In patients with end-stage renal disease, hip fractures are
associated with a high risk of mortality and frequent
complications1-3.
The reported one-year mortality rates for elderly patients without end-stage
renal disease who sustain a femoral neck fracture have ranged from 15% to
40%4-6.
The corresponding rate for patients who are on chronic hemodialysis because of
renal disease is
50%3. Hip fractures
in this patient group are frequently complicated by problems such as hematoma,
prolonged wound drainage, infection, implant failure, and nonunion. Most
studies of hip fractures in patients with end-stage renal disease have
involved small numbers of subjects and have not analyzed the effects of
fracture type or treatment method on
outcome2,3,7.
Our goals were to clarify the range of clinical outcomes after treatment of
a hip fracture in patients with end-stage renal disease who are on chronic
hemodialysis and to evaluate how the type of fracture and treatment affect the
outcome.
The study involved patients with end-stage renal disease who were
admitted to two hospitals affiliated with our institution for the treatment of
a femoral neck or intertrochanteric fracture between 1994 and 2003. The
inclusion criteria were a diagnosis of chronic renal failure, dependence on
hemodialysis, no functioning renal transplantation before the fracture, an
ability to walk independently, and no history of hip disease or fracture. The
exclusion criteria were treatment with peritoneal dialysis; a functioning
renal transplantation before the fracture; an ability to walk that is
dependent on the use of a cane, crutches, or walker; and a history of hip
disease or fracture.
For each of the twenty-nine patients (thirty-two hip fractures) identified,
one independent investigator (O.S.), reviewed the medical records and recorded
the sex of the patient, age at the time of fracture, etiology of the renal
disease, total time on hemodialysis prior to hip surgery (hemodialysis
duration), mechanism of injury, and types of initial and secondary surgical
procedures. The details of hemodialysis, the status of any renal transplant,
and all data concerning the medical condition of the patient were gathered
from the hospital records and were analyzed by a nephrologist (N.O.). None of
the patients had a functioning renal transplant at the time of fracture, so
confounding factors such as prolonged use of immunosuppressive drugs or
prednisone were not analyzed.
The fractures were divided into three groups for analysis according to the
type of fracture and treatment. Group 1 included intertrochanteric fractures
treated with internal fixation by a sliding hip screw or proximal femoral nail
(PFN; Synthes, Bettlach, Switzerland); Group 2, femoral neck fractures treated
with osteosynthesis with use of three or four cannulated screws; and Group 3,
femoral neck fractures treated with hemiarthroplasty. The fractures in Group 2
were classified according to the system described by
Garden8, in which
type-1 and 2 fractures were considered nondisplaced and type-3 and 4 fractures
were considered displaced.
Baskent University Hospital is a certified transplantation and dialysis
referral center. The patients included in this study were followed
concurrently by the nephrologists and the internists. All of the patients had
been undergoing hemodialysis in the same institution and continued to do so
after surgery. All subsequent renal transplantations were carried out by the
general surgical team in the same institution. The medical management of these
patients was performed by the nephrologists to ensure that all patients were
in the best medical condition prior to surgery. For the prevention of
osteodystrophy, all patients had vitamin-D supplementation, monitoring of the
serum parathormone level and phosphorus level, and treatment with serum
calcium for electrolyte imbalance. Pneumatic compression devices and
injections of low-molecular-weight heparin were used as prophylaxis against
deep-vein thrombosis. Prophylactic antibiotic treatment with a
second-generation cephalosporin was initiated preoperatively and was continued
for twenty-four hours. The patients who underwent osteosynthesis were
mobilized with partial weight-bearing the day after surgery, and full
weight-bearing was allowed after union was confirmed radiographically. Those
who underwent arthroplasty were mobilized with full weight-bearing in the
early postoperative period. A physiotherapist working as a member of the
orthopaedic team ensured adherence to a standardized rehabilitation
protocol.
Data were gathered from the records for the following parameters:
postoperative hematoma with drainage that was longer than four days, deep-vein
thrombosis (confirmed with Doppler ultrasound), infection (confirmed with
cultures), iatrogenic nerve injury (tested clinically and confirmed with
electromyography), fracture nonunion, osteonecrosis, implant failure
(confirmed radiographically), and mortality.
