Between January 1992 and December 1993, all patients undergoing
total knee arthroplasty by one of the authors at one of three university-affiliated
teaching hospitals (Tulane University Hospital, Veterans Affairs
Medical Center, New Orleans, and Veterans Affairs Medical Center,
Alexandria, Louisiana) were considered for this study. The indication
for the operation was osteoarthritis that was severe enough to warrant
total knee arthroplasty after an adequate trial of nonoperative
therapy. Exclusion criteria included a previous tibial osteotomy
or an operation involving the extensor mechanism, a history of septic
arthritis or osteomyelitis, severe medical disability limiting the
ability to walk, disabling joint disease in another lower-extremity
joint, inflammatory arthropathy, and severe deformity. Severe deformity
was defined as varus, valgus, or a flexion contracture of >15°.
During the enrollment period, 137 patients were evaluated for possible
inclusion in the study. Forty patients (fifty-four knees) met one
of the exclusion criteria and were not offered enrollment. Ninety-seven
patients who met the inclusion criteria were offered enrollment
in the study, and eighty-nine patients (121 knees) agreed to participate.
This represented 92% of all eligible patients. Three early
postoperative deaths left 118 knees in eighty-six patients for analysis
in the previous report12.
Of these eighty-six patients, eight died, three had a severe stroke
that prevented further follow-up, and two had failure of the total
knee replacement due to late deep infection; this left a total of
seventy-three patients (102 knees) available for follow-up. Sixty-seven
(92%) of the seventy-three patients and ninety-three (91%)
of the 102 knees were followed. Evaluations consisted of determination
of a Knee Society clinical score and completion of a patient questionnaire
including a series of questions relating to patellofemoral symptoms
by sixty-four patients, and completion of a telephone questionnaire
by three patients. The same nurse performed all evaluations in a
double-blind fashion, independent of any physician involvement.
The institutional review boards of all of the involved hospitals
approved the study protocol, including the consent forms. Patients
agreed to be blinded to which procedure they had received to reduce
bias in their responses to follow-up questionnaires. All patients
received the same posterior-cruciate-sparing prosthetic components
(Miller-Galante II; Zimmer, Warsaw, Indiana), and all operations
were performed by, or under the direct supervision of, one of the authors.
Randomization was accomplished by opening a randomly selected
envelope in the operating room after all femoral and tibial cuts
had been made and immediately prior to patellar preparation. When
a patient was to have a bilateral arthroplasty, the first knee received
the treatment indicated by the envelope and the contralateral knee
received the other treatment. All procedures were performed with
a uniform approach and technique, as previously described12. If the patella subluxated during
passive range-of-motion testing (the so-called no-thumbs test13), a lateral retinacular release
was performed. When resurfacing was performed, a cemented, three-peg,
all-polyethylene component was utilized. Calipers were used to ensure
restoration of the preoperative thickness within 1 mm12. When resurfacing was not performed,
patelloplasty was carried out. This included removal of osteophytes,
smoothing of fibrillated cartilage, and drilling of eburnated bone.
Study evaluations were conducted preoperatively and at follow-up
visits conducted at six months, at twelve months, and annually thereafter.
All physical examinations were performed by the same trained nurse-clinician,
and radiographs were repeated. At all preoperative and postoperative
visits, a Knee Society clinical score was recorded. These scores
were obtained for each knee in a double-blind fashion.
To explore the relationship between patient weight and the clinical
result, the percentage by which the patient’s weight exceeded
the maximum allowed for a "large frame" on the Metropolitan
Life Insurance Company (1983) weight-for-height tables14 was calculated. Analysis of covariance
was performed to assess the relationship of the Knee Society clinical
score to obesity and the presence or absence of patellar resurfacing, which
were used as independent variables. The preoperative score was used
as a covariant, to adjust for any preoperative differences among
the subjects.
Patient satisfaction was assessed with use of questionnaires. All
patients completed detailed questionnaires preoperatively and at
follow-up visits. Of particular interest in this study population
were items related to patellofemoral joint function. The instruments
included visual analog scales for pain and function and assessments
of patient satisfaction. Questionnaire items asked to what degree
knee symptoms interfered with activities of daily living, work,
and recreation. The ability to climb stairs, rise from a chair,
and exit an automobile were specifically assessed, as was the presence
or absence of anterior knee pain as a means of identifying symptoms
related to the patella.
