The Journal of Arthroplasty Vol. 21 No.

3 2006

A Proposed Classification of Supracondylar Femur

Fractures Above Total Knee Arthroplasties

Edward T. Su, MD, Erik N. Kubiak, MD, Hargovind DeWal, MD,

Rudi Hiebert, BS, and Paul E. Di Cesare, MD

Abstract: A new classification system is proposed for supracondylar femur fractures

above total knee arthroplasties based on fracture location relative to the femoral
component. Radiographs of 28 cases were evaluated and classified according to the
proposed system by 12 physicians: 3 trauma specialists, 3 adult reconstruction
specialists, 3 musculoskeletal radiologists, and 3 orthopaedic residents. The same
12 physicians reevaluated the same 28 cases 3 months later. The mean reliability
coefficient for all observers was 0.74 (substantial agreement). The coefficient for
reproducibility after 3 months was 0.85 (almost perfect). The power of the study was
80%. The proposed classification system is easy to use and has good interobserver
reliability among orthopaedic residents, orthopaedic attendings—trauma and
reconstruction—and radiologists. Intraobserver reliability was also excellent at
3 months. Key words: periprosthetic, femur, fracture, total knee arthroplasty,
radiographs, classification.
n 2006 Elsevier Inc. All rights reserved.

Supracondylar periprosthetic femur fractures above
total knee arthroplasties are an uncommon com-
plication after total knee arthroplasty, with a
reported incidence of 0.3% to 2.5% [1-3]. These
fractures are most common in elderly patients with
osteoporosis and such other risk factors as rheu-
matoid arthritis, neurological disorders, chronic
steroid therapy, femoral anterior cortical notching,
and revision knee arthroplasty [1,4,5]. The usual
mechanism of injury is a fall directly onto the
flexed knee.
From the Department of Orthopaedic Surgery, Musculoskeletal
Research Center, NYU-Hospital for Joint Diseases Orthopaedic Institute,
New York, New York.
Submitted June 30, 2003; accepted May 15, 2005.
No benefits or funds were received in support of the study.
Reprint requests: Paul E. Di Cesare, MD, Department of
Orthopaedic Surgery, Musculoskeletal Research Center, NYU-
Hospital for Joint Diseases Orthopaedic Institute, 301 East 17th
Street, Room 1501, New York, NY 10003.
n 2006 Elsevier Inc. All rights reserved.
0883-5403/06/1906-0004$32.00/0
doi:10.1016/j.arth.2005.05.022
There are several interpretations of what con-
stitutes bthe supracondylar region of the distal
femur.Q Neer et al [6] in 1967 defined it as the
lower 3 in (7.62 cm) of the femur (see Table 1).
Culp et al [4] specified 9 cm proximal to the knee
joint line as the cutoff point [7]. Sisto et al [2]
included fractures occurring within 15 cm proxi-
mal to the knee joint line. For the purposes of this
paper, we define a supracondylar periprosthetic frac-
ture as one occurring within 15 cm of the joint line
or, in the case of a stemmed component, less than
5 cm from the proximal end of the femur.
Treatment of supracondylar periprosthetic frac-
tures includes both nonoperative means (skeletal
traction, bracing) and surgery (external fixation,
open or closed reduction with internal fixation,
revision arthroplasty with or without distal femoral
allografts). Supracondylar periprosthetic fracture
classification systems—for example, those devised
by DiGioia and Rubash [8], Chen et al [5], and
Rorabeck and Taylor [9,10] (see Tables 2-4)— are
helpful for distinguishing fractures amenable to

