ARTICLE IN PRESS

Journal of Biomechanics 37 (2004) 1361–1369

Three-dimensional load measurements in an external fixator

.
K. Seide*, N. Weinrich, M.E. Wenzl, D. Wolter, C. Jurgens
Berufsgenossenschaftliches Unfallkrankenhaus (Trauma Hospital) Hamburg, Bergedorfer StraX e 10, 21033 Hamburg, Germany
Accepted 16 December 2003

Abstract

On the basis of a six-degree-of-freedom adjustable fracture reduction hexapod external fixator, a system which can be used for
measuring axial and shear forces as well as torsion and bending moments in the fixator in vivo was developed. In a pilot study on 9
patients (7 fresh fractures and 2 osteotomies of the tibia), the load in the fixator during the healing process was measured after 2, 4, 8
and 12 weeks and at fixator removal. The measured values enabled both the type of fracture to be determined as well as the
monitoring of the healing process. In well-reduced type A3 fractures small axial (direction of the bone axis) forces were found in the
fixator. A2, B2 and C3 fractures showed distinct axial forces, which decreased during the healing process, according to an increasing
load transfer over the bone. Bending moments in the fixator showed good correspondence with the clinical healing process, except in
the case of a C3 fracture. A combination of bending moment and axial force proved to be particularly suitable to assess fracture
healing. In transverse fractures, the well-known resorption phenomenon of bone in the fracture gap at approximately 4 weeks was
detected by the system. Compared with other external fixator load measurements in vivo, the hexapod offers the advantage of being
able to measure all forces and moments in the fixator separately and with a relatively simple mechanical arrangement. In our
opinion, it will be possible to control fracture healing using this system, thereby minimizing radiation exposure from radiographs.
Furthermore, the measurement system is a step towards the development of external fixator systems that enable automatic
adjustments of the callus mechanical situation (‘‘automatic dynamization’’) and inform the patients about the optimal weight
bearing of their extremity (‘‘intelligent fixator’’).
r 2004 Elsevier Ltd. All rights reserved.

Keywords: Load measurements; Fracture healing; External fixator; Hexapod

1. Introduction investigated in robotics and has been applied in position

Monitoring of the fracture healing process is routinely
done by radiographic examinations which permit form
and intensity of callus calcification to be determined.
However, after primary demonstration of an exact
reduction, the mechanical properties of the osteosynth-
esis are the main determinants for therapeutic decisions.
Estimation of mechanical callus strength from radi-
ological images depends largely on the expertise of the
surgeon, but in some cases the identification of healing
problems by radiological means is very difficult.
A new system for in vivo three-dimensional load
measurements during fracture healing in an external
fixator is introduced in this study. It is based on a hexapod
mechanism also known as parallel robot manipulator
or Stewart platform. This mechanism has been well
*Corresponding author. Tel.: +49-40-7306-2701; fax: +49-40-7306-
2703.
E-mail address: k.seide@buk-hamburg.de (K. Seide).
control applications, for example, in flight simulators for
the three-dimensional positioning of the cockpit (Stewart,
1965). The hexapod mechanism in an external fixator
allows a computer-assisted universal three-dimensional
fracture reduction or deformity correction (Seide and
Wolter, 1996; Seide et al., 1999). By adding 6 uniaxial
force-measuring elements, the hexapod was extended into
a six-spacial-degree-of-freedom measuring fixator.
The goal was to construct a system for monitoring
fracture healing by mechanical measurement and thus
minimize radiation exposure. Therefore parameters
which optimally describe an emerging callus had to be
identified. Furthermore, it had to be shown that
different fracture types in patients can be distinguished
using the system. The hypothesis of the study was that
conclusions drawn from the mechanical parameters
measured are consistent with the decisions of the
treating surgeons made on the basis of the correspond-
ing routine radiographs.

0021-9290/$ - see front matter r 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jbiomech.2003.12.025
 ARTICLE IN PRESS
1362 K. Seide et al. / Journal of Biomechanics 37 (2004) 1361–1369

2. Material and methods

The system is based on a modified ring fixator.

