Femoral condylar lift-off in vivo in total knee
arthroplasty
D. A. Dennis, R. D. Komistek, S. A. Walker, E. J. Cheal, J. B. Stiehl
From the Rocky Mountain Musculoskeletal Research Laboratory, Denver, USA
W e carried out weight-bearing video radiological
studies on 40 patients with a total knee
arthroplasty (TKA), to determine the presence and
magnitude of femoral condylar lift-off. Half (20) had
posterior-cruciate-retaining (PCR) and half (20)
posterior-cruciate-substituting (PS) prostheses. The
selected patients had successful arthroplasties with no
pain or instability. Each carried out successive
weight-bearing knee bends to maximum flexion, and
the radiological video tapes were analysed using an
interactive model-fitting technique.
Femoral lift-off was seen at some increment of knee
flexion in 75% of patients (PCR TKA 70%; PS TKA
80%). The mean values for lift-off were 1.2 mm with a
PCR TKA and 1.4 mm with a PS TKA. Lift-off
occurred mostly laterally with the PCR TKA, and
both medially and laterally with the PS TKA.
Separation between the femoral condyles and the
articular surface of the tibia was recorded at 0°, 30°,
60° and 90° of flexion. Femoral condylar lift-off may
contribute to eccentric polyethylene wear, particularly
in designs of TKA which have flatter condyles.
Coronal conformity is an important consideration in
the design of a TKA.
J Bone Joint Surg [Br] 2001;83-B:33-9.
Received 5 October 1999; Accepted after revision 24 March 2000
Retrieval studies of tibial inserts after total knee arthroplasty
(TKA) have shown many examples of eccentric wear of the
polyethylene. The distribution of force across the knee
D. A. Dennis, MD
R. D. Komistek, PhD
S. A. Walker, MS
Rocky Mountain Musculoskeletal Research Laboratory, 2425 South Colo-
rado Boulevard, Suite 280, Denver, Colorado 80222, USA.
E. J. Cheal, PhD
Johnson & Johnson Professional Inc, 325 Paramount Drive, Raynham,
Massachusetts 02767, USA.
J. B. Stiehl, MD
Midwest Orthopaedic Biomechanical Laboratory, 2015 E Newport, #703,
Milwaukee, Wisconsin 53211, USA.
Correspondence should be sent to Dr D.A. Dennis
©2000 British Editorial Society of Bone and Joint Surgery
0301-620X/00/110632 $2.00
VOL. 83-B, NO. 1, JANUARY 2001
during midstance is approximately 70% to 75% medially
1-3
and 25% to 30% laterally, but this may be substantially
4,5
different if femoral condylar lift-off occurs. This may
lead to: 1) eccentric loading; 2) premature wear of polyeth-
ylene; 3) excessive load on subchondral cancellous bone;
and 4) premature loosening of the prosthesis. Nilsson and
6
Kärrholm carried out a roentgen stereophotogrammetric
analysis (RSA) of 45 patients after insertion of either a
cemented or an uncemented TKA. They assessed micro-
movement of the components and showed that in the pres-
ence of condylar lift-off asymmetrical subsidence and tilting
of the tibial component could occur, suggesting uneven
loading. Our aim was to determine if condylar lift-off
occurs during weight-bearing deep knee bends in vivo.
Patients and Methods
Medial and lateral femorotibial condylar separations were
determined in the knees of 40 patients with a TKA. A
posterior-cruciate-retaining (PCR) TKA with a relatively
flat posterior lipped polyethylene insert had been intro-
duced in 20 patients, and a posterior-cruciate-substituting
(PS) prosthesis (Press-Fit Condylar design; Johnson &
Johnson Professional Incorporated, Raynham, Massachu-
setts) in a further 20. The coronal conformity of the PCR
and PS TKA designs was similar. The patients were chosen
randomly from a group with clinically successful arthro-
7
plasties (Hospital for Special Surgery Knee Scores, PCR =
92.7, PS = 96.9 points), who had no ligamentous laxity or
pain. Table I gives the clinical details of the two groups. We
performed a statistical analysis comparing these groups
using Student’s t-test and multivariate regression analysis
to correlate clinical parameters with condylar lift-off. All
procedures had been carried out by a single surgeon using
a standardised technique. Identical techniques for liga-
mentous balancing were used in both groups. Recession of
the posterior cruciate ligament was undertaken when indi-
cated. The femoral component was introduced with its axis
of rotation as parallel as possible to the transepicondylar
axis. The tests were performed at a minimum of one year
after operation.