Of the sixty-seven patients, sixty-four (eighty-eight knees) returned
for reevaluation and three (five knees) were evaluated by telephone
questionnaire. A Knee Society clinical score was not determined
for these five knees. Thus, although a total of forty-seven knees
with the patella resurfaced and forty-six without patellar resurfacing
were analyzed, Knee Society clinical scores were obtained for forty-four
knees in each group. Five patients (five knees) who had originally
had the arthroplasty without resurfacing had undergone a revision to
resurface the patella. Their scores and questionnaires were analyzedwith
those of the nonresurfaced group.
Descriptive statistics (averages, medians, and so on) were used
to summarize the data. Because some individuals contributed scores
for both knees, generalized estimating equation regression methods
were used to test for significance while controlling for the dependence
between scores. Both categorical and continuous outcomes were analyzed
with use of generalized estimating equation regression methods,
and preoperative scores were controlled for when possible. Significance
was defined as p < 0.05. All analyses were conducted with
use of SAS software (version 8; SAS Institute, Cary, North Carolina).
Complications
There were no acute infections (within twenty-four months postoperatively).
Two late hematogenous infections necessitated a reoperation, and
the two patients were excluded from the study. There were no revisions
for aseptic component loosening.
Seven (12%) of the original sixty knees with a nonresurfaced patella
were subsequently resurfaced, all because of anterior knee pain.
Six of the resurfacings had been done by the two to four-year follow-up
examination, and one had been done by the five to seven-year follow-up
examination. Two of the patients with a subsequent resurfacing were
not reevaluated for the present study: one could not be contacted,
and the other had had a major stroke, preventing further evaluation. The
seven revisions were accomplished without operative complications,
and there was an initial decrease in the anterior pain in six of
the seven knees. Prior to the patellar resurfacing, the anterior
pain was rated as 8 on a scale of 1 to 10 in all seven knees. The
average pain rating after the resurfacing was 2.3 at the two to
four-year follow-up examination, but the rating had deteriorated
in four of the five reevaluated patients (average, rating 7.4) five
to seven years following the original total knee arthroplasty. The
average duration of follow-up after the subsequent patellar resurfacing
was 36.8 months (range, twelve to forty-eight months).
Nine knees with a resurfaced patella were painful anteriorly
at the time of final follow-up, with an average pain rating of 7.2. There
was no sign of patellar subluxation or maltracking, and these knees
were treated nonoperatively. None of the resurfaced knees underwent
revision for pain, but eight of the nine knees with a resurfaced
patella that were painful anteriorly were rated as 4. There were
no patellar fractures or dislocations and no episodes of patellar
component loosening among the resurfaced patellae.
Clinical Results
Knee Society Clinical Score
The average preoperative Knee Society clinical score was 88.5
points (median, 89 points; range, 0 to 142 points) (Table I). The average
score for pain was 44.5 points (median, 44 points; range, 0 to 74
points), and the average score for function was 41.8 points (median,
45 points; range, -20 to 60 points). The average Knee Society clinical
score at the time of final follow-up was 165.4 points (median, 185
points; range, 47 to 200 points). The average score for pain was
88.4 points (median, 93 points; range, 41 to 100 points), and the
average score for function was 77.1 points (median, 92.5 points; range,
-20 to 100 points). The average Knee Society clinical score for
the resurfaced knees was 87.4 points (median, 89 points; range,
33 to 132 points) preoperatively and 161.6 points (median, 179.5
points; range, 47 to 200 points) postoperatively. The average Knee
Society clinical score for the nonresurfaced knees was 89.6 points
(median, 89 points; range, 0 to 130 points) preoperatively and 169.1
points (median, 191.5 points; range, 52 to 200 points) postoperatively.
There was no significant difference between the resurfaced and nonresurfaced
knees with respect to the overall Knee Society clinical score (p = 0.36)
or the pain (p = 0.77) or function (p = 0.16) subscore.
There was also no significant difference in the Knee Society scores
or subscores between obese and nonobese patients or among patients
with different grades of chondromalacia.
The average range of motion at the time of final follow-up was
101° (median, 105°; range, 65° to 130°). This was a decrease from
the average of 111° (median, 105°; range, 80° to 140°) at the time
of the two to four-year follow-up. There was no significant difference
in the average range of motion between the knees with a resurfaced
patella and those with a nonresurfaced patella (p = 0.63).