Instead of parallel interconnecting longitudinal bars,
three pairs of ball joints are attached at the proximal
and distal ring. The joints are connected by six linear
actuator elements (‘‘distractors’’) in the form of
circularly arranged triangles (Figs. 1 and 2). Bone
fixation was performed either by Kirschner wires
(Ilizarov ring fixator configuration) or with Schanz’
screws. The mechanism was also mounted secondarily to
AO, Hoffman or Orthofix fixators.
Force sensors (Burster, measuring range 1000 N with
an accuracy of 1% of the final value and a maximum
load of 3000 N) were inserted in line with the distractor
elements of the hexapod fixator. This insertion resulted
in measuring distractors, which were mechanically
compatible with the original non-measuring ones. The
sensors were connected to a PC-based measurement
system (National Instruments, DAQ AT-MIO-64F-5).
A measuring software was developed using Microsoft
Visual Basics.
The peculiarity of the hexapod is based on the fact
that the ball joints that connect the rings to the linear
actuators or force sensor elements are not blocked. So
bending, torsional or shear load transfer is theoretically
not possible along these elements and there is exactly
one load component (axial) carried and measured by
each unit. Additionally, according to the theory of the
hexapod kinematics, there is no interrelationship be- Fig. 2. Patient during measurement of fixator forces and moments
under axial load of the leg.
tween the units and thus exactly six independent degrees
of freedom are allowed.
From the force values FD1 yFD6 measured in the six
distractor elements, the components of the load L in the as well as the moments Mx ; My ; Mz with respect to the
fixator, i.e., the axial force Fz and the shear forces Fx ; Fy three spacial axes can be calculated. The calculation
provides transformation of six-degree-of-freedom linear
force values to 6-degree-of-freedom spatial force and
moment values. The transformation was mathematically
realized with a matrix multiplication of the distractor
force vector with a 6
6 transformation matrix:
0 1
Fx
B C
B Fy C
B C
BF C
B z C
L ¼B C
B Mx C
B C
BM C
@ yA
Mz
0 10 1
f1x f2x f3x f4x f5x f6x FD1
B CB C
B f1y f2y f3y f4y f5y f6y CB FD2 C
B CB C
B f f2z f3z f4z f5z f6z C B C
B 1z CB FD3 C
¼B CB C: ð1Þ
B m1x m2x m3x m4x m5x m6x CB FD4 C
B CB C
Bm CB C
@ 1y m2y m3y m4y m5y m6y A@ FD5 A
Fig. 1. Measuring hexapod external fixator.
m1z m2z m3z m4z m5z m6z FD6
 ARTICLE IN PRESS
K. Seide et al. / Journal of Biomechanics 37 (2004) 1361–1369
z In order to facilitate the evaluation of the quantities
Mz
fulcrum K oi
Fz
ri
Mxy Fi
Di
callus
bone
ui
y and moments were calculated from the distractor values.
x explanation for the relatively high error value is that the
Fground
Fig. 3. Sketch of the experimental model and illustration of the
parameters used in the mathematical computations of the measure-
ment software.
The elements of the matrix depend on the geometrical
arrangement of the hexapod in the individual case.
fix ; fiy ; fiz in Eq. (1) correspond to the components
of a unit vector f i in the direction of the measured force,
i.e., the direction of the respective distractor. For the
calculation of f i ; first the forward kinematics is solved,
which means the computation of the ring position of the
hexapod from ring sizes, distractor lengths and ball joint
positions relative to the rings (Seide et al., 2002). Then
the positions of the joints oi and ui in the global
coordinate system are computed (Fig. 3).1 From this f i is
given as
f i ¼ oiDui i with Di ¼ joi  ui j:
For the computation of the moments the definition of a
fulcrum is necessary. The software is laid out in such a
way that moments can be computed both with respect to
the ring center point and with respect to a clinically
determined point (e.g., fracture location, assuming an
idealized solid connection between bone and ring). The
position vector of this point is designated as K. Thus
the lever arm ri for the forces working on a distractor i is
the spacial difference between K and the upper joint
position oi :
ri ¼ oi  K:
Due to the definition of the moment this leads to
m i ¼ ð oi  K Þ
f i
with the components mix ; miy ; miz ; to be used in Eq. (1).
1
o corresponds to German ‘‘oberer Ring’’, which means upper ring,
u corresponds to German ‘‘unterer Ring’’, which means lower ring
(Seide et al., 2002).
1363
(i.e., by eliminating the direction), the shear forces Fx ; Fy
and the bending moments Mx ; My were summarized as
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Fxy ¼ Fx2 þ Fy2 ; ð5Þ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Mxy ¼ Mx2 þ My2 : ð6Þ
The accuracy of the overall system, i.e., mechanical
arrangement and software, was investigated. A typical
Fxy hexapod configuration for the lower leg (symmetrical
assembly, 150 mm ring diameter, 130 mm ring distance)
was exposed to defined forces and moments in a
material testing machine (Zwick1455). Fixator forces
This procedure showed maximum errors of 11%. An
sensors available in the clinic’s laboratory had a range of
1000 N (measuring error 1%) each, resulting in an
unfavorable approx. 5000 N measuring range for the
whole system. Also, the involvement of 6 sensors and
errors from determination of the mounting parameters
of the hexapod have to be considered. The accuracy was
considered appropriate, comparing well with deviations
caused by biological influences normally encountered in
clinical investigation. Additionally, the force-measuring
distractors were routinely calibrated every month to
avoid errors caused for example by mechanical damage.
Clinical measurements were performed on 9 patients
(Table 1), in which fresh fractures (o2 weeks) or
osteotomies with acute correction (i.e., without callus
distraction osteoneogenesis) were treated solely with an
external fixator. Typically, non-complicated crural
fractures are stabilized internally when possible or by