Under radiological surveillance, the patients carried out
three successive deep knee bends from full extension to
maximum flexion in the sagittal plane as if picking up an
33
34 D. A. DENNIS, R. D. KOMISTEK, S. A. WALKER, E. J. CHEAL, J. B. STIEHL

Table I. Details of the 20 patients who had PS TKA and the 20 who had PCR TKA
PS PCR
Gender
M 14 7
F 6 13
Mean age in years 70.3 65.8
Mean flexion in degrees
Preoperative 112.1 118.4
Postoperative 106 101.6
Mean coronal alignment
Preoperative 2.66 varus (14 varus to 17 valgus) 0.93 varus (10 varus to 12 valgus)
Postoperative 5.6 (3 to 8 valgus) 5.6 (3 to 7 valgus)

object from the floor, with no constraints applied to either
the hip, knee, ankle or foot. A Siremobil 2000 Digital x-ray
image intensifier (Siemens, Isolin, New Jersey) was used
and the images were recorded on a workstation computer.
They were then stored on videotape for subsequent redigiti-
sation using a frame grabber. The position of contact
between the femur and the tibia was determined using a
8,9
three-dimensional (3D) model-fitting technique. The 3D
computer-aided design (CAD) solid models of the femoral
and tibial components were overlaid on to the two-dimen-
sional (2D) radiological images using a 3D model-fitting
technique as described below (Fig. 1). The 3D image was
then rotated to a frontal view and the distances from the
medial and lateral femoral condyles to the tibial tray were
measured (Fig. 2). The difference between medial and
lateral measurements was considered to represent condylar
lift-off if it was greater than 0.75 mm. Images were ana-
lysed at 0°, 30°, 60° and 90° of flexion.
3D model fitting. Using a model developed from a
technique of 3D overlay which has been described pre-
8,9
viously, the relative position of the components was
recorded in 3D from a single-perspective image by manip-
ulating a CAD model in 3D space. A 3D unit was then
created on a Silicon Graphics Indigo workstation (Moun-
tainview, California) in C++ using the Open Inventor
Toolkit library (Mountainview, California). AutoCAD (San
Rafael, California) was used to extract measurements.
Individual radiological frames were digitised at specified
degrees of flexion The operator manipulated the models to
create a correct fit when the silhouettes of the femoral and
tibial components matched the corresponding components
in the radiological image perfectly. The position of each
component was then recorded and measurements were
made using a CAD-modelling program. The process was
repeated at each angle of flexion and the kinematics
determined.

Fig. 1
The 3D solid models of the femoral and tibial components are precisely fitted over the 2D radiological images. Once the best fit has
been obtained the femoral component is rotated relative to its tibial counterpart into a pure sagittal and then a pure frontal view.

THE JOURNAL OF BONE AND JOINT SURGERY

 FEMORAL CONDYLAR LIFT-OFF IN VIVO IN TOTAL KNEE ARTHROPLASTY
Fig. 2
After the components have been rotated to the frontal
view, the distances from the medial (YM) and the lateral
(YL) femoral condyles to the tibial tray are measured.
Lift-off is measured and recorded as the angle !.
The error-analysis images were corrupted with noise (left) to create a ‘worse-case scenario’.
The quality of the in vivo images was better (right).
An error analysis was undertaken by taking radiographs
of components mounted on an apparatus with six degrees
of freedom. Accurate positioning of the components was
achieved using rotational and translational stages with an
accuracy of 15 arc sec and 0.01 mm, respectively. The
components were set in an initial position and then rotated
and translated to known values. Radiological images of
the components were then created at each setting. The 3D
model-fitting process was used for each setting of the
rotational and translational stages to determine the relative
position of the components. A second dynamic test was
undertaken to determine the effect of movement. Noise,
having the same interference frequency as soft tissue and
bone, was added to each error-analysis image (Fig. 3).
Thus, the quality of the error-analysis images was less
good than that of those in vivo. The components were
examined radiologically at a variable speed of between 0.5
Fig. 4a
VOL. 83-B, NO. 1, JANUARY 2001
35
Fig. 3
and 0.3m per second. The translational error of the 3D
9
model-fitting technique was less than 0.5 mm. In order to
account for unknown variables, separation values of
0.75 mm (50% safety factor) were chosen to represent
the presence of femoral condylar lift-off. A second valida-
tion was used by radiographs in the frontal plane (Fig. 4).
