Demographic Data
Over a seven-year period, we studied all patients who presented to The
Shoulder Injury Service of The New Royal Infirmary of Edinburgh, Edinburgh,
United Kingdom, with an acute posterior dislocation of the shoulder and a Neer
two, three, or four-part proximal humeral
fracture4, in which
there was a displaced fracture of the humeral head. All of these patients
received their initial assessment in our affiliated emergency department,
which provides the only acute musculoskeletal trauma service for adults in the
local population, which is representative of most European cities, in that it
is stable and predominantly urban.
We excluded five patients who presented to our service between seven and
twenty-six weeks after the injury with a chronic posterior
fracture-dislocation. For the purposes of this study, we considered only
patients in whom the humeral head was fully dislocated from contact with the
glenoid fossa, and we excluded twenty-three patients in whom the fractured
humeral head was rotated or subluxated posteriorly but was not fully
dislocated. In addition, we included only patients who had a complex
fracture-dislocation1,
in which there was either an isolated fracture through the neck of the humerus
(a Neer two-part
fracture4) or a neck
fracture associated with fractures through one or both tuberosities (a Neer
three-part or four-part
fracture4). We
excluded an additional sixty patients during the period of the study who
sustained a so-called simple posterior
dislocation1, in
which there was either no associated proximal humeral fracture or just an
osteochondral fracture of the humeral head (a reverse Hill-Sachs lesion) or
there was an isolated fracture of the lesser tuberosity. An additional two
elderly patients (eighty-five and eighty-nine years old) with a
fracture-dislocation were excluded from further analysis, as they were too
medically frail to undergo any form of surgery for the injury and detailed
imaging of the injury was not performed.
After these exclusions, there were twenty-six patients (twenty-eight
shoulders) who had sustained acute complex posterior fracture-dislocations of
the proximal aspect of the humerus (two patients had bilateral simultaneous
injury). We used age and gender-specific local census data to assess the
overall annual incidence of posterior fracture-dislocation in our local
population, since our institution provides the only source of acute
orthopaedic trauma care to the stable population of the local community.
Initial Treatment Protocol and Assessment of the Anatomy of the
Injury
All patients in the series were referred acutely from our affiliated
emergency department to our orthopaedic service, and no attempt at closed
manipulation of the shoulder was made. All patients were seen within the first
two days after the injury, and the average time to presentation in the
emergency department was four hours (range, two to forty-six hours) after the
injury. Preoperatively, we made trauma radiographs (anteroposterior and
"modified axial"
radiographs30) and
computerized tomograms. Intraoperatively, one of us (C.M.R.) documented the
anatomy of the injury, recording the location and extent of the key fracture
lines through the humeral head and tuberosities, and the extent of comminution
of these fragments. On the basis of the radiographic and operative findings,
one of us (C.M.R.) classified the injuries using the Neer
classification4. We
also produced additional descriptive information, which was based on the
fracture and soft-tissue injury.
Operative Protocol
All patients were treated operatively by a single surgeon (C.M.R.). The
operation was performed with the patient under general anesthesia, and
prophylaxis with a single-dose broad-spectrum antibiotic was administered. A
shoulder-strap skin incision and modified deltoid-splitting surgical approach
was used in all patients, with identification and protection of the anterior
branch of the axillary nerve as it traversed the distal extent of the
deltoid-splitting
incision31. The
empty glenoid was accessed through the rotator interval, with use of the long
head of the biceps tendon as a landmark to direct the arthrotomy toward the
superior pole of the osseous glenoid. We identified four key steps in the
operation.
Step 1: Relocation
The humeral head was disimpacted from the posterior part of the glenoid rim
under direct vision. A bone lever was inserted between the posterior part of
the glenoid rim and the engaged humeral head to disimpact it. Once the humeral
head had been freed from the posterior part of the glenoid, relocation was
easily achieved under direct vision, by rotating it forward to face the
glenoid. The vascularity of the reduced humeral head was assessed by
inspection of its residual capsular attachments and identification of any
active arterial back-bleeding from its fractured cancellous
surface32.
