0198-0211/98/1908-0555$03.
00/0
FOOT & ANKLE INTERNATIONAL
Copyright © 1998 by the American Orthopaedic Foot and Ankle Society, Inc.
Ankle Fracture Classification: A Comparison of Reliability of Three X-ray
Views Versus Two
Michael E. Brage, MD.: Matthew Rockett, D.P.M.,t Robert Vraney, MD.,::j: Robert Anderson, MD.,§ and
ABSTRACT
Our hypothesis was that malleolar ankle fractures could
be classified with two radiographic views as reliably as
with three views. Four different observers independently
evaluated 99 sets of ankle radiographs. The examiners
classified the ankle fractures by using both the Lauge-
Hansen and Danis-Weber systems. The interobserver
and intraobserver variations were analyzed by kappa
statistics.
With regard to intraexaminer reliability, the examiners
demonstrated excellent accord in classifying the frac-
tures in the Danis-Weber system with either three views
or two views. The kappa values were comparable. In the
Lauge-Hansen system, three examiners demonstrated
excellent accord and one examiner demonstrated good
accord in classifying the fractures. Similar kappa values
were generated when examiners classified fractures
with either three views or two views. With regard to
interexaminer reliability, good to excellent accord was
demonstrated overall among the four examiners when
they used the Danis-Weber system with either three
views or two views. The examiners were in good agree-
ment when they used the Lauge-Hansen system. Similar
kappa values were generated whether the examiners
used three views or two views.
Three radiographic views are usually ordered for eval-
uation of an acute ankle injury. Previous studies have
shown that only two views are needed for diagnosis of a
malleolar ankle fracture. This study demonstrates that
malleolar ankle fractures can be classified with two
views, lateral or mortise, with a reliability as good as that
achieved with three views. The best agreement is
achieved with lateral and mortise views.
University of Chicago Hospitals and Clinics, Chicago, Illinois
60637.
* Assistant Professor of Surgery, UCSD Medical Center, San
Diego, California. To whom requests for reprints should be ad-
dressed at UCSD Medical Center, 200 West Arbor Drive, 8894, San
Diego, CA 92103.
t Resident in Podiatry.
:j: Resident in Orthopaedic Surgery.
§ Resident in Emergency Medicine.
II Assistant Professor of Biostatistics.
555
Downloaded from fai.sagepub.com at CARLETON UNIV on May 10, 2015
Alicia Toledano, Sc.D.JJ
Chicago, Illinois
INTRODUCTION
In general, the management of ankle fractures in-
cludes diagnosis, classification, and treatment. Radio-
graphs of the injured ankle are a prerequisite to
management. Major textbooks recommend three ra-
diographic views of the ankle for proper examination:
the AP, mortise, and lateral views. 2 ,16 ,17
Recently, investigators have challenged the need for
three radiographic views in the diagnosis of malleolar
ankle fractures. 3 ,13 ,14 Vangsness and colleagues 13 re-
ported that when they diagnosed malleolar fractures
of the ankle, the overall accuracy of two views was not
statistically different from the accuracy of three views.
Other investigators are in agreement, citing two views
as adequate for the radiological examination of a frac-
tured ankle.3 ,14
Ankle fractures are most commonly classified ac-
cording to the systems of Lauge-Hansen and Danis-
Weber. To our knowledge, whether malleolar ankle
fractures can be classified reliably from two radio-
graphic views has not been investigated. Our hypoth-
esis was that malleolar ankle fractures could be clas-
sified with two radiographic views as reliably as with
three views, regardless of the classification system
used.
MATERIALS AND METHODS
We obtained a computer-generated list of ankle
fractures treated either surgically or in the outpatient
clinics at the University of Chicago from 1994 through
the present. There were 99 sets of ankle radiographs
chosen for this study after careful scrutiny for appro-
priate AP, lateral, and mortise views.
Four different observers, including an attending or-
thopaedic surgeon (examiner one), a third-year podi-
atry resident (examiner two), an emergency room phy-
sician (examiner three), and a third-year orthopaedic
resident (examiner four), independently and under
identical conditions, classified the ankle radiographs.
556 BRAGE ET AL. Foot & Ankle InternationallVol. 19, No. 8/August 1998