405
406 The Journal of Arthroplasty Vol. 21 No. 3 April 2006
Table 1. Neer et al [6] Classification
Type I Undisplaced (b5 mm displacement and/or
b58 angulation)
Type II Displaced N1 cm
A With medial femoral shaft displacement
B With internal femoral shaft displacement
Type III Displaced and comminuted
nonoperative treatment from those requiring sur-
gery; they do not, however, distinguish among
operative treatments.
Surgical repair for these fractures has become
more common as the need to mobilize patients soon
after injury to limit morbidity and mortality has
become better recognized. Several operative treat-
ments have been used to stabilize these special
fractures, including but not limited to intramedul-
lary nails placed both retrograde and antegrade,
buttress plates, fixed-angle plates, and external
fixators. Selection of the appropriate device relies
heavily upon the location of the fracture relative to
the total knee component. This crucial point is
missed by current classification schemes.
For a classification system to be useful, it must be
reliable and reproducible, facilitate interaction
among medical professionals, and aid in improving
conclusions that can be drawn from research. We
have created a new classification system for
periprosthetic femur fractures that should act as a
guide for the treatment of these fractures. The
supracondylar periprosthetic femur fracture classi-
fication proposed herein is straightforward, con-
sisting of 3 types based on the location of the
fracture relative to the femoral component (see
Fig. 1) as determined from standard anteroposterior
and lateral radiographs.
Materials and Methods
Twelve observers were recruited, 3 each of
fellowship-trained trauma specialists, fellowship-
trained adult reconstruction specialists, fellow-
Table 2. DiGioia and Rubash [8] Classification
Group I Extra-articular, undisplaced
(b5 mm displacement or
b58 angulation)
Group II Extra-articular, displaced
(N5 mm displacement or
N58 angulation)
Group III Severely displaced
(loss of cortical contact)
or angulated (N108); may have
intercondylar or T-shaped component
Table 3. Chen et al [5] Classification
Type I Nondisplaced (Neer type II)
Type II Displaced and/or comminuted (Neer types II and III)
ship-trained musculoskeletal radiologists, and
orthopaedic residents.
Initially, observers classified 28 consecutive cases
of periprosthetic femur fractures above total knee
arthroplasties from good-quality anteroposterior
and lateral radiographs collected by the first author
(ES, who was not among the observers; cases were
excluded in which the knee component was
stemmed or had an intercondylar box) according
to the new classification system outlined in Fig. 1.
Cases were identified from our emergency room
logs from 1996 to 2002.
Subsequently, the observers were shown the
same 28 radiographs in random order 3 months
later. Observers were not initially informed that
they were to be retested, and radiographs were
made unavailable for review during the 3-month
interval. No time limit was allotted for review of
the radiographs. No questions were allowed while
the radiographs were being reviewed or during the
interval between reviewing sessions.
Statistical Analysis
Study results were analyzed using the j statistic
according to standard criteria described by Landis
and Koch [11], where 0.00 to 0.20 = slight
agreement, 0.21 to 0.40 = fair agreement, 0.41 to
0.60 = moderate agreement, 0.61 to 0.80 = sub-
stantial agreement, and 0.81 to 1.00 = almost
perfect agreement.
Power analysis of the results was performed for
evaluation of intraclass correlation coefficients [12].
Results
Interobserver Reliability
There was perfect agreement among the 4 observ-
er groups for 61% of the fractures (Table 5). The
mean reliability coefficient (j statistic) for ortho-
paedic trauma specialists was 0.85 (almost perfect
Table 4. Rorabeck and Taylor [10] Classification
Type I Undisplaced fracture; prosthesis intact
Type II Displaced fracture; prosthesis intact
Type III Displaced or undisplaced fracture;
prosthesis loose or failing
 Periprosthetic Femur Classification ! Su et al 407

Fig. 1. The 3 types of fracture in the proposed classification system: type I: fracture proximal to the femoral component;
type II: fracture originating at the proximal end of femoral component and extending proximally; type III: fracture in
which any part of the fracture line can be seen distal to the upper edge of the anterior flange of the femoral component.

agreement); for the adult reconstruction specialists,
0.73 (substantial agreement); for the musculoskel-
etal radiologists, 0.64 (substantial agreement); and
for the residents, 0.80 (substantial agreement).
Overall, the mean reliability coefficient for all
observers was 0.74 (substantial agreement).
Intraobserver Reproducibility
correlation coefficient of 0.59. The intraclass corre-
lation coefficient of 0.74 was statistically significant-
ly greater than a lower boundary of 0.59 that by
convention is approximately the demarcation from
bmoderateQ to bsubstantialQ reliability [7].
Discussion

At the second assessment 3 months later, intra-
class correlation between the first and second review
was 0.85 with a lower 95% confidence limit of 0.78
for all physicians combined (Table 6). The mean
reliability coefficient (j statistic) for the orthopaedic
trauma specialists was 0.91 (almost perfect); for the
adult reconstruction specialists, 0.80 (substantial
agreement); for the musculoskeletal radiologists,
0.82 (almost perfect agreement); and for the resi-
dents, 0.88 (almost perfect agreement).
Power Analysis
Twenty-eight samples reviewed by 12 physicians
had 80% statistical power to demonstrate that the
observed intraclass correlation coefficient of 0.74
was statistically distinguishable from an intraclass
Table 5. Interrater Correlation Coefficients
A useful classification system must facilitate com-
munication, education, research, and treatment. To
these ends, the ideal classification system would
ensure both reliable reproduction of evaluations
among observers and reproducible evaluations of a
single observer. This is particularly true in orthopae-
dics for fracture patterns that are seen infrequently
and for which there is limited shared experience
among treating physicians. Classification systems
can guide operative treatment, allowing surgeons
to effectively pool cases and discuss appropriate
treatment options for specific situations. A reliably
reproducible classification scheme facilitates research
by allowing treatment methods to be appropriately
compared. The proposed classification scheme guides
Table 6. Intraclass Correlation Coefficients
(After Repeat Review at 3 Months)