Step 2: Fracture Reduction and Provisional Stabilization
We anatomically reduced and provisionally stabilized the humeral head using
two or three Kirschner wires, to restore rotational stability and allow an
assessment of residual posterior instability. Judging the accuracy of the
reduction of the primary fracture was often difficult in injuries in which
marginal impaction had produced an acute osteochondral fracture of the humeral
head (a reverse Hill-Sachs lesion). However, the presence of the intact
posterior capsule and periosteal hinge was a useful aid to reduction, serving
as a template for "closing the door" to achieve the reduction.
Step 3: Assessment of Stability and Adjuvant Stabilization
Procedures
Residual osteochondral defects of the anterior aspect of the humeral head
(reverse Hill-Sachs lesions), which persisted after provisional reduction of
the head and engaged the posterior part of the glenoid rim with the shoulder
in neutral rotation, were considered to be a potential source of posterior
instability. Large residual osteochondral fragments with retained soft-tissue
attachment were elevated, the metaphyseal defects were packed with morselized
femoral head allograft, and the fractures were fixed with so-called positional
screws1,33-36.
When the fragments were larger, comminuted, or loose, a deep-frozen femoral
head osteochondral allograft was sculpted to fill the
defect1,33,35,37.
It was our experience that these procedures restored stability
intraoperatively through a full range of rotation of the shoulder, and repair
of the posteroinferior capsulolabral avulsion was not required.
Step 4: Definitive Fracture Stabilization
Once we had achieved temporary fixation of the fracture and restored
shoulder stability, definitive internal fixation of the fracture was
performed. The complexity of the internal fixation was determined by the
number of key fracture fragments. We stabilized two-part fractures of the
anatomic neck with two, three, or four partially threaded cancellous lag
screws, dependent on the length of the metaphyseal spike attached to the
humeral head. The screws were inserted perpendicular to the fracture line in
the humeral head, through the intact lateral proximal humeral metaphysis. The
humeral head component of the more complex three-part and four-part fractures
was reduced to the tuberosities with use of either lag screws or interosseous
sutures. This composite was then stabilized to the humeral shaft with use of a
cloverleaf plate and screws. Locking plate fixation was not used in any
patients because of the good bone quality and the large size of the fracture
fragments.
The rotator interval arthrotomy was loosely approximated without tension,
after which a layered closure was performed, with particular care being taken
to repair the area where the deltoid was partially detached from the acromion.
All patients wore a shoulder sling with the arm held in neutral abduction,
neutral flexion, and neutral rotation and with the elbow flexed 90°. A
foam pad or pillow was used to restrict the patient from internally rotating
the shoulder beyond neutral for six weeks. Pendulum exercises and elbow
range-of-motion exercises were commenced immediately after surgery, and
active-assisted range-of-motion exercises were begun at two weeks, but
abduction of the shoulder beyond 90° was avoided. Isometric exercises of
the rotator cuff and graduated, active range-of-motion exercises under
physiotherapy supervision were begun after sling removal and were continued
for at least six months after the operation.
Assessment of the Outcome After Reconstruction
The two chief outcome measures were clinically or radiographically apparent
fracture-related complications and the functional outcome, which was assessed
by three scoring systems and by restoration of the normal activities of daily
living. We evaluated all patients of our shoulder injury clinic at one, six,
twelve, and twenty-four weeks after the injury and at one and two years after
the injury.
At each of the follow-up visits, we assessed the functional outcome using
the Short Form-36 (SF-36) general health
measure38,39,
the upper limb-specific Disabilities of the Arm, Shoulder and Hand (DASH)
score40, and the
shoulder-specific Constant
score41. Prior to
the consultation, the patient completed the questionnaire component of these
assessments using a printed booklet. The SF-36 scores were compared with those
for an age and sex-matched group of healthy
volunteers39.
