po

dRickard Branemark c, John H. Evans a,b, Mark J. Pearcy a,b

a

reect the dierent strategies used in controlling the prosthetic knee joint.

used for over a century, problems of pain in the residual too short to support the use of a conventional socket.In part to overcome these problems, a surgical technique

(osseointegration) has been developed to allow a prosthesisto be directly anchored to the bone through a xationbased on a titanium implant. A coupling device (the abut-ment) is attached to the implant, while its distal end

* Corresponding author. Address: Institute of Health and BiomedicalInnovation, Queensland University of Technology, Victoria Park Road,Kelvin Grove Qld 4059, Australia.

E-mail address: wc.lee@qut.edu.au (W.C.C. Lee).

Clinical Biomechanics 22 (2Interpretations. This study increases the understanding of biomechanics of bone-anchored osseointegrated prostheses. The loadingdata provided will be useful in designing the osseointegrated xation to increase the fatigue life and to rene the rehabilitation protocol. 2007 Elsevier Ltd. All rights reserved.

Keywords: Transfemoral amputation; Prosthetics; Osseointegration; Transducer; Activities of daily living

1. Introduction

The most common method of attaching a prosthesis to aresidual limb is by means of a prosthetic socket, often withsome suspensory devices to retain the socket when notload-bearing. Although this attachment method has been

limb and skin breakdown sometimes arise (Gallagheret al., 2001; Hagberg and Branemark, 2001; Mak et al.,2001). High pressure applied from the prosthetic socketto the soft tissue of a residual limb that is not adapted totolerate load has been suggested as the cause of the painand skin breakdown. In addition, some residual limbs areSchool of Engineering Systems, Queensland University of Technology, Brisbane, Australiab Institute of Health and Biomedical Innovation, Queensland University of Technology, Brisbane, Australia

c Centre of Orthopaedic Osseointegration, Sahlgrenska University Hospital, Goteborg, Swedend Department of Prosthetics and Orthotics, Sahlgrenska University Hospital, Goteborg, Sweden

Received 12 December 2006; accepted 12 February 2007

Abstract

Background. Direct anchorage of a lower-limb prosthesis to the bone through an implanted xation (osseointegration) has been sug-gested as an excellent alternative for amputees experiencing complications from use of a conventional socket-type prosthesis. However,an attempt needs to be made to optimize the mechanical design of the xation and rene the rehabilitation program. Understanding theload applied on the xation is a crucial step towards this goal.

Methods. The load applied on the osseointegrated xation of nine transfemoral amputees was measured using a load transducer,when the amputees performed activities which included straight-line level walking, ascending and descending stairs and a ramp as wellas walking around a circle. Force and moment patterns along each gait cycle, magnitudes and time of occurrence of the local extrema ofthe load, as well as impulses were analysed.

Findings. Managing a ramp and stairs, and walking around a circle did not produce a signicant increase (P > 0.05) in load comparedto straight-line level walking. The patterns of the moment about the medio-lateral axis were dierent among the six activities which mayKinetics of transfemoral amxation performing comm

Winson C.C. Lee a,b,*, Laurent A. Frossar0268-0033/$ - see front matter 2007 Elsevier Ltd. All rights reserved.doi:10.1016/j.clinbiomech.2007.02.005utees with osseointegratedn activities of daily living

a,b, Kerstin Hagberg c, Eva Haggstrom d,

www.elsevier.com/locate/clinbiomech

007) 665673

iomprotrudes through the soft tissue to provide the attachmentfor the external prosthesis. In addition to alleviating theskin problems and residual limb pain (Sullivan et al.,2003), studies have also shown that amputees using trans-femoral osseointegrated prostheses enjoy a greater rangeof hip motion and better sitting comfort compared to thesocket-type (Hagberg et al., 2005). They can walk furtherand be more active than using a conventional prosthesis(Robinson et al., 2004; Sullivan et al., 2003), and can haveimproved sensory feedback (through osseoperception)(Branemark et al., 2001). External components of the pros-thesis can be attached to and detached from the abutmenteasily and the alignment is faithfully preserved. To date,there are over 80 transfemoral amputees world-wide whohave been tted with the transfemoral osseointegrated x-ation developed by Dr. R. Branemark (Branemark et al.,2001). However, factors such as steroid medication oranti-tumour chemotherapy which may interfere with bonehealing, heavy smoking or diabetes which may increase riskof bone sepsis, and body weight in excess of 100 kg are pos-sibly contraindications to the tting of lower-limb osseoin-tegrated prostheses (Robinson et al., 2004).

