The

BY AUGUSTUS

Four
A. AND

Biomechanical
WHITE III, WAYNE From the M.D., 0. DR. MED. SC4, SOUTHWICK,

Stages
MANOHAR NEW M.D4,

of Fracture
M. HAVEN, PANJABI, TEK. CONNECTICUT

Repair*t
LIC., DR. TECH4,

Engineering Surgery.

Laboratory Yale

for

Musculoskeletal School

Diseases.
New

Section

of

Orthopaedit

University

of Medicine,

Haven

ABSTRACT:

Based

on

analysis

of

the

torque-angle

weighing
dardized

between
surgical

four

and

4.5
an

kilograms,
osteotomy

using
was

a standone at

curves and roentgenographic findings in fifty-three healing tibial fractures in rabbits tested in torsion to failure, four biomechanical stages of fracture healing were defined, as follows: Stage I failure through original fracture site, with low stiffness; Stage II
-

procedure

the mid-shaft of each tibia below the point where the tibia is joined by the fibula. Parallel Kirschner wires were placed two above and two below each osteotomy site, and the experimental fracture was reduced and held by specially constructed clamps ronment was quantitatively
constant compression of

failure through original fracture site, with high stiffness; Stage Ill failure partially through original fracture site and partially through intact bone, with high stiffness; and Stage IV failure entirely through
-

by

which controlled

the

mechanical enviand monitored. A

(two kilograms

twenty

newtons

intact related torque gressed nal

both

bone, with high stiffness. These stages corwith the progressive increases in the average and energy absorption to failure as healing proand also with the average times since the origifracture. that this system of staging will provide

force) tibia,

was applied alternating

to both tibiae, while on the right or left the side in sequential animals, a cyclic load of twenty newtons (two of cyclic two hours seven days fifth, from sixth, two to

superimposed

experimental It is hoped
a standard

kilograms force) was also applied. The schedule loading was fifty-five counts per minute for in the morning and two hours in the afternoon, a week. At seventh, sixteen putated, until some and time eighth during weeks the third, fourth, after osteotomy,

by which important variables related to ultimate strength of healing fractures can be correlated and an objective way to predict delayed unions and non-unions and to determine the level of activity that is safe for patients with a healing fracture. Existing classifications of the stages in fracture healing

animals labeled,

were killed and their hind limbs were amand frozen at -20 degrees centigrade

are based on biochemical observations, histological appearances, roentgenographic criteria, clinical judgment, or some combination of these. We are not aware of any
classification in which stages of fracture repair are based

tested at a later date. The surgical procedures were carried out under general anesthesia using surgical asepsis. The mid-shaft of the tibia was exposed, and a transverse osteotomy was made using an oscillating saw. The bone was cooled with refrigerated Ringers solution the cut was completed, duced anatomically kilograms force) clamps attached serted were transversely After the made as the osteotomy was done. After the experimental fracture was reand the twenty-newton (two was Kirschner applied wires through that were the in-

on objective the healing torque-angle ure at healing

measurements bone tested

to

of the mechanical properties failure. By analysis of made after we identified

of the failfour

curves and roentgenograms fracture sites in rabbits,

compression to the four through operation, the

clearly delineated stages in a study originally designed to investigate the effects of cyclic loading on the rate of fracture healing. The histological findings in the study and the effects of the cyclic loading 12 are reported elsewhere. The purpose of this presentation is to describe the four ical stages of fracture repair evidence and rationale Experimental
Design

the bones. two oblique reduction. Each

roentgenograms rabbit was then and stomach. while the hind

to check

fully supported The forelimbs

on a sling beneath the chest were able to touch a platform

and to present on which they Methods

the biomechanare based.

