The study group consisted of 2000 consecutive primary Charnley
total hip arthroplasties that were performed on 1689 patients at
one institution from March 10, 1969, through September 27, 1971.
Eight hundred and twenty-eight men (996 hips) and 861 women (1004
hips) were included in the study. The mean age of the patients at
the time of surgery was 63.5 years (range, twenty-four to ninety-two
years). Forty-three of the hips were in patients who were younger
than forty years old, 144 were in patients who were forty to forty-nine
years old, 476 were in patients who were fifty to fifty-nine years old,
779 were in patients who were sixty to sixty-nine years old, 497
were in patients who were seventy to seventy-nine years old, and
sixty-one were in patients who were eighty years of age or older.
The diagnosis that led to the total hip arthroplasty was osteoarthrosis
in 1647 hips; rheumatoid arthritis in 166 hips; developmental dysplasia
in 116 hips; osteonecrosis of the femoral head in 103 hips; acute
fracture, nonunion of a fracture, or posttraumatic degenerative
disease in 207 hips; and miscellaneous causes (including a slipped
capital femoral epiphysis and Legg-Calvé-Perthes disease)
in twelve hips. Two hundred and fifty-one hips had more than one
diagnosis.
The operative approach was transtrochanteric in all cases. An all-polyethylene
socket (inside diameter, 22.25 mm) was cemented into the pelvis
after anchoring holes were made in the ilium, ischium, and pubis.
The femoral canal was prepared with broaches, and no pulsatile lavage
was used. A smooth-surfaced stainless-steel monoblock, so-called
flat-back Charnley component with a 22.25-mm head (DePuy International, Leeds,
United Kingdom) was cemented into the femur with use of hand-packed
radiopaque methylmethacrylate cement and no plug in the medullary
canal. Perioperative antibiotics were administered routinely.
Patients were followed prospectively and were asked to return to
our institution at one year, two years, five years, and every five
years thereafter for an interview and a clinical examination. When
this was not possible, patients were asked to answer a standardized
letter questionnaire and to send radiographs. Patients who did not
answer the letter questionnaire were contacted by telephone and
interviewed with use of a standardized telephone questionnaire.
Patients were specifically queried about whether they had had additional
surgery on the hip. If they had had additional surgery at another
institution, a request was made to obtain operative reports and information
concerning the indication for the surgery, the surgical findings,
and the components that were revised. The most recent follow-up
evaluation was conducted by interview and examination for 309 hips,
letter questionnaire for 1042 hips, and telephone questionnaire
for 649 hips. Dates of death were obtained either by direct communication
with the patient’s family or by way of the Social Security
network.
Implant survivorship was estimated with use of the Cox proportional-hazards
model and was adjusted for correlated data—that is, for
two hips in patients who underwent bilateral total hip arthroplasty17. Patients were censored at death
or at revision. End points considered in these analyses included
reoperation, removal or revision of an acetabular and/or
femoral component for any reason, and revision or removal of an
acetabular and/or femoral component for aseptic loosening.
For each end point, the risk factors that were considered included
gender, underlying diagnosis, and age (decade of life) at the time
of the total hip arthroplasty. Ninety-five percent confidence intervals
were calculated for the survivorship estimates. The rate of survival until
death was also compared with the expected rate of survival in an
age-and-gender-matched population of white United States citizens18. A one-sample log-rank test was
used to test for significance19.
Twelve hundred and twenty-eight patients (1459 hips) had died
by the time of the twenty-five-year follow-up after the total hip
arthroplasty. For the first twenty years after the procedure, our
patients lived longer than the predicted life spans of an age and
gender-adjusted population of white citizens in the United States
(the group most similar to the study group) (p < 0.0001).
After twenty years, the rate of survival of the study cohort was
poorer than predicted (Fig. 1).
