See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/51031805

Total knee arthroplasty in stiff knee compared with flexible knees

Article  in  MUSCULOSKELETAL SURGERY · April 2011

DOI: 10.1007/s12306-011-0099-6 · Source: PubMed

CITATIONS READS

8 284

5 authors, including:

Matteo Fosco Mattia Filanti

S.Clemente Hospital, Mantova, Italy Istituto Ortopedico Rizzoli
28 PUBLICATIONS   218 CITATIONS    7 PUBLICATIONS   88 CITATIONS   

SEE PROFILE SEE PROFILE

Luca Amendola Lucia Savarino

Ospedale Maggiore Carlo Alberto Pizzardi di Bologna Istituto Ortopedico Rizzoli
64 PUBLICATIONS   734 CITATIONS    142 PUBLICATIONS   3,570 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

pediatric NONUNION View project

Improve abrasion properties and increased useful life of UHMWPE hip and knee prosthesis, using electron beam irradiation View project

All content following this page was uploaded by Domenico Tigani on 06 June 2014.

The user has requested enhancement of the downloaded file.

Musculoskelet Surg (2011) 95:7–12
DOI 10.1007/s12306-011-0099-6
ORIGINAL ARTICLE
Total knee arthroplasty in stiff knee compared with flexible knees
M. Fosco • M. Filanti • L. Amendola •
L. M. Savarino • D. Tigani
Received: 18 November 2009 / Accepted: 14 February 2011 / Published online: 9 March 2011
Ó Springer-Verlag 2011
Abstract This retrospective study was done to evaluate
the results of total knee arthroplasty performed on 32
patients with stiff knee, having a preoperative arc of
movement between 0° and 50° (average 30°). This group of
patients were matched with a group of 32 flexible knees,
randomly selected from the same cohort of patients who
underwent knee arthroplasty in our ward. At a mean fol-
low-up of 4.5 years (min 2, max 11 years), seven patients
of the stiff group reported complications (21.8% overall):
four prosthetic infection that successively underwent
removal of the implant, one skin necrosis 4 months after
the intervention, one early contracture and one late stiff-
ness of the knee. In the control group, in two cases, there
was substitution of the implant due to periprosthetic
infection. At the end of the study period, the clinical
evaluation was not possible in four patients of the stiff and
in two patients of the control group who underwent revi-
sion of the prosthetic components. An excellent or good
clinical result was obtained in 92% of stiff group and in
96% of the control group patients. Although the final
results achieved in these patients are worse than those of
patients with flexible knee due to disadvantageous preop-
erative conditions and high complication rate, our results
demonstrate the efficacy of the arthroplasty procedure as
treatment of stiff knee.
M. Fosco (&)
M. Filanti
L. Amendola
D. Tigani
Department of Orthopaedic Surgery, Rizzoli Orthopaedic
Institute, University of Bologna, Via Pupilli 1,
40136 Bologna, Italy
e-mail: matteo.fosco@email.it
L. M. Savarino
Laboratory of Orthopaedic Pathophysiology and Regenerative
Medicine, Bologna, Italy
Keywords Stiff knee
Total joint replacement

Knee infection
Range of motion
Introduction
Usually, a stiff knee is considered as a knee with less than
50° of range of motion (ROM) [1–6]. As the amount of
flexion achieved postoperatively after a total knee arthro-
plasty is determined mainly by the preoperative ROM [7],
stiff knee could be considered a challenging issue, partic-
ularly to obtain a functional postoperative ROM.
There are in these joints other surgical problems that
regard primarily surgical exposure and correct ligament
balancing, often difficult to obtain due to an extensive
release of soft tissues [2, 8–11]. Despite effective surgical
techniques and modern prosthetic models, due to under-
lying problems, the results achieved in these patients are
often worse than those in patients with wide arc of
movement, and frequent complications usually influence
the final result. Although only a suboptimal clinical and
functional outcome is reachable in these patients, they
show an improvement by the arthroplasty procedure better
than patients with flexible knee.
The aim of our study was to present our experience
treating stiff or ankylosed knee joint with TKA, comparing
the results obtained in these patients with those obtained in
patients with flexible knees.
Materials and methods
From March 1997 to August 2009, at first ward of the Rizzoli
Orthopedic Institute in Bologna have been performed 861
primary total knee arthroplasty. Retrospectively analysing
123
8 Musculoskelet Surg (2011) 95:7–12