a secondary exchange of the external fixator by an
ð2Þ internal implant. So, the investigated cases represent a
selected sample (e.g., polytrauma, open fractures, severe
soft-tissue damage) and the number of patients available
for the measurement was therefore limited. In 7 cases
fractures and in 2 cases osteotomies of the tibia were
given. An assembly with Kirschner wires in Ilizarov’s
technique was used in 6 cases, and an assembly with
Schanz’ screws in 3 cases. For Ilizarov’s assembly, full
weight bearing was allowed; for Schanz’ screws assem-
bly, 10 kg partial load bearing was permitted, according
to our usual clinical practice.
In all cases, a fracture reduction had been performed
with the hexapod mechanism (Seide and Wolter, 1996;
ð3Þ
Seide et al., 1999). The hexapod elements were ex-
changed by lighter standard fixator elements (long-
itudinal rods, joints) after the primary reduction as well
ð4Þ
as after each measurement session and remounted
before the next. This was performed for better patient
comfort, as the measurement elements with its
cables were not suitable for outpatient activities.
Additionally, the remaining measurement sensors would
 ARTICLE IN PRESS
1364 K. Seide et al. / Journal of Biomechanics 37 (2004) 1361–1369
Table 1
Patients
Diagnosis (classification after AO) Time to fixator
removal (days)
1 Osteotomy proximal lower leg (equals A3) 112
2 Osteotomy proximal lower leg (equals A3) 169
3 1 Open shaft fracture middle third (A3) 153
4 1 Open shaft fracture middle third (A3) 93
5 1 Open fracture proximal lower leg (A2) 105
6 Oblique shaft fracture middle third (A2) 144
7 Oblique shaft fracture middle third (A2) 108
8 2 Open oblique shaft fracture middle/ 130
distal third (B2)
9 2 Open shaft fracture middle/distal third 270
with long comminution zone (C3)
have superposed the fracture zone when taking radio-
graphs in many cases. Mounting parameters of the
hexapod were noted and the hexapod was re-mounted
for each session in an identical manner.
The hexapod elements were mounted force free
(controlled by measurement, hanging leg). The standard
fixator rods were removed afterwards. The measuring
hexapod took over the fixator load completely during
the measurements, including inherent stresses of the
bone-soft-tissue-fixator mounting.
The measurements were performed 2, 4, 8 and 12
weeks after the trauma or the osteotomy. If the fixator
remained longer than 12 weeks, a further measurement
was made before removal of the fixator. An influence of
the measurements on the healing process was excluded.
Neither were fragment position and fixator assembly
modified during the measurements, nor were the results
of the measurements used for clinical decisions.
Measurements were performed under axial and
bending loads of the leg. The external axial force was
measured with a one-degree-of-freedom vertical ground
sensor positioned under the heel. For the application of
axial loads (direction of the bone axis) the patient was
instructed to stand relaxed with a stretched leg and to
put weight on the heel, slowly increasing and decreasing
the ground force. Three repetitions were performed. The
total measuring period was 20 s. The scanning rate was
0.2 s. The patients were instructed to limit the load so
Clinical decision after Further course of healing, remarks
12 weeks
Leave fixator in place Later discussion of X-rays concluded that
consolidation could have been assumed already
after 8 weeks
Leave fixator in place Consolidation in fixator, measurement at time of
fixator removal was not possible
Leave fixator in place Consolidation in fixator with partial bone defect,
temporary brace with 10 kg weight bearing after
fixator removal
Fixator removal Fracture not consolidated at fixator removal,
because of a pin consolidation in brace
infection and
application of a brace
Fixator removal, full Immediate mobilization without crunches or
weight bearing other safety measures
Leave fixator in place Consolidation in fixator, then immediate
mobilization without crunches or other safety
measures after fixator removal
Fixator removal and 4 weeks plaster cast with full weight bearing for
full weight bearing precautionary reasons
Leave fixator in place Consolidation in fixator with partial bone defect,
temporary brace with 10 kg weight bearing after
fixator removal
Leave fixator in place Cancellous bone graft necessary, consolidation in
fixator, after fixator removal cast for
precautionary reasons
that no pain appeared. Maximal loads applied were
individually different and ranged between 100 and
400 N, found in fresh and healed fractures, respectively.
For the application of the bending load, the patient was
placed supine with a free-hanging fixator and the heel
supported by the ground sensor. At the fracture gap
location, a ventro-dorsal force was applied manually.
The load was limited in such a way that no pain in
the fracture arose (up to 60 N). The effective lever arm,
i.e., the distance between the point of force application
and the heel was noted. The measurements under
bending loads were not performed in the first 4 patients,
since this measurement procedure was later included in
the protocol. The measurement at 2 weeks was missing
for two patients because the clinical condition of the
patients after a polytrauma did not allow for a
measurement in the laboratory. In one patient the
measurement before the fixator removal could not be
completed due to organizational reasons.
The patients were informed in detail (‘‘informed
consent’’) and participated in very motivated manner
in the sessions, which lasted about 2 h.
The evaluations were performed with the Excel
spread-sheet program (Microsoft). For the measure-
ments under axial load, linear regression lines were
computed with respect to the ground force. The
regression coefficient (upward slope gradient) obtained
describes the stiffness of the bone–fixator system. Each
 ARTICLE IN PRESS
K. Seide et al. / Journal of Biomechanics 37 (2004) 1361–1369 1365