The outcome for each group was analyzed, and early (within the first month
after hip fracture repair) and late complications (after the first month) were
recorded. Hematoma, infection, thromboembolism, and iatrogenic nerve injury
were classified as early complications, and nonunion, osteonecrosis, and
implant failure were considered late complications. Early mortality (death in
the first year after repair of the hip fracture) and late mortality (death
after the first year) were analyzed as well.
Five types of statistical analysis were performed. Survival analysis was
carried out with use of the Kaplan-Meier test, and the cumulative survival
rates (and standard error) of the groups were compared with use of the
log-rank test. The Fisher exact test was used to compare the complications
between diabetic and nondiabetic patients. The effect of nonunion on mortality
was compared with the Fisher exact test for all patients managed with internal
fixation (Groups 1 and 2). The accepted metric for the statistical analysis of
mortality was the total rate of mortality (early and late). The Levene test
revealed that age, sex, and hemodialysis duration were homogeneously
distributed among the groups, and the influence of these variables on
mortality was assessed with use of logistic regression analysis. The
chi-square test was used to compare the overall complication rates among the
groups. All calculations were performed with use of SPSS for Windows software
(version 11.5; SPSS, Chicago, Illinois), and alpha was set at 0.05.
During the enrollment period, the cases of forty patients with
forty-five hip fractures were identified and retrospectively reviewed. Of the
forty patients, eleven were excluded from the study for the following reasons:
a history of core decompression of the fractured hip because of osteonecrosis
(two patients), treatment with peritoneal dialysis (four patients), and a
functioning renal transplant at the time of fracture (five patients). The
study group comprised the remaining twenty-nine patients with thirty-two hip
fractures. There were fifteen men and fourteen women, with a mean age of
fifty-seven years (range, nineteen to eighty-one years). The etiologies of
end-stage renal disease were diabetes mellitus (ten patients), hypertension
(five patients), a combination of chronic glomerulonephritis and
pyelonephritis (five patients), a combination of nephrolithiasis and urethral
stenosis (four patients), amyloidosis (two patients), polycystic kidney
disease (one patient), and tuberculosis (one patient); the cause was unknown
in one patient. The mean duration of hemodialysis (the total time on
hemodialysis prior to hip surgery) for the twenty-nine patients was 6.8 years
(range, six months to twenty years). Seven patients had renal transplantation
after treatment of the hip fracture; however, all but one of the transplants
failed because of vascular complications in conjunction with complications of
immunosuppressive agents. The patient with the functioning graft had been on
hemodialysis for only six months when he sustained the hip fracture. He
received the kidney transplant after the hip fracture had healed.
Of the twenty-nine patients, eleven had intertrochanteric fractures and
eighteen had femoral neck fractures. Three patients had a bilateral femoral
neck fracture. All of the intertrochanteric fractures and seventeen of the
twenty-one femoral neck fractures were caused by trauma. The other four were
stress fractures of the neck (an undisplaced femoral neck fracture without a
history of trauma). Egol et al. described stress fractures as insufficiency
fractures resulting from impaired bone
quality9. Since we
could not elicit a history of trauma, we defined these complete but
nondisplaced fractures as stress fractures.
Of the eleven intertrochanteric fractures in Group 1, two were treated with
a proximal femoral nail (PFN), and nine were treated with a sliding hip screw
(Dynamic Hip Screw; Synthes). The thirteen femoral neck fractures (four
nondisplaced and nine displaced) in the ten patients in Group 2 were all
treated with percutaneous screw fixation. The other eight femoral neck
fractures (Group 3) were treated with unipolar hemiarthroplasty (five hips) or
bipolar hemiarthroplasty (three hips).
Complications and Outcome
Group 1 (Eleven Hips)
Five patients had no complications
(Table I). The early
complications were postoperative hematoma with prolonged drainage (two hips)
and infection (one hip), and none of the patients had deep venous thrombosis
or thromboembolism. The late complications were nonunion in five hips. Of the
five hips with nonunion, two had been treated with a proximal femoral nail and
three with a sliding hip screw. One of the patients with a nonunion had had an
early wound infection, and one had had a postoperative hematoma. The hip with
infection and nonunion was débrided in a second operation thirteen days
after the initial surgery. This controlled the infection, but no union of the
fracture was seen at ten months. Of the other four hips with nonunion, one was
in a patient who was lost to follow-up after eight months and three were in
patients who died before a second surgery was performed.
