Journal of Biomechanics 31 (1998) 741745

Representative assessment of long bone shaft biomechanical

properties: an optimized testing method
Jos A.M. Bramer!,*, Robbert H Barentsen", Maarten vd Elst!, Elly S.M. de Lange#,
Peter Patka!, Henk J.Th.M. Haarman!
! Department of Traumatology, University Hospital Vrije Universiteit, Amsterdam, The Netherlands
" Department of Clinical Physics and Engineering, University Hospital Vrije Universiteit, Amsterdam, The Netherlands
# Department of Epidemiology and Biostatistics, University Hospital Vrije Universiteit, Amsterdam, The Netherlands
Received in final form 17 June 1998

Abstract
Whole bone bending tests are commonly used in mechanical evaluation of long bones. Reliable information about the midshaft can
only be obtained if the bending moment is uniformly distributed along the shaft, and if the distribution of the bending stress is not
adversely influenced by rigid clamping of the bone ends. A testing device was developed to determine bending stiffness of long bones in
24 directions, perpendicular to the bone axis. For optimal distribution of bending moment and stress, four-point bending was
performed, and bone ends were simply supported, not rigidly clamped. The method was validated by repeated testing of a stainless
steel rod, and a sheep femur. Leftright ratios were assessed twice in 2 groups of 5 sheep: one control group, and one group in which
the left femur was stabilized with a stainless steel interlocking nail for 2.5 yr, after a midshaft osteotomy. Test results obtained with the
steel rod reproducibly were close to predicted values. Measurements with the sheep femurs were reproducible and precise for 3 of the
4 parameters of the bending test. Stiffness parameters were significantly higher in the operated sheep than in the control group. We
conclude that the method described here provides accurate and reproducible information, which is representative for the long bone
shaft. ( 1998 Elsevier Science Ltd. All rights reserved.
Keywords: Animal experiments; Long bone shaft; Mechanics; Four-point bending; Stress distribution

1. Introduction
The rigidity of long bones reveals differences if bending
tests are done in several directions of the transverse plane
(Ruff and Hayes, 1983; Lovejoy et al., 1976). These differences are even more pronounced after fracture fixation
(Foux et al., 1993). Foux developed a method to assess
the distribution of the bone rigidity by performing threepoint bending tests in 24 directions perpendicular to the
bone axis (Foux et al., 1990). A disadvantage of threepoint bending tests is the local deformation of the bone at
the site where the force is applied, resulting in an underestimation of the Youngs modulus (Turner, 1993). Moreover, the bending moment is maximal at this site, which
will have a major effect on the test results. If a four-point

* Corresponding author. Tel.: 31-20-6151301; fax: 31-20-6151301.

0021-9290/98/$19.00 ( 1998 Elsevier Science Ltd. All rights reserved.
PII S0021-9290(98)00101-8

bending test is used, the bending moment will be uniform

between the applied forces (Fig. 1) and the weakest part
of the shaft will determine the outcome of the test (Timoshenko and Goudier, 1970). Moreover, the influence of
shearing force will be reduced (Torzilli et al., 1981).
Furthermore, test results will be influenced by the
method of fixating the bone ends in the testing device.
The theory of elasticity shows that rigid clamping of
a beam in a bending device, will result in maximal stress
near the points of fixation, and minimal stress in the
middle. More appropriate distribution of the stress, with
the maximum in the middle, will be achieved if the ends
are allowed translation in the plane perpendicular to
the plane of bending (Fig. 2) (Timoshenko and Goudier,
1970; Griffel, 1966).
In this report we describe a device for the assessment of
the rigidity of long bones in a four-point bending test in
24 planes, in which the bone ends are not rigidly fixed but
simply supported. This results in a uniform distribution

742

J.A.M. Bramer et al. / Journal of Biomechanics 31 (1998) 741745

Fig. 1. Bending moment distribution along the length of a beam in three- and four-point bending.

