There is strong evidence linking excessive body weight to degenerative
joint disease of the
knee1-7.
Consequently, a large proportion of patients who undergo total knee
arthroplasty are obese or morbidly
obese8-13.
Many authors believe that a high body weight will lead to a less-than-optimal
outcome of total knee arthroplasty as a result of increased stress across the
components and increased load on the surrounding
bone9,14,15.
Several studies have implicated excessive weight as a negative predictor of
success of total knee
arthroplasty8-10,12,14,16-19,
whereas others have indicated that obesity is not a negative predictor of knee
arthroplasty
outcomes11,13,20-27.
Winiarsky et al.19
compared the outcomes of fifty total knee arthroplasties with cement in forty
morbidly obese patients (mean body mass index, 44) with the outcomes of 1768
similar procedures in nonmorbidly obese patients (mean body mass index,
28)19. At
approximately five years after the operation, the morbidly obese patients had
lower objective and functional Knee Society scores as well as higher rates of
wound-healing problems and other perioperative complications. In contrast,
Spicer et al.26
found similar ten-year prosthetic survival rates after 385 arthroplasties in
326 obese patients (body mass index, 30) and 425 arthroplasties in 371
nonobese patients.
The purpose of the present study was to compare the clinical and
radiographic results of total knee arthroplasties performed in obese patients
with those of arthroplasties performed with the same prosthesis in nonobese
patients.
Between September 1, 1991, and December 31, 1996, 772 total knee
arthroplasties were performed with the Duracon total knee prosthesis
(Stryker-Howmedica-Osteonics, Allendale, New Jersey). After institutional
review board approval was obtained for the study, seventy-eight knees in
sixty-eight patients who were obese (defined as a body mass index of =30)
at the time of the surgery and who had been followed for a minimum of five
years were identified from a database of all patients. An additional eight
knees in eight obese patients were identified, but four of those patients had
died and four had been lost to followup and hence were excluded from the
study. All of those eight knees were functioning well at the time of the
latest follow-up (mean duration, four years; range, two to five years).
The body mass index equals a person's weight in kilograms divided by his or
her height in meters squared and correlates well with total body
fat28. Obesity is
defined as a body mass index of =30 kg/m2, and morbid obesity is
defined as a body mass index =40 kg/m2 (overweight is defined as
a body mass index =27
kg/m2)28-30.
With use of a list of the obese patients ordered by duration of follow-up
(with the patient with the longest duration listed first), each patient was
directly matched with the first nonobese, control patient (with a body mass
index of <30), hand-selected from a computerized database, who was
appropriately matched with respect to the preoperative diagnosis, age at
surgery (within ten years), duration of follow-up (within two years), and
whether he or she had had a unilateral or bilateral arthroplasty. An attempt
was made to match the patients by gender, but this was unsuccessful because
there were too many obese women. A separate statistical analysis comparing the
results of men and women in the entire group of patients (obese and nonobese)
as well as in the individual groups revealed no significant difference in
outcome between the two sexes. The authors were blinded to the outcomes at the
time of the match. All patients had been followed for a minimum of five years.
Seventy-eight knees in sixty-eight nonobese patients were included in the
study. Demographic data for each group are summarized in
Table I. Follow-up data were
obtained by means of a blinded, retrospective review of the computerized
database, charts, and radiographs as well as with telephone conversations. The
mean duration of follow-up was eighty months (range, sixty to 123 months) for
the obese group and eighty-three months (range, sixty to 123 months) for the
nonobese group.
The operative technique was the same in all patients. All tibial and
patellar components were cemented, and all patellae were resurfaced. A
cementless femoral component (a so-called hybrid total knee replacement) was
used when there was qualitatively good bone stock and excellent bone cuts had
been made. All of the components were cemented in thirty-six knees (46%) in
the obese group and forty knees (51%) in the nonobese group, and a hybrid
replacement was performed in forty-two knees (54%) in the obese group and
thirty-eight knees (49%) in the nonobese group (p = 0.71).
The postoperative activity level of all patients was assessed at the time
of the latest follow-up but was utilized to describe the entire activity level
throughout the postoperative period. A description of the scoring system can
be found in Table II.