Each examiner evaluated the radiographs 6 times at

2-week intervals; twice with all three views, twice with
mortise and lateral views, and twice with AP and lat-
eral views. The patient data were obscured.
Before classifying the radiographs, the examiners
met to review the classification systems. The fracture
patterns described by Lauqe-Hansen'' " ? and by
Danis-Weber4 ,1 5 are illustrated in Figures 1-7.
The examiners classified each radiograph by using
both classification systems. Two extra categories,
"not classifiable" and "not fractured," were added.
This gave 17 possibilities for classification according
to the Lauge-Hansen system: supination-adduction
(SA) stages I-II, supination-eversion (SE) stages I-IV,
pronation-abduction (PA) stages I-III, pronation-ever-
sion (PE) stages I-IV, pronation-dorsiflexion (PO)
stages I-II, not classifiable, and not fractured. For the
Danis-Weber system, five possibilities existed: A, B,
C, not classifiable, and not fractured.
The inter- and intraobserver variations were ana-
lyzed by kappa statistics.' Kappa calculates a statis-
tical measure of interrater agreement in the case of
more than two outcomes when the number of raters
is fixed. The kappa-statistic measure of agreement is
Fig. 1. Supination-adduction fracture. Roman numerals indicate
scaled to be zero when the amount of agreement is
the stages of the injury. what would be expected to be observed by chance
and to be one when there is perfect agreement. For
intermediate values, we defined kappa values greater
than 0.75 as showing excellent accord, values be-
tween 0.75 and 0.50 as showing good accord, and
values below 0.50 as showing poor accord." Interob-

Fig. 2. Supination-eversion fracture (left),

Only the usual bony injuries are depicted,
The fibula fracture (right) extends posteri-
orly and obliquely.

Downloaded from fai.sagepub.com at CARLETON UNIV on May 10, 2015

Foot & Ankle InternationallVol. 19, No. 8/August 1998 ANKLE FRACTURE CLASSIFICATION 557

Fig. 3. Pronation-abduction frac-

ture (left). The fibula fracture (right)
is frequently comminuted.

Fig. 4. Pronation-eversion frac-

ture (left). The fibula fracture (right)
is high above the joint.

Downloaded from fai.sagepub.com at CARLETON UNIV on May 10, 2015

558 BRAGE ET AL. Foot & Ankle Internationai/Vol. 19, No. 8/August 1998

•
Type A
Fig. 5. Type A fracture in the Danis-Weber system. The fibula
fracture lies below the level of the tibial plafond.

server kappas were calculated as weighted averages

of kappas for each subgroup versus all others. We
also used kappa statistics to assess agreement of the
Danis-Weber and Lauge-Hansen classification sys-
tems made from two radiographic views with those
made from three views, assuming that the classifica-
tion based on three views is the truth. Type B
Fig. 6. Type B fracture in the Danis-Weber system. The oblique
RESULTS
fibula fracture starts at the level of the joint and extends proximally
and posteriorly.
Interobserver Reliability

Interobserver reliability with the Danis-Weber clas-

sification system was good for classifications based
on three radiographic views and was excellent for
those based on two views (Table 1). Interobserver
reliability with the Lauge-Hansen classification system
(by fracture pattern and by stages within the fracture
pattern) was good, with kappas for classifications
made from two views slightly higher than those made
from three views (Table 1).
traobserver reliability for all observers with the Lauge-
Hansen fracture patterns was also excellent (Table 2).
Similar results were obtained for all 17 possibilities for
the Lauge-Hansen classification system (Table 2). In
all cases, the average reliability for classifications
based on two views was at least as high as that for
classifications based on three views.