Fig. 2. Bending stress distribution along the length of a beam in case of fixated and simply supported ends.

of the bending moment, and a more appropriate distribution of the bending stress along the shaft.

deformation was introduced. After each test the cups

were taken out of the rings, turned 15 and placed in the
device again for the next test. This was repeated 24 times,
untill a full revolution was made.

2. Device design
The long bone was placed with the ends in cylindrical
metal cups. The axis was centred. Fixation took place by
filling the cups with a low melting point Bismuth alloy (A
301, Degussa, Wolfgang, Germany; melting point 47C)
in the liquid state. The outside of the cups consisted of 24
small facets, corresponding with the 24 facets of the rings
in which they were placed (Fig. 3). These rings had a lug
on two sides which was placed on a saddle.
The device was positioned on a bending machine
(Hounsfield H5000M, Hounsfield Test Equipment, England) which recorded load and deflection. A four-point
bending test was performed by applying two equal forces
at the edges of both cups. The cups were allowed translation in the horizontal plane perpendicular to the plane of
bending. Bending was performed with a constant crosshead speed of 1 mm/minute. Deflection was measured at
the probe where the bending machine applied the force
(F in Fig. 3). The test was nondestructive, no plastic

3. Validation of the method

The method was validated in four experimental settings:
(A) A stainless-steel rod (asi no. 316), with a diameter of
8 mm, was tested 5 times in the device.
(B) The entire procedure, including fixation of the bone
ends in the cups, and mounting of the device, was
repeated 4 times with one sheep femur.
(C) Both femurs of five healthy adult sheep (controls)
were tested to assess leftright differences. The entire
procedure was repeated for a second time to assess
reproducibility of the method (duplo tests).
(D) Both femurs of five operated sheep were tested. These
sheep received a stainless-steel interlocking nail 2.5 yr
earlier stabilizing a midshaft osteotomy of the left
femur. Before testing, the nail was removed. Four
femurs of two sheep were tested twice (duplo tests).

J.A.M. Bramer et al. / Journal of Biomechanics 31 (1998) 741745

743

where x is the distance from the center of the lug to the

point where the force was applied, y half the length of the
long bone shaft, F the applied bending force, d the displacement of the probe and I the moment of inertia.
For each femur the 24 EI-values were plotted in polar
coordinates and elliptical regression analyses was performed, resulting in an ellipse. The ellipses of both femurs
of one sheep were plotted in the same coordinate set,
centralized and the ellipse of the right femur was mirrored in the -axis.
With the use of semimajor axis (a), semiminor axis (b),
and angle of inclination (a) of both ellipses (Fig. 4), the
following parameters were calculated:
Stiffness index:

SI"

EItestbone
EIcontralateral

.
minimal

This is the ratio of the stiffness of the testbone versus the

contralateral bone in the direction in which this ratio is
minimal.
Fig. 3. Device for bending test in 24 directions: (1) tested bone;
(2) metal cups and (3) corresponding rings; (4) lug at the side of the
rings; (5) saddle; (6) testprobes for applying force.

All tests were performed 8 weeks after the sheep were

sacrificed. The femora were stored in 70% alcohol, and
kept wet during testing (Linde and Sorensen, 1993). All
experiments were approved by the DEC (Animal Experiment Committee) and carried out in accordance with the
Dutch regulations of Animal Welfare.

4. Processing of the test results

Area ratio:
(ab)testbone
AR"
.
(ab)contralateral
This ratio represents the total stiffness of the testbone, as
compared to the contralateral bone.
Flatness ratio:
(b/a)testbone
FR"
.
(b/a)contralateral
This represents the relative distribution of the stiffness of
the bone in different directions.
Inclination difference:

After linear regression on the loaddeflection curves,

24 loaddeflection quotients were obtained for each femur. Because the main contents of the cups consisted
of the bismuth alloy, the elasticity of the alloy
(4]109 N m~2) was assumed for this part of the device.
The flexural rigidity (EI) of the bone in every direction
tested could be determined by the following equation,
derived from beam theory.
Equation for calculating flexural stiffness (EI):
1 Fx2y
1 Fx3
d" 2
# 2
(EI)bone 3(EI)cups

(EI)

bone

"

[F/d]1@2x2y
F
x3
1!
d 6(EI)
cup3

C CD

ID"atestbone!acontralateral .
This is the difference between the angles the semiminor
axis make with the anteriorposterior plane.