regression calculation included three load cycles (100 0.8 1 A3

values, 20 s, as described above). To allow judgement of 2 A3
the repeatability of measurements, the standard error of 3 A3
0.6 4 A3
the regression coefficient was determined. The calcu-

Fz/Fground
lated regression coefficient values were charted against
0.4
the healing time. The fixator bending moment under
external bending load was charted with respect to the
external bending load so that the evaluation was 0.2
performed in the mode of a relative bending stiffness.
The y-axis-intersections of the regression lines are to 0
be interpreted as offset stresses. These stresses, which 2 4 8 12 fixator
are, e.g., due to primary pressure or tension on the (a) weeks removal
fracture site, were excluded by evaluating the slope
0.8
values of the regression lines in this study.
5 A2
6 A2
0.6 7 A2

Fz/Fground
3. Results 8 B2
9 C3
0.4
3.1. Forces and moments under axial load of the
extremity
0.2

Linear relationships between ground force Fground and

bending moment Mxy as well as ground force Fground and
axial force Fz in the fixator were found in the range of
the applied loads (R2 between 0,7 and 0,99, po0:01).
The regression coefficients were calculated. These were
found to be particularly suitable for the clinical
measurements to monitor fracture healing. Different
behaviors of these parameters were found for different
fracture types and different healing progression (Figs. 4,
6, 8 and 9). The standard errors of the regression
coefficients were relatively small, due to the high number
of 100 single measurements included in each regression
coefficient. The applied method of evaluating the
regression between external load and fixator seems to
be a favorable approach to enhance the accuracy in the
clinical experiment, as compared to single readings.
Axial fixator forces were small in A3 (transverse)
fractures over the entire healing process (Fig. 4a). In all
cases less than 10% of the ground force load was
transferred by the fixator; the axial load transfer was
performed by the fractured bone. In A2, B2 and C3
fractures distinct axial loads at the beginning and a
decreasing load transfer during fracture healing ap-
peared in the fixator (Fig. 4b). In 5 cases a slight increase
around the fourth week was noted.
In the case of the C3 (comminuted, Fig. 5) fracture,
64% of the external axial load was transferred by the
fixator at the beginning. Since the fracture was probably
not load carrying at this time, for the other 36% a load-
carrying function of the soft tissues (Schmidt et al.,
1995) has to be assumed. The decrease of the load
transfer in the fixator, i.e., an increase of the load
transfer over the healing bone, was also found in this
case, in particular, between the fourth and the twelfth
week (Fig. 4b). Contrary to the other cases 15%
0
2 4 8 12 fixator
(b) weeks removal
Fig. 4. Axial force in the fixator under axial load of the leg during the
course of fracture healing. Each point represents the regression
coefficient between fixator axial force and ground force. In type A3
fractures values increase at 4 weeks and are then small (a). In A2 and
B2 fractures larger and in a C3 fracture (b) maximum values are found,
decreasing during the healing process. Error bars represent the
standard error of the regression coefficient.
remained after 12 weeks. This correlated with the
clinical decisions: a further fixator treatment and a
cancellous bone graft in the later course of healing were
necessary.
The bending moment in the fixator under axial load of
the leg also showed a clear dependence on the type of
fracture (Fig. 6a and b). Even more clearly than for the
axial force measurements three of the four A3 cases
showed an increase of the fixator bending load between
the second and the fourth week and a decrease there-
after.
In two cases (patients 1 and 7) a high bending stability
of the fracture resulted already after 8 weeks. In cases 5
and 7 (Fig. 7) maximally 1.5 Nmm/N resulted at 12
weeks. Correlating with this measured high callus
stability were the radiographic findings and the clinical
decisions to immediately mobilize under full load after
fixator removal. The other patients received a temporary
brace after fixator removal for 3 months as a safety
measure to protect from peak loads. In patient 6, a low
value at 12 weeks and a slight further decrease under
ongoing fixator treatment was observed. Here, the
measurements showed that fixator removal would have
already been possible after 12 weeks. From the
 ARTICLE IN PRESS
1366 K. Seide et al. / Journal of Biomechanics 37 (2004) 1361–1369

