-point improvement in mean DASH and CS scores in the
operative group. This is generally considered to be below the
minimal clinically relevant difference (ten to fifteen points) for
these scales. The reasons for this marginal overall difference,
despite the fact that these significant complications (nonunion
and symptomatic malunion) are lower, are probably multi-
factorial. The number of individuals in the nonoperative group
who developed nonunion or symptomatic malunion was forty-
six of 200, a rate of 23%. Although most of these individuals
were symptomatic with significantly lower outcome scores
(i.e., a mean of sixteen points worse on the DASH score in the
nonunion patients in the study by Virtanen et al.13), this rate
may result in only a modest average decrease in long-term
outcome scores for the group as a whole, since most patients
treated nonoperatively do well. It is clear that prognostic
indicators that would identify the individuals most at risk
for developing one of these negative outcomes would be very
useful in refining operative indications. The number needed to
treat for the prevention of nonunion and symptomatic mal-
union was 4.6; a lower number (minimizing the number of
patients exposed to surgery to benefit one person) would be
preferable.
Overall complication rates were higher in the nonoperative
group than in the operative group (42% versus 29%, p = 0.0075).
Many of these complications (e.g., nonunion), require operative
intervention that is greater in magnitude (operative time, re-
covery time, complication rate) than the original intervention
(primary fixation) being investigated. The commonest compli-
cation in the operative group was hardware irritation (from an
implanted plate or prominent pin) that resulted in hardware
O P E R AT I V E V E R S U S N O N O P E R AT I V E C A R E OF DI SP L ACE D
M I D S H A F T C L AV I C U L A R F R A C T U R E S
removal. Theoretically, this complication could be minimized by
less prominent implants or improved insertion techniques.
Our review shows that 15% of 200 patients in the nonop-
erative group developed a nonunion, which is significantly greater
(p = 0.001) than the 1% rate of nonunion in the 212 patients of
the operative group. The number needed to treat for one person to
benefit from operative treatment in relation to nonunion was 7.6.
Another complication significantly more common in the non-
operative group is symptomatic malunion (seventeen of 200 in
the nonoperative group versus 0 of 212 in the operative group, p <
0.001). However, with the data available, we were not able to draw
any specific conclusions as to which patients are most likely to
suffer from one of these significant complications.
Our meta-analysis shows that primary fixation of dis-
placed midshaft clavicular fractures has an early effect on de-
creasing pain and improving function. This is demonstrated by
the significantly improved DASH scores (15 to 20 points) at six
weeks and three months after injury, higher Constant scores at
six weeks and three months, and significantly better pain scores
at three weeks (for example, at five, nineteen, and thirty-three
days after injury in the study by Witzel)12,13,15. In the Witzel
study15, all patients were asked to return to moderate activity two
months after injury; 80% of the operatively treated patients were
able to achieve this, whereas only 55% of the patients random-
ized to the nonoperative group were physically active at this level.
Two of the studies in which these outcome scores were used
detected a significantly greater improvement at earlier intervals,
but not necessarily at the later (one to two-year) intervals12,17.
The DASH scores that were used to evaluate the return to daily
activities within the first three months were significantly better
for the operative group in the study by Smekal et al.17. The only
significant difference between the two groups in the study by
Judd et al. was at the three-week point, in favor of the operative
group, with use of the SANE and L’Insalata scores16.
Our study has some limitations common to all meta-
analyses, and other specific limitations. The conclusions drawn
from a meta-analysis are only as unbiased and accurate as the
data that are entered. We have attempted to provide conclu-
sions based only on high-quality data from Level-I randomized
clinical trials; quasi-randomized trials, single-arm prospective
studies, and retrospective reviews were excluded. Bias is in-
herent in many analyses that focus on a specific population or
geographic area. By including all studies available, including
those from multiple countries reported in multiple languages,
we believe that our conclusions are applicable to most popu-
lations. Homogeneity, or the consistency of the patients in the
analyzed studies, the treatments used, and the outcomes re-
ported, is critical to the reliability of conclusions drawn in a
meta-analysis. While our population was homogeneous in
terms of major demographic parameters, there was significant
heterogeneity in terms of the complication rates due primarily
to one study that had a high (48%) complication rate in the
operative group (probably due to the implant used)16. There
was no evidence of heterogeneity in the other specific compli-
cations we examined (nonunion and symptomatic malunion).
Lastly, despite our best efforts in using multiple search methods
 683
TH E JO U R NA L O F B O N E & JO I N T SU RG E RY J B J S . O RG
d
V O LU M E 94-A N U M B E R 8 A P R I L 18, 2 012
d d
and content experts, it is possible that we have not detected an
eligible existing trial, the results of which may be applicable to
our meta-analysis.
