According to the admission and surgery logs of the Mayo Clinic,
Rochester, Minnesota, 2472 patients were treated for hip fracture
between January 1988 and December 1998. Of these patients, 1035
were treated for intertrochanteric or subtrochanteric fracture.
Clinical records and radiographs that had been made at the time
of the injury or immediately postoperatively for these patients
were reviewed, and fifty-five patients with a reverse obliquity
fracture pattern were identified. Fifty-three patients
were treated with open reduction and internal fixation, and two
had primary prosthetic replacement. The choice of implant was left
to the discretion of the treating surgeon. Three patients died within
the first thirty-six days, and three were lost to follow-up. None
of those six patients were known to have complications related to
the implant. The remaining forty-nine patients were followed
until union occurred or a revision operation was performed. The
mean duration of clinical follow-up was eighteen months (range,
three to sixty-seven months).
Approval for this retrospective study was obtained from our institutional
review board. All patients were interviewed and examined by us personally. An
assessment of function was based upon pain, living situation, need
for walking aids, need for analgesics, and walking capacity. Fractures
were judged to be clinically healed when pain-free walking and range
of motion of the hip were possible.
Forty-six patients (94%) had a complete set
of radiographs, including initial diagnostic radiographs, immediate
postoperative radiographs, and final radiographs. Three patients
did not have a complete set of radiographs, but clinical notes clearly
documented radiographic union; thus, these patients were included
in the review. The mean duration of radiographic follow-up was fifteen months
(range, three to sixty months). Radiographs were reviewed for fracture
translation in the anterior-posterior and medial-lateral planes.
Fractures were judged to be healed radiographically if bridging
callus was evident on three of four cortices as seen on two views.
All patients were followed until radiographic signs of union were
present.
Fractures were classified according to the Orthopaedic Trauma
Association scheme (Fig. 2)7. There
were no nondisplaced fractures. Twenty-nine (59%)
were type 31.A3.1 and twenty (41%) were type 31.A3.3. All
had a fracture line extending proximally into the greater trochanter. Four-part
reverse obliquity fractures were distinguished from standard intertrochanteric
fractures by examining the obliquity of the fracture line separating
the femoral shaft from the proximal part of the femur.
Sixteen patients were treated with a 135° sliding hip screw;
fifteen, with a 95° blade-plate; ten, with a 95° dynamic condylar
screw; three, with a cephalomedullary nail (two, with a gamma nail
and one, with a reconstruction nail); and three, with an intramedullary
hip screw. Two patients treated with primary endoprosthetic replacement
were not included in the analysis of the results. At the discretion
of the treating surgeon, eight patients had bone-grafting at the
index procedure; one autograft and seven allografts were used.
Results were analyzed according to implant type, implant position,
fracture type, quality of reduction, and use of bone graft at the
index operation. Reduction was judged to be anatomic if <4
mm of displacement was present between the major fracture fragments.
Small fragments involving only the lesser trochanter were not considered
in the grading of the quality of reduction. Implant position was classified
according to published recommendations and manufacturer recommendations
for implant position8. For example,
sliding hip screws with a lag screw placed above the center of the
femoral head as seen on the anteroposterior radiograph or anterior
to the center of the femoral head as seen on the lateral radiograph
were considered to be in poor position. Dynamic condylar screws
and proximal screws of cephalomedullary nails were judged by similar
criteria. Only blade-plates placed deep into the femoral head within
1 cm of the subchondral bone were considered to be in good position9.
All patients received antibiotic prophylaxis preoperatively and
postoperatively as well as anticoagulation with low-dose warfarin
(target prothrombin international normalized ratio, 1.5 to 2.2)
or low-molecular-weight heparin postoperatively. Early weight-bearing
was not permitted when a fixed-angle device had been used; however,
when an intramedullary or sliding-screw device had been implanted,
weight-bearing was advanced as tolerated at the discretion of the
treating surgeon. Patient survival was compared with the expected
survival of an age and gender-matched population (West North Central
Whites10) with use of a one-sample
log-rank test11. The
overall two-year cumulative probability of patient survival was
estimated with the Kaplan-Meier method12.
Discrete factors were compared with both patient survival and the
rate of reoperation with use of a chi-square test or, if
appropriate, a Fisher exact test. A p value of £0.05 was
considered to be significant in all tests.
Fifty-five reverse obliquity intertrochanteric hip fractures
were identified; they accounted for 5.3% of the 1035 intertrochanteric
and subtrochanteric hip fractures and 2.2% of all 2472
hip fractures treated at our Level-One Trauma Center during the study
period. The forty-nine patients with follow-up
data had a mean age of seventy-eight years (range, seventeen
to ninety-eight years) at the time of the fracture. There
were thirteen male patients and thirty- six female patients.
