Total joint replacement surgery is widely recognized as one
of the most cost-effective interventions in medicine today.
The impact that these procedures have on quality of life has been well
documented1. New implants are
designed and manufactured with the expectation that they will be
equivalent or superior to existing products. Regardless
of how much testing is done prior to product release, in reality
our patients are the final testing grounds for new devices. Only
in our patients do all of the variables that affect outcome—that
is, design issues, material issues, manufacturing issues, patient-related
factors, and surgical technique—finally come into play.
As a result, premature failures can and do occur.
We then have to ask ourselves two questions. First, is the magnitude
of the problem of premature failure serious enough to warrant the
expenditure of resources? Second, if so, what is the optimal way
to disseminate information in order to have a timely and positive
effect on the problem?
To answer the first question, we have only to look at the revision burden
in the United States. Review of Medicare data shows that in 1996
and 1997 revisions accounted for approximately 17% to 18% of
all total hip replacement procedures2.
Despite improvements in surgical technique and in the designs of
prostheses, revision rates do not seem to be declining over time.
In 1993 the revision burden was approximately 16.8%, and
it increased to approximately 17.3% in 19972. There are numerous examples of implants
and technologies that did not perform as expected for a variety
of reasons. Examples include carbon-fiber-reinforced polyethylene,
Hylamer, heat-pressed polyethylene, and metal-backed patellae,
to name just a few. It is also important to remember that the data
on revisions reflect procedures performed in Medicare beneficiaries,
who in general are older than sixty-five. It is likely that the
revision burden in our younger patients is even higher; however,
there is currently no accurate way to document this.
If one accepts that our revision burden is too high and that
this burden is related at least in part to products that do not
perform as intended, we then have to answer the second question:
how do we disseminate information in the orthopaedic community? There
are three potential basic methodologies for studying implant performance.
These include randomized clinical trials, meta-analyses,
and retrospective case series. Prospective, randomized clinical
trials remain the gold standard for evaluation of medical interventions.
However, when it comes to joint replacement procedures, these trials
simply are not practical. A recent issue of Orthopedics Today lists
more than twelve manufacturers of hip implants, with more than fifty
femoral components alone. Performance of a randomized clinical trial
to examine the relative effectiveness of currently available systems
in the United States would be not only cost-prohibitive but also
of limited value. As an example, Rorabeck et al. performed a randomized
clinical trial of hip replacements3.
They compared the results of arthroplasties performed with and without
cement and demonstrated the beneficial aspects of hip replacement.
However, their data currently are of limited value as they relate
to the performance of a specific implant and, since that study,
the manufacturer has made changes to virtually every aspect
of the system. The two types of metal-backed acetabular
components that were used have both been redesigned, the sterilization
method for the polyethylene has changed, and the ion-implanted titanium
femoral heads that were used were replaced with cobalt-chromium heads.
Only the tapered cementless titanium-alloy femoral component has
remained unchanged.
Meta-analysis is another technique with which to analyze
a given technology or intervention. With meta-analysis,
the literature is critically reviewed and studies that meet specific
inclusion criteria are included with the hope that the sample size,
and thus the validity of the conclusions, can be increased.
One common inclusion criterion for meta-analysis
is randomization. Since randomized clinical trials evaluating the
relative efficacy of hip and knee replacement designs have not
been, and will not be, carried out in large numbers, meta-analysis
is not going to be of major help in this arena.
What we are then left with are retrospective case series. There is
no doubt that these are of value, bringing to light both positive
and negative factors that influence outcome. When it comes to implant
performance, however, it is important to realize that there are
limitations to retrospective case series, which most often represent
the experience of a single surgeon or center. These surgeons and
centers are often recognized subspecialized experts in the field
of joint replacement surgery, and thus their experiences do not
necessarily reflect what occurs in the general orthopaedic community, where
the majority of procedures are done by surgeons who perform less
than fifteen hip replacements per year. In addition, the surgery
may be done by the innovator, which introduces potential bias. Finally,
and perhaps most importantly for the orthopaedic community and our patients,
is the fact that retrospective case series are not reported in a
timely enough fashion to serve as an early warning system for technologies
with problems.