The predicted measurements of lift-off were within
0.5 mm of the actual values derived by digitising the
frontal images.
Results
There was lift-off in 30 of the 40 patients during a deep
knee bend. Examples of condylar lift-off, as determined by
the image-matching process, are shown in Figures 5 and 6.
There was lift-off at some increment of flexion in 14 of the
patients (70%) with a PCR TKA and 16 of those (80%)
Frontal radiographic images of a)
a PS and b) a PCR (right) TKA
during a deep knee bend. Both
knees showed condylar lift-off.
Fig. 4b
36 D. A. DENNIS, R. D. KOMISTEK, S. A. WALKER, E. J. CHEAL, J. B. STIEHL

Fig. 5
Sagittal (a) and frontal (b) im-
ages of a PCR TKA showing
condylar lift-off.

Fig. 6

Sagittal (a) and frontal (b) im-

ages of a PS TKA showing con-
dylar lift-off.

Table II. Percentage incidence of condylar lift-off for the 20 PCR and the with a PS TKA. In the PCR TKA group there was lift-off
20 PS patients on the lateral side in 13 and on the medial side in two. In
Flexion angle (degrees) the PS TKA group there was lift-off on the lateral side in
0 30 60 90 12 and on the medial side in nine. In two patients in the
PCR subjects PCR-TKA group and five in the PS-TKA group there was
Medial 0 0 10 0 lift-off both medially and laterally at differing increments
Lateral 25 25 25 25 of knee flexion. In these patients there was lift-off from one
PS subjects condyle and then at a different angle of flexion condylar
Medial 20 10 30 20 contact would be re-established followed by lift-off of the
Lateral 25 30 30 25 other femoral condyle. During a deep knee bend, these
patients would thus have a rocking movement of their knee
Table III. The percentage of patients having maximum lift-off after PCR
in the frontal plane.
and PS TKA For both the PCR and PS TKA designs, lift-off occurred
Flexion angle (degrees) most commonly at 60° of flexion (Table II), and this was
0 30 60 90 present in 35.0% of PCR and 60.0% of PS TKAs. The
PCR subjects
maximum lift-off was also observed at 60° of flexion in
Medial 0 0 10 0
both TKA designs (Table III). Five (25.0%) patients with a
Lateral 15 10 15 20
PS TKA and five (25.0%) with a PCR TKA had maximum
lift-off at 60° of flexion. For those with a PCR TKA the
PS subjects
maximum lift-off was 1.8 mm (mean 1.2 mm) and for those
Medial 5 0 20 5
with a PS TKA 2.7 mm (mean 1.4 mm). For those with lift-
Lateral 10 10 5 20
off, the average was 1.1 to 1.2 mm for the PCR TKA group
THE JOURNAL OF BONE AND JOINT SURGERY
 FEMORAL CONDYLAR LIFT-OFF IN VIVO IN TOTAL KNEE ARTHROPLASTY
Table IV. Mean amount of lift-off (mm) at each flexion angle for those
patients experiencing lift-off
Flexion angle (degrees)
0 30 60 90
PCR knee 1.2 1.1 1.2 1.2
PS knee 1.4 1.3 1.4 1.8
Both knees 1.3 1.2 1.3 1.6
at each angle of flexion and 1.3 to 1.8 mm for the PS-TKA
group (Table IV).
Clinical correlation. Statistical analysis of both groups
using Student’s t-test revealed no statistically significant
differences in flexion and coronal alignment before and
after operation. Patients with a PS TKA were significantly
older (70.3 v 65.8 years) than those with a PCR TKA
(p = 0.05).
Multivariate regression analysis showed no statistically
significant correlation between the presence of lift-off and
the flexion or alignment in the coronal plane before and
after operation.