Specific limitations of our study include the fact that one
of the six randomized trials we analyzed was published in ab-
stract form only. While we were able to obtain additional in-
formation from this author, the lack of detailed information and
the peer review process limits the assessment of this study. A
sensitivity analysis, using only the five peer-reviewed studies, did
not show any difference in our conclusions. The variety of dif-
ferent outcome measures used limited our ability to combine
outcome scores and make more definitive conclusions; it may
also have resulted in a decrease in our ability to identify a true
difference where one actually existed. For example, five of six
studies reported the nonunion rate and we were able to show
a clear decrease in nonunion rate in the operative group; how-
ever, we were unable to do so as conclusively for the DASH or
Constant scores that only three of the studies employed. Two
operative techniques are currently used for primary fixation of
midshaft fractures of the clavicle (plate fixation and intramed-
ullary pin insertion), and both were used in the trials we studied;
it should be noted that a recent randomized clinical trial com-
paring locked intramedullary nailing versus plating for displaced
midshaft fractures of the clavicle showed a 100% union rate in
both groups with similar outcome scores a mean of twelve
months after injury26. However, the lack of a standard operative
protocol must be considered a weakness of our meta-analysis;
although both techniques have been associated with excellent
results in a variety of reviews, it may be that there are funda-
mental differences in outcome between the two techniques12-17.
In this meta-analysis of randomized clinical trials com-
paring the effect of primary operative fixation versus sling
treatment of displaced midshaft clavicular fractures in young
active individuals, certain benefits of fixation are noted, espe-
cially early in the period after injury. A more rapid return to
function, a decreased complication rate (especially with regard
to the more serious negative outcomes of nonunion and
References
1. Robinson CM. Fractures of the clavicle in the adult. Epidemiology and classifi-
cation. J Bone Joint Surg Br. 1998;80:476-84.
2. Postacchini F, Gumina S, De Santis P, Albo F. Epidemiology of clavicle fractures.
J Shoulder Elbow Surg. 2002;11:452-6.
3. Neer CS 2nd. Nonunion of the clavicle. J Am Med Assoc. 1960;172:1006-11.
4. Rowe CR. An atlas of anatomy and treatment of midclavicular fractures. Clin
Orthop Relat Res. 1968;58:29-42.
5. Hill JM, McGuire MH, Crosby LA. Closed treatment of displaced middle-third
fractures of the clavicle gives poor results. J Bone Joint Surg Br. 1997;79:
537-9.
6. McKee MD, Pedersen EM, Jones C, Stephen DJ, Kreder HJ, Schemitsch EH, Wild
LM, Potter J. Deficits following nonoperative treatment of displaced midshaft cla-
vicular fractures. J Bone Joint Surg Am. 2006;88:35-40.
7. Nowak J, Holgersson M, Larsson S. Sequelae from clavicular fractures
are common: a prospective study of 222 patients. Acta Orthop. 2005;76:
496-502.
8. Crenshaw AH. Fractures of the shoulder girdle, arm and forearm. In: Crenshaw
AH, editor. Campbell’s operative orthopaedics. 8th ed. St. Louis: Mosby Year Book;
1992. p 989-1053.
9. Poigenfürst J, Rappold G, Fischer W. Plating of fresh clavicular fractures: results
of 122 operations. Injury. 1992;23:237-41.
O P E R AT I V E V E R S U S N O N O P E R AT I V E C A R E OF DI SP L ACE D
M I D S H A F T C L AV I C U L A R F R A C T U R E S
symptomatic malunion), and marginally superior long-term
functional outcome scores were observed with operative in-
tervention. It is clear that there is a specific subset of individuals
with a completely displaced midshaft fracture of the clavicle
who will benefit from fixation. However, this information
should not be used to justify an indiscriminate approach to
surgical fixation of all clavicular fractures. Patients with a com-
pletely displaced midshaft clavicular fracture may be counseled
that they will be at a higher risk of sustaining nonunion and
symptomatic malunion if the fracture is treated nonoperatively,
but that there is no clear evidence that surgical treatment will
improve their long-term function in general. They should also be
counseled that, approximately 75% of the time, a completely
displaced clavicular fracture that is treated nonoperatively will heal
with few, if any, long-term consequences. Surgeons can use this
information appropriately to help patients make optimal deci-
sions in a shared decision-making process. The patient must
weigh this information against the risks of anesthesia, wound
infection, nonunion with an operation, and the potential need for
hardware removal.