limbs were unsupported cyclic loading device promazine) was given operative appeared

and accessible for applying the to the wires. Thorazine (chloroccasionally in the immediate postbut otherwise ate well. the animals

period for restlessness to be comfortable and Preparation to testing ankle,

muscle procedure

In
*

thirty-three
Delta Award

adult
paper,

female
read

New
in part

Zealand

rabbits
Meeting

Specimen Prior knee

ing testing

Kappa

at the Combined

the

tibiae was

were
the pairs

disarticulated removed,
of Kirschner

at the leavwires

of the Orthopaedic Research Society and The American Academy of Orthopaedic Surgeons, New Orleans, Louisiana, January 30, 1976. t This work was supported by U.S. Public Health Grant ROl AMI6O3I and by The Crippled Childrens Aid Society. New Haven, Connecticut. : Yale University School of Medicine, 333 Cedar Street, New Hayen, Connecticut

and
the

and
intact

the skin
between

carefully drying
the

to prevent from

the bone

specimens
as well

from

out during
healing fracture

the tis-

as to protect

06510.

accidental

damage

during

dissection
OF BONE AND

of the
JOINT

soft

188

THE JOURNAL

SURGERY

THE

FOUR

BIOMECHANICAL

STAGES

OF

FRACTURE

REPAIR

189 torque-angle was defined Therefore, curve up to the by the torque-angle a single number

sue. The Kirschner

portions of the tibia wires were removed

proximal and distal to the and the two ends, including polyester resin to specimen in the of the prepared at 90 degrees

the shaded area point of failure. curve, which was

under Stiffness

the

the wires, were potted ensure precise fit and torsion-testing machine. specimens to each Torsion were other. then

in quick-setting fixation of each Roentgenograms made in two

non-linear.

could not represent the true stiffness of the bone; we used the ratio of the maximum torque to the angle at that point (Fig. 1) to define the stiffness Results of these bones.

planes

Test

The bones were tested in a torsion-testing machine (A. H. Burstein, New York, N.Y.) which was connected to a recording oscilloscope (Techtronic type 56 1 B). Using
this experimental setup, four mechanical measurements of

One hypothesis

of the goals of this investigation that the rate of fracture healing

was to test is influenced

the by

the mechanical environment. To crease in strength in the fractured were determined, one tibia having stant compression and alternating compression. the other, The

do this the rates of intibiae of each rabbit been subjected to conto both results, constant and reported

the strength of each sional loading were torque

stiffness which the

tibia while determined. angle (2),

The Maximum torsion

it was These energy

torque angle

subjected to torwere: maximum (3),

the the torque number

(1),
(4)

torsion
(Fig. bone

absorption
was was

and
at

1). failed.

elsewhere 12, showed that when there was no difference in healing ical environments used. Therefore, were
ing had

all tibiae were compared rates in the two mechanall of the healing bones considering with
group

of degrees prior
ing had very

of torsional Most
low

deformation ofthe bones

and

in the before
therefore

test days

bone
angular

just
de-

put

into
been

one
applied.

group
This

without combined
combined

whether cyclic
was

to failure.

thirty
had

of healtorque. would curves to 8 deenergy of

compression

or compression

loaddivided

stiffness

formations

of as much

as 40 degrees

at maximum

Since angular deformations not be clinically acceptable, for these weak bones grees of deformation, absorption, deformation. comparisons maximum and stiffness

greater than 8 degrees the load-deformation only up torque,

according to healing time into six subgroups studied after three, four, five, six, seven, and of healing.
Strength Measurements as a Function
of Ti,ne

of fractures eight weeks

were considered and the maximum were computed

at 8 degrees

Fifty-three

of

the

sixty-six

bones

were

tested

and

This procedure, we believe, permitted valid of these weak bones with those that had deformations in the vicinity of 8 degrees. The of the energy in Figure put 1 by

analyzed. Thirteen bones were lost or for technical reasons could not be analyzed. The mean values plus or minus one standard deviation of the three measurements of strength are listed in Table I. The average newton-meters (one newton-meter maximum = ten torque kilograms in

energy absorption was used as a measure into the bone until failure. It is represented
l.5

force-centimeter) as a function of healing time in days was probably the best indicator of strength. Up to about twenty-three days, the torque reflected the so-called strength of the soft tissues connecting the fragments and
MAXIMUM TORQUE (1)

I .0

the surrounding bone. As healing progressed, the torque strength increased and reach a value of about 1 .5 newtonmeters at eight weeks. The energy absorption, the work done on the bone to deform it up to the point of failure, reached a value of about 0.075 newton-meter and weeks. at eight weeks. to deThe level average stiffness off at some time rose sharply after eight then seemed Additional

TORQUE
(Newton

ENERGY

ABSORPTION

(3)

Meters)
TABLE 0.5
CHANGES IN STRENGTH

I
PARAMETERS OF OF TIME

EXPERIMENTAL

FRACTURES

AS A FUNCTION

No.
HealingTime ( Weeks) 3 0 5 0 Bones

of

Torque (Nm) Av. S.D. Av.