30 1 A3
2 A3
25 3 A3

Mxy/Fground (Nmm/N)
4 A3
20

15

10

0
2 4 8 12 fixator
(a) weeks removal

30 5 A2
6 A2
25 7 A2

Mxy/Fground (Nmm/N)
8 B2
20 9 C3
15

10

0
2 4 8 12 fixator
(b) weeks removal
Fig. 5. Anteroposterior and lateral radiographs of a C3 fracture, 12
Fig. 6. Bending load in the fixator under axial load of the leg during
weeks after trauma (patient 9). The Fracture is not consolidated, the
the course of fracture healing. Each point represents the regression
load measurements show a decreasing, but still significant load transfer
coefficient between fixator bending moment and ground force. Values
in the fixator. Note that the radiographs do not show the hexapod
decrease in A3 (a), A2 and B2 fractures (b). In contrast, consistently
elements, since these were mounted before and replaced by standard
small bending moments are seen in C3 fractures (b). Error bars
rods after the measurement sessions in the lab (see text).
represent the standard error of the regression coefficient.

measuring curves for patient 1, it had to be assumed that
a fixator removal would have been possible after 8
weeks. A renewed discussion of the radiographs with the
treating surgeon confirmed this assumption.
In contrast, in two cases (patients 3 and 4) a flattening
of the curve was observed. In case 4 the highest
measurement value of all patients at 12 weeks
(9 Nmm/N) correlated with a pin infection, which then
led to the removal of the fixator. Likewise, in case 3 a
relatively high bending moment was present after 12
weeks. The fixator was left in place.
A characteristic pattern showed the comminuted
fracture (C3, patient 9). Here the fixator bending
moment was in a low range over the entire healing
process. In this case only the combination with the
measuring curve of the axial force Fz showed that the
fracture was slowly consolidating.
Measurements with the hexapod allow the determina-
tion of shear forces in the fixator. Values of the order of
magnitude 1/20 of the axial load (Fig. 8) resulted.
Fracture consolidation was found to reduce shear forces
similar to the other parameters. For example, patient 7
showed a definite decrease at 8 weeks and patient 8
between 12 weeks and fixator removal. Both these
decreases correlate to the curves found in the bending
moments.
3.2. Bending moment under bending load of the extremity
In 5 cases a bending load case was examined in
addition to the axial load case (Fig. 9). In all patients a
decrease of the bending moment transferred in the
fixator was found during fracture healing. This corre-
sponds to a shift in the bending stiffness in the overall
system from the fixator to the bone. The inter-individual
comparison indicated that the courses of the values were
similar to the corresponding curves under axial load.
However, the temporary increase in instability in cases 3
and 4 at 4 weeks is not apparent. In the case of patient 4,
in whom the fixator had to be removed because of a pin
infection, an increase of the fixator bending load was
found parallel to pin loosening. As under axial load, the
C3-fracture (patient 9) showed a spacial behavior with
almost constant values up to 8 weeks, but followed by a
slight decrease. Patient 8, who showed distinctly more
fixator bending moment than the other patients at 12
 ARTICLE IN PRESS
K. Seide et al. / Journal of Biomechanics 37 (2004) 1361–1369 1367

1 1 A3
3 A3
0.8 6 A2
4 A3

My/Mload
0.6 8 B2
9 C3
0.4

0.2

0
2 4 8 2 fixator
weeks removal

Fig. 9. Bending moment in the fixator under bending load of the leg
during the course of fracture healing. Each point represents the
regression coefficient between bending moment measured in the fixator
and applied bending moment, with respect to the transverse axis. The
results correspond to the measurements of the bending moment under
axial load. However, the temporary instability in A3 fractures after 4
weeks is not apparent. Error bars represent the standard error of the
regression coefficient.

Fig. 7. Anteroposterior and lateral radiograph of an A2 fracture, 12

weeks after trauma (patient 7). The fracture is consolidated, and the
load measurements show small forces and moments in the fixator. weeks under axial load, also showed persisting high
Note that the radiographs do no show the hexapod elements, as these values until 12 weeks under bending load.
were mounted before and replaced by standard rods after the
measurement sessions in the lab (see text).