Intraobserver Reliability Agreement of Two Views with Three Views

Intraobserver reliability was excellent for all observ- The agreement of the classifications based on two
ers with the Danis-Weber classification system (Table views with those based on three views among the
2). With one exception of good agreement, the in- examiners tended to be excellent for Danis-Weber

Downloaded from fai.sagepub.com at CARLETON UNIV on May 10, 2015

Foot & Ankle Internationai/Vol. 19, No. 8/August 1998
Type C
Fig. 7. Type C fracture in the Danis-Weber system. The fibula
fracture lies well above the level of the ankle joint.
classifications on both the first and second readings
(Table 3). Agreement of the classifications for two
views with those for three views and Lauge-Hansen
fracture patterns was excellent for two observers on
both readings and for one additional observer on one
reading and good otherwise (Table 3). Agreement of
the classifications based on two views with those
based on three views and staging within the Lauge-
Hansen fracture patterns was good for two observers
on both readings and for one additional observer for
lateral and AP views and excellent otherwise (Table 3).
Downloaded from fai.sagepub.com at CARLETON UNIV on May 10, 2015
ANKLE FRACTURE CLASSIFICATION 559
TABLE 1
Interobserver Reliability for Four Observers Classifying 99
Ankle Images
Three Lateral and Lateral
views mortise and AP
Danis-Weber
First reading 0.6860 0.8616 0.8040
Second reading 0.7221 0.8146 0.7760
Lauge-Hansen (fracture patterns)
First reading 0.5766 0.6448 0.6277
Second reading 0.5969 0.6616 0.6208
Lauge-Hansen (stages within the fracture patterns)"
First reading 0.5925 0.6301 0.6160
Second reading 0.6337 0.6461 0.6189
B Based on categories used by observers.
DISCUSSION
The management of malleolar ankle fractures in-
volves diagnosis, classification, and treatment. Three
radiographic views of the ankle are usually ordered:
the AP, the mortise, and the lateral. The AP radiograph
is taken in the long axis of the foot, with the beam
centered on the ankle joint. The lateral view is taken
with the foot in a neutral position, with the x-ray beam
directed perpendicular to the medial malleolus. The
mortise view is taken with the leg internally rotated
until the fifth metatarsal is inclined 10 to 15° from the
plane perpendicular to the x-ray cassette." In general,
the AP and mortise views allow identification of gross
lateral subluxation of the talus and of transverse or
oblique fractures of the malleoli. The lateral view is
useful in determining the fracture configuration of the
lateral malleolus and in demonstrating fractures of the
anterior and posterior tibial articular surfaces.
Currently, there is a dispute to determine whether
the acutely injured ankle requires three radiographic
views for proper evaluation. Wallis 14 retrospectively
reviewed 945 patients whose ankles were radio-
graphed because of acute trauma. His objective was
to determine whether the mortise view provided suffi-
cient additional information to warrant its inclusion in
the routine radiographic series for acute ankle trauma.
First, he examined the AP and lateral films for soft
tissue swelling and bony injury. Then he inspected the
mortise view for the presence of bony injury and to get
a subjective opinion to determine whether a previously
noted fracture was better visualized. Of the 128 frac-
tures identified, 95.3% could be identified on the AP
and lateral views. Six additional fractures were identi-
fied only on the mortise view, and all were avulsion
fractures of the distal fibula. No major fracture was
missed when only the AP and lateral views were used.
Cockshott and colleaques" prospectively evaluated
242 patients examined at their hospital within 24 hours
560 BRAGE ET AL. Foot & Ankle International/Vol. 19, No. 8/August 1998