Of the 2000 hips, 1942 (97%) were followed for at least twenty-five
years or until a revision operation, removal of the component, or
death, and 1980 (99%) were followed for at least twenty
years or until a revision operation, removal of the component, or
death. The longest follow-up period was 28.4 years. Of the 372 unrevised
hips in patients alive at the last follow-up evaluation, 323 had
been followed for at least twenty-five years and thirty-two had
been followed for twenty to twenty-four years. The mean age of the
patients still living at the end of the study was eighty years at
the time of their last follow-up evaluation. Five hundred and forty-one
of the 2000 hips were in patients who subsequently lived at least
twenty-five years after the operation. For patients who died, the
mean length of time between the last clinical follow-up examination and
death was 0.6 year; it was more than two years for only fifty-one
patients.
During the twenty-five-year study period, 296 hips were known
to have had a reoperation. The type of surgery was revision or removal
of a component in 242 hips and a reoperation without revision or
removal of a component (for example, trochanteric wire removal)
in fifty-four hips. Of the hips treated with removal or revision
of a component, 151 had the reoperation because of aseptic loosening.
The acetabular component alone was removed or revised in forty-five
hips, the femoral component alone was removed or revised in ninety-five
hips, and both components were removed or revised in 102 hips. Thus,
the acetabular component was removed or revised in a total of 147
hips (because of aseptic loosening in 100), and the femoral component
was removed or revised in a total of 197 hips (because of aseptic
loosening in 118). Among the forty-seven acetabular components that
were removed or revised for reasons other than aseptic loosening, the
reason was recurrent dislocation in nineteen hips, deep infection
in seventeen hips, fracture of a component in seven hips, and miscellaneous
reasons in four hips. Among the seventy-nine femoral components
that were removed or revised for reasons other than aseptic loosening,
the reason was fracture of the prosthetic stem in forty-five hips,
deep infection in seventeen hips, recurrent dislocation or subluxation
in eleven hips, periprosthetic femoral fracture in four hips, and
miscellaneous reasons in two hips.
For the 2000 hips, the twenty-five-year rates of survivorship free
of reoperation, free of revision or removal of the implant for any
reason, and free of revision or removal for aseptic loosening were
77.5% (95% confidence interval, 74.9% to 80.3%),
80.9% (95% confidence interval, 78.4% to
83.0%), and 86.5% (95% confidence interval,
84.0% to 88.4%), respectively (Fig. 2, Table I).
Survivorship free of revision for aseptic loosening was better for
the female patients than for the male patients: at twenty-five years,
the survivorship was 91.0% (95% confidence interval,
88.2% to 93.4%) for hips in women and 80.9% (95% confidence
interval, 76.9% to 84.8%) for hips in men (p < 0.0001).
With this end point, women had better twenty-five-year survivorship
of both the acetabular component (93.0% [95% confidence
interval, 90.2% to 95.2%] compared with 86.0% [95% confidence
interval, 82.0% to 89.4%] for the men;
p < 0.0001) and the femoral component (93.2% [95% confidence
interval, 90.8% to 95.3%] compared with
85.8% [95% confidence interval, 82.4% to
88.8%] for the men; p < 0.0001).
At twenty-five years, survivorship free of revision for any reason
(p < 0.0001) and survivorship free of revision for aseptic loosening
(p < 0.0001) were strongly associated with the age of the
patient at the time of the total hip arthroplasty; the rate was
better for older patients. For each decade earlier in life at which
the procedure was performed, the survivorship free of revision for
aseptic loosening was poorer (Fig. 3, Table II). No patient who was eighty years
of age or older at the time of surgery had a subsequent revision
for aseptic loosening of the implant. For younger patients, survivorship
of the acetabular component free of revision for aseptic loosening
was poorer than survivorship of the femoral component to the same
end point. In contrast, for older patients, survivorship of both
the acetabular and the femoral component free of revision for aseptic
loosening was high (Figs. 4 and 5, Table II).