Assessment of range of motion and muscle power was made by an independent
research physiotherapist using a standardized protocol. All of the data were
transferred to an SPSS database (SPSS, Chicago, Illinois), which was used in
all subsequent data analysis. We also recorded the time of return to work or
normal daily activities after the injury.
Patients underwent radiographic follow-up evaluations (anteroposterior and
so-called modified axial
radiographs30) at
each review. One of us (C.M.R.) measured the head-shaft inclination
angle42 on the
radiographs that were made after the injury, immediately after surgery, and at
three months, six months, one year, and two years after the injury, to assess
fracture displacement after reconstruction. We also measured the degree of
residual tuberosity displacement, in millimeters, adjusting for magnification
artifact.
All radiographs were evaluated for humeral head osteonecrosis,
osteoarthrosis, and fracture-healing. Assessment of fracture-healing was a
problem, as the near anatomic reduction made it difficult to assess internal
callus formation, and none of the fractures produced external callus. We
therefore arbitrarily deemed fractures to be united when the patient reported
no shoulder pain or only mild activity-related discomfort, and there was no
radiographic evidence of loss of the initial fracture reduction or evidence of
implant loosening or breakage on radiographs. We graded osteonecrosis
according to the symptoms (severity of pain, function, and range of movement)
and the radiographic extent of humeral head involvement and collapse.
All information gathered in the study was recorded and analyzed with use of
the SPSS software package (version 12; SPSS, Chicago, Illinois). We examined
differences in epidemiological patterns of the different injury mechanisms
using the Mann-Whitney U test for continuous data and the Fisher exact test
for categorical data. The sequential change in functional scores over time was
analyzed with use of the Wilcoxon matched-pairs test, and we sought to examine
the factors that influenced these scores with use of linear regression
analysis. The general health scores (SF-36) were individually compared with
age and sex-matched control values for the normal
population39 with
use of the Mann-Whitney U test.
Epidemiology
The incidence of posterior fracture-dislocations was 0.6 per 100,000
population per year during our study period. The mean age of the twenty-six
patients was fifty-three years (range, twenty-six to eighty-five years), and
there were nineteen men and seven women. The majority of the patients were
middle-aged men, with sixteen (62%) of the twenty-six patients between forty
and sixty-five years of age. The age and sex-specific incidence of these
injuries in the general population reflected this trend, with the peak
incidence occurring in the middle-aged patients and with a much lower
incidence in younger patients and the elderly.
Eleven patients (including the two patients with bilateral injury)
sustained the injury during a seizure. The seizures occurred in three patients
with a known history of epilepsy, and they were induced by alcohol or drug
withdrawal in four patients, by hypoglycemia in three patients with diabetes
mellitus, and by hypoxia during an asystolic cardiac arrest in one patient.
The mean age of these patients was fifty-one years (range, twenty-six to
seventy-six years), and there were ten men and one woman. These patients,
therefore, had a slightly, but not significantly, lower mean age than the
remainder of the study patients.
The twelve injuries produced during falls occurred in a slightly older
population of individuals (mean age, fifty-seven years; range, thirty-two to
eighty-five years) with an equal gender distribution. Ten of those injuries
were sustained in falls from a height of >2 m (seven patients fell down
stairs or off a ladder, and three fell from a building or wall), while the
other two were simple domestic falls from a standing height or below. The
three other fracture-dislocations all occurred in men (mean age, forty-five
years; range, thirty-two to sixty-three years) during road traffic accidents,
and all were isolated injuries.
There were no open fractures or clinically detectable neurovascular
injuries related to the fracture-dislocations in this series. All injuries
were confined to the shoulder, and no patient had other injuries involving the
musculoskeletal or other systems.