4. Discussion

Clinical in vivo applications of mechanical measure-

0.1 1 A3 ments in the process of the fracture healing have been
2 A3 .
published. Typically, the techniques used (Jorgensen,
0.08 3 A3
4 A3
1972; Burny et al., 1984; Cunningham et al., 1987; Evans
Fxy/Fground

0.06 et al., 1988) measure bending moments in unilateral

external fixators with sensors on longitudinal rods. With
0.04 the Fraktometer (Claes, 1991) the fracture gap move-
ment is unidirectionally measured. Compared with these
0.02
in vivo load measurements in external fixators, the
0 hexapod offers the advantage that the fixator load can
2 4 8 12 fixator be measured in all six spacial degrees of freedom, which
(a) weeks removal
are 3 forces and 3 moments. Measurements of fracture
healing in six degrees of freedom have been realized so
0.1 5 A2 far in some very special internal implants (nail (Schnei-
6 A2
0.08 7 A2 der et al., 1990), spine fixator (Rohlmann et al., 1995)).
8 B2 The characteristic of the hexapod (also known as
Fxy/Fground

9 C3
0.06 ‘‘parallel robot mechanism’’) is based on the fact that
exactly one degree of freedom is carried and measured
0.04 by each measurement unit. According to the theory of
the hexapod kinematics no interrelation between the
0.02
elements is given. So, there are six degrees of freedom in
0 space on the one hand and six independent degrees of
2 4 8 12 fixator freedom load carrying/measuring elements on the other.
(b) weeks removal
A mathematical relationship is given between the linear
Fig. 8. Shear forces in the fixator under axial load of the leg for A3 (a), force measurements in the hexapod fixator and the six
A2, B2, and C3 fractures (b) during the course of fracture healing.
load degrees of freedom (axial, shear, bending, tor-
Each point represents the regression coefficient between fixator shear
force and ground force. Values are of the order of magnitude 1/20 of sional) in the Cartesian coordinate system.
the axial load. Error bars represent the standard error of the regression The fixator load situation is different from other
coefficient. fixators, where, e.g., in a ring fixator 3 rods are fixed on
 ARTICLE IN PRESS
1368 K. Seide et al. / Journal of Biomechanics 37 (2004) 1361–1369
the rings in all degrees of freedom, summing up to 18
degrees of freedom and thus resulting in prestresses
between the rods. Common clinical experience is that
these stresses can be substantial.
In the performed measurements a distribution be-
tween the applied limb load and the forces and moments
in the fixator was determined. As a measure for the
callus formation, it would be most interesting, to
compute an ‘‘absolute fracture stiffness’’ (Evans et al.,
1988). For this a mechanical model would have to be
applied. Theoretically, measurement values depend on
dimensions and material properties of callus, Kirschner
wires and/or Schanz’ screws as well as the carrying
elements of the fixator. However, as opposed to in vitro
or animal experiments, many simplifying assumptions
would have to be made in such calculations, when one
deals with real clinical situations, e.g., fracture gaps
found in a clinical situation are considerably irregular
and linearity for axial force in assemblies with tensioned
wires cannot be presumed. Also, the authors are aware
of the fact that soft tissues play an additional mainly
unknown role in the load distribution. Therefore, only a
summarizing parameter rather than a biomechanically
precisely defined callus stiffness can be measured. So,
the main question to be answered by this study was,
whether, despite these theoretical limitations, the
hexapod fixator load measurements could be useful
as—relative—indicators of the fracture healing progress.
In our opinion—with reservation due to the relatively
small number of measured patients—this can be
assumed.
The optimal measurement parameter was dependent
on the type of the fracture. In transverse fractures (A3)
the fixator bending moment was most predictive,
whereas the fixator axial load gave nearly no relevant
information. In contrast, in a comminuted fracture
fixator axial load was found to be the only meaningful
predictor. The evaluation of the healing process by
means of a combination of the parameters bending
moment and axial force in the fixator under axial load of
the extremity was possible in all examined patients. So,
the measurements suggest that two well-selected degrees
of freedom are the minimum criteria necessary for the
mechanical monitoring of fracture consolidation. The
measurements agreed with the radiologically diagnosed
stability and the treatment measures taken after 12
weeks (e.g., removal of the fixator, continuing treatment
in the fixator, unprotected or protected weight bearing),
which were accomplished without knowledge of the
measurement results.