the clinical and radiological data of patients, we have iden- We did not implant patellar prosthesis in any patient, in
tified 42 patients who had a knee ROM less than 50°. Among those we performed removal of osteophytes, remodelling
them, 32 patients with a follow-up exceeding 2 years (stiff of the articular surface and partial peripheral denervation.
group) were identified for our study. From the first day, postoperatively in all cases was
These patients were matched with a control group of 28 started the same rehabilitation protocol; in patients under-
patients, as 32 knees totally with a ROM of more than 50°, going TTO flexion of the knee has been limited at 70° for
randomly selected from the same cohort of patients oper- the first 4 weeks.
ated of knee arthroplasty at our ward. The match was made
for age at surgery, weight-bearing alignment of the limb, Control group
class of disability, preoperative American Knee Society
(AKS) total score and length of follow-up. The control group included 28 patients for a total of 32
knees (24 women and four men), with an average age of
Stiff group 67 years (range 49–81 years), involvement of right knee in
17 and left knee in 15 patients. ROM of the knee averaged
The 32 patients of the stiff group (20 women and 12 men) 86.8° (range 70°–110°). There were no significant differ-
had a mean age of 65 years (range 44–82 years), right knee ences between the two groups regarding the age at surgery,
involved in 18 and left knee in 14 cases. The mean ROM weight-bearing alignment of the limb, class of disability,
was 29.7°: it was less than or equal to 5° in seven patients, preoperative AKS total score and length of follow-up
while was between 6° and 50° in the remaining 25 patients. (Table 1). As surgical exposure was used a medial para-
The preoperative diagnosis was degenerative osteoarthritis patellar incision and a median arthrotomy. It was not
in 20 patients, posttraumatic arthritis in four patients, necessary to use special exposure techniques, except the
rheumatoid arthritis in four patients, haemophilic arthrop- release of the retinaculum performed in nine cases. Using
athy in two patients, arthropathy secondary to septic the same surgical criterion as in stiff group, in 29 cases,
arthritis, healed 50 and 60 years before, in the remaining posterior stabilized prosthesis (Nex Gen PS, Zimmer Inc.,
two patients. According to the clinical distinction of Knee Warsaw, Ind, USA) has been used, in three cases cruciate
Society [12], 16 patients were considered to be Class A, retaining prosthesis (Nex Gen CR, Zimmer Inc., Warsaw,
8 patients Class B and 8 patients Class C. The mean Ind, USA). In no case has been implanted patella pros-
mechanical alignment of the knee was evaluated with thesis. Then patients were subjected to the same rehabili-
radiograms of the lower limbs on a long plate and was 5.1° tation protocol of the study group.
varus (range 10° valgus–25° varus).
For the arthroplasty procedure in all patients were used a Postoperative care
medial parapatellar incision and a median arthrotomy [13],
except in two cases in which skin incision was carried out In the early postoperative period, in all patients were
on existing scars of past surgeries. started exercises of passive mobilisation of the knee joint
To facilitate articular exposure in all patients was encouraging them to perform active knee flexion exercises
performed lateral retinaculum release, in 10 patients
together with a quadriceps snip [14]; in six patients was
performed a tibial tubercle osteotomy (TTO). Sequential Table 1 Preoperative clinical data
soft-tissue release was done to correct the angular Stiff (SD) Control (SD) Significance
deformity; if posterior cruciate ligament appears to be
Age 65 (11) 67 (12.1) NSa
functional and balanced, cruciate retaining (CR) prosthe-
Alignment 5.1 varus (5.3) 5.2 varus (5.9) NSa
sis was used, conversely a posterior stabilized (PS) model
ROM 29.7 (17.6) 86.8 (11.1) P \ 0.001b
was the implant of choice. If during insertion of trial
Patient category A(16); B(8); C(8) A(16); B(8); C(8) NSc
components, the knee was unstable in both coronal and
sagittal plane, a hinged prosthesis was used. In this way, AKS
in two cases was implanted a cruciate retaining prosthesis AKSS 31.3 (17) 29.3 (13.7) NSb
(Nex Gen CR, Zimmer Inc., Warsaw, Ind, USA), and in AKFS 34 (17.8) 37.8 (17.5) NSa
27 patients, a posterior stabilized prosthesis (Nex Gen PS, AKStot 63.2 (32.5) 64.3 (27) NSb
Zimmer Inc., Warsaw, Ind, USA), while in the remaining SD Standard deviation
three patients was implanted a hinged prosthesis: one a
Oneway anova
Endo-Model (Waldermar Link GmbH & Co, Hamburg, b
Mann–Whitney
Germany) and two Nex Gen RHK (Zimmer Inc., Warsaw, c
Pearson’s chi-square was evaluated by Montecarlo method for
Ind, USA). small samples