Pathological Anatomy of the Injury
All patients had a displaced primary fracture of the anatomic neck of the
humerus, propagating from the area of an osteochondral fracture of the
anterior aspect of the humeral head (a reverse Hill-Sachs lesion), which was
engaged on the posterior part of the glenoid rim, with the humeral head facing
posteriorly. The posterior dislocation of the humeral head avulsed the
posteroinferior capsulolabral soft-tissue sleeve and produced a reverse
Bankart lesion (a capsulolabral avulsion of the posteroinferior glenoid rim)
in all patients. However, the capsule and periosteal sleeve were in continuity
posteriorly in all patients, and the anatomic neck fracture hinged on these
structures. We distinguished three fracture subtypes, determined by the extent
of secondary fracture lines in the tuberosities.
Type 1: Neer Two-Part Anatomic Neck Fracture (Figs.
1-A,
1-B, and
1-C)
In six fracture-dislocations (in five men and one woman, with an average
age of forty-eight years; range, thirty-one to sixty-three years), there was
no associated tuberosity fracture and the lateral wall of the proximal part of
the humerus remained intact (a Neer two-part anatomic neck fracture). In four
of the shoulders, a long medial metaphyseal spike of bone was attached to the
anatomic neck fracture.
Type 2: Neer Three-Part Fracture of the Lesser Tuberosity (Figs.
2-A,
2-B, and
2-C)
In five fracture-dislocations (all in men with an average age of fifty-one
years; range, thirty-one to seventy-three years), there was, in addition to
the anatomic neck fracture, an isolated fracture through the lesser
tuberosity. This propagated from a fracture line in the region of the reverse
Hill-Sachs lesion (a Neer three-part fracture of the lesser tuberosity).
Type 3: Neer Three-Part Fracture with Composite Tuberosity
"Shield" Fragment (Figs.
3-A,
3-B,
3-C and 3-D)
In seventeen fracture-dislocations that occurred in fifteen patients (nine
men and six women, with an average age of forty-eight years; range, thirty-one
to sixty-three years), there was, in addition to the anatomic neck fracture, a
fracture involving both tuberosities. In all seventeen shoulders, the
tuberosities formed a composite "shield" fragment, as described by
Edelson et al.43,
through an intact periosteal sleeve (a Neer three-part fracture, since the two
tuberosities were minimally displaced from each other). However, in thirteen
of the seventeen shoulders, there were one or more vertical intertubercular
fracture lines, resulting in a "shattered shield" configuration,
which remained a functional composite through the intact periosteal sleeve. In
sixteen of the seventeen shoulders, the tuberosity fracture line exited
distally in the lateral aspect of the metaphysis in the region of the surgical
neck of the humerus. In one of the seventeen shoulders, there was propagation
of the tuberosity fracture into the proximal part of the diaphysis. That
patient also had a metaphyseal spike of bone attached to the anatomic neck,
which resulted in a complete secondary fracture in the proximal part of the
humeral diaphysis.
Surgical Treatment
Operative reduction and internal fixation was performed at an average of
fourteen hours (range, six to thirty-five hours) after referral from the
emergency department. The average duration of surgery was 116 minutes (range,
sixty-five to 190 minutes), with average operative blood loss of 300 mL
(range, 150 to 1000 mL).
Following the relocation and provisional reduction of the humeral head,
eleven shoulders had residual posterior glenohumeral instability with the
shoulder in neutral rotation. According to our protocol, we treated six
shoulders (one Type-1, three Type-2, and two Type-3 injuries) by elevating the
osteochondral fragments of the anterior aspect of the humeral head (Figs.
4-A and 4-B) and packing the
defect with morselized allograft. The impaction fragments were comminuted or
loose in five shoulders (one Type-1, two Type-2, and two Type-3 injuries), and
a sculpted femoral head osteochondral allograft was used to fill the defect.
All patients had residual capsular attachments and active arterial
back-bleeding on direct observation of the relocated humeral head, and, in all
of them, we followed the protocol of internal fixation, determined by the
configuration of the secondary fracture lines, as described above (Figs.
4-A, 4-B,
5-A, and 5-B).
Results of Treatment
Complications
Three shoulders developed early postoperative superficial wound erythema,
which was attributed to wound infection (although it was not bacteriologically
confirmed), and it resolved after a five to seven-day course of broad-spectrum
antibiotic therapy. There were no wound hematomas or deep wound
infections.