There were eighteen right and thirty-one left fractures.
Forty-five fractures (92%) were caused by a fall,
and four fractures were caused by a motor-vehicle accident. Five patients
had an associated injury, and 82% of the patients had at
least one major medical comorbidity.
Thirty-two (68%) of the forty-seven
reverse obliquity fractures treated with internal fixation healed and
did not require another operation. The mean time to clinical union
for the fractures that healed primarily was 4.1 months (range, two
to nine months), and the mean time to radiographic union was 5.1
months (range, two to twenty months).
The overall rate of failure of fracture-healing or fixation was
fifteen (32%) of forty-seven. The union rate was
seven of sixteen for the hips treated with a sliding hip screw,
thirteen of fifteen for those treated with a 95° blade-plate, seven
of ten for those treated with a 95° dynamic condylar screw, two
of three for those treated with a cephalomedullary nail, and three
of three for those treated with an intramedullary hip screw. The
failure mode of the sliding hip screw was medial translation of
the distal fragment and loss of proximal fixation resulting in either
nonunion or screw cutout (Fig. 3). The failure mode of the blade-plate was
nonunion with plate fracture in both hips. The failure mode of the
dynamic condylar screw was cutout of the screw from the femoral
head in two patients and nonunion without cutout in one.
The fixed 95° angle devices (the blade-plate and the dynamic
condylar screw) led to union in twenty of twenty-five hips
and outperformed the sliding hip screw (p = 0.023). The
performance of the blade-plate alone was also superior to that of
the sliding hip screw (p = 0.023). With the numbers available, no
difference in the rate of union could be demonstrated between the
sliding hip screw and the dynamic condylar screw (p = 0.248)
or between the blade-plate and the dynamic condylar screw (p = 0.358).
The results within the sliding-hip-screw group were analyzed.
Three sliding hip screws had poor placement (the lag screw was high
or anterior in the femoral head), and all three failed. Six of the
thirteen sliding hip screws with good placement also failed. Even
when the sliding hip screws were well placed and had resulted in
anatomic fracture reduction, the failure rate was high (two of five).
All failures of the sliding hip screws were due to loss of proximal fixation
resulting in cutout or nonunion.
Radiographic review of medial- lateral fracture translation
for the group as a whole showed no measurable translation of the
proximal fracture fragment in thirty-two (68%)
of the forty-seven patients. There was no noticeable translation
in the anterior-posterior plane in any hip. Of the seven healed
fractures treated with a sliding hip screw, six showed translation,
with the mean final translation in the medial- lateral
plane being 66% (range, 0% to 80%) of
the femoral diameter at the level of the fracture (Fig. 4).
Thirteen (68%) of nineteen fractures with a large posteromedial
fragment, suggesting an unstable posteromedial buttress (type 31.A3.3),
healed compared with nineteen (68%) of twenty-eight
fractures without a large posteromedial fragment (type 31.A3.1).
Fracture reduction was graded as anatomic in twenty- three
hips and nonanatomic in twenty-four. Four anatomically
reduced fractures (17%) and eleven nonanatomically reduced
fractures (46%) had a failure of treatment (p = 0.060).
Eleven (26%) of forty-two hips in which the implant
was in a good position and four (80%) of five with a poor implant
position had a failure of treatment. This difference was significant
(p = 0.023).
Seven of the eight fractures that had been treated with bone
graft at the initial operation healed, while twenty-five
of the thirty-nine fractures that were not treated with
bone graft healed. This difference was not significant (p = 0.41).
Two patients underwent primary prosthetic replacement and were
followed for five and six years. At the time of the last review,
neither had pain or had had a revision. Radiographs revealed no
evidence of prosthetic loosening.
Fifteen fractures treated with internal fixation failed to heal
or had a failure of fixation. The mean time to failure was eight
months (range, one day to twenty-six months). In six, the
failure mechanism was loss of proximal fixation with the implant
cutting out of the femoral head (four sliding hip screws and two
dynamic condylar screws cut out). Eight hips had nonunion and loss
of proximal fixation without cutout of the implant from the femoral head.
One patient treated with a reconstruction nail had a nonunion that
required bone-grafting without revision of the fixation.
Two patients were diagnosed with a nonunion but did not undergo
a reoperation because they had major medical comorbidities and low
functional demands. Six failures with loss of fixation and cutout
through the femoral head were revised to a hemiarthroplasty with
internal fixation of the greater trochanter. Six patients had revision
of the internal fixation with bone-grafting; autograft was used
in five and allograft, in one. One patient with a well-fixed reconstruction
nail underwent autogenous bone-grafting. Two patients had a revision
to a 135° hip screw with cerclage wire to prevent translation; one,
a revision to a 150° hip screw with cerclage wire; two, a revision
to a blade-plate; and one, a revision to a dynamic condylar screw.