Discussion
Numerous kinematic analyses of the knee have shown
femoral condylar lift-off in RSA, cadaver and ligament
2,3,5,6,10-20 10
laxity studies. Bourne et al, in an evaluation of
the normal and osteoarthritic knee in vitro, showed that the
lateral condyle can be completely unloaded in the osteo-
arthritic knee aligned in 3.1° of varus. Similarly, Kostuik et
14
al analysed 15 cadaver knees under various loading con-
ditions and showed that there is unloading of one condyle if
the knee is positioned in 3° of either varus or valgus. The
effect of lift-off is further seen in retrieval analyses of
21
failed tibial components. Lewis et al reviewed a series of
TKA failures with polyethylene wear and recorded a pre-
dominance of wear posteromedially. We have shown that
lift-off, after PCR-TKA, primarily occurs laterally resulting
in high polyethylene stresses medially.
There have been numerous descriptions of the abduction
and adduction moments at the knee during normal gait. It
has been shown that an abduction moment occurs at heel
strike, but quickly reverses to an adduction moment
22
throughout the remaining stance phase. This adduction
moment has been shown to measure between 36 and 50 N,
5,23,24
increasing if there is a coexisting varus deformity.
During the midstance phase of gait, the medial compressive
1-3
load increases to 70% to 75% of the load across the knee.
There are many compensatory mechanisms for resisting
this adduction moment including: 1) the redistribution of
condylar loads; 2) co-contraction of antagonist muscle
groups; 3) increased tension in the lateral soft tissue; 4)
increased tension in the cruciate ligaments; 5) increased
swaying of the body in the lateral direction; 6) decreased
stride length; and 7) a decreased inversion moment at the
5
ankle brought about by out-toeing. Although these com-
VOL. 83-B, NO. 1, JANUARY 2001
37
pensatory mechanisms aid stability of the knee, they may
lead to increased joint reaction forces, and if the compensa-
tory mechanisms are inadequate, lift-off may occur.
Increased soft-tissue laxity in the frontal plane would
5
enhance the likelihood of lift-off. Burstein carried out a
study to determine the varus and valgus stabilising struc-
tures in the knee and showed that the primary stabilisers are
the collateral ligaments while the cruciate ligaments serve
as secondary stabilisers. Although the cruciate ligaments
contribute only 25% of the varus-valgus stability of the
knee when compared with the collateral ligaments, the
adduction moment increases in the absence of the cruciate
3
ligaments. Noyes et al analysed the adduction moment at
the knee in 32 patients with an absent anterior cruciate
ligament (ACL) and coexisting varus deformity, using gait
analysis. They showed a 35% increase in the adduction
moment and a 51% increase in the soft-tissue tension on the
lateral side of the knee in these patients.
15
Markolf et al analysed 35 patients with ACL-deficient
knees in non-weight-bearing conditions in vivo. The results
were compared with 49 patients with clinically normal
knees. They showed that the varus-valgus laxity in the ACL-
deficient knees increased by 36% at full extension and by
19% at 20° of flexion. Other studies have concluded that the
varus-valgus laxity is greater in knees with a TKA than in
normal knees, and that the absence of the posterior cruciate
12,18
ligament (PCL) leads to an increase in joint laxity.
There is controversy as to whether static radiological
tibiofemoral alignment reflects the adduction
2,3,11,13,17,19,20
moment. We found no correlation between
the frontal (coronal) tibiofemoral alignment and the pres-
ence of lift-off, although no major malalignment was pres-
ent in any of the patients.
Our study has shown that lift-off commonly occurs in
both PCR and PS TKAs, usually at angles of flexion greater
than 60°. Lift-off occurred both medially and laterally in
subjects with a PS TKA, but was observed predominantly
on the lateral side in those with a PCR TKA. This may be
related to the presence or absence of the cruciate ligaments.
In the knee with intact cruciate ligaments, the ACL origi-
nates at the lateral femoral condyle while the femoral
attachment of the PCL is medial. We suggest therefore that
the ACL acts as a check, limiting lift-off laterally, with the
PCL resisting lift-off medially. In the PS TKA with both
cruciate ligaments resected, the incidence of lift-off was
similar both medially and laterally. In PCR TKA, lift-off
predominantly occurred laterally, since the retained PCL
may resist medial lift-off.
The cause of lift-off is multifactorial and factors affect-
ing its incidence include accuracy of ligamentous balanc-
ing at operation, the position of the components, the
rotation of the femur relative to the tibia (abduction/adduc-
tion in early to mid-flexion, internal/external rotation in
mid to full flexion), the direction of the resultant muscle
force across the knee and the dynamic stabilising effect of
the supporting musculature.