5. Results
(A) In repeated testing of the steel rod in the device,
the mean value of flexural stiffness was

744

J.A.M. Bramer et al. / Journal of Biomechanics 31 (1998) 741745

Fig. 4. Ellipses of both femurs of operated sheep no. 4, plotted in the same set of coordinates. a"semi-major axis (e.g. 93.5 N m2 for the left, and
71.4 N m2 for the right femur) b"semi-minor axis (e.g. 86.4 N m2 for the left, and 68.4 N m2 for the right femur). a"angle of the semi-minor axis with
AP-plane (inclination) (e.g. 145.62 for the left, and 122.7 for the right femur).

31.98 N m2 (standard deviation 2.02), whereas

the calculated theoretical value was 35.39 N m2.
(B) Repeated performance of the test with one sheep
femur resulted in mean values of 108.30 N m2 for
the semimajor axis (S.D. 4.04), and 101.25 N m2
for the semiminor axis (S.D. 4.15) of the ellipse.
The mean area was 109.72 N2m4 (S.D. 7.39), and
the mean flatness was 0.94 (S.D. 0.04). Inclination
showed a mean value of 123.00 (SD 49.99).
(C)/(D) In the bending tests with the control sheep, the
Stiffness Index, Area Ratio, and Flatness Ratio
all approached 1, as expected (Table 1). In the
operated sheep, the stiffness index and the aria
ratio were significantly higher (p"0.009, Mann
Whitney test). The flatness Ratio in the operated
sheep approached 1, not differing from this value
in the control group (p"0.917). The Inclination
difference appeared to be highly variable within
the two test groups, and the difference between
these groups was not significant (p"0.754).
From the seven duplo tests of five control and two
operated sheep, the standard deviation of the difference
of the duplo, and the repeatability coefficient were calculated for each parameter (Bland and Altman, 1986).

For the Stiffness Index, Area Ratio, and Flatness Ratio

the standard deviations were acceptable (0.07, 0.15, and
0.04, respectively) as were the repeatability coefficients
(0.14, 0.30, and 0.08). However, the standard deviation
and repeatability coefficient of the inclination difference
were extremely high (64.27 and 128.54).

6. Discussion
In vivo, long bones are mostly exposed to torsion and
bending forces (Ruff and Hayes, 1983; Raftoupoulos and
Qassem, 1983; Bertram and Biewener, 1988), representing the most common cause of long bone fractures
(Lovejoy et al., 1976; Evans et al., 1951; Alms, 1961). The
majority of fractures occur in the middle third of the shaft
(Dencker, 1965). Biomechanical evaluation in animal experiments should produce accurate information about
the whole shaft, especially the middle third.
In the presented method this was achieved by using
a four-point bending test without rigid fixation of the
bone ends. This way the bending moment was uniform
along the whole length of the shaft, and the bending
stress was maximal in the shaft between the applied
forces, and minimal at the proximal and distal ends. No