Energy (Nm) S.D.

Stiffness (Nm/Deg.) Av. S.D.

0.22

0.09

.016

.007

.028

.012

4
5

14
4

0.35
0.75

0.31
0.30

.019
.040

.010
.015 .047

.081
.126 .264

.106
.057 .092

ANGULAR
FIG.

DISPLACEMENT
1 displacement one to four.

(Degrees)
graph showing

6 7
8

4 16
13

1.13 1.33
1.40

0.51 0.52
0.48

.065 .072
.071

.032
.042

.257
.322

.100
.074

A typical

four
VOL.

strength
59-A,

computer-plotted torque-angle parameters numbered from

NO. 2, MARCH 1977

190

A.

A.

WHITE

III,

M.

M.

PANJABI,

AND

W.

0.

SOUTHWICK

Rabbit
56

#54

TORQuE (NT-Il)

10 (DEGREE)

ftCUt5R FIG.

DISPLACEMENT

(DECREES)

ANGULAR DISPLOCEMEOT FIG.

3 of one rabbit after

in

2
two

Torque-angle

graphs
demonstrating

of the tibiae
a dramatic

twenty-six

days

A composite torque-angle graph of six bones representative of the entire healing period. The numbers on the graphs indicate days of healing time. As healing progresses, there is an increase in the strength of union shown by the changes in the torque-angle graphs.

of healing,

difference

the patterns

of the

curves. The left bone is stronger and has a hard-tissue type of curve, while the right bone is weaker and has a soft-tissue type ofcurve. The right leg was subjected to compression and cyclic loading and the left. to compression only. In the whole series, there was no consistent difference associated with the two types of treatment
12

tails
where Changes

of the
10

temporal

changes

in strength

are

given

elsesoft-tissue cated and at approximately hard-tissue phases twenty-six were sharply demardays by to twenty-seven

in the

Torque-Angle

Graphs

The
six these time

changing

patterns
bones with

of the
healing

torque-angle
times

graphs
ranging increased

of
from

an abrupt torque-angle (Fig. torque-angle right tibia

soft-tissue failed failure The with

change, dramatically plots of a pair of bones tibiae after twenty-six were with
pattern) stiffness

exemplified by the from the same rabbit days of healing had

representative graphs in both

3). These

twenty-one

to fifty-six
it is evident the initial

days
that and

are
the the

shown
stiffness part final

in Figure
of healing.

2. From
with

curves that (26 R) failed

type of failure high relatively

markedly different. The low stiffness (a rubbery,

and (a the left tibia (26 type L) of hard-tissue

The after
bones days,

earliest
with healing soft with

evidence
times had

of returning days
of twenty-one type

strength of healing.
to of phase

was
twenty-six behavior

seen In the
-

twenty-one
the

to twenty-four
tissue a rubbery

pattern). Four Biomechanical Stages of Healing

that
sociated

is, a low forty-nine

type

torque
a low

and
stiffness.

a large

angular
later

deformation
of healing,

as-

The

distinct

change

in the load-deformation

curve

at to

In the

after
tissue

to fifty-six
-

of behavior

days, the bones had a hardthat is, a high torque and small with a high stiffness. The

twenty-six to twenty-seven a much stiffer, hard-tissue evident clinically Both the physician when and

days from a rubbery quality type of resiliency is sometimes

angular

deformation

associated

fractures are treated in traction. the patient may observe that the

FIG.

FIG.

6 occurred failure failure failure

FIG.