Estimated Lower 95% Estimated Lower 95%

Correlation Confidence Limit Correlation Confidence Limit

All observers 0.74 0.65 All observers 0.85 0.78

Trauma specialists 0.85 0.76 Trauma specialists 0.91 0.86
Adult reconstruction 0.73 0.58 Adult reconstruction 0.80 0.65
specialists specialists
Musculoskeletal radiologists 0.64 0.47 Musculoskeletal radiologists 0.82 0.72
Orthopaedic residents 0.80 0.69 Orthopaedic residents 0.88 0.81
408 The Journal of Arthroplasty Vol. 21 No. 3 April 2006
the surgeon’s operative treatment choice with a high
degree of interobserver reliability and intraobserver
reproducibility, thus greatly improving communica-
tion among health care specialists.
In our opinion, fracture location in relation to the
femoral knee component is the best determinant
of appropriate operative treatment. Although intra-
medullary nails have become increasingly popular
for the treatment of these fractures, they are not
always appropriate. In the authors’ experience,
type I fractures are usually most amenable to
treatment with an antegrade or retrograde intra-
medullary nail (if the femoral component has an
open box to facilitate nail placement). We do not
recommend using metal cutting tools to make a
hole in a closed box to facilitate the use of an
intramedullary nail as the resultant metal particles
may contribute to third body wear. Type II
fractures will likely require treatment with either
a fixed-angle device or retrograde supracondylar
nail (only if the femoral box is open). Type III
fractures may be treated by either a fixed-angle
device (if the segment of bone remaining can
accommodate fixation) or revision arthroplasty
with a stemmed femoral component, possibly in
combination with a distal femoral allograft. If there
is loosening of the femoral component, revision
arthroplasty with a stemmed femoral component
may be required, regardless of the location of the
fracture; however, some surgeons might favor
treatment of the fracture first, with later revision
of the components. The proposed classification
system is easy to use and has good interobserver
reliability and intraobserver reproducibility. Appli-
cation of the proposed classification system toward
appropriate surgical management must await care-
ful scrutiny of clinical results; this system is in use
at our institution and has proven helpful for
guiding treatment for periprosthetic fractures
above total knee components and may help inves-
tigators compare results with different fracture
treatments to determine optimum fracture therapy
in the future.
References
1. Merkel KD, Johnson Jr EW. Supracondylar fracture
of the femur after total knee arthroplasty. J Bone
Joint Surg Am 1986;68:29.
2. Sisto DJ, et al. Treatment of supracondylar fractures
following prosthetic arthroplasty of the knee. Clin
Orthop 1985;265.
3. Ritter MA, et al. Anterior femoral notching and
ipsilateral supracondylar femur fracture in total knee
arthroplasty. J Arthroplasty 1988;3:185.
4. Culp RW, et al. Supracondylar fracture of the femur
following prosthetic knee arthroplasty. Clin Orthop
1987;212.
5. Chen F, et al. Management of ipsilateral supra-
condylar femur fractures following total knee arthro-
plasty. J Arthroplasty 1994;9:521.
6. Neer CS, Grantham SA, Shelton ML. Treatment of
supracondylar fractures following prosthetic arthro-
plasty of the knee. A study of one hundred and ten
cases. J Bone Joint Surg Am 1967;49:591.
7. Cain PR, et al. Periprosthetic femoral fractures
following total knee arthroplasty. Clin Orthop
1986;205.
8. DiGioia III AM, Rubash HE. Periprosthetic frac-
tures of the femur after total knee arthroplasty. A
literature review and treatment algorithm. Clin
Orthop 1991;135.
9. Rorabeck CH, Taylor JW. Periprosthetic fractures of
the femur complicating total knee arthroplasty.
Orthop Clin North Am 1999;30:265.
10. Rorabeck CH, Taylor JW. Classification of peripros-
thetic fractures complicating total knee arthroplasty.
Orthop Clin North Am 1999;30:209.
11. Landis JR, Koch G. The measurement of observer
agreement for categorical data. Biometrics 1977;
33:159.
12. Figgie MP, et al. The results of treatment of supra-
condylar fracture above total knee arthroplasty. J
Arthroplasty 1990;5:267.