All patients were evaluated preoperatively and postoperatively with the
Knee Society objective rating
scale31 at the time
of the latest follow-up. Ratings of excellent (90 to 100 points) and good (80
to 89 points) were considered to indicate success, whereas ratings of fair (70
to 79 points) and poor (less than 70 points) were considered to indicate
failure. Additionally, knees revised or in need of revision because of aseptic
loosening, infection, or polyethylene wear or that showed signs of
radiographic loosening were considered failures.
Patients were also evaluated for the presence of comorbidities,
perioperative complications, wound-healing complications, and patellofemoral
symptoms. The status of the patellofemoral joint at the time of the latest
follow-up was graded as described by Stern and
Insall12, with
grade 0 indicating no symptoms referable to the knee; grade I, mild pain when
climbing stairs; and grade II, moderate-to-severe pain when rising from a
chair or limiting stair-climbing.
Initial and subsequent postoperative radiographs were examined for changes
or progression of abnormalities. Measured parameters included zonal interface
lucencies and preoperative and postoperative
alignment31.
The obese group was divided into nonmorbidly obese and morbidly obese
subgroups to determine the effects of increasing obesity on outcome. The mean
body mass index (and standard deviation) for the obese patients was 35.3
± 4.2 (range, 30.0 to 47.0). Eleven patients (with twelve knees) in the
obese group were morbidly obese, with a mean body mass index of 43.2 ±
2.3 (range, 40.0 to 47.0). The mean body mass index for the nonobese group was
26.2 ± 2.5 (range, 17.6 to 29.8)
(Table I).
Data Analysis
The clinical and radiographic outcomes in the obese group were compared
with those in the nonobese group. In addition, the outcomes in the nonmorbidly
obese and morbidly obese subgroups were stratified and independently compared
with those in the nonobese group.
Parametric and nonparametric statistical analysis, with use of the Computer
Program for Epidemiological Analysis (PEPI) software package (version 2.03;
USD, Stone Mountain, Georgia) and Statistics Calculator (version 5.0; Statpac,
Minneapolis, Minnesota), was employed to compare the groups. The significance
of differences between groups was determined with the Pearson chi-square test
(with use of the Yates correction), Wilcoxon-Mann-Whitney test, Fisher exact
test, likelihood ratios, and Student t test. Kaplan-Meier survivorship curves
were generated to analyze differences in time to prosthetic failure between
the obese and nonobese patients as well as among the morbidly obese,
nonmorbidly obese, and nonobese patients. A p value of <0.05 was considered
significant.
Sixty-nine (88%) of the seventy-eight knees in the obese group were
considered to have a successful outcome at the time of the latest follow-up.
Four knees in four obese patients required revision and thus were considered
failures, and five knees in five obese patients were considered failures
because of a fair or poor Knee Society objective score. In comparison,
seventy-seven (99%) of the seventy-eight knee replacements in the nonobese
group were successful, and there were no revisions in that group. At the time
of the latest follow-up, there was a significant difference in the success
rates between the knees in the obese group and those in the nonobese group (p
= 0.02) (Table II).
Kaplan-Meier survivorship analysis revealed similar rates of prosthetic
survival between the obese and nonobese groups until between sixty and eighty
months, when the decreased survival rate in the obese group became apparent.
At eighty months, the obese group had a6n 87.7% ± 5.4% (standard error)
chance of prosthetic survival (95% confidence interval, 72.1% to 95.1%), with
a reoperation, clinical failure, and radiographic failure as the end points,
and the nonobese group had a 98.7% ± 1.9% chance of prosthetic survival
(95% confidence interval, 87.9% to 99.9%)
(Fig. 1). The curves were not
continued beyond eighty months because the confidence intervals became quite
large at that point.
Stratification of knees in the obese group into morbidly and nonmorbidly
obese subgroups revealed a lower success rate when those subgroups were
compared with the nonobese group. Ten of the twelve knee replacements in the
morbidly obese subgroup were successful at the time of the latest follow-up,
whereas fifty-nine (89%) of the sixty-six knees in the nonmorbidly obese
subgroup and seventy-seven (99%) of the seventy-eight knees in the nonobese
group were successful. The rate of success in the nonobese patients was
significantly higher than the rate in the nonmorbidly obese patients (p =
0.02).