Fig.
ture

4: Roentgenograms

Fig. 5: Roentgenograms
site. Fig. 6: Roentgenograms fracture site and in the Fig. 7: Roentgenograms the intact bone.

of Stage-I fracture healing before (left) and after (right) testing. Rubbery low-stiffness failure of Stage-Il fracture healing before (left) and after (right) testing. Hard-tissue, high-stiffness of Stage-Ill intact bone. of Stage-IV fracture fracture healing healing before before (left) (left) and after and after (right) (right) testing. testing. Hard-tissue, Hard-tissue, high-stiffness high-stiffness

at the fracture site. occurred at the fracoccurred occurred both at the in

entirely

THE

JOURNAL

OF

BONE

AND

JOINT

SURGERY

THE

FOUR

BIOMECHANICAL

STAGES

OF

FRACTURE

REPAIR

191 curve illustrated testing partially fails in Figure 2, and by the roentgenoshown in Figure
5. a

fracture time.
stiffness

becomes It is this
resistance

stuck
change
to the

during a rather from the springy,

short period of elastic, lowresis-

tern like the failure mode grams made Stage

twenty-seven-day like the one before and bone after

more

rigid,

high-stiffness

tance stuck.

that

the

clinician testing

feels one

when expects

the

fracture

becomes to occur at

111: The

through

the origithrough the hard-tissue

In torsional

the failure

nal experimental previously intact pattern

in Figure

fracture site and partially bone with a high-stiffness, forty-nine-day

example

the weakest part of the structure. tures in our rabbits tested with
occurred in one of three ways,

In the healing tibial fractorsional loading, failure

with the failure lines in one

like

the

torque-angle
failure

curve
mode

pattern
is seen

2. An

of a Stage-Ill

in the roentgenograms and after

Stage

in Figure site of failure fracture and

6 which

were

made

before

of the following tal fracture

healing

positions: osteotomy
fracture

purely
and

through site); partially

partially

the experimenthrough
through the

(healing

the
in-

testing. 1V: The

is not related occurs with illustrated the one we

to the origin Figure shown on 2. the if

experimental

inal experimental pattern like The failure

roentgenograms

a high-stiffness

tact bone; or entirely through the intact bone, not involving the healing experimental fracture. Based on these characteristics of healing bone tested to failure, we established the following four biomechanical stages
Stage

the fifty-six-day curve mode here resembles

in Figure 7.

Having

identified

these

stages

reasoned and useful, measurements The bones

that

of fracture
I:

repair. the original rubbery experipattern,

mental

fracture

The bone fails through site with a low-stiffness,

.

they are valid and are to be meaningful should correlate with the quantitative strength as well as the time since fracture.

they of were

and a torque-angle curve like the one for the twenty-oneday curve in Figure 2 The failure mode of a bone in this stage of healing is illustrated by the roentgenograms made before mental and
Stage

after

failure

shown

in Figure

4. the original experihard-tissue pat-

therefore grouped into the four biomechanical stages on the basis of the test data, and the average maximum torque and average energy absorption to failure for each group were determined (Figs. 8 and 9). Although there was considerable variation, a distinct trend progressive increase in strength from When the average times after fracture was evident in the Stage I to Stage IV. for the bones in each

II: fracture

The bone fails through site with a high-stiffness,

of the four stages were compared, the same progression was evident (Fig. 10). As demonstrated by the histograms (Figs. 8, 9, and 10), the four stages correlated closely with strength
MAX MUM TORQUE 1.5 ENERGY .10

and

healing

time,

and

the

statistical

correlations

of the differences by t test were energy absorption

shown in these histograms as determined significant (Table II), except those for in the later stages of healing.

Ui

I-

I0

Ui

IUi

z
0 I-

0.5

z
.05

Ui

0 Stage n=
Histogram biomechanical
the average

I (JO)

II (7)
FIG.

III
(15)
8

0 Stage
(21) Histogram
biomechanical the average

I
(10)

II
(7)
FIG.

III
(15)
9

IV
(21)

showing
stages

average
of fracture

maximum
healing. Vertical

torques
lines

for
above

the four and below

showing average energy absorptions for the four stages of fracture healing. Vertical lines above and below
points represent one standard deviation from the mean.

points
NO.

represent

one standard
1977

deviation

from

the mean.

VOL.

59-A,

2, MARCH

192
TABLE
SIGNIFKANCE ENERGY OF DIFFERENCES ABSORPTIONS. AND BETWEEN HEALING

A. II

A.