As with amputees using socket-type prostheses, thosewho use osseointegrated prostheses undergo a rehabilita-tion process, which involves incremental static loadingson the abutment to prepare the bone to tolerate forcestransmitted from the implant when performing essentialactivities. Understanding the load experienced during var-ious activities might help rene the rehabilitation process.Previous studies have also indicated certain gait deviationsamong transfemoral amputees in walking on level anduneven terrains (Jaegers et al., 1995; James and Oberg,1973; Murray et al., 1983; Schmalz et al., 2007), in spiteof the improved prosthetic knee joint designs in lockingand bending mechanics. In addition, mechanical failuresof the abutment sometimes occur after long use or as theresult of excessively high magnitude load application usu-ally induced by a fall (Sullivan et al., 2003). The abutmentis designed to fail in order to protect the bone from over-load, but attempts can be made to optimize the strengthof the xation and to develop safety devices to protectthe xation and the bone with the understanding of theloads developed during common daily activities. To renethe rehabilitation program as well as to develop the xationsystem, safety devices and dierent prosthetic componentsto address the mechanical problems and improving walkingability, it is important to have a comprehensive under-standing of the load applied on the xation.

Over the past two decades, loadings applied at the distalend of prosthetic sockets have been studied (DiAngeloet al., 1989; Stephenson and Seedhom, 2002; Nietertet al., 1998; Berme et al., 1975; Frossard et al., 2003).The load has been calculated using inverse dynamicsrelying on the motion of the prosthesis captured by amotion analysis system and the ground reaction forces

666 W.C.C. Lee et al. / Clinical Bmeasured by a force plate (DiAngelo et al., 1989; Stephen-son and Seedhom, 2002). The load can also be measureddirectly using appropriate load transducers (Berme et al.,1975; Nietert et al., 1998). Frossard and colleagues (Fros-sard et al., 2003) measured the direct load applied at thetransfemoral socket end using a commercial load trans-ducer mounted between the prosthetic knee joint and thesocket, and suggested that direct load measurement couldimprove accuracy and allow measurement to be taken forunlimited walking steps. In addition, direct measurementallows loadings to be measured for any type of activityand on any terrain.

Although the application of such direct measurementtechniques to transfemoral amputees using osseointegratedprostheses has been reported (Frossard et al., 2001), thedata were limited to one subject only. Comprehensiveunderstanding of the load applied on osseointegrated xa-tion during level walking is important. It is also crucial tounderstand the load in various activities of daily living suchas climbing stairs and walking inclines. Because of the lossof some musculature and joint mobility, functionaldemands may increase dramatically when performing dailyactivities. In addition, amputees may employ dierent load-ing strategies to help them manage dierent activities, all ofwhich may pose a potential danger to the structural integ-rity of the xation system and the bone.

The aim of this study is to use a direct measurementmethod to compare the load applied on the osseointegratedxation of nine transfemoral amputees performing severalactivities of daily living including managing ramps, stairsand walking around a circle which are believed to be themost commonly performed activities during daily living.

2. Methods

2.1. Participants and prostheses

A total of two female and seven male unilateral transfe-moral amputees tted with osseointegrated xation partic-ipated in this study. The demographic details of eachsubject are summarized in Table 1. All participants havebeen walking with the xation for at least one year, andcan walk 200 m independently without additional walkingaids. Load measurement took place in a clinical environ-ment at Sahlgrenska University Hospital, Gothenburg,Sweden where the participants were recruited. Humanresearch ethical approval was received from the Queens-land University of Technology. Written consent wasobtained from all participants.