of having sustained ankle injuries. Radiographs re-

1'-000CO
,....coco vealed fractures in 29% of the patients. The AP and
,....l!)0
COO'lCO lateral films identified all of the fractures. These inves-
cicici
tigators concluded that the mortise view provided no
additional diagnostic information.
C\JC\JCO
Vangsness and colleagues 13 determined the accu-
000'l
C\Jl!)0 0'lC\J1'-
'<1'00 '<1'00 racy and agreement of the diagnosis of ankle fractures
C\J C\J C\J C\J C\J C'J
among four examiners who twice evaluated 123 sets
of ankle radiographs: once with all three views and
once with only the lateral and mortise views. The
I'-O'll'-
CO CO CO
COI'-CO
COO'lCO
COI'-C\J
COCOO'l overall accuracy for all physicians and all types of
0000
CO CO CO
Ll'i0Ll'i
COO'lCO
""':cci""':
COCOO'l malleolar fractures ranged from 92.5% to 98.4%. The
accuracy of using two radiographic views was within
the 95% expected threshold of accuracy for three
views. No strong differences among the physician
observations were noted. The investigation of missed
fractures found all of them to have a subtle, nondis-
ell C\Jl!)to COl!)CO 0'l0C")
0..
0..
COl!)O'l
tol'-l!)
00e;;
00C\J
C\JC\JO
l!)1'-C\J
COl!)l!)
O'll'-I'-
placed fracture configuration. These authors con-
CO CO CO 000'l COI'-CO CO coco
~ cicici r=-r-=ci cicici cicici cluded that the lateral and mortise views alone were
adequate for initial ankle fracture diagnosis or screen-
COCOO'l I'-C")CO o O'l 0
ing.
l!)1'-l!) coo,.... OCOto
After the diagnosis is made, the ankle fracture is
o~o OOci0
l!)coto
cir--:r--:
to coco to to CO
classified. Our purpose in this study was to determine
whether ankle fractures can be classified with two
..
C'II.S!
001'-
000'>
cici0
000'l
,.... ,....
CO to CO
COO'lO'l
r--:Ll'iLl'i
COO'lO'l
radiographic views as reliably as with three views.
Both the Danis-Weber and Lauge-Hansen classifica-
w~
...I ._ tion systems were evaluated because these two sys-
m:: tems are commonly used. The examiners were at dif-
<l:.g
I- .-
(jj ferent levels of training, and two were in fields of
a::
medicine other than orthopaedics. All of the physi-
~ l!) I'-C\J,....
cians in the study are involved in practices in which
~ ell
0.. 00"l" l!)1'-l!) they routinely evaluate acute ankle injuries.
..c 0.. OOCO O'lC")CO
o 000'l I'-COCO
~ ~r=o cicici The examiners attained good reliability with the
....f!
.5 Lauge-Hansen system. The kappa values generated
were similar but tended to be slightly higher when the
examiners used two radiographic views. The examin-
ers achieved excellent reliability with the Danis-Weber
system when given two x-ray views and good reliabil-
C")I'-I'- 000'> "l"COI'- C")I'-O'l
0'l0'l0'l 000'> 0'l0'l0'l 0'l0'l0'l ity when given the three views. Thus, the classification
N00 cidOO C'5Ll'i0 N0cO
0'l0'l0'l 0
,.... 0,....0 ' l 0'l0'l0'l O'l O'l O'l ]5 of malleolar ankle fractures among examiners was
reliable regardless of whether two or three radio-
~g> graphic views were used.
As individuals, the examiners achieved excellent in-
II
Qm traobserver reliability by using two or three x-ray views
and the Danis-Weber system. The examiners also
achieved good to excellent results with the Lauge-
Hansen system. Thus, the intraobserver classification
of malleolar ankle fractures was reliable with either two
or three radiographic views.
Greater reliability was achieved with the Danis-We-
ber system than with the Lauge-Hansen system. Our
explanation is that the Danis-Weber system is easier
to use because it is simpler and it includes fewer
fracture patterns and stages. Nevertheless, the exam-

Downloaded from fai.sagepub.com at CARLETON UNIV on May 10, 2015

 6l
o
......
Qo
::t>
::J
Q:
<D
s
CD
:3
tll

TABLE 3
g.
Agreement Between Classifications Made with Two and Three Views ~
::::::
Lauge-Hansen Lauge-Hansen (stages within ~
:-
Danis-Weber
(fracture patterns) the fracture patterns) .....
!O
Po' (%) P/ (%) Kappa Po' (%) Pe' (%) Kappa Po' (%) P/ (%) Kappa
~
Examiner 1. First Reading 00
Lateral and mortise viewsv/ 95.96 70.38 0.8636 91.92 56.62 0.8137 85.86 25.17 0.8110 s
Lateral and AP viewsx/ 95.96 70.37 0.8636 90.91 55.35 0.7964 81.82 24.86 0.7580 S
Second Reading lii......
Lateral and mortise viewsx/ 92.93 67.99 0.7791 92.93 56.63 0.8370 89.90 25.49 0.8644
Lateral and AP views-c/ 91.92 70.98 0.7215 93.94 55.50 0.8638 89.90 25.37 0.8646 ~
Examiner 2. First Reading 00