All revisions with internal fixation healed without additional surgery.
Functional Data
Before the hip fracture, thirty-eight patients (78%) walked
without support, two (4%) required a cane, and nine (18%)
required a walker. Before the injury, thirty-five patients
(71%) lived independently, nine (18%) were in
a nursing home, and five (10%) lived at home but were dependent
on others.
At the time of the last clinical review, forty-eight patients
(98%) had little or no pain, one (2%) had moderate
pain, and none had severe pain. Forty-six patients (94%)
used no analgesics, and three (6%) used non-narcotic analgesics
for occasional discomfort. Ten patients (20%) were unable
to walk, twenty-four (49%) could walk indoors
only, seven (14%) could walk one to six blocks, one (2%)
could walk greater than six blocks, and seven (14%) could
walk an unlimited distance. Twelve patients (24%) needed
no walking aids, three (6%) used a cane occasionally, four
(8%) needed a cane full-time, and twenty (41%)
needed a walker. At the time of the last review, fifteen patients
(31%) lived alone, and eight (16%) were dependent
on others. After all reconstructive procedures had been performed,
twenty-one (53%) of forty patients who had been
able to walk independently (with no or cane support) preoperatively
required a walker full-time or were unable to walk. Twenty (57%)
of the thirty-five patients who had lived alone preoperatively
lost their functional independence postoperatively and resided in
a nursing home or an assisted-living facility.
Medical Complications and Mortality
Twenty-eight (51%) of the fifty-five
patients were alive at the time of this review. The mortality rate was
4% (two of fifty-five patients) at thirty days, 19% (ten
of fifty-two patients) at one year, and 33% (seventeen
of fifty-two patients) at two years; all of these rates were significantly
higher than the expected mortality rate for West North Central Whites10
(p < 0.001). There were three intraoperative complications.
Marked hypotension developed in one patient, necessitating rapid
completion of the procedure. Two nondisplaced greater trochanteric fractures
became displaced during insertion of a cephalomedullary
nail. There were no intraoperative deaths. Twenty-one patients
(38%) had one or more postoperative medical complications.
These included congestive heart failure in six, cardiac arrhythmia
in three, deep venous thrombosis in three, pneumonia in three, myocardial
infarction in two, pulmonary emboli in two, hematoma in one, delirium
tremens in one, and severe inappropriate antidiuretic hormone secretion
syndrome in one. There were no infections.
Reverse obliquity intertrochanteric fractures of the femur are
recognized as biomechanically different from standard intertrochanteric
fractures. In this review, they were found to be uncommon but not rare
fractures, accounting for about 5% of all intertrochanteric
and subtrochanteric fractures. We found that the rate of failure
of internal fixation for this fracture pattern was higher than the
rates in most reports of internal fixation of intertrochanteric fractures
with use of modern internal fixation devices2-5.
This finding was implant-specific, with 95° fixed-angle devices
associated with better results than sliding hip screws. These data
corroborate and expand information provided by Henry et al.13, who reported that seventeen reverse
obliquity fractures (a prevalence of 16.6% in their series)
treated with a sliding hip screw had a significantly higher failure
rate than standard intertrochanteric fractures treated with the
same implant (24% compared with 3%, p = 0.007).
They recommended consideration of primary prosthetic replacement
for these fractures in elderly, osteoporotic patients.
Our review of a large consecutive series of hip fractures allowed
us to calculate the prevalence of reverse obliquity hip fractures
and to review the results of treatment of a large number of these uncommon
fractures. Several different types of fixation devices were used
in this series, which is both a strength and a weakness of the study.
The inclusion of various fixation devices allows analysis of trends
in the results of the use of each type of device, but because this
was not a prospective, randomized series comparing the various devices
it provides only qualitative, and perhaps biased, comparisons. Not
all of the fractures were anatomically reduced or had ideal implant
placement. One of the values of this study is the demonstration
that poor implant placement had a strong negative effect on the
outcome of these fractures. There was also a trend suggesting the
negative effect of fracture malreduction on the outcome.
The retrospective nature of this study leads to some limitations
in interpretation. From the operative reports, we were unable to
determine the exact reduction techniques used in each case or the
extent of soft-tissue stripping that occurred. Kinast et al.14 reported that indirect reduction
techniques can favorably influence the rate of union of proximal femoral
fractures treated with a blade-plate, obviating the need for bone-grafting.
Additionally, bone quality, which is probably a factor in fixation
failure, could not be reliably assessed in this retrospective review.