A fifty-two-year-old man with a Type-2 posterior fracture-dislocation had
an acute redislocation of the humeral head develop three days postoperatively.
At the initial reconstruction, a large defect of the anterior part of the
humeral head had been treated by elevation and bone-grafting. At revision
surgery, the redislocation was found to be due to the collapse of the elevated
segment of the humeral head, with reengagement of the posterior part of the
glenoid on the resultant defect. The humeral head defect was filled with a
segmental femoral head allograft, which restored intraoperative stability of
the shoulder. This patient subsequently had a stiff, pain-free shoulder, with
radiographic evidence of osteonecrosis in 50% of the humeral head at one year
after the injury.
Postoperatively, the mean head-shaft inclination angle was 133° (range,
128° to 144°). In twenty-five shoulders, both tuberosities were
anatomically reduced, whereas three shoulders with fractures of the greater
tuberosity had 5, 9, and 15 mm of displacement. No subsequent loss of
reduction or fixation failure was seen in the shoulders that had radiographic
follow-up at two years. One shoulder (described above) had partial
osteonecrosis of the humeral head develop, and two others had changes
consistent with osteoarthrosis of the humeral head (reduction of the
glenohumeral joint space without signs of segmental collapse or osteonecrosis
of the humeral head). These changes had not progressed further, despite
follow-up extending between three and four years in these patients, and none
had had further surgery at the time of the latest follow-up. According to our
definition of fracture union, all fractures were united by the time of the
one-year assessment. In the ten of the eleven shoulders in which bone grafts
had been used, the area of the graft was incorporated without radiographic
signs of collapse at the two-year follow-up evaluation.
Three patients underwent hardware removal for symptoms of mild ache and
weather sensitivity, without radiographic features of osteoarthrosis or
osteonecrosis between twelve and fifteen months after surgery. The symptoms in
all three patients resolved after hardware removal. An additional two patients
had symptoms of subacromial impingement develop after they had regained full
shoulder movement six months after the injury. The symptoms resolved in both
after a subacromial injection of corticosteroid and local anesthetic, although
one patient had a recurrence of symptoms at ten weeks after the injection. His
symptoms were attributed to mechanical impingement from the internal fixation,
and they resolved following hardware removal and acromioplasty at one year
after the injury. No patient, other than the one described above, had
recurrent subluxation or dislocation of the shoulder following
reconstruction.
Functional Outcome
Despite the fact that reminders to attend our clinic were made by telephone
or sent by mail, three patients defaulted from additional follow-up after the
three-month assessment, while one patient defaulted after the six-month
assessment. None of these patients received further medical treatment for the
fractures in our local health authority, and none had complications from
surgery at the time of the last follow-up assessment. Some of the remaining
twenty-two patients (twenty-four fractures) missed one or more follow-up
appointments, but we were able to obtain a final two-year follow-up assessment
for all of these patients.
The median DASH scores for the group showed continued significant
improvement within the first two years after surgery (p < 0.001). Although
the median Constant scores improved significantly within the first year after
surgery (p < 0.001), there was no significant difference between the median
scores at one and two years. At two years after surgery, the median Constant
score was 83.5 points (interquartile range, 75.5 to 88.5 points) and the
median DASH score was 17.5 points (interquartile range, 12 to 19 points). The
majority of the functional loss on these scores was attributable to loss of
internal rotation of the shoulder. With the numbers available, there was no
significant difference in outcome between injured shoulders on the dominant
side and those on the nondominant side.
