All patients who had had a hemiarthroplasty at the Mayo Clinic
between 1974 and 2001 were identified retrospectively in a joint registry. All
patients treated with a hip arthroplasty at our institution were entered in
the database, and they were followed at two to three months, one year, two
years, five years, and each subsequent five-year interval until revision or
death. This evaluation was done with a clinical examination, standardized
letter, or telephone questionnaire. An accuracy of 95% has been reported for
this database9. The
patients' charts were reviewed to determine the type of surgery, comorbidities
at the time of the surgery, subsequent dislocations, and etiology of the
dislocations.
The patients were divided into two groups, depending on the indication for
the surgery. Patients in whom the hemiarthroplasty was performed following
resection of a tumor involving the femoral head or neck or the proximal part
of the femoral shaft constituted one group and were compared with patients
treated with a hemiarthroplasty for any other etiology. The latter group was
recently described and analyzed in
detail8.
Of the 2132 proximal femoral arthroplasties performed between 1974 and
2001, 1812 were done in patients with no tumor involvement. Within that group,
1335 (73.7%) were performed because of a fracture of the proximal part of the
femur; 208 (11.5%), because of osteonecrosis of the femoral head; 128 (7.1%),
because of nonunion of the proximal part of the femur; forty-one (2.3%),
because of degenerative joint disease; ninety-seven (5.4%), because of a
failed previous operation, such as open reduction and internal fixation of a
proximal femoral fracture or an osteotomy; two (0.1%), because of rheumatoid
arthritis; and one (0.06%), because of synovial chondromatosis. During this
same time span, 320 hemiarthroplasties were performed for tumor-related
conditions. The median age at the time of the surgery was 77.0 years (range,
twelve to ninety-nine years) for the patients without tumor involvement at the
hip and 61.0 years (range, eleven to ninety-one years) for those with tumor
involvement. The age difference between these two groups was significant (p
< 0.001).
Of the patients without tumor involvement at the hip, 1282 (71.0%) were
female and 530 (29.0%) were male, whereas the group with tumor involvement
comprised 176 female patients (55.0%) and 144 male patients (45.0%). The
gender distribution was significantly different between the two groups (p <
0.001).
The cumulative probability of dislocation and patient survival was
estimated with use of the Kaplan-Meier survival
method10. The Cox
proportional hazards survival method was used to assess the significance of
the effect on the dislocation rate of a tumor at the surgical site,
preservation of the greater trochanter, and the extent of the osseous
resection11. The
alpha level was set at 0.05 for significance.
The malignant processes leading to the hemiarthroplasties involved
the femoral head, femoral neck, peritrochanteric region, proximal part of the
femoral shaft, or more than one of these areas, and included metastatic tumors
(195 patients; 60.9%), multiple myeloma (thirty-one; 9.7%), chondrosarcoma
(twenty-three; 7.2%), osteosarcoma (fourteen; 4.4%), and other primary tumor
subtypes (thirty-one; 9.7%). The tumor type was not specified for twenty-six
hips (8.1%).
Thirty-two of the 1812 patients without tumor involvement and eleven of the
320 patients with tumor involvement of the femur had a hip dislocation after
the hemiarthroplasty. Most of the dislocations in the patients with tumor
involvement occurred without trauma and were managed with closed reduction
(Table I). The cumulative
probability of dislocation at ten years after the hemiarthroplasty was 10.9%
(95% confidence interval = 0% to 22.5%) for the patients with tumor
involvement and 2.1% (95% confidence interval = 1.2% to 3.1%) for those
without tumor involvement. The risk of dislocation was increased threefold
when the hemiarthroplasty was performed for a condition related to tumor
involvement (hazard ratio = 3.2 [95% confidence interval = 1.5% to 6.7%], p =
0.002) (Fig. 1).
The mean time (and standard deviation) from the surgery to the dislocation,
in the patients who had a dislocation, was 3.4 ± 5 years (median,
thirty-seven days; range, zero days to 16.6 years) in the group without tumor
involvement and 2.3 ± 3.9 years (median, twenty-four days; range, one
day to 10.4 years) in the group with a tumor-related condition.
In the group with tumor involvement, eight operations were performed
between 1974 and 1979; 118, between 1980 and 1989; 162, between 1990 and 1999;
and thirty-two, between 2000 and 2001. The surgical decade did not correlate
with the dislocation rate, with the numbers available.
One hundred and seventy-six of the tumor resections were performed in the
area of the femoral neck; twenty-seven, in the peritrochanteric area; and 117,
distal to the lesser trochanter. A statistical analysis comparing the
dislocation rates following resections distal to the lesser trochanter with
those following more proximal resections showed no significant association
between the level of bone resection and the postoperative dislocation rate,
with the numbers available (p = 0.4).