The survivorship curves revealed that, at eighty months, there was a 91.7%
± 11.8% (standard error) chance of prosthetic survival (95% confidence
interval, 47.9% to 99.2%) in the morbidly obese subgroup
(Fig. 2), an 83.6% ±
8.7% chance (95% confidence interval, 58.7% to 94.8%) in the nonmorbidly obese
group, and a 98.7% ± 1.9% chance (95% confidence interval, 87.9% to
99.9%) in the nonobese group.
There were no significant differences in the mean preoperative Knee Society
objective scores among the groups (Table
II), but the mean preoperative functional scores differed
significantly between the morbidly obese and nonobese groups (p < 0.02).
There was a significant difference in the mean postoperative Knee Society
objective scores between the obese and nonobese groups
(Table II) (p < 0.05). There
was also a significant difference in the mean changes in the objective Knee
Society score (postoperative minus preoperative score) between those groups (p
< 0.05), although this may be slightly skewed as the obese group started
with a 2-point higher mean objective score. The difference in the mean
postoperative Knee Society scores between the morbidly obese and nonobese
groups was significant as well (p = 0.04). There were no differences in
revision and infection rates between the nonobese and obese groups. However,
the difference in revision rates became significant when the obese group was
stratified into morbidly and nonmorbidly obese subgroups and the morbidly
obese group was compared with the nonobese group (p = 0.02).
The clinical function of the patellofemoral articulation at the time of the
latest follow-up in the obese group was grade 0 in forty-five knees (58%),
grade I in twenty-three knees (29%), and grade II in ten knees (13%). In the
nonobese group, forty knees (51%) had grade-0 function; twenty-eight knees
(36%), grade-I; and ten knees (13%), grade-II
(Table II). Statistical
analysis revealed no difference in patellofemoral scores among the subgroups,
with the numbers available. The rates of patellar complications were also
similar among the groups. One morbidly obese patient required a patellar
revision at 103 months postoperatively, and one nonobese patient required
repair of a traumatic rupture of a patellar ligament.
There was no difference in activity levels or the rate of perioperative
complications between the obese and nonobese groups
(Table II). No patient was
bedridden (activity level of 0), and only two patients (one obese and one
nonobese patient) had an activity level of 5.
The obese and nonobese patients had similar rates of hypertension,
clinically relevant coronary artery disease, and cancer. The obese group had a
higher prevalence of diabetes mellitus (eight of sixty-eight patients compared
with zero of sixty-eight patients in the nonobese group) (p = 0.02).
The radiographic results are presented in the Appendix. No knee that had a
good or excellent clinical outcome in either group had impending radiographic
failure. There were similar rates of nonprogressive radiolucencies in the
three groups, and no knee in any group showed progressive radiolucencies. The
postoperative knee-alignment measures were virtually identical among the
groups (see Appendix).
Complications
The rates of perioperative complications, including problems with primary
wound-healing, were similar in all of the groups. Nine knees in nine obese
patients were classified as failures (Table
II). Four were considered to be failures because they required
revision, whereas the other five were failures because of a fair or poor Knee
Society objective score.
Perioperative complications: Two obese patients (two knees) had
perioperative complications (a deep vein thrombosis and a wound dehiscence in
one patient and a footdrop in the other), whereas none did in the nonobese
group. At the time of the latest follow-up, both patients with perioperative
complications had a successful Knee Society objective score.
Postoperative complications: Of the four obese patients who
required revision, one underwent the reoperation at thirty-six months because
of unremitting pain; approximately twelve months later, he underwent a
neurectomy to treat continued pain. At forty-four months after the revision,
the Knee Society objective score for this patient was poor (65 points) because
he continued to have pain at rest. The three other revisions that were done in
obese patients were performed because of loosening of a tibial component at
fifty-five months after the index arthroplasty, to exchange the polyethylene
insert and accomplish a lateral patellar release at 103 months, and to treat a
chronic infection at eighty months. All three of those patients eventually had
a successful outcome, with Knee Society objective scores of =90 points. In
addition, five knees in five other obese patients were considered failures
because of persistent pain that led to a fair or poor Knee Society objective
score. None of those knees showed radiographic signs of impending failure, and
no additional surgical treatment was undertaken. In the nonobese group, one
knee was considered a failure because of persistent pain, which led to a poor
Knee Society objective score of 62 points at sixty months postoperatively.