At two years, nineteen patients reported no pain in their shoulder(s), two
had mild pain, which was not activity related, and one reported moderate pain
on use of the arm. The average combined forward flexion of the twenty-four
shoulders was 172° (range, 60° to 180°), and the average combined
abduction was 169° (range, 45° to 180°). Twenty-two of the
twenty-four shoulders had regained full external rotation (with the shoulder
in neutral flexion and abduction) at two years after surgery; the other two
shoulders had lost 10° and 24° of external rotation. Sixteen of the
twenty-four shoulders had regained full internal rotation (with the shoulder
in neutral flexion and abduction) at two years after surgery; the remaining
six patients had lost between 5° and 45° of internal rotation. No
patient had evidence of rotator cuff weakness or impingement or signs of
posterior glenohumeral joint instability on clinical testing at the two-year
follow-up assessment.
The patient who had required a reoperation for resubluxation of the
shoulder was an important outlier from the remainder of the cohort in terms of
his poor functional scores and poor range of shoulder movement. The functional
scores and range of shoulder movement for the remainder of the cohort were
homogeneous. Although the median Constant and DASH scores were slightly better
for the Type-1 fracture-dislocation subtype, compared with Types 2 and 3, this
difference did not reach significance, with the numbers available (p = 0.09).
In addition, we were unable to identify any risk factors from the demographic
information or morphological data on the fractures, which were independently
predictive of the functional outcome or range of motion at two years after
surgery.
The pain, physical functioning, and role-physical components of the SF-36
scores showed significant improvements within the first six months after
surgery (p < 0.05), and the latter two parameters improved significantly
within the first year (p < 0.05). These parameters were significantly lower
than those of the age and sex-matched controls at three and six months (p <
0.05), but they were not significantly different thereafter. The scores for
the six other subcategories did not change significantly within the first year
after surgery and were not significantly different from those of age and
sexmatched controls. The eight subcategory SF-36 scores were not significantly
different from those of age and sex-matched controls at two years.
All eight patients who had been employed in a sedentary job prior to the
injury returned to their full work duties and normal daily activities by one
year. Of the ten patients who had been regularly employed in a manual job
prior to the injury, eight (five within the first year) had returned to their
previous work duties by two years after the injury, while the remaining two
patients had changed to lighter, nonmanual work. The four patients who had not
been in regular employment prior to the injury had returned to normal domestic
and recreational pursuits within the first year after the injury. The
employment status of the remaining four patients was unknown, as they had
defaulted from follow-up.
Our study demonstrates that complex posterior fracture-dislocations are a
rare but important group of proximal humeral fractures, which occur
predominantly in middle-aged men during seizures or high-energy injuries.
These injuries should therefore be considered separately from the majority of
low-energy osteoporotic proximal humeral fractures, which occur predominantly
in elderly women.
Most previous isolated case reports or small case series of these injuries
have emphasized their
rarity14,44-46,
and they have often been considered together with other three and four-part
fractures of the proximal part of the humerus or with so-called simple
posterior dislocations, with no fracture of the anatomic neck or tuberosities.
However, it is preferable to consider posterior fracture-dislocations as a
separate entity because, in medically fit patients, they invariably benefit
from operative
treatment1,45.
Using detailed evaluation of the radiographic and the operative findings,
we were able to identify a consistent pattern of injury. The humeral head was
fractured in the region of the anatomic neck in all patients, with the primary
fracture line propagating from the region of an anterior osteochondral
fracture of the humeral head, which was locked onto the posterior part of the
glenoid rim. Similar radiographic findings have been reported
previously43,47,
and it is likely that the anatomic neck fracture occurs after dislocation
because of the propagation of the osteochondral fracture as it impacts on the
"anvil" of the posterior part of the glenoid
rim43. This type of
injury is analogous to the Type-I anterior fracture-dislocation, which
propagates from the base of a posterior osteochondral fracture, as the head
dislocates anteriorly, and impacts on the anterior aspect of the
glenoid28,43.
Three injury subtypes were distinguished by the location of their secondary
fracture lines, ranging from the Neer two-part fracture in which the
tuberosities remained intact (a Type-1 injury) through the Neer three-part
fracture of the lesser tuberosity (a Type-2 injury) to the most complex Neer
three-part fracture with a composite "shield" fragment of the
greater and lesser tuberosity (a Type-3 injury). There were minor differences
in outcome among these subtypes, which did not reach significance. However,
recognition of the subtypes was important because it served as a guide to the
technique of internal fixation that we used.