123
Musculoskelet Surg (2011) 95:7–12
and straight leg raises. Pain during motion was controlled
with peripheral nerve block catheter or intravenous pain
killers and antithrombotic profilaxis was done still full
weight bearing was gained. During discharge, all patients
had a ROM of at least 50° and were able to walk with aid
of crutches. At home were then carried out intensive
exercises of active and passive mobilisation of the knee
with progressive increase of the articularity, except for
patients who underwent TTO in which was observed lim-
itation of flexion to 70° for the first 4 weeks. Patients of
both groups were followed in the outpatient clinic for at
least 2 years, with controls after 6 weeks, 6 and 12 months
after intervention, then yearly. At each follow-up was
performed a systematic assessment that included a clinical
examination with Knee Society scores [12] and weight-
bearing radiographs of both knees in which according to the
Knee Society evaluating system [15] were measured the
mechanical axis of the operated limb and the different AP
and lateral angles of the prosthetic components (Fig. 1).
All complications following the operation were noted,
based on the classification of Epps CH [16].
Statistical analysis
The SPSS programme was used for the statistical analyses.
A P value of 0.05 was considered statistically significant.
The sample size was analysed based on an expected infe-
rior clinical result for the stiff group. An inferior result was
defined as 10 points difference in AKS score postopera-
tively. We included 32 patients in each group detecting
such a difference with one-sided testing (a = 0.05 and
power of 90%). Tests used to compare both the study
Fig. 1 Right knee osteoarthritis in a patient with rheumatoid arthritis (a). In (b) is shown correct positioning of the prosthetic components
without radiographic signs of loosening 2 years postoperatively
9
groups were oneway anova test, Mann–Whitney test,
Pearson’s chi-square test and Fischer’s test. The data for
this investigation were collected and analysed in compli-
ance with the procedures and policies set forth by the
Helsinki Declaration, and all patients gave their informed
consent. The study was authorised by the local ethical
committee.
Results
Clinical results
The mean follow-up for the stiff group was 57.2 months
(range 24–133 months) and 55.2 months (range 24–130
months) for the control group.
At the end of the study period, four patients of the stiff
and two patient of the control group underwent revision of
the prosthetic components; the clinical evaluation was
possible in 28 out of 32 stiff group patients, in 30 out of 32
control group patients (Table 2).
In the stiff group, postoperative complications were
reported in seven patients (21.8% overall): four patients were
complicated of infections of the prostheses, treated with
femoral–tibial fusion in three cases, and prosthetic revision
in the other one. One patient presented a skin necrosis
4 months after the arthroplasty procedure: soon mobilisation
of the knee was stopped; then the patient obtained complete
wound healing after a partial skin grafting. One patient who
regained less than 90° of functional flexion was manipulated
under general anaesthesia at 4 months follow-up. One
patient with stiff knee at 20 months postoperatively under-
went arthroscopic debridement and manipulation.
123
10 Musculoskelet Surg (2011) 95:7–12