It is interesting that the measured reduction of
stability after 4 weeks in transverse osteotomies and
transverse fractures can be explained by the results of
experimental investigations of fracture gap movement in
sheep. For fracture gap widths between 0.5 and 1 mm
and relative movements of 18%, a temporary increase of
the interfragmentary movement was found. A correlat-
ing bone absorption was determined histologically
(Claes et al., 1989).
Bending measurements under external bending loads
showed the course of callus stiffness, comparable to
those under axial load. However, the former were less
sensitive, e.g., the temporary instability in A3 fractures
after 4 weeks was not apparent. Thus a conclusion of the
study is that the latter is to be recommended for clinical
practice. However, it is also possible that the mode of
bending load application in the study was not optimal
and a special load frame might improve the accuracy as
well as the sensitivity of these measurements.
A difference between earlier investigations with
external fixators and the investigations with the hexapod
fixator is given. Cases examined with the ‘‘Fraktometer’’
showed maximum gap widths of 5–6 mm in type A and
type B fractures and average gap widths of 10 mm in
cases with type C fractures or delayed healing fractures
(Wentzensen, 1995). Due to the exact reduction, which is
typically performed when the hexapod is used (Seide
et al., 1999), a fragment contact has to be assumed.
Therefore, in A3 fractures, i.e., with an almost
transverse gap, only a small axial load was measured
in the fixator, which means that the axial load was
transferred by the bone from the beginning. However,
the fixator was substantially loaded on bending under
axial load of the extremity, which is explainable by
‘‘buckling’’ in the fracture site. This is increasingly
stabilized during the healing process through endostal
and periostal callus. The measured fixator bending loads
decreased accordingly. The hypothesis of an axial
deformation in the fracture gap under load, which
decreases during fracture healing, can be supported with
type B and C fractures by the hexapod measurements. In
A2 (simple oblique) fractures, a mechanism of the
interfragmentary movement in the sense of sliding and
tilting is conceivable, which explains the measured
simultaneous changes of the axial force and the bending
moment in the fixator during the healing process.
In vivo interfragmentary movements by patients with
ring and hexapod fixators have been documented and
reported by Duda et al. (Duda et al., 2002, 2003). The
interfragmentary movement rather than the load in
the fracture site has been shown to be important for the
healing process (Claes et al., 1997). For this reason it has
to be discussed, whether the deformation or the load
needs to be measured, in order to utilize the measure-
ment values concerning the influence and optimization
of the callus’ mechanical situation. In contrast to
fracture healing measurements in transverse osteotomies
under laboratory conditions in animals (Claes et al.,
1997) callus deformation values are more difficult to
access in humans, because their measurement requires
invasive techniques, e.g., the insertion of additional
Schanz’ screws. On the other hand, there is a mechanical
 ARTICLE IN PRESS
K. Seide et al. / Journal of Biomechanics 37 (2004) 1361–1369
relationship between deformation and loads, depending
on the stiffness parameters of the involved structures
(bone, fixator elements, callus). Thus, either measure-
ment should be appropriate. This should be especially
true in relative measurements during the course of
healing.
After developing the measuring external fixator a
pilot study in nine patients was carried out. The
measured parameters resulted in curves in good agree-
ment with the healing process. This leads to the
conclusion that it will be possible to control fracture
healing using a measurement fixator system, minimizing
the use of radiographs with their radiation exposure.
However, this has to be proven in further investigations.
The future goal is the development of intelligent
implants and external fixators, which automatically
adjust load distribution in the callus according to the
results of measurements (‘‘automatic dynamization’’)
and thus optimize fracture healing. In addition, it can
inform the patients about the optimal load of their
extremity (‘‘intelligent fixator’’, Wolter, 1991; Wolter
and Seide, 1997). A precursor of such systems was an
external fixator provided with force sensors for the
application of controlled longitudinal forces in an
external compression osteosynthesis (Muller . et al.,
1982). Duda (Duda et al., 1998) developed a method
for the description of the spacial rigidity of a fixator
system over a rigidity matrix of 36 elements. It is