Patient Satisfaction and Questions Regarding
Patellofemoral Function
In all groups, the scores related to patellofemoral function improved
from 2 or 3 (on a scale of 0 ["impossible"] to
10 ["no problem"]) preoperatively
to 8 or 9 postoperatively.
Patients were asked: "Are you satisfied with the results
of your operation?" Overall, the answer was "yes" for
eighty-six (92%) of the ninety-three knees. Forty-three
(93%) of the forty-six arthroplasties without resurfacing
of the patella and forty-three (91%) of the forty-seven
with resurfacing of the patella resulted in patient satisfaction;
there was no significant difference between these results (generalized
estimating equations, p = 0.67).
There was also no difference in the patient’s postoperative ability
to get in and out of an automobile (p = 0.36) or a chair (p = 0.94)
or to negotiate stairs (p = 0.99) between those who had
resurfacing and those who did not. The ratings of the degrees of
difficulty were virtually identical between the two groups (Table II).
Anterior Knee Pain
Thirty-nine (42%) of the ninety-three knees were painful anteriorly
before the operation, and thirty-three (85%) of them had
this symptom relieved by the operation. As expected, patients with
preoperative anterior knee pain were found to have lower preoperative
Knee Society pain scores (average, 79.7 points; median, 82 points;
range, 0 to 132 points) than those who did not have preoperative
anterior knee pain (average, 94.7 points; median, 94 points; range,
40 to 129 points) (p < 0.0082), but this was not reflected
in the function scores (an average, median, and range of 39.7, 50, and
-20 to 60 points for the patients with anterior knee pain compared
with 43.3, 45, and 0 to 70 points for those without it, p = 0.35).
Patients with preoperative anterior knee pain were not found to
have significantly different postoperative Knee Society clinical
scores (average, 163.9 points; median, 180 points; range, 56 to
200 points) compared with those without preoperative anterior knee
pain (average, 166.4 points; median, 192 points; range, 52 to 200
points) (p = 0.55).
Two (9%) of twenty-two knees with anterior pain before resurfacing
of the patella continued to have anterior pain after the operation,
but seven (28%) of twenty-five knees without anterior knee
pain before resurfacing of the patella had anterior pain after the
operation. Thus, a total of nine (19%) of forty-seven arthroplasties
with patellar resurfacing were followed by anterior knee pain. Four
of the seventeen knees with anterior pain before the arthroplasties
without patellar resurfacing continued to have anterior pain after
the operation, but new anterior pain developed in four (14%)
of the twenty-nine knees that had not had anterior pain preoperatively.
Thus, a total of eight (17%) of the forty-six knees without
patellar resurfacing had postoperative anterior pain. There was
no significant difference in the prevalence of postoperative anterior knee
pain when patients with a resurfaced patella and those with a nonresurfaced
patella were compared (generalized estimating equations, p = 0.79).
Four of the seventeen knees with anterior pain postoperatively were
not satisfactory to the patients, whereas only three (4%) of
the seventy-six knees without anterior pain were not satisfactory
to the patients; this suggested that patients without anterior knee
pain were significantly more likely to be satisfied with the results
of the surgery (generalized estimating equations, p = 0.01).
Patients without anterior knee pain were 7.5 times more likely to
be satisfied (odds ratio = 7.49, 95% confidence
interval = 1.50 to 37.42).
Patients with a Bilateral Procedure
Twenty-three (96%) of the twenty-four patients with
a bilateral arthroplasty were satisfied with the results for both
the resurfaced and the nonresurfaced patella. (This analysis excludes
two patients with bilateral arthroplasty in whom the nonresurfaced
patella was revised to a resurfaced patella.)
When asked to compare their knees, five (21%) of the
patients with a bilateral procedure responded that they preferred
the resurfaced side, seven (29%) preferred the nonresurfaced
side, and twelve (50%) expressed no preference. The average
magnitude of the difference, on a scale of 1 ("a little")
to 10 ("a lot"), was 5.7 for those who preferred
the nonresurfaced side and 7.0 for those who preferred the resurfaced
side.
With the numbers available, we found no significant differences
between resurfaced and nonresurfaced patellae with respect to the
Knee Society pain (p = 0.77), function (p = 0.16),
and total (p = 0.36) scores or the assessments of patellofemoral
function (p = 0.36, 0.94, and 0.99). Patients who had had
replacement with resurfacing on one side and replacement without
resurfacing on the other did not express a clear preference for
either side.