Table 2 Clinical outcomes Table 3 Radiological outcomes

Stiff (SD) Control (SD) Significance Stiff (SD) Control (SD) Significance

AKS Mechanical 2.2° valgus (2) 3.4° valgus (1.5) NSa

AKSS 85.2 (14.5) 92.6 (11.4) P = 0.003 b alignment
AKSS% 76.8 (25.3) 89.3 (16.5) P = 0.001b a 96.5° (2.9) 96.9° (3.4) NSa
AKFS 70.8 (25.1) 84 (20.6) P = 0.034b b 89.3° (1.9) 87.9° (7.1) NSa
AKFS% 54.9 (40) 76.3 (26.7) P = 0,03b c 3.6° (2) 3° (1.4) NSa
AKStot 157.2 (35.2) 176.6 (29.4) P = 0.008b r 86.5° (2.7) 87.2° (2.4) NSa
AKStot% 66.5 (29.6) 83.4 (19.1) P = 0.006b Loosening
ROM 87.4 (27.6) 106.6 (11.4) P = 0.003 b Femoral N = 2(1); 3(1) N = 1(1); 2(2) NSb
ROM% 64.4 (26.3) 59.9 (33.7) NSb Tibial AP N = 1(2); 2(2) N = 1(1); 2(1); 4(1) NSb
Complications 7/32 2/32 NSc Tibial LL N = 1(3); 3(1) N = 1(2) NSb
Follow-up 57.2 (37.9) 55.2 (34.4) NSa SD Standard deviation, a Femoral component anteroposterior,
b Tibial component anteroposterior, c Femoral component sagittal,
SD Standard deviation, % Percentage of improvement
r Tibial component sagittal
a
Oneway anova a
Oneway anova
b
Mann–Whitney b
Pearson’s chi-square was evaluated by Montecarlo method for
c
Fischer’s exact test small samples

At the last follow-up in the stiff group, the mean AKS
knee score (AKSS) was 85.2 points (range 38–97 points),
improvement of 76.8% from preoperative evaluation; the
clinical result was excellent or good in 26 cases, fair in one
and poor in one case. The functional score (AKFS) also
increased postoperatively from 34 (range 0–70 points) to
70.8 points (range 25–100 points, increase 54.9%). Func-
tionally the result was good in 16, fair in three and poor in
10 patients. The range of motion postoperatively for the
stiff group was 87.4° (range 25°–125°, increase 64.4%).
In the control group, there were two prosthetic infec-
tions (6.2% overall) that subsequently underwent substi-
tution of the implant. For the other patients of this group,
the final AKSS was 92.6 points mean (range 40–100 points,
increase 89.3%), being excellent or good in 29 cases and
poor in one; the AKFS was 84 points mean (range 15–100
points, increase 76.3%), as excellent in 25, fair in one and
poor in four cases; all these results were statistically dif-
ferent from the stiff group (Table 2). The range of motion
gained postoperatively for the control group was 106.6°
(range 90°–120°), being statistically different, while the
mean increase was of about 60%, and this was not statis-
tically different from the stiff group.
Radiological results
The radiological results for both groups are given in
Table 3. No significative differences were noted between
the two groups neither with respect to alignment of the
mechanical axis nor for angles of positioning of the pros-
thetic components. There was no patient, neither in the
study nor in the control group, with scores of radiolucent
line significant for prosthetic loosening [11].
Discussion
The arthroplasty procedure in a stiff or ankylosed knee joint
could be a challenging situation for the orthopaedic surgeon,
both for difficult surgical exposure and for complications
that often affect these patients, during and after surgery.
However, also applying strict rehabilitation protocols, in
most cases only a fair result is reachable, compared to
patients with flexible knee [3, 4, 6, 17–21]. In the last
20 years, various studies have been dedicated to describe
the results obtained by treating these patients with knee
arthroplasty procedure. Nevertheless, most of them include
only a small number of patients [1, 8], while the only one
series dedicated to comparative study between stiff and
flexible knee lacks statistical power [9]. Mullen JO, con-
sidering a stiff knee as one with less than 90° arc of flexion,
found no differences between a limited cohort of stiff knee
and a control group with normal arc of flexion; however,
without affecting a match control, they do not consider the
possibility of selection bias between the two groups.
We agree with other authors that a stiff knee should be
considered as a knee with less than 50° of ROM [1–6]. In
our series, we would like to compare results of total knee
arthroplasty between two groups of patients, thinking that
selection bias should be minimised by known prognostic
factors. That is why we conducted a case control study
between two groups with the same number of cases and
matched for age, mechanical limb alignment, patient cat-
egory, preoperative AKS score and length of follow-up.
Furthermore, all the patients of the two groups have been
treated surgically by the same senior author (D.T.), thus
utilising always the same surgical criterion during exposure
and soft-tissue release.