The effect of preservation of the greater trochanter at the time of the
surgery was analyzed as well. The greater trochanter was not preserved at the
time of the surgery in 128 patients with tumor involvement, it was reattached
in four, it was partially preserved in thirteen, and it was completely
preserved in 175. With the numbers available, the dislocation rate was not
found to be significantly different in the patients in whom the greater
trochanter had been preserved, but there was a trend toward a more favorable
outcome when the greater trochanter had been preserved (hazard ratio = 3.5
[95% confidence interval = 0.9% to 13.3%], p = 0.06)
(Fig. 2).
The ten-year survival rate in the group with a tumor-related condition (89%
[95% confidence interval = 78% to 100%]) was significantly lower than that in
the group without a tumor (98% [95% confidence interval = 97% to 99%]) (p <
0.01). The deaths of the patients with a non-tumor-related arthroplasty
occurred at a median of 4.2 years (range, three days to twenty-five years)
postoperatively, and the deaths of the patients with a tumor-related condition
occurred at a median of 8.8 months (range, zero days to 11.8 years
postoperatively).
There have been few studies of the long-term dislocation rates
following
hemiarthroplasty6,12-19.
The current study demonstrated a significantly higher ten-year dislocation
rate (of 11%) following hemiarthroplasties for tumor-related conditions. This
study also demonstrated that, when dislocations occurred following
hemiarthroplasties for tumor-related indications, they occurred earlier in the
postoperative course than did dislocations following hemiarthroplasties
performed for non-tumor-related conditions. This finding was probably due to
the often limited survival of the patients with tumor involvement.
The patient population treated with a hemiarthroplasty for a
non-tumor-related condition differed from that treated for a tumor-related
indication with respect to age and gender. Overall, the group with a
tumor-related condition was younger and the patients were more frequently
male. It has been reported that women and older patients are more likely to
have a dislocation following a hip
arthroplasty20.
Therefore, it is interesting that, in the current study, a population with
younger patients and more men had a higher dislocation rate. It appears that
factors associated with surgery for tumor-related conditions affect the
dislocation rate more than do age and gender.
Resection of the greater trochanter, extensive soft-tissue resection, and
the level of the bone resection were assumed to be other potential factors
predisposing patients to dislocation. Preservation of the greater trochanter
should lower the dislocation rate because the abductor mechanism is kept
intact, and we did find a trend favoring a lower dislocation rate when the
greater trochanter had been preserved (p = 0.06). In the first years of the
follow-up period, the dislocation rate for patients with tumor involvement who
had had the greater trochanter preserved equaled the rate for patients without
tumor involvement. After four years, however, the dislocation rate for
patients with tumor involvement and a preserved greater trochanter increased
beyond that for patients without tumor involvement
(Fig. 3).
Interestingly, the level of the osteotomy (neck, intertrochanteric, or
subtrochanteric) did not influence the dislocation rate in the patients with
tumor involvement. However, since the overall number of dislocations was
fairly low, it is conceivable that differences would be distinguishable with a
greater statistical power. This may be especially true of the influence of
subtrochanteric resections (the number of which was limited in the study)
since, by definition, subtrochanteric resections include the greater
trochanter.
The preservation of the greater trochanter seems to be a dominant factor in
determining the stability of the position of the hemiprosthesis, and it seems
to be more important than the osteotomy site and the extent of soft-tissue
resection. This observation contradicts experiences with total hip
arthroplasties in which the soft-tissue resection and the osteotomy site
seemed to influence the dislocation
rate21-23.
The importance of the greater trochanter probably lies in the preservation of
the biological insertion sites of the abductors, which constitute an important
stabilizing force against dislocating forces around the
hip21. The
abductors are usually reattached to the components, the fascia, or other
muscle groups when the greater trochanter cannot be preserved. This maneuver
does not seem to provide stability of the same quality as that provided by
preservation of the original insertions.
Not surprisingly, the postoperative survival of the patients with tumor
involvement was strikingly shorter than that of the patients without tumor
involvement. This affected the power of the study to analyze causative factors
for long-term dislocations. Because of the limited life expectancy of patients
treated with hemiarthroplasty for tumor-related conditions, long-term
dislocation rates might not be as important as mid-term and short-term
dislocation rates. Dislocations and the subsequent necessity for additional
treatments, which are sometimes surgical, affect the quality of life of these
patients who have a limited life span. Therefore, we think that, when
hemiarthroplasty is performed in patients with tumor involvement, preservation
of the greater trochanter should be attempted whenever it is justifiable
according to the principles of oncologic surgery. ?