For the whole study group, the twenty-five-year survivorship of
the acetabular component free of removal or revision for any reason
(87.0% [95% confidence interval, 84.6% to 88.9%])
was a little better than that of the femoral component to the same
end point (84.6% [95% confidence interval, 82.3% to
86.7%]), whereas the twenty-five-year survivorship free
of revision for aseptic loosening was almost identical for the acetabular
component (89.9% [95% confidence interval, 87.5% to
91.8%]) and the femoral component (89.8% [95% confidence
interval, 88.0% to 91.6%]). These rates
became similar by the twenty-five-year mark, although femoral survivorship
declined earlier than acetabular survivorship did (Table III). There were
more femoral revisions than acetabular revisions for aseptic loosening
in the first fifteen years of the study (seventy-seven femoral revisions
compared with forty-four acetabular revisions), but there were fewer
femoral revisions than acetabular revisions in the last ten years
of the study (forty-one femoral revisions compared with fifty acetabular
revisions). Thus, the rate of femoral revision for aseptic loosening
was nearly linear, or decreased with time, whereas the rate of acetabular
revision for aseptic loosening was low during the early follow-up
period but increased in the later period (Fig. 6).
The diagnosis leading to the primary total hip arthroplasty had an
effect on the survivorship of the prosthesis. Patients with rheumatoid
arthritis had significantly better twenty-five-year survivorship
free of revision for any reason (91.8% [95% confidence
interval, 83.0% to 95.2%]) compared with
the rest of the group (80.1% [95% confidence
interval, 77.4% to 82.4%]) (p = 0.047).
Patients with developmental dysplasia of the hip had significantly
poorer twenty-five-year survivorship free of revision for any reason
(65.9% [95% confidence interval, 53.9% to
74.7%]) compared with the rest of the group (82.4% [95% confidence
interval, 79.6% to 84.5%]) (p = 0.0006). Developmental
dysplasia of the hip had an adverse effect on survivorship free
of revision for any reason for both the acetabular component (74.3% [95% confidence
interval, 61.7% to 82.7%] compared with
88.3% [95% confidence interval, 85.8% to
90.2%] for the rest of the cohort; p = 0.0008)
and the femoral component (74.5% [95% confidence
interval, 63.7% to 82.6%] compared with
85.4% [95% confidence interval, 83.2% to
87.6%] for the rest of the cohort; p = 0.015).
Multivariate survivorship analysis was performed with an end point
of aseptic loosening of either the femoral or acetabular component,
the femoral component alone, or the acetabular component alone for
the risk factors of age (continuous variable), male gender, and
diagnosis (relative to osteoarthrosis). Highly significant associations
were identified (p < 0.001) for age and gender (Table IV). Patients
who were older at the time of arthroplasty were at a lower risk
of needing revision for aseptic loosening (the odds ratio was 0.5
for each ten-year increase in patient age), whereas men were at
a higher risk (odds ratio, 2.7). Compared with patients with osteoarthrosis,
patients with inflammatory arthritis were at a lower risk of needing
revision for aseptic loosening (odds ratio, 0.3). In contrast, patients
with dysplasia had a higher risk of needing revision for aseptic
acetabular loosening (odds ratio, 2.1) (Table IV).
To determine the factors that govern component survivorship over
a long period of time, we followed a large series of consecutive
primary total hip arthroplasties performed with the same type of
well-designed prosthesis. The inclusion of 2000 hips permitted us
to analyze the cohort by age, gender, and underlying diagnosis with
a sufficient number of hips in each category to provide meaningful
and statistically valid comparisons. By following the cohort for
a minimum of twenty-five years or until a revision or the death
of the patient, we were able to determine the lifetime risk of revision
for most of the patients and, in the smaller cohort still living
at the end of the study period, to determine survivorship of the
implant until the patients had reached a mean age of eighty years.
The high rate of follow-up, the large number of patients, and the
fact that most patients were followed until death or to within a mean
of 0.6 year of the time that they died improved the accuracy and
validity of the survivorship estimates.
Although not all patients were able to return to the treating institution
for follow-up assessment, the end points selected in this study
allowed determination of implant survival with use of a questionnaire
and telephone contact and are thus independent of clinical examination.
For these end points, the clinical information obtained from responses
to the questionnaire has been demonstrated to be significantly comparable
with that obtained through physician interaction20.
The mean age of the patient population in whom total hip arthroplasty
was performed between 1969 and 1971 is similar to that of the present-day
patient population treated with total hip arthroplasty; however,
other features of the present-day population may be different. With
the success of total hip arthroplasty, more patients are at the
very youngest and very oldest ends of the age spectrum. Furthermore,
at present, patients with multiple medical comorbidities and patients with
more complex and difficult acetabular and femoral anatomic problems
are considered candidates for arthroplasty, whereas they were not
at the beginning of this study.