38 D. A. DENNIS, R. D. KOMISTEK, S. A. WALKER, E. J. CHEAL, J. B. STIEHL
These findings emphasise the importance of coronal
femorotibial curvature and conformity in the design of a
16
TKA. Miller et al in a laboratory analysis, evaluated peak
tibial polyethylene stresses under eccentric loading in both
flat-on-flat and more coronally curved femorotibial
implants. They showed substantially higher polyethylene
stresses in flat-on-flat designs, often exceeding the yield
strength of polyethylene, whereas designs with increased
coronal curvature and conformity proved less sensitive to
eccentric loading.
The results of our study also add support for the use of
metal-backed tibial components with central stems in
order to reduce the peak subchondral cancellous bone
23
load should femoral condylar lift-off occur. Bartel et al
performed a finite-element analysis to study the effect of
metal backing on peak subchondral cancellous stresses,
and reported a reduction of 16% to 39% with metal-
backed compared with all-polyethylene tibial components
when measured under eccentric loading conditions. This
25
analysis is supported by clinical studies. Ritter et al
evaluated 200 cases of metal-backed PCR TKA, of a
single, flat-on-flat condylar design (Anatomical Grad-
uated Components; Biomet Corporation, Warsaw,
Indiana). A survival rate of 98% was observed at ten
26
years. Faris et al reviewed 538 cases of the same
prosthetic design with the exception that an all-polyethyl-
ene tibial component was used. At a mean follow-up of
only 4.2 years, 50 tibial components had been revised
because of medial tibial collapse and loosening. In an
additional 155 knees there was a radiolucency in the
medial tibial plateau measuring 1 mm or more. We sug-
gest that lateral condylar lift-off may have played a role in
the premature failure of these all-polyethylene flat-on-flat
tibial components. Excessive medial loading is associated
with lateral lift-off. In the absence of metal backing of the
tibial component, these high loads become transmitted to
the underlying medial subchondral bone resulting in col-
lapse and loosening.
Limitations of our investigation include the lack of accu-
rate assessment of polyethylene wear, deformation, and the
sagittal anteroposterior curvature of the tibial polyethylene
insert, although no gross wear or deformation was noted on
plain radiographs.
We have shown that femoral condylar lift-off can occur
after both PCR and PS TKA. This results in eccentric
loading and may lead to premature wear of the polyethyl-
ene and prosthetic loosening. Our findings suggest the
importance of both coronally curved and conforming
femorotibial geometry and metal backing of the tibial
components to improve the longevity of the arthroplasty.
Based on similar findings with other current designs of
TKA, the presence of lift-off is not unique to the implants
27,28
used in this study. Further research is required to assess
the contribution of surgical technique as well as the design
of the implant in affecting the incidence and magnitude of
lift-off.
We wish to acknowledge DePuy and Johnson & Johnson Company for
their financial support of our research.
One or more of the authors have received or will receive benefits for
personal or professional use from a commercial party related directly or
indirectly to the subject of this article.
References
1. Hsu RWW, Himeno S, Coventry MB, Chao EYS. Normal axial
alignment of the lower extremity and load-bearing distribution at the
knee. Clin Orthop 1990;255:215-27.
2. Johnson F, Leitl S, Waugh W. The distribution of load across the
knee; a comparison of static and dynamic measurements. J Bone Joint
Surg [Br] 1980;62-B:346-9.
3. Noyes FR, Schipplein OD, Andriacchi TP, Saddemi SR, Weise M.
The anterior cruciate ligament-deficient knee with varus alignment: an
analysis of gait adaptations and dynamic joint loadings. Am J Sports
Med 1992;20:707-16.
4. Andriacchi TP. Dynamics of knee malalignment. Orthop Clin North
Am 1994;25:395-403.
5. Burstein AH. Biomechanics of the knee. In: Insall JN, ed. Surgery of
the Knee. New York: Churchill Livingstone, 1984:21-30.
6. Nilsson KG, Kärrholm J. Increased varus-valgus tilting of screw-
fixated knee prostheses: stereoradiographic study of uncemented ver-
sus cemented tibial components. J Arthroplasty 1993;8:529-40.
7. Insall JN, Hood RW, Flawn LB, Sullivan DJ. The total condylar
knee prosthesis in gonarthrosis: a five to nine-year follow-up of the
first one hundred consecutive replacements. J Bone Joint Surg [Am]
1983;65-A:619-28.