123
Musculoskelet Surg (2011) 95:7–12
Fig. 2 Left knee osteoarthritis in sequelae of osteomyelitis (a) and 3 years after the arthroplasty procedure (b). Considering the extensive soft-
tissue release and the axial deviation, a hinged prosthesis with intramedullary stems were used
Nevertheless, our study presents several limitations: the
retrospective nature of the study and the lack of match
between the two groups for model of prosthetic implants
used; in fact in our series in three cases, it was necessary to
use hinged prosthesis, due to severe instability of the joint.
Cause to underlying disease, stiff knees usually do not
show problems of instability; so that treating these patients,
cruciate retaining or posterior stabilized prosthesis were the
implants of choice in the majority of the cases, while more
constrained implants were reserved only for cases that
required severe soft-tissue dissection. On the contrary,
Bhan et al. [3] recommend using constrained condylar
prostheses in all patients with ankylosed knees, in order to
substitute for deficient or absent collateral ligaments. Our
general approach to preserve as much bone and soft tissues
as possible, thus implanting only minimally constrained
models whenever possible and with the possibility to
change during surgery, agree with previous reports for the
management of patient with limited articularity of the knee
[2, 4, 5, 10, 11]. Analysing our series, we found like similar
experiences [2–4, 6] that patients with different prosthetic
model did not gain different clinical results, neither have
different incidence of complications, although a compari-
son of results is prone to inadequate power. There is little
published information regarding results of TKAs in stiff
knees compared to flexible knees. The majority of authors
agree on considering worse results achieving in stiff knees;
nevertheless, by our match control study, it appears an
important data: the stiff and control group, although having
homogeneous preoperative clinical (AKSS) and functional
(AKFS) scores (Table 1), gained different results since the
arthroplasty procedure. The results were better for patients
of control group with an improvement of 89.3% for the
11
AKSS (from 29.3 to 92.6 points) and of 76.3% for the
AKFS (from 37.8 to 84 points). Differently in the stiff
group, the increase rate was of 76.8% for the AKSS (from
31.3 to 85.2) and of 54.9% for the AKFS (from 34 to 70.8
points).
As expected by analysing the literature [7–9, 17], the
final ROM was better in mobile knees than in stiff or
ankylosed knees, with 106° against 87° degrees, respec-
tively. Nevertheless, the percentage of ROM improvement
was greater in stiff group: an increment of 64.4% in the
stiff group and 59.9% in the control group, thus demon-
strating the efficacy of TKA also in these difficult cases;
our results of postoperative ROM (mean 87.4°, range 25°–
125°) are comparable to those of the literature in which is
reported a mean postoperative arc of flexion in stiff knees
that range from 74° to 103°[4, 6, 11, 17–21].
Another important issue in patients with stiff knees is
the high complication rate, with the most frequent prob-
lems reported that are deep infective process and superfi-
cial wound complications. In our series, although
nonsignificant statistical outcomes, the incidence of com-
plications is higher in patients with stiff knee (21.8% of
cases) than in patients with flexible knees (6.2% of cases).
Even though the complication rate could appear inade-
quate, our results join with those of previous studies, in
which the reported complications rate ranged from 6 to
66% [4, 6–8, 18–20].
The increased risk of infective complications in stiff
knee is connected to extensive soft-tissue exposure, longer
operating time and wound problems often associated with
previous incision [22, 23]. One case in our series showed,
before the arthroplasty procedure, a complex wound
situation, due to previous intervention of membrane
123
12
arthroplasty of the knee 40 years before. This patient
4 months postoperatively developed a skin necrosis and
needed a myocutaneous flap reconstruction (Fig. 2).
Blasier and Matthews [16] think that the area of skin
most sensitive and prone to wound problems is that over-
lying the upper tibia. In this contest, tibial tubercle oste-