The mean duration of follow-up for Group 1 was fourteen months, and five of
the eleven patients died. Four of them died in the first year after hip
fracture surgery. These patients included one who died from respiratory
failure caused by pneumonia two days after the operation, two who died of
cardiopulmonary arrest at four and six months, and one who died seven months
after surgery because of infection-related complications. The remaining
patient died of cardiopulmonary arrest seventeen months after surgery.
For this group, the cumulative survival rate (and standard error) was 51%
± 16.3%, and the mean survival time was 20.1 ± 4 months.
Group 2 (Thirteen Hips)
Two patients had no complications (Table
II). This group included all three patients with a bilateral
femoral neck fracture. The types of early morbidity observed were deep wound
infection (two hips) and postoperative hematoma (three hips). Eleven hips had
late complications, including nonunion in nine hips and osteonecrosis in
two.
For all nine hips with a displaced femoral neck fracture treated with
osteosynthesis, the outcome was nonunion. Of the nine hips with nonunion, two
were mechanical failures and seven were considered as biological failures.
Osteosynthesis led to union in all four nondisplaced fractures, but two of the
hips had osteonecrosis develop. Two hips with nonunion had a deep infection,
and both were treated with a Girdlestone resection arthroplasty. One of the
remaining seven hips with an aseptic nonunion was the left hip of a patient
with bilateral fracture. A vascularized iliac bone graft was implanted, and
union was achieved three months later. Of the other six fractures with aseptic
nonunion, one was treated with Girdlestone resection arthroplasty (Figs.
1-A,
1-B, and
1-C) and the other five were in
four patients (one with bilateral nonunion) who died before any secondary
intervention was performed.
The mean duration of follow-up for Group 2 was 34.7 months, and the overall
mortality rate was 50%. Three of the ten patients in Group 2 died in the first
year after hip fracture surgery. They included one patient who died of a
cerebrovascular accident two months after hip fracture repair, one who died of
cardiopulmonary arrest at three months, and one who died at ten months because
of peritonitis contracted through peritoneal hemodialysis. The patients who
had died at two and three months after surgery had radiographic signs of
mechanical failure of the fracture fixation, leading to nonunion before their
death. Two patients died of cardiopulmonary arrest in the late period (at
twenty-six and sixty-three months) after surgery. For this group, the
cumulative survival (and standard error) was 42% ± 17.6% and the mean
survival time was 44.6 ± 10 months.
Group 3 (Eight Hips)
Seven of the eight hips with a femoral neck fracture had no complications
(Table III). The remaining
patient had deep venous thrombosis and a sciatic nerve palsy develop. There
were no late complications in this group.
The mean duration of follow-up for Group 3 was 20.8 months, and three of
the eight patients died. All three deaths occurred early, and all were due to
cardiopulmonary arrest. The latest death in Group 3 occurred at eleven months
postoperatively. For this group, the cumulative survival rate was 62.5%
±17.1% and the mean survival time was 32.3 ± 6.3 months.
Comparison of the Groups
The mean duration of follow-up for this series was twenty-three months, and
the overall mortality rate for the twenty-nine patients was 45%. A comparison
of the cumulative survival rates in the groups, with use of the log-rank test,
revealed no significant differences (Table
IV). A comparison of the complications between the diabetic and
nondiabetic patients, with use of the Fisher exact test, revealed no
significant difference, with the numbers studied (p = 0.064). Nonunion had no
effect on the rate of mortality (the total number of deaths in Groups 1 and 2)
(p = 0.387). Regression analysis of age, sex, and total hemodialysis duration
in relation to mortality risk revealed that only age had a significant
influence on mortality (p = 0.019) (Table
V). Comparison of the overall complication rates showed that the
patients in Group 3 (hemiarthroplasty for femoral neck fracture) had
significantly fewer postoperative complications than those in the other two
groups (p = 0.039).