8. Dennis DA, Komistek RD, Hoff WA, Gabriel SM. In vivo knee
kinematics derived using an inverse perspective technique. Clin
Orthop 1996;331:107-17.
9. Hoff WA, Komistek RD, Dennis DA, Gabriel SA, Walker SA. A
three dimensional determination of femorotibial contact positions
under in vivo conditions using fluoroscopy. J Clinical Biomechanics
1998;13:455-70.
10. Bourne RB, Finlay JB, Papadopoulos P, Rorabeck CH, Andreae P.
In vivo strain distribution in the proximal tibia: effect of varus-valgus
loading in the normal and osteoarthritic knee. Clin Orthop
1984;188:285-92.
11. Brugioni DJ, Andriacchi TP, Galante JO. A functional and radio-
graphic analysis of the total condylar knee arthroplasty. J Arthroplasty
1990;5:173-80.
12. Dorr LD, Ochsner JL, Gronley J, Perry J. Functional comparison of
posterior cruciate-retained versus cruciate-sacrificed total knee arthro-
plasty. Clin Orthop 1988;236:36-43.
13. Goh JCH, Bose K, Khoo BCC. Gait analysis study on patients with
varus osteoarthritis of the knee. Clin Orthop 1993;294:223-31.
14. Kostuik JP, Schmidt O, Harris WR, Wooldridge C. A study of
weight transmission through the knee joint with applied varus and
valgus loads. Clin Orthop 1975;108:95-8.
15. Markolf KL, Kochan A, Amstutz HC. Measurement of knee stiff-
ness and laxity in patients with documented absence of the anterior
cruciate ligament. J Bone Joint Surg [Am] 1984;66-A:242-53.
16. Miller GJ, Perry W, Goll C. Congruency and varus/valgus loading
effect on prosthetic knee contact stress. Combined ORS San Diego,
CA, 1995.
17. Prodromos CC, Andriacchi TP, Galante JO. A relationship between
gait and clinical changes following high tibial osteotomy. J Bone Joint
Surg [Am] 1985;67-A:1188-94.
18. Soudry M, Walker PS, Reilly DT, Kurosawa H, Sledge CB. Effects
of total knee replacement design on femoral-tibial contact conditions.
J Arthroplasty 1986;1:35-45.
19. Wang CJ, Walker PS. The effects of flexion and rotation on the
length patterns of the ligaments of the knee. J Biomech
1973;6:587-96.
20. Weinstein JN, Andriacchi TP, Galante J. Factors influencing walk-
ing and stair climbing following unicompartmental knee arthroplasty.
J Arthroplasty 1986;1:109-15.
21. Lewis P, Rorabeck CH, Bourne RB, Devane P. Posteromedial tibial
polyethylene failure in total knee replacements. Clin Orthop
1994;299:11-7.
22. Dennis DA. Patellofemoral complications in total knee arthroplasty.
Orthopaedic Knowledge Update. Hip and Knee Reconstruction,
1995:283-9.
THE JOURNAL OF BONE AND JOINT SURGERY
 FEMORAL CONDYLAR LIFT-OFF IN VIVO IN TOTAL KNEE ARTHROPLASTY
23. Bartel DL, Burstein AH, Santavicca EA, Insall JN. Performance of
the tibial component in total knee replacement: conventional and
revision designs. J Bone Joint Surg [Am] 1982;64-A:1026-33.
24. Schipplein OD, Andriacchi TP. Interaction between active and
passive knee stabilizers during level walking. J Orthop Res
1991;9:113-9.
25. Ritter MA, Worland R, Saliski J, et al. Flat-on-flat, nonconstrained,
compression moulded polyethylene total knee replacement. Clin
Orthop 1995;321:79-85.
VOL. 83-B, NO. 1, JANUARY 2001
39
26. Faris MP, Ritter MA, Keating EM, Meding JB. Mid term follow-up
of all-poly tibial component in flat-on-flat design. Procs AAHKS
eighth Annual Meeting, 1998.
27. Stiehl JB, Dennis DA, Komistek RD, Keblish PA. In vivo kinematic
analysis of a mobile bearing total knee prosthesis. Clin Orthop
1997;345:60-6.
28. Stiehl JB, Komistek RD, Dennis DA, Paxon RD, Hoff WA. Fluoro-
scopic analysis of kinematics after posterior-cruciate-retaining knee
arthroplasty. J Bone Joint Surg [Br] 1995;77-B:884-9.