WHITE

III,

M.

M.

PANJABI,

AND

W.

0.

SOUTHWICK

THE TIMES

MEAN OF

MAXIMUM THE FOUR

TORQUES, STAGES

FRACTURE

60
Time

Stages Compared

Maximum Torque

Energy

Absorption
1

Healing

I-Il
Il-Ill Ill-IV I-Ill ll-lV
* .

:1: 1:
*

t
*

40
>-

*
*

t p <

< >

0.01 0.05 0.05

(significant). (significant). (not significant).

20

1: p

Discussion The
tween validity tion. has stages The aroused of the and

precise,
stages potential need healing for

objectively
and the

determined
correlations

demarcation
of the strength

bede-

0Stage n
=

I (10)
average stages

H
(7)

ifi
(15)

Iv
(21)
deviations for

terminations

and healing

time

make

us optimistic
proposed of bone tests
13,59,h1,

about
classificastrength

the

usefulness non-destructive interest by defined

of the

Histogram showing the four biomechanical

healing times and standard of fracture healing.

considerable

and of such

the

four
tests

our

biomechanical recognizable objective

more
stages

roentgenographic
of healing becomes

information
available,

relevant
it may be

to the
possible

four
to

should
tests. by

be
The

of value
end of Stage examination

in the

development
more

clinical
clinically If

I is readily or by

recognize
given time

and

identify
clinical

the stages
situation.

of fracture

healing

at any

physical

tests6-9.

in the

References
I
.

ABENDSCHEIN.

WALTER,
.

and

2.
3.
4.

BRASH,
CAMPBELL,

J . I.

J.

and SKORECKI, N., and JURIST.

WHITE.

HYATT, G. J .: Termination

CRELIN, Healing
DOHERTY,

E. S.:
Fractures.

5.
6.

W.
Long

P.:
Bones.
ACKE:

(To B0vILL,

J. M.: A. A.; PANJABI, be published.)

E. G.; and

W.: Ultrasonics and Selected Physical Properties of Bone. Clin. Orthop. . 69: 294-301, 1970. ofthe Modulus ofElasticity of Bone by a Vibration Method. Med. Biol. Eng. , 8: 389-393, 1970. Mechanical Impedance of the Femur: A Preliminary Report. J. Biomech., 4: 319-322, 1971. M. M.; and SOUTHWICK, W. 0.: Correlations of Microscopic Anatomy with Physical Properties
WILSON,

of of

Human
JERNUERGER.

J. Biomech.,
Measurement

7: 559-561, of Stability

E. L.: 1974.
of

Evaluation
Fractures.

of the

Use

of

Resonant

Frequencies

to Characterize Scandinas

Physical

Properties

Tibial

A Mechanical

Method.

Acta

Orthop.

ca.

Supplementum

135,

1970.

J. M.: In Viva Determination of the Elastic Response of Bone. I. Method of Ulnar Resonant Frequency Dctermination. Phys. Med. Biol.. 15: 417-426. 1970. 5. LEWIS, J. L.: GOLDSMITH. W.: and WONG, A.: Impact Response of Long Bones as a Diagnostic Tool br Fracture Healing. Proc. 26th Annual Conference on Engineering in Medicine and Biology. p. 45, 1973. 9. MATTHEWS. L. S.; KAUFER, HERBERT; and SONSTEGARD, D. A.: Manual Sensing of Fracture Stability: A Biomechanical Study. Acta Orthop. Scandinavica, 45: 373-381, 1974. 10. PANJABI. M. M., and WHITE, A. A.: Temporal Changes in the Physical Properties of Healing Bone Fractures. (To he published.) I I. THOMPSON, G. T.: In Vivo Determination of Bone Properties from Mechanical Impedance Measurements. Abstract from Aerospace Medical Association Annual Science Meeting, Las Vegas, Nevada, May 1973. 12. WHITE. A. A.; PANJAB1, M. M.; and SOUTHWICK, W. 0.: Effects ofCompression and Cyclical Loading on Fracture Healing. A Quantitative Biomechanical Study. (To be published.)
JURIST,

7.

THE

JOURNAL

OF

BONE

AND

JOINT

SURGERY