Patients on chronic hemodialysis have a high mortality rate because
of the complications of end-stage renal disease. The multiorgan insufficiency
and metabolic imbalance that occur in patients on chronic hemodialysis cause
immunocompromise, which places them at an increased risk for
infection10,11.
Pericarditis, cardiac tamponade, congestive heart failure, coronary artery
disease, platelet dysfunction, duodenal ulcers, and iatrogenic complications
related to fluid administration (fluid overload) or inappropriate medications
are also frequently encountered in patients with renal failure, causing
mortality. Byrne et al. studied the one, three, and five-year survival rates
for 95,394 patients on
hemodialysis12.
They found that the one-year cumulative survival rate for individuals who were
on hemodialysis was nearly 60% in patients who were eighty years old or older.
In our study of twenty-nine patients on chronic hemodialysis who underwent
repair of a hip fracture, we observed a 45% overall mortality rate in a mean
post-operative follow-up time of twenty-three months. In a study of four
patients with chronic renal failure who had a hip fracture and were treated
nonsurgically, Schaab et al. reported that all of the patients died within the
first year because of a lack of
mobility13. Tierney
et al. investigated thirteen hip fractures in twelve patients with end-stage
renal disease3.
Twelve hips were treated surgically, and six patients died within one year
after the fracture. The authors identified hip fracture as the reason for the
higher mortality rate in this group compared with the rates reported for
patients with end-stage renal disease without a hip fracture. In another study
of eight patients with end-stage renal disease who had nine hip fractures that
were treated surgically, the first-year mortality rate was
38%2. Those authors
concluded that the rate was not related to operative complications, and they
stressed that a team approach involving an expert nephrologist and early
aggressive mobilization could reduce early complications and early
mortality.
Most previous studies of hip fracture in patients with end-stage renal
disease have not analyzed how the fracture type or treatment method affects
the outcome. Researchers have grouped all of these patients together under the
label "hip fractures treated
surgically."2,3,7
To our knowledge, the present study is the first to evaluate the effects of
the type of hip fracture and treatment method on outcome in patients on
hemodialysis.
The eleven patients on hemodialysis in our study who had an
intertrochanteric fracture (Group 1) had high rates of early morbidity (three
patients) and nonunion (five patients). This rate of nonunion is much higher
than the 2% rate reported in the literature for intertrochanteric fractures in
patients without renal
failure14-16.
Screw purchase in the femoral head was judged to be strong by the surgeon
during fracture fixation, and cement augmentation was not deemed
necessary17,18.
However, given the high rate of mechanical failure in our series, this might
have been a reasonable adjunct. Another alternative for treating these
fractures is proximal femoral replacement. Several studies have compared
proximal femoral replacement with internal fixation for intertrochanteric
fractures19,20.
In those series, both long-term and short-term results demonstrated an obvious
advantage of fixation over replacement, with a shorter operative time, less
blood loss, fewer units of blood transfused, a lower mortality rate, and lower
hospital costs. Although those series did not specifically study patients
undergoing hemodialysis, fixation seems to be the first choice in the
treatment of intertrochanteric fractures. The reported one-year mortality
rates after intertrochanteric fracture for the general population have ranged
from 10% to
30%21,22.
These figures are comparable with the rate of early mortality (four patients)
that we observed in the eleven patients with end-stage renal disease who had
intertrochanteric fractures.
The surgical treatment alternatives for femoral neck fractures are
osteosynthesis and arthroplasty. A letter by Tzamaloukas et al. analyzed the
findings of two different studies and recommended operative treatment of
femoral neck fractures despite the high risk of preoperative and postoperative
complications23.
Currently, as far as we know, there are no large series in the literature that
provide results of osteosynthesis for femoral neck fractures in patients on
long-term hemodialysis, but reports have noted poor results related to
infection, nonunion, and
osteonecrosis23-27.
We observed a high complication rate (eleven of thirteen hips) in our group
that was treated with osteosynthesis. In all nine hips with a displaced
femoral neck fracture treated with this method, the result was a nonunion.