Although closed, percutaneous, or arthroscopically assisted techniques to
reduce the posterior dislocation have been
described44,48-50,
there is a substantial risk of additional soft-tissue injury or displacement
of the humeral head. We believed that reduction could be accomplished safely
only under direct vision, and therefore we performed open relocation through a
modified deltoid-splitting surgical
approach14, with
identification and protection of the anterior branch of the axillary
nerve31. The
improved access provided by this
approach14,
compared with the deltopectoral approach, was a key factor in facilitating the
relocation of the dislocated humeral head and its subsequent reduction and
internal fixation.
We assessed the stability of the relocated humeral head in all patients
intraoperatively, as it was apparent that the defect of the anterior aspect of
the head was often of sufficient size to reengage on the posterior part of the
glenoid rim and produce an acute redislocation. If the reduction was unstable,
we used either elevation of the osteochondral fragment with bone-grafting (as
for a depressed tibial plateau fracture) or a sculpted allograft to fill the
defect. With use of these methods, there was only one acute redislocation and
the bone grafts underwent satisfactory radiographic incorporation.
Many techniques to stabilize posterior fracture-dislocations have been
described, and they range from simple
manipulation44 and
minimally invasive
techniques48,51
through open reduction and internal
fixation14,27
to
arthroplasty19,20,52.
Evaluation of their relative merits is difficult, owing to the rarity of these
injuries, and level-I evidence is lacking. Although arthroplasty is the most
commonly advocated primary
treatment19,20,
recent studies have suggested that the results are often
suboptimal22-26.
As posterior fracture-dislocations tended to occur in middle-aged individuals,
with higher functional expectations, we adopted a policy of using operative
internal fixation to preserve the humeral head. It was our hypothesis that
this provided an opportunity for the patient to regain a more functional
shoulder and to avoid the longer-term risks associated with the use of an
arthroplasty. However, our patients were exposed to the risks inherent in a
humeral head-preserving reconstruction, most notably osteonecrosis and
segmental collapse of the humeral head, requiring secondary reconstructive
surgery.
Despite the limited conclusions that can be drawn from a case series study,
our operative technique yielded favorable outcomes, with a low prevalence of
complications. The results are superior to those recently reported for
hemiarthroplasty22-26,
although several patients required additional treatment for shoulder pain and
impingement symptoms during the first year after reconstruction. These
complications were often improved or resolved by secondary interventions, such
as subacromial steroid injection, hardware removal, or acromioplasty.
The low rates of nonunion and osteonecrosis after reconstruction in our
study are similar to those described after open reduction and internal
fixation of Type-I anterior
fracture-dislocations28.
Several factors may have provided a favorable environment for continued
perfusion of the humeral head after its reduction and internal fixation. In
all shoulders, the posterior capsule and periosteal sleeve "hinge"
were intact, providing a potential source of vascularization. The
intertubercular groove was uninjured in Type-1 and Type-2 injuries, and the
periosteal sleeve was intact in this area in the Type-3 variant. The
anterolateral ascending branch of the anterior circumflex artery, which runs
in this anatomic location and is the main source of humeral head
perfusion53, may
therefore have been intact. Finally, most patients were middle-aged and
medically fit, which may also have improved the prospects for preservation of
blood flow to the humeral head.
This study had several limitations. There was no control group of patients
treated by other techniques, such as hemiarthroplasty, with whom to compare
our results. Given the rarity of this injury, it would require a large
multicenter, randomized, controlled trial to adequately power a study to
compare the relative merits of our approach with other techniques. Secondly,
all of the injuries in our study were treated by a single surgeon, and
concerns might be expressed over the generalizability of our findings. Despite
the two-year minimum follow-up period, our study is a short-term
investigation. Reappraisal of the outcome in these patients in the longer term
would be appropriate, to ascertain whether the initially encouraging results
are sustained over time. ?