J.A.M. Bramer et al. / Journal of Biomechanics 31 (1998) 741745

local deformation of the midshaft was introduced, because the loads were not applied directly to the bone.
The stiffness index, area ratio, and flatness ratio proved to be useful in comparing leftright differences in
different groups of animals. These ratios approached 1 in
the control group, indicating no difference between left
and right femur. This symmetry of mechanical properties
was demonstrated before (Kersey et al., 1994; Mather,
1967; Sumner et al., 1988). It results in the contralateral
femur being the ideal control. The stiffness index and
area ratio were significantly higher in the operated sheep.
The flatness ratio approached 1 in both the controls and
the operated sheep, meaning that there was no preferential direction of rigidity. This might be explained by the
intramedullary fixation of the fractures used here, making stress shielding and other possible effects similar in all
directions. In the duplo tests, the stiffness index, the area
ratio, and the flatness ratio appeared to be very reproducible. The inclination difference showed a large
variation and a poor reproducibility, which might be
explained by the fact that the flatness ratio always approached 1. This means that the ellipse of each bone
approached a circle, making inaccuracy in assessment of
the direction of the semiminor axis more likely. If there is
no preferential direction of the stiffness, the inclination
difference seems not a very relevant parameter. However,
if flatness ratios would not equal 1, this parameter might
produce more reproducible and useful results.
We conclude that the described method accurately and
reproducibly determines mechanical properties of the
long bone shaft. By optimal distribution of bending moment and bending stress, appropriate information can be
obtained about the complete shaft.
Acknowledgements
The authors wish to acknowledge the Biomaterials
Group of the University of Leiden, the Netherlands,
Klaas Boshuizen of the Dept. of Clinical Physics and
Engineering, and Ger Vink and the other workers of
the Clinical Animal Experimental Laboratory of the
Vrije Universiteit Amsterdam for their support in this
study.

745

References
Alms, M., 1961. Fracture mechanics. Journal of Bone Joint and Surgery
43B, 162166.
Bertram, J.E., Biewener, A.A., 1988. Bone curvature: sacrificing strength
for load predictability? Journal of Theoretical Biology 133(1), 7592.
Bland, J.M., Altman, D.G., 1986. Statistical methods for assessing
agreement between two methods of clinical measurement. Lancet,
307310.
Dencker, H., 1965. Shaft fractures of the Femur. A comparative study of
the results of various methods of treatment in 1003 cases. Acta Chir
Scand. 130, 173184.
Evans, F., Pederson, H., Lissner, H., 1951. The Role of tensile stress in
the mechanism of femoral fractures. Journal of Bone Joint and
Surgery 33A, 485501.
Foux, A., Black, R.C., Uhthoff, H.K., 1990. Quantitative measures for
fracture healing: An in-vitro biomechanical study. Journal of Biomechanics and Engineering 112, 401406.
Foux, A., Uhthoff, H.K., Black, R.C., 1993. Healing of plated femoral
osteotomies in dogs, A mechanical study using a new test method.
Acta Orthop. Scand. 64(3), 345353.
Griffel, W., 1966. Handbook of formulas for stress and strain. Fred.
Ungar Publishing Co, New York.
Kersey, R.C., Szivek, J.A., Sacoman, D.M., 1994. Symmetry of biomechanical properties in canine femora. Journal of Applied Biomaterials 5, 99101.
Linde, F., Sorensen, H.C.F., 1993. The effect of different storage
methods on the mechanical properties of trabecular bone. Journal of
Biomechanics 26, 12491252.
Lovejoy, C.O., Burstein, A.H., Heiple, K.G., 1976. The biomechanical
analysis of bone strength: A method and its application to platycnemia. American Journal of Physical Anthropology 44, 489506.
Mather, B.S., 1967. The symmetry of the mechanical properties of the
human femur. Journal of Surgical Research 7, 222225.
Raftopoulos, D.D., Qassem, W., 1983. Three dimensional curved beam
stress analyses of the human femur. American Journal of Physical
Anthropology 60, 383400
Ruff, C.B., Hayes, W.C., 1983. Crosssectional geometry of Pecos
Pueblo femora and tibiae- a biomechanical investigation: II. Sex, age,
side differences. American Journal of Physical Anthropology 60(3),
383400.
Sumner, D.R., Turner, T.M., Galante, J.O., 1988. Symmetry of the
canine femur: implications for experimental sample size requirements. Journal of Orthopaedics Research 6(5), 758765.
Timoshenko, S.P., Goudier, J.N., 1970. Theory of Elasticity. McGrawHill, London.
Torzilli, P.A., Takebe, K., Burstein, A.H., Heiple, K.G., 1981. Structural
properties of immature canine bone. Journal of Biomechanical
Engineering 103, 232238.
Turner, C.H., 1993. Measurements of Youngs modulus in bending tests
can be highly inaccurate (comment). Journal of Bone Joint and
Surgery 11, 462463.