Hylamer polyethylene is a good example of a material with problems.
Hylamer was a form of polyethylene in which the crystallinity of
the material and the size of the crystalline regions were increased.
The objective was to increase the strength and fatigue resistance
of the polymer by creating larger crystallites and to increase the
proportion of the polymer that was in the stronger crystalline phase.
Laboratory testing of the altered polyethylene demonstrated an 8% lower mean
wear rate and fatigue-crack-growth tests indicated a reduction
in the crack-growth rate, which were hypothesized to increase its
fatigue life4. Despite the in
vitro results, however, clinical performance was not as expected.
In retrospect, accelerated polyethylene wear was noted as early
as one to three years after surgery. The material was released in
late 1990, potential problems were first reported in the literature
in 19965, and marketing was discontinued
in 1997. Press-fit sockets with a hydroxyapatite coating applied
to a relatively smooth surface is another example
of a situation in which information concerning potential problems could
have been disseminated in a more timely manner. Hydroxyapatite-coated
femoral components have had an excellent track record; however, hydroxyapatite
on a macrotextured acetabular component did not fare as
well. In a retrospective series6,
the failure rate was 22% at five to nine years
after surgery. The implant was released to the market in 1991; the
report on the magnitude of the problem did not come out until 1998,
long after the product was removed from the market.
Recognizing that a lack of timely dissemination of information
is currently a problem, we have to ask how it can be done better. One
answer is a national joint replacement registry, the goal of which
is threefold. First, a national joint replacement registry
must accurately define the epidemiology of joint replacement surgery
in a particular patient population. This goal seems trivial, but
in fact it is not. To understand the importance of reported failures,
we need to know the number of prostheses that are being implanted.
We currently have only estimates about what implant technologies
are being used in which patients and about the true revision rate.
As documented in Sweden, a national registry can provide accurate
data regarding which prostheses are being implanted and which prostheses
are being removed7.
The second goal of a national registry is to provide timely information
for the orthopaedic community on the outcomes of joint replacement.
The end point obviously has to be clear and unequivocal. In a national
registry, the end point for failure is revision surgery. An implant
that is being revised has by definition failed; that is not controversial. A
national registry may be less accurate in documenting the cause
of failure, but it provides information as to where potential problems
might be and directs future analysis toward problem areas. Since
information is gathered in real time and can be analyzed on an ongoing
basis, it has the potential to bring to our attention problems that
may be occurring long before they would be reported to the orthopaedic
community by traditional methods. This would obviously benefit all
of the stakeholders in this arena, including surgeons, manufacturers,
payers, and most importantly patients. From this standpoint, a national
registry can serve as an effective postmarket surveillance tool,
a responsibility that technically lies with the United
States Food and Drug Administration. However, surveillance by the
Food and Drug Administration rarely extends beyond three years, which
in the world of joint replacement is inadequate.
The third general goal of a national joint replacement registry
is to identify risk factors for a poor outcome and, through continuous
feedback to participating centers and surgeons, to improve outcomes.
Again, to determine if this is a realistic goal, we can turn to
the Swedish Hip Registry, which was established in 1979. The stated
goals of the Swedish Hip Registry were to describe the outcomes
of primary hip replacement and to provide information regarding serious
complications. The hypothesis of its founders was that, by sharing
information on outcomes, individual hospitals would be able to compare
their results with national averages. Units with poor
outcomes would improve by following the best example. This is now
referred to as evidence-based medicine. The success of
this registry in improving the quality of hip replacement in Sweden
has been well documented8,9. The
registry identified specific technical factors that are associated
with improved outcomes of femoral replacement with cement,
including porosity reduction and plugging of the femoral canal.
Once identified, this information was provided to the
orthopaedic community, leading to a change in practice. The prevalence
of use of a distal femoral plug increased from approximately 30% in
1982 to >95% in 1987. Over the past two decades,
the percentage of revision hip replacements has been cut
in half, and currently it is approximately 8% in Sweden compared
with 18% in the United States.