This may be due to both biological failure of the bone and mechanical failure
of the implant. In these fractures, renal osteodystrophy and osteoporosis can
lead to biological and mechanical failure. No implant can withstand the
mechanical forces in the hip if fracture union does not occur, and eventual
mechanical failure is inevitable. Therefore, it is difficult to differentiate
between a biological or mechanical cause of failure in these complex cases.
Two series have reported a mechanical advantage of dynamic hip-screw fixation
over cannulated screws in patients with
osteoporosis28,29.
This may also be true for patients undergoing chronic hemodialysis. In our
study, all four nondisplaced fractures of the femoral neck united with
internal fixation; however, two of the patients later had osteonecrosis
develop. Both patients had mild functional impairment and pain. Nonunion is
not a major problem in nondisplaced femoral neck fractures, and it is
difficult to predict which patients will have clinically important
osteonecrosis develop later. Although arthroplasty is a reliable alternative,
in order to prevent long-term complications, such as infection and loosening,
internal fixation may be accepted as the treatment of choice for nondisplaced
femoral neck fractures.
The nonunion rate in the osteosynthesis group (nine of thirteen hips) is
much higher than the previously reported rates for these fractures in patients
without end-stage renal
disease30. The mean
duration of follow-up for our study group that was treated with osteosynthesis
was 34.7 months, and the overall mortality rate was 50%, which was also much
higher than the rates that have been documented for patients without renal
failure who had femoral neck fractures treated with
osteosynthesis31-33.
The literature contains conflicting findings with regard to the outcome of
hip arthroplasty in patients with renal
failure7,34-37.
Lieberman et al.34
recorded infections in three of sixteen hips, and Naito et
al.35 reported
infections in two of fifteen hips. Considering the high rates of deep wound
infection in this patient group, numerous authors have concluded that
arthroplasty should be reserved for patients who are expecting to undergo or
have already undergone successful renal
transplantation34,36,37.
Gualtieri et al. reported good to excellent results in six of eight patients
on long-term hemodialysis who had femoral neck fractures treated with
arthroplasty and were followed for two to eight years
postoperatively36.
In that small series, two hips required revision and two patients died.
Sakalkale et al. analyzed the cases of twelve patients with a hip fracture
who were either on long-term hemodialysis or had received a renal transplant,
and all of whom had undergone total hip
arthroplasty37.
They reported that there were two deep wound infections, and seven of the
patients died. A decreased immune response and repeated bacteremia in patients
undergoing chronic hemodialysis may increase the risk of
infection10,34,35.
None of our patients on hemodialysis who underwent arthroplasty had an
infection develop. They also had no hematomas, possibly because of the
meticulous control of bleeding intraoperatively and the usage of fractionated
heparin instead of standard heparin for hemodialysis in the perioperative
period.
Comparison of the overall complication rates in our study revealed that the
rate for the patients with a femoral neck fracture treated by hemiarthroplasty
was significantly lower than those for both other groups (p = 0.039). The
group treated with osteosynthesis had a high frequency of nonunion. We also
analyzed the effect of nonunion on mortality. Although there was a tendency
toward higher mortality rates in patients with a nonunion, this was not
significant, with the numbers studied (p = 0.387). These patients were
severely osteoporotic because of renal failure, and most of them had renal
osteodystrophy at the time of fracture repair. The poor bone quality and
vascular changes that are typical of patients with renal failure might have
contributed to the high rate of failure with internal fixation. Our analysis
revealed no significant difference among the groups with respect to cumulative
survival proportions. As expected, older age was found to significantly
influence mortality risk.
In conclusion, hip fractures in patients with end-stage renal disease who
are on chronic hemodialysis are associated with high rates of complications
and death. However, surgical treatment can be successful with thorough
preoperative medical preparation and meticulous postoperative
follow-up2.
Osteosynthesis is an acceptable method for treating intertrochanteric and
nondisplaced femoral neck fractures in these patients; however, all of the
displaced femoral neck fractures that were treated this way resulted in
nonunion. Considering this, we recommend arthroplasty as the treatment of
choice for displaced femoral neck fractures in patients with end-stage renal
disease who are on long-term hemodialysis. ?