We undertook this study to evaluate the effects of obesity on one type of
total knee implant that has been highly successful in the general
population11. The
results suggest that obesity has a negative effect on the outcome of total
knee replacement. At a mean of approximately seven years postoperatively, the
obese group had a significantly lower rate of success than did the nonobese
group. Stratification of the obese group into nonmorbidly obese and morbidly
obese subgroups revealed significant differences in revision rates and
postoperative objective and functional scores when those subgroups were
compared with the nonobese group. Kaplan-Meier survivorship analysis revealed
similar rates of prosthetic survival between the two cohorts until between
sixty and eighty months, at which time the decrease in the survivorship in the
obese group became apparent (Fig.
1). The survival analysis of the morbidly and nonmorbidly obese
groups showed a similar pattern of failure
(Fig. 2).
Several reports have described the adverse effects of obesity on the
outcomes of total knee
arthroplasty8-10,12,14,16-19.
Stern and Insall12
evaluated the results of 257 knee arthroplasties in 182 patients at a mean of
four years (range, two to seven years) postoperatively and found a higher
prevalence of patellofemoral symptoms in obese patients. Thirty percent
(thirteen) of forty-three moderately to severely obese patients reported
patellofemoral symptoms, whereas 14% (thirty-one) of 214 underweight to mildly
obese patients did so (p < 0.03). In the present study, obese and nonobese
patients had similar rates and severities of patellofemoral symptoms and
patellofemoral joint-related complications. This finding is in contrast to the
higher prevalence of patellofemoral symptoms experienced by obese patients in
the studies by Stern and
Insall12, Pritchett
and Bortel18, and
Griffin et al.14.
This difference may be explained in part by the fact that the prosthesis used
in this study had features that were specifically intended to minimize
patellofemoral
complications32,33.
Other authors have reported similar overall results in obese and nonobese
patients. In the study by Spicer et
al.26, the ten-year
prosthetic survival rates were similar for obese and nonobese patients, but
obese patients had a higher rate of focal osteolysis on radiographic
analysis29. Spicer
et al. reported slightly lower postoperative Knee Society scores in the obese
group but found the absolute improvement in Knee Society scores to be
independent of body mass index. This observation is in contradistinction to
the findings in the present study, in which obese patients had a smaller mean
improvement in the Knee Society score (31 compared with 37 points, p = 0.01;
Table II). In another report,
Mont et al. compared the results of fifty cementless knee arthroplasties in
obese patients (body mass index, >30) with those of fifty cementless total
knee arthroplasties in a directly matched nonobese group and found no
significant difference in Knee Society scores (p = 0.75), bead-shedding, or
progressive radiolucencies at a mean of sixty-five months (range, twenty-four
to 144 months)
postoperatively11.
Other
studies13,20-25
have revealed similar short-term outcomes between obese and nonobese patients.
We believe that the results of the present study differ from those of studies
showing similar outcomes between obese and nonobese patients because of the
longer follow-up of directly matched patients in our study.
We also found no differences in activity levels or perioperative
complication rates between the groups. In addition, the radiographic results
were similar among the groups, which is in contrast to the findings of Spicer
et al., who reported a focal osteolysis rate in obese patients that was five
times greater than that in nonobese
patients26, and the
observations by Griffin et al., who found a higher prevalence of
nonprogressive radiolucent lines in obese
patients14.
Some of the findings in our study are limited by the small numbers of
patients available. In addition, the patients could not be directly matched
for gender because of the disproportionately higher number of women in the
obese group. Nevertheless, the two groups were matched for age, diagnosis,
duration of follow-up, and unilaterality or bilaterality, and they had similar
preoperative Knee Society objective and functional scores.
In summary, the key findings in the present study are that obese patients
had a lower success rate and lower postoperative Knee Society scores and,
therefore, lower satisfaction levels. Overall, it appears that obesity has a
negative impact on the results of total knee arthroplasty, with morbid obesity
having an even more dramatic negative effect with regard to revision rates
after total knee arthroplasty.
Tables describing findings with regard to radiolucency and alignment of the
implants are available with the electronic versions of this article, on our
web site at
(go to the article citation and click on "Supplementary Material")
and on our quarterly CD-ROM (call our subscription department, at
781-449-9780, to order the CD-ROM).