One of the concerns of the orthopaedic community in the United States
is that a national registry will drive patients to major centers.
This in fact has not been the case in Sweden. The percentage of
primary hip replacements done in rural hospitals in Sweden increased
from <20% in 1981 to approximately 40% in
1999. At the same time, outcomes improved substantially at the more
poorly performing centers.
Other attempts have been made in the United States to ensure
the safety of medical devices entering the market. The Safe Medical Devices
Act of 1990 was enacted by Congress on November 28, 1990. The intent
of this law was to ensure that medical devices are safe and effective
by monitoring so that the Food and Drug Administration learns about
serious problems and quickly removes defective devices. Technically,
this act requires that facilities in which medical devices are to
be used report to the Food and Drug Administration problems that "contribute
to the death, serious illness or serious injury of the
patient." Under this act, orthopaedic surgeons are required
to identify device problems and then supply information identifying
the device and the nature of the problem.
In June 1993, the Food and Drug Administration introduced MedWatch.
This program had three basic components: (1) medical device reporting
(MDR) by manufacturers, (2) an effort that encourages reporting
by doctors and other health-care professionals, and (3) a user-facility
reporting requirement as mandated by the Safe Medical Devices Act. This
regulation defines an injury as any event requiring medical or surgical
intervention to preclude impairment of a body structure
or function. This definition includes all revisions of
hip and knee replacements, but obviously complete reporting is not occurring
as only a small percentage of revision arthroplasties are actually
being reported. It should be noted that the Food and Drug Administration
does not have adequate staff to investigate the few reports that
they currently receive, and certainly it would not be able to handle properly
reports on the thousands of hip and knee revisions that occur each
year.
Despite its failure to achieve the stated goals, MedWatch includes a
critical component relating to confidentiality. Under the Freedom
of Information Act any MedWatch report is subject to public disclosure,
but only after the report has been purged of confidential details,
including the names of patients and health-care professionals. The
concept of confidentiality is critical for the success of any joint
replacement registry. Participants must be assured that the registry
will not be used in a punitive way. In a litigious society such
as ours, legislation may be required to further protect the integrity
of a national joint replacement registry to ensure that the data
are used as intended—to serve as an early warning system
for premature device failure and to improve outcomes for our patients.
Currently, there are joint replacement registries in Sweden,
Finland, Norway, Denmark, Hungary, and Canada, and registries are
being evaluated in New Zealand, Australia, and England.
In England, the development of a national registry was mandated
by the government in response to a recent problem with
a hip replacement system that had an exceedingly high failure rate
and was subsequently recalled. The registries in these countries
are financed by the government, as one would likely be in the United
States.
The establishment of a national joint replacement registry with federal
support requires consideration of both practical and political realities.
Practical challenges include an examination of costs, construction
and management of the data, governance of the database, and access
to the data. Political challenges involve review of applicable federal
laws as they relate to confidentiality and privacy rules as well
as to discoverability.
While the goals and objectives of a national joint replacement
registry seem consistent with the mission of several federal
agencies that are concerned about providing good-quality patient care,
cost is an obvious issue. Although a national registry would be
expected to have a high initial operating cost, the potential cost-savings
in the long run could be substantial. Barrack et al.10 estimated that the cost of a routine
revision total hip replacement was approximately $18,000.
For a complex revision, the cost goes up to as high as $30,000.
If it is estimated that 10% of revisions are complex, the average
cost of all revisions in the United States would be approximately $19,200.
As 200,000 hip arthroplasties are performed annually in the United States
and the revision burden is 18%, 36,000 revisions are performed
annually. If the number of revision hip replacements could be decreased
by 5% (1800), an estimated cost-savings of more
than thirty million dollars could be realized.
The evidence that a national joint replacement registry would benefit
our patients is strong. These registries have the capacity to define
the epidemiology of joint replacement, to serve as an early warning
system for premature device failure, and to identify factors that
are associated with both good and poor outcomes. On the basis of
evidence, practices can be driven toward following the best example through
feedback to hospitals and surgeons. The time has come to initiate
such an effort in the United States.