Amputees were tted with their regular prosthetic com-ponents, as presented in Table 1, with the load transducersubstituted for the adaptor which connected the Rotasafeto the knee joint. Rotasafe is a safety device, based on aclutch, which prevents excessive torque on the abutment.A compromise was made for three participants who couldnot retain a Rotosafe due to the lack of space to t the loadtransducer. The transducer was tted by a prosthetist who

echanics 22 (2007) 665673replicated the usual alignment of the prosthesis for eachamputee.

Table 1Subject characteristics

Subjectnumber

Gender(M/F)

Age(years)

Height(m)

Total massa

(kg)Side of amputation(R/L)

Footwear Prostheticfoot

Prostheticknee

Rotasafe

1 F 57 1.63 61.1 R Sandals Totalconcept

Total knee Yes

2 M 50 1.81 74.3 L Sandals TruStep Total knee No3 M 59 1.89 87.1 R Leather

shoesTruStep Total knee No

4 F 49 1.58 53.3 R Sandals Totalconcept

Total knee Yes

5 M 41 1.77 96.6 R

6 M 26 1.78 90 R

7 M 46 1.99 99.5 L8 M 50 1.82 99.8 R

9 M 45 1.72 80.4 R

Mean 47 1.78 82.5

prosthesis.

W.C.C. Lee et al. / Clinical Biomechanics 22 (2007) 665673 6672.2. Apparatus

The technique used to measure the load is similar to theone described in previous studies (Frossard et al., 2003;Frossard et al., 2001). A six-channel load transducer(Model 45E15A; JR3 Inc., Woodland, USA) was used tomeasure directly the 3-dimensional forces and momentsapplied to the abutment. The transducer was mounted tocustomized plates that were positioned between the abut-ment and the prosthetic knee. The transducer was aligned

Standard deviation 9.7 0.12 16.8

a The total mass includes body mass plus the mass of the instrumentedso that its vertical axis was co-axial with the long (L) axisof the abutment and femur. The other axes correspondedto the anatomical antero-posterior (AP) and medio-lateral(ML) direction of the abutment as depicted in Fig. 1.

Fig. 1. Example of a typical prosthetic leg setup used to directly measurethe forces and moments applied on the xation of transfemoral amputee(left: front view, right: side view). A commercial transducer (A) wasmounted to specially designed plates (B) that were positioned between theadaptor (C) connected to the xation (D) and the knee mechanism (F).Forces acting along the AP, ML and L axes were denotedas FAP (anterior was positive), FML (lateral was positive),and FL (compression was positive), respectively. Momentsabout the three axes were denoted as MAP (lateral rotationwas positive), MML (anterior rotation was positive) andML (external rotation was positive), respectively. The max-imum capacity was 2273 N for FL, 1,136 N for FAP andFML, and 130 Nm for moments about the three axes. Accu-racy was 0.1% of the maximum capacity. Each channel wassampled at 200 Hz. A wireless modem (Ricochet ModelRunningshoes

C-walk Total knee No

Leathershoes

Carbon copy C-leg Yes

Sandals C-walk Total knee YesLeathershoes

Flex foot GaitMaster Yes

Runningshoes

TruStep Total knee Yes21062; Metricom Inc., Los Gatos, USA) was used to trans-mit data from the transducer to a nearby laptop computer.

2.3. Protocol

Approximately 15 min of practice with the instrumentedprosthetic leg was allowed before load measurement toensure subject condence and comfort. Then, the partici-pants were asked to perform each of the activities includ-ing: straight-line level walking; walking upstairs,downstairs, upslope, and downslope; and walking along a

Table 2Descriptions of the six various activities performed during directmeasurement of load

Activities Descriptions

Levelwalking

Level walking along a level, straight-line walkway

Downslope Descending a 6.5 degrees of slopeUpslope Ascending a 6.5 degrees of slopeDownstairs Descending stairs of 30 cm height 