During the first twenty years after total hip arthroplasty, patients
were found to have a lower rate of death than was predicted from
United States population-based survivorship data. This finding probably
in part represents a selection bias imposed by the decision to perform
surgery: because total hip arthroplasty was elective in most patients,
the sicker patients in the population would have been excluded from
the cohort on the basis of operative risk. It is also possible,
however, that patients who underwent hip replacement were able to
stay more active and hence to improve their overall health. Greater longevity
of patients who received elective total hip arthroplasty was previously
demonstrated in one other study21,
and we consider this finding to be important. As surgeons anticipate
the needed durability of joint prostheses, they need to be aware
that patients may outlive general population-based estimates. More
than twenty years after hip replacement, the overall survivorship
of our cohort of patients was poorer than that predicted for the
United States population. The relevance of this finding is not known;
however, the long-term rates of death may be increased because of
comorbidities or inactivity in the subgroup of patients who require
total hip arthroplasty at a young age (for example, patients with
rheumatoid arthritis).
As previously demonstrated at twenty years9,15,
survivorship free of revision for aseptic loosening continued to
be nearly linear over the first twenty-five years and the failure
rate did not decrease exponentially with time. As a cohort of patients
ages, the decreasing levels of activity and demands probably partially
offset the effects of the prolonged time in service on the prosthesis.
In addition, as patients age, the threshold for the performance
of another operation rises, since the decision to perform surgery
is made subjectively by the patient and surgeon on the basis of
the relative risks and benefits of the procedure.
Younger age has been recognized to have a negative effect on the
durability of a total hip prosthesis in most previous studies22-29, but not in all30. Previous studies have also demonstrated
that, in younger patients, cemented sockets fare worse than cemented
femoral implants31-34. Our results
confirm and extend those findings by providing specific survivorship
information for the acetabular and femoral implants for each decade
of life. This experience shows that the performance of cemented
acetabular and femoral components is dependent on the age of the
patient at the time of implantation but that the effect of age is
more profound on the performance of the acetabular component. Gender
has been shown to have an effect on the durability of total hip prostheses9,15,26, and this finding was confirmed
in the present study. Compared with women, men had a twofold increase
in the rate of implant failure from aseptic loosening.
The rates of survivorship free of revision for aseptic loosening were
virtually identical for the acetabular and femoral components. The
acetabular component was revised for aseptic loosening more frequently
than was the femoral component in young patients, but the femoral
component required revision for aseptic loosening more frequently
than the acetabular component in older patients. More femoral than
acetabular components were revised during the first fifteen years
of the study, whereas more acetabular than femoral components were
revised during the last ten years of the study.
Except for aseptic loosening, the most common reason for femoral
revision in this series was fracture of the femoral component. The
fractures are explained in part by the metallurgy of the first-generation
implants. However, the fracture rate was higher in this series than
in some others in which the same implants were used9. This difference is probably due
to a specific technique used at our institution during the time
of the study. Trochanteric wires were implanted by placing drill-holes
into the lateral aspect of the femur and allowing the metallic femoral
implant to deflect the drill. This technique frequently led to a
notch in the lateral aspect of the femoral stem, which served as
a stress riser through which many of the broken stems subsequently fractured.
Specific diagnoses were associated with higher rates of revision.
In our study as well as in others, patients with rheumatoid arthritis
did better than the rest of the cohort, probably because of a reduced
level of activity27,34-37. Conversely,
patients with developmental dysplasia of the hip exhibited poorer
implant survivorship, a finding that may be explained in part by
abnormalities in acetabular and femoral bone geometry in dysplastic
hips.
By rigorously defining the patient populations at risk for revision
and by quantitating the risks to the acetabular and femoral components
independently in each patient population, the information in this
paper helps to identify patients at risk for failure of Charnley
total hip arthroplasty. This information also helps to identify
patient populations that may benefit most from newer technologies
intended to improve on the remarkable results of Charnley total
hip arthroplasty.