Appendix
A table showing population demographics is available
with the online version of this article as a data supplement
at jbjs.org. n
Robbin C. McKee
Daniel B. Whelan, MD, FRCS(C)
Emil H. Schemitsch, MD, FRCS(C)
Michael D. McKee, MD, FRCS(C)
Division of Orthopaedics, Department of Surgery,
St. Michael’s Hospital and the University of Toronto,
800 – 55 Queen Street East, Toronto,
ON M5C 1R6, Canada.
E-mail address for M.D. McKee: McKeeM@smh.ca
10. McKee MD, Seiler JG, Jupiter JB. The application of the limited contact dynamic
compression plate in the upper extremity: an analysis of 114 consecutive cases.
Injury. 1995;26:661-6.
11. Mueller M, Burger C, Florczyk A, Striepens N, Rangger C. Elastic stable intra-
medullary nailing of midclavicular fractures in adults: 32 patients followed for 1-5
years. Acta Orthop. 2007;78:421-3.
12. Canadian Orthopaedic Trauma Society. Nonoperative treatment compared with
plate fixation of displaced midshaft clavicular fractures. A multicenter, randomized
clinical trial. J Bone Joint Surg Am. 2007;89:1-10.
13. Virtanen KJ, Paavola MP, Remes VM, Pajarinen J, Savolainen V, Bjorkenheim
JM. Nonoperative versus operative treatment of midshaft clavicle fractures: a ran-
domized controlled trial. Read at the 75th Annual Meeting of the AAOS; 2010 Mar
9-12; New Orleans, LA. Paper no 331.
14. Smith CA, Rudd J, Crosby LA. Results of operative versus nonoperative treat-
ment for 100% displaced midshaft clavicle fractures: a prospective randomized
clinical trial. Read at the 16th Annual Open Meeting of the American Shoulder and
Elbow Surgeons; 2000 Mar 18; Orlando, FL. Paper no 31.
15. Witzel K. [Intramedullary osteosynthesis in fractures of the mid-third of the
clavicle in sports traumatology]. Z Orthop Unfall. 2007;145:639-42. German.
16. Judd DB, Pallis MP, Smith E, Bottoni CR. Acute operative stabilization versus
nonoperative management of clavicle fractures. Am J Orthop (Belle Mead NJ).
2009;38:341-5.
 684
TH E JO U R NA L O F B O N E & JO I N T SU RG E RY J B J S . O RG
d
V O LU M E 94-A N U M B E R 8 A P R I L 18, 2 012
d d
17. Smekal V, Irenberger A, Struve P, Wambacher M, Krappinger D, Kralinger FS. Elastic
stable intramedullary nailing versus nonoperative treatment of displaced midshaft clavic-
ular fractures-a randomized, controlled, clinical trial. J Orthop Trauma. 2009;23:106-12.
18. Constant CR, Murley AH. A clinical method of functional assessment of the
shoulder. Clin Orthop Relat Res. 1987;214:160-4.
19. Hudak PL, Amadio PC, Bombardier C. Development of an upper extremity out-
come measure: the DASH (disabilities of the arm, shoulder and hand) [corrected].
The Upper Extremity Collaborative Group (UECG). Am J Ind Med. 1996;29:602-8.
20. Williams GN, Gangel TJ, Arciero RA, Uhorchak JM, Taylor DC. Comparison of the
Single Assessment Numeric Evaluation method and two shoulder rating scales.
Outcomes measures after shoulder surgery. Am J Sports Med. 1999;27:214-21.
21. L’Insalata JC, Warren RF, Cohen SB, Altchek DW, Peterson MG. A self-administered
questionnaire for assessment of symptoms and function of the shoulder. J Bone Joint
Surg Am. 1997;79:738-48.
O P E R AT I V E V E R S U S N O N O P E R AT I V E C A R E OF DI SP L ACE D
M I D S H A F T C L AV I C U L A R F R A C T U R E S
22. Detsky AS, Naylor CD, O’Rourke K, McGeer AJ, L’Abbé KA. Incorporating vari-
ations in the quality of individual randomized trials into meta-analysis. J Clin Epi-
demiol. 1992;45:255-65.
23. Bhandari M, Richards RR, Sprague S, Schemitsch EH. The quality of reporting of
randomized trials in the Journal of Bone and Joint Surgery from 1988 through 2000.
J Bone Joint Surg Am. 2002;84:388-96.
24. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials.
1986;7:177-88.
25. Review Manager (RevMan) [Computer Program]. Version 5.0. Copenhagen: The
Nordic Cochrane Centre, The Cochrane Collaboration, 2008.
26. Ferran NA, Hodgson P, Vannet N, Williams R, Evans RO. Locked intra-
medullary fixation vs plating for displaced and shortened mid-shaft clavicle
fractures: a randomized clinical trial. J Shoulder Elbow Surg. 2010;19:
783-9.