Obesity, the degree of patellar chondromalacia, and preoperative
anterior knee pain did not predict either a lower postoperative
knee score or postoperative anterior knee pain. These three factors
have commonly been considered as key in the decision whether to
resurface the patella, when the so-called selective patellar resurfacing
approach is utilized9-11,15. We
did not find any differences in the ability to climb stairs, to
rise from a chair, or to exit an automobile when patients were analyzed
according to the degree of patellar chondromalacia. In fact, the
six surviving patients with grade-IV16 chondromalacia
who had not been treated with resurfacing had a relatively high
average Knee Society score of 170.2 points, and none subsequently
required patellar resurfacing. Although it appears that patients
with grade-IV chondromalacia with a nonresurfaced patella scored
well, only a small number were studied and the importance of this
finding is uncertain.
Preoperative anterior knee pain seems a logical reason to resurface
the patella, particularly since patellar resurfacing relieved anterior
pain in thirty-three (85%) of thirty-nine knees in the
present study. Although previously published information has suggested
anterior knee pain as a reason for resurfacing9,11,
we found that preoperative anterior knee pain was not predictive
of postoperative anterior knee pain. Moreover, most postoperative
anterior knee pain was of new onset. There is an approximately equal
likelihood that anterior knee pain will develop postoperatively
regardless of whether patellar resurfacing is performed. The prevalence
of anterior knee pain in this study was consistent with the rates
in previously reported studies17-19.
Brinker et al. reported an average Knee Society clinical score of
180.2 points for 200 knees of asymptomatic older individuals (average
age, 59.6 years)20. If the average
postoperative knee score of 165.4 points in our study is normalized
to this rating, a score of 91.8% (165.4 of 180.2 points)
is obtained, verifying the excellent clinical results attained by
the study group as a whole. The normalized scores for the nonresurfaced
and resurfaced groups were 93.8% (169.1 of 180.2 points)
and 89.7% (161.6 of 180.2 points), respectively.
At an average of 70.5 months postoperatively, the results in the
present study were analogous to the results in our short-term follow-up
study in that no difference in any parameter studied was demonstrated
between patients with a resurfaced patella and those with a nonresurfaced
patella. The overall prevalence of anterior knee pain increased
from 10% at the two to four-year follow-up interval to
18% in the present study. There were ten new patients with
anterior knee pain, a significant increase from the number of patients
with such pain in the earlier study (exact binomial probability,
p = 0.011). This increase demonstrated that the pain-free
period could be limited. Of the ten knees with new anterior knee pain,
seven had a resurfaced patella. The difference between the knees
with resurfacing and those without resurfacing was not significant
in this regard (p = 0.34), illustrating that the occurrence
of anterior knee pain is a dynamic process regardless of whether
the patella is resurfaced or not. Of the seventeen patients who
had anterior knee pain at five to seven years, six had had anterior
knee pain preoperatively and seven had had this symptom at two to
four years. Postoperative anterior pain is probably clinically important
in light of the fact that four of the seventeen knees with the symptom
were not satisfactory to the patients compared with three (4%)
of the seventy-six knees without anterior pain (generalized estimating equations,
p = 0.01).
A potential major disadvantage of not resurfacing the patella is
the possibility that it will subsequently require resurfacing. Subsequent
resurfacing was performed in seven (12%) of the original
sixty knees in this series. Clinical improvement was seen initially
in six of the seven knees; however, anterior knee pain recurred
in four of the five patients who were evaluated more than five years
after the original procedure. Patients with anterior knee pain and
a nonresurfaced patella should be advised of the substantial risk
of recurrence of anterior knee pain despite subsequent resurfacing.
Conversely, there are fewer options available for the treatment
of anterior pain in a knee with an already resurfaced patella, as
isolated revision of the patellar component has been reported to
be fraught with complications21.
The results of the present study may be specific, to some degree,
to the implant and the surgical techniques that were utilized. It
seems clear, however, that anterior knee pain remains an important
clinical issue following total knee arthroplasty. The exact etiology
remains elusive, and the effects of implant design, surgical technique,
and alteration of patellar tracking on the prevalence of postoperative
anterior knee pain are yet to be clearly defined22-25.