Lateral and mortise viewsx/ 97.98 70.62 0.9312 97.98 61.15 0.9480 88.89 25.55 0.8508
Lateral and AP viewsx/ 95.96 70.72 0.8620 95.96 60.62 0.8974 77.78 26.44 0.6979
Second Reading
Lateral and mortise views'x/ 97.98 70.62 0.9312 97.98 61.15 0.9480 87.88 25.30 0.8377
Lateral and AP viewsx/ 96.97 70.81 0.8962 93.94 60.52 0.8465 77.78 25.66 0.7011
Examiner 3. First Reading
Lateral and mortise views x/ 92.93 70.37 0.7614 86.87 57.37 0.6920 81.82 27.90 0.7478
Lateral and AP viewsx/ 92.93 69.62 0.7673 88.89 61.53 0.7111 76.77 30.28 0.6668
Second Reading
:to>
Lateral and mortise views x/ 94.95 75.15 0.7968 87.88 61.61 0.6843 78.79 29.53 0.6990 Z
Lateral and AP viewsx/ 95.96 74.40 0.8422 85.86 63.13 0.6165 77.78 30.78 0.6790 A

Downloaded from fai.sagepub.com at CARLETON UNIV on May 10, 2015

r
Examiner 4. First Reading m
Lateral and mortise views x/ 93.94 68.11 0.8100 87.88 62.09 0.6803 71.72 27.62 0.6092 T1

Lateral and AP viewsx/ 91.92 68.00 0.7474 84.85 60.61 0.6154 72.73 26.53 0.6288
Second Reading
~
--i
Lateral and mortise viewsx/ 95.96 71.22 0.8596 93.94 65.38 0.8249 80.81 28.45 0.7318 C
::D
Lateral and AP vtewsx/ 95.96 72.00 0.8557 92.93 66.84 0.7868 81.82 29.44 0.7423 m
'Po =observed agreement, Pe = agreement expected due to chance. o
y compared with all three views. ~
(j)
:;:;
g
o
z