otomy as described by Whiteside and Ohl [23], which
consists in medializing the tubercle and reattaching it in a
higher position, could increase the tension of the overlying
skin flaps, thus increasing risk of skin necrosis. We expe-
rienced this complication in a patient who had an extensive
approach associated with TTO.
Conclusions
Stiff knee represents a challenge for the orthopaedic sur-
geon, due to difficulties to obtain a good articular exposure
and high complication rate compared to flexible knee.
Although optimal clinical and functional results seem to be
difficult to obtain, we have showed with our limited cohort
of patients that a better improvement on ROM of patients
with stiff knee is reachable, permitting a suboptimal clin-
ical and functional outcome. Anyway, high risk of com-
plications remains performing arthroplasty even by an
expert surgeon, so patients should be advised that the
outcome could not be guaranteed and that they would need
one or more revisions in their lifetime.
Acknowledgments We thank Mrs Elettra Pignotti for the statistical
support.
Conflict of interest None.
References
1. Aglietti P, Windsor RE, Buzzi R, Insall JN (1989) Arthroplasty
for the stiff or ankylosed knee. J Arthrop 4:1–5
2. Rajgopal A, Ahuja N, Dolai B (2005) Total knee arthroplasty in
stiff and ankylosed knees. J Arthrop 20:585–590
3. Bhan S, Malhotra R, Kiran EK (2006) Comparison of total knee
arthroplasty in stiff and ankylosed knees. Clin Orthop Relat Res
451:87–95
123
View publication stats
Musculoskelet Surg (2011) 95:7–12
4. Montgomery WH, Insall JN, Haas SB, Becker MS, Windsor RE
(1998) Primary total knee arthroplasty in stiff and ankylosed
knees. Am J Knee Surg 11:20–23
5. McAuley JP, Engh GA (2003) Constraint in total knee arthro-
plasty. J Arthrop 18:51–54
6. Kim YH, Kim JS (2009) Does TKA improve functional outcome
and range of motion in patients with stiff knees? Clin Orthop
Relat Res 467:1348–1354
7. Gatha NM, Clarke HD, Fuchs R, Scuderi GR, Insall JN (2004)
Factors affecting postoperative range of motion after total knee
arthroplasty. J Knee Surg 17:196–202
8. Bradley GW, Freeman MA, Albrektsson BE (1987) Total pros-
thetic replacement of ankylosed knees. J Arthrop 2:179–183
9. Mullen JO (1983) Range of motion following total knee arthro-
plasty in ankylosed joints. Clin Orthop Relat Res 179:200–203
10. Kim YH, Cho SH, Kim JS (1999) Total knee arthroplasty in bony
ankylosis in gross flexion. J Bone Joint Surg Br 81:296–300
11. Kim YH, Kim JS, Cho SH (2000) Total knee replacement after
spontaneous osseous ankylosis and takedown of formal knee
fusion. J Arthrop 15:453–460
12. Insall JN, Dorr LD, Scott WN (1989) Rationale of the knee
society clinical rating system. Clin Orthop Relat Res 248:13–14
13. Insall JN (1971) A midline approach to the knee. J Bone Joint
Surg Am 53:1584–1586
14. Garvin KL, Scuderi G, Insall JN (1995) Evolution of the quad-
riceps snip. Clin Orthop Relat Res 321:131–137
15. Ewald FC (1989) The knee society total knee arthroplasty roe-
tgenographic evaluation and scoring system. Clin Orthop Relat
Res 248:9–12
16. Epps CH (1994) Complications of prosthetic knee arthroplasty.
In: Epps CH Jr (ed) Complications in orthopaedic surgery, 3rd
edn. Lippincott, Philadelphia, pp 1056–1086
17. McAuley JP, Harrer M, Ammeen D, Gerard E (2002) Outcome of
knee arthroplasty in patients with poor preoperative range of
motion. Clin Orthop Relat Res 404:203–207
18. Bae DK, Yoon KH, Kim HS, Song SJ (2005) Total knee
arthroplasty in stiff knees after previous infection. J Bone Joint
Surg Br 87:333–336
19. Naranja RJ Jr, Lotke PA, Pagnano MW, Hanssen AD (1996)
Total knee arthroplasty in a previously ankylosed or arthrodesed
knee. Clin Orthop Relat Res 331:234–237
20. Schurman JR 2nd, Wilde AH (1990) Total knee replacement after
spontaneous osseous ankylosis: a report of three cases. J Bone
Joint Surg Am 72:455–459
21. Scott RD, Siliski JM (1985) The use of a modified V-Y quadri-
cepsplasty during total knee replacement to gain exposure and
improve flexion in the ankylosed knee. Orthopedics 8:45–48
22. Dennis DA (1997) Wound complications in total knee arthro-
plasty. Instr Course Lect 46:165–169
23. Whiteside LA, Ohl MD (1990) Tibial tubercle osteotomy for
exposure of the difficult total knee arthroplasty. Clin Orthop Relat
Res 260:6–9