conceivable to develop a fixator with controllable
rigidity matrix using motors and flexible elements in a
hexapod and adjust it as a function of fracture type and
measured healing status. Further, by applying the
hexapod system to control the acting forces during a
callus distraction (Aronson and Harp, 1994; Wolfson
et al., 1990), the distraction time and quality of the
developing bone could be optimized.
Acknowledgements
The authors wish to thank the Biomechanics Section
of the Technical University Hamburg-Harburg for help
in manufacturing the sensor-elements and Mrs. Sarah
Strametz for correcting the translation.
References
Aronson, J., Harp, J.H., 1994. Mechanical forces as predictors of
healing during tibial lengthening by distraction osteogenesis.
Clinical Orthopaedics and Related Research 301, 73–79.
Burny, F., Donkerwolcke, M., Bourgois, R., Domb, M., Saric, O.,
1984. Twenty years experience in fracture healing measurement
with strain gauges. Orthopedics 7, 1823–1826.
Claes, L., 1991. Die Messung der Knochenheilung bei Fixateur-
externe-Osteosynthesen mit dem Fraktometer FM 100. Chirurg 62,
354–355.
1369
.
Claes, L., Wilke, H.J., Rubenacker, S., 1989. Interfragmentary strain
and bone healing—an experimental study. 35th annual meeting,
Orthopaedic Research Society, February 6–9, 1989, Las Vegas,
Nevada.
Claes, L., Augat, P., Suger, G., Wilke, H.J., 1997. Influence of size and
stability of the osteotomy gap on the success of fracture healing.
Journal of Orthopedic Research 15, 577–584.
Cunningham, J.L., Evans, M., Harris, J.D., Kenwright, J., 1987. The
measurement of stiffness of fractures treated with external fixation.
Engineering in Medicine 16 (4), 229–232.
Duda, G.N., Kirchner, H., Wilke, H.J., Claes, L., 1998. A method to
determine the 3-D stiffness of fracture fixation devices and its
application to predict inter-fragmentary movement. Journal of
Biomechanics 31, 247–252.
Duda, G.N., Sollmann, M., Sporrer, S., Hoffmann, J.E., Kassi, J.-P.,
Khodadadyan, C., Raschke, M., 2002. Interfragmentary motion in
tibial osteotomies stabilized with ring fixators. Clinical Orthopae-
dics and Related Research 396, 163–172.
Duda, G.N., Bartmeyer, B., Sporrer, S., Taylor, W.R., Raschke, M.,
Haas, N.P., 2003. Does partial weight bearing unload a healing
bone in external ring fixation? Langenbecks Archive 388,
298–304.
Evans, M., Kenwright, J., Cunningham, J.L., 1988. Design and
performance of a fracture monitoring transducer. Journal of
Biomedical Engineering 10, 64–68.
.
Jorgensen, T.E., 1972. Measurements of stability of cruaral fractures
treated with Hoffmann osteotaxis. Acta Orthopaedica Scandina-
vica 43, 207–218.
.
Muller, K.H., Bowe, K.-H., Becker, J., 1982. Kraftbestimmung bei
externer Kompressionsosteosynthese am Frakturspalt des
Menschen. Unfallheilkunde 85, 321–337.
Rohlmann, A., Bergmann, G., Graichen, F., Weber, U., 1995. In vivo
measurement of implant loads in a patient with a fractured
vertebral body. European Spine Journal 4, 347–353.
.
Schmidt, H.G.K., Wolter, D., Sasse, S., Schneider, E., Schumann, U.,
.
Jurgens, C., 1995. Die Lastverteilung im Ringfixateur bei
segmentalem Knochendefekt. In: Wolter, D., Hansis, M., Have-
mann. D. (Hrsg.) (Eds.), Externe und interne Fixateursysteme.
Springer, Berlin, pp. 37–43.
Schneider, E., Michel, M.C., Genge, M., Perren, S.M., 1990. Loads
acting on an intramedullary femoral nail. In: Bergmann, G.,
Rohlmann, A., Graichen, F. (Hrsg.) (Eds.), Implantable Telemetry
in Orthopaedics. Forschungsermittlung der FU, Berlin.
Seide, K., Wolter, D., 1996. Universelle dreidimensionale Korrektur
und Reposition mit dem Ringfixateur unter Anwendung der
Hexapod-Anordnung. Unfallchirurg 6, 422–424.
Seide, K., Wolter, D., Kortmann, H.R., 1999. Fracture reduction and
deformity correction with the hexapod Ilizarov fixator. Clinical
Orthopaedics and Related Research 363, 186–195.
.
Seide, K., Wollnack, J., Weinrich, N., Jurgens, C., 2002. Theorie und
Software des Hexapod Fixateurs externe. Biomedizinische Technik
47, 326–333.
Stewart, D., 1965. A platform with six degrees of freedom. Proceedings
of the Institute of Mechanical Engineers 180, 371–378.
Wentzensen, A., 1995. In-vivo-Messung der Knochenheilung bei
Fixateur-externe-Osteosynthesen. In: Wolter, D., Hansis, M.,
Havemann, D., (Hrsg.) (Eds.), Externe und interne Fixationssys-
teme. Springer, Berlin, pp. 107–111.
Wolfson, N., Hearn, T.C., Thomason, J.J., Armstrong, P.F., 1990.
Force and stiffness changes during Ilizarov leg lengthening. Clinical
Orthopaedics and Related Research 125, 58–60.
Wolter, D., 1991. Das intelligente Implantat. Abstraktband BIOME-
CHANIKA 1991, Hamburg.
Wolter, D., Seide, K., 1997. Das intelligente Implantat. In: Schneider
E. (Hrsg.) (Eds.), Biomechanik des menschlichen Bewegungsappar-
ats. Vol. 261, Hefte zur Zeitschrift Der Unfallchirurg, pp. 236–240.