U'1
Q)
......
562 BRAGE ET AL.
iners did achieve good reliability with the Lauge-Han-
sen system; this contrasts with results from a previous
study. In that study, Thomsen and colleagues 12 eval-
uated the observer variation of four examiners inde-
pendently classifying 94 ankle radiographs by using
the Lauge-Hansen and Danis-Weber systems. Their
results showed that the examiners had poor reliability
when classifying fractures into the stages of the
Lauge-Hansen system. The difference between this
study and ours may be in the quality of the radio-
graphs used. Thomsen and colleagues 12 noted that
not all of their radiographs were of high quality. The
radiographs in our study were carefully chosen for
appropriate AP, lateral, and mortise views.
The examiners achieved better reliability with the
Danis-Weber system when using two views than they
did when using three views. The classifications in the
Danis-Weber system are based solely on the location
of the fibula fracture. We believe that, occasionally, the
obliquity of the fibula fracture may manifest differently
on the AP view or on the mortise view. This obliquity
may, at times, have caused the fracture to be classi-
fied differently for a particular examiner. This circum-
stance improved the interobserver results, but it did
not affect the results of the examiners as individuals.
Vangsness and colleagues 13 concluded that the lat-
eral and mortise views alone were sufficient for diag-
nosis of malleolar ankle fractures. In our study, when
we classified ankle fractures, the lateral and mortise
views gave better reliability among the examiners, but
they did not affect the intraobserver reliability. Neither
the examiner's specialty nor the examiner's level of
training affected our results.
Currently, we recommend using the lateral and mor-
tise views for classifying ankle fractures. The use of
two views instead of three will save time and money
and will reduce the patient's exposure to x-rays.
REFERENCES
1. Altman, D.G. (ed.): Practical Statistics for Medical Research,
1st Ed., Vol. 1. London, Chapman and Hall, 1991, pp. 403-409.
Downloaded from fai.sagepub.com at CARLETON UNIV on May 10, 2015
Foot & Ankle Internationai/Vol. 19, No. 8/August 1998
2. Chapman, M.W.: Fractures and fracture-dislocations of the
ankle. In Mann, RA, and Coughlin, M.J. (eds.), Surgery of the
Foot and Ankle, 6th Ed., Vol. 2. St. Louis, C.V. Mosby, 1993, pp.
1439-1464.
3. Cockshott, W.P., Jenkin, J.K., and Pui, M.: Limiting the use of
routine radiography for acute ankle injuries. Can. Med. Assoc.
J., 129:129-131, 1983.
4. Danis, R. (ed.): Les Fractures Malleolaires. Theorie et Pratique
de L'osteosynthese. Paris, Masson et Cie, 1949, pp. 133-165.
5. Goergen, T.G., Danzig, L.A., Resncik, D., and Owen, C.A.:
Roentgenographic evaluation of the tibiotalar joint. J. Bone
Joint Surg., 59A:874-877, 1977.
6. Lauge-Hansen, N.: Fractures of the ankle: analytic historic
survey as the basis of new experimental, roentgenologic, and
clinical investigations. Arch. Surg., 56:259-317, 1948.
7. Lauge-Hansen, N.: Fractures of the ankle, II: combined exper-
imental-surgical and experimental-roentgenologic investiga-
tions. Arch. Surg., 60:957-985, 1950.
8. Lauge-Hansen, N.: Fractures of the ankle, IV: clinical use of the
genetic roentgen diagnosis and genetic reduction. Arch. Surg.,
64:488-500, 1952.
9. Lauge-Hansen, N.: Fractures of the ankle, V: pronation-dorsi-
flexion fracture. Arch. Surg., 67:813-820, 1953.
10. Lauge-Hansen, N.: Fractures of the ankle, III: genetic roent-
genologic diagnosis of fractures of the ankle. Am. J. Roentge-
nol., 71:456-471, 1954.
11. Svanholm, H., Starklint, H., Gundersen, H.J.G., Fabricius, J.,
Barlebo, H., and Olsen,S.: Reproducibility of histomorpho-
logic diagnoses with special reference to the kappa statistic.
Acta Pathol. Microbiol. Immunol. Scand., 97:689-698, 1989.
12. Thomsen, N.B., Overgaard,S., Olsen, L.H., Hansen, H., and
Nielson, S.T.: Observer variation in the radiographic classifica-
tion of ankle fractures. J. Bone Joint Surg., 73B:676-678, 1991.
13. Vangsness, C.T., Carter, V., Hunt, T., Kerr, R., and Newton,
E.: Radiographic diagnosis of ankle fractures: are three views
necessary? Foot Ankle Int., 15:172-174, 1994.
14. Wallis, M.G.: Are three views necessary to examine acute ankle
injuries? Clin. Radiol., 40:424-425, 1989.
15. Weber, B.G.: Die Verletzungen des oberen Sprunggelenkes,
2nd Ed., Bern, Verlag Hans Huber, 1972.
16. Weber, M.J.: Ankle fractures and dislocations. In Chapman,
MW. (ed.), Operative Orthopaedics, 2nd Ed., Vol. 1. Philadel-
phia, J.B. Lippincott, 1993, pp. 731-745.
17. Wilson, F.C.: Fractures and dislocations of the ankle. In Rock-
wood, CA, Jr., and Green, D.P. (eds.), Rockwood and Green's
Fractures in Adults, 2nd Ed., Vol. 2. Philadelphia, J.B. Lippin-
cott, 1975, pp. 1665-1701.