Since the mean time to failure in this series was eight months,
more than twice the minimal follow-up period, more failures
may have occurred had all patients been followed for a longer period.
There were no nondisplaced fractures in our series. In addition,
all fractures had evidence of an additional fracture line, which
was usually nondisplaced, extending proximally into the greater trochanter.
This may be a consideration when the surgeon contemplates intramedullary
fixation of these fractures with use of a starting point in the greater
trochanter. Two patients in our series sustained displacement of
a previously nondisplaced fracture during intramedullary nailing;
however, this had no adverse effect on the functional outcome for
these patients. Two patients in our series had a four-part fracture
with a displaced greater-trochanter component, which is not subclassified
by the current Orthopaedic Trauma Association scheme. Surgeons treating
four-part proximal femoral fractures should carefully scrutinize
preoperative radiographs to assess the primary fracture obliquity
when selecting implants for internal fixation.
Sliding hip screws have proven to be successful implants for
the treatment of intertrochanteric fractures of the proximal part
of the femur. The key to the success of these devices is controlled
postoperative impaction of the fracture to a stable configuration2-5. This concept requires that the
direction of compression be perpendicular to the major fracture
line, a condition present in most intertrochanteric fracture patterns
(Fig. 5).
The application of this concept to reverse obliquity fractures is
suspect because sliding of the proximal fragment and medialization
of the distal fragment can lead to fracture distraction (Fig. 6). Under these
circumstances, there is no medial buttress, the implant acts as
a load-bearing device, and subsequent loss of proximal
fixation can occur. Treatment of these fractures with a sliding
hip screw led to a 56% failure rate (nine of sixteen) in this
series. The most common mode of failure was medialization of the
distal fragment and loss of proximal fixation with nonunion or cutout
of the lag screw superiorly. This is the mode of failure predicted
on the basis of the biomechanics of sliding hip screws and of this
unique fracture pattern.
As in previous reports of the use of sliding hip screws for intertrochanteric
fractures3,6,15, implant position
influenced the success of the internal fixation in our series. All
three sliding hip screws that were in a poor position failed. Notably, however,
six of the thirteen hip screws placed in an ideal position failed
as well. Most of the fractures that united after treatment with
a sliding hip screw had marked medial displacement of the distal
fragment, averaging 66% of the femoral diameter at the level
of the fracture. These fractures appear to have "won the
race" between fracture-healing and fixation failure. Even
when healing occurs, however, the resultant deformity of the proximal
part of the femur with this degree of translation could complicate
later prosthetic replacement, if needed (Fig. 4).
On the basis of biomechanical arguments against sliding hip screws,
fixed-angle devices have been advocated for the treatment of reverse
obliquity fractures16,17, but
we are not aware of any published clinical data supporting their
use. This paper provides evidence that fixed-angle devices can be
used to treat these fractures with a high likelihood of success.
The blade-plate performed well in our series; it was successful
in thirteen of fifteen cases. The dynamic condylar screw was successful
in seven of ten cases. Thus, fixed-angle devices (successful in twenty
of twenty-five cases) were superior to the sliding hip
screw (successful in seven of sixteen cases), and the difference
was significant (p = 0.023). On the basis of this information,
surgeons at our institution now use fixed-angle devices to treat most
reverse obliquity intertrochanteric fractures. Blade-plates have
theoretical advantages over 95° dynamic condylar screws because
blade-plates provide more resistance to rotation of the proximal fragment
and also do not allow the proximal fragment to slide laterally.
Our data did not demonstrate a difference in the performance of
these two devices.
The experience with intramedullary fixation of these fractures
was limited in this series. Intramedullary fixation has the theoretical
advantage of a shorter lever arm for the fixation device, and it
has less potential for fracture collapse and limb-shortening when
it is used for unstable intertrochanteric fractures. Recent prospective,
randomized studies have shown the potential benefits of intramedullary devices
in the treatment of unstable proximal femoral fractures18,19. All three hips in which the
fracture was treated with a cephalomedullary device in this series healed,
but in two a previously nondisplaced fracture of the greater trochanter
was displaced during nail insertion.
Devices that were not used in this study also may be considered
for the treatment of these fractures. Special plates have been designed
to allow axial compression to occur at the fracture while the sliding
screw is locked. Preliminary reports about these devices have been
encouraging20,21. A trochanteric
blocking plate22 has also been
designed to prevent sliding of the proximal fragment when a sliding
hip screw is used.
In conclusion, reverse obliquity fractures of the intertrochanteric
region of the femur are challenging to treat. Although union or
a successful reconstructive procedure was achieved in most patients, the
mortality rates in our elderly population were high and the functional
results were generally poor. Better results were achieved when a
fixed-angle fixation device was properly applied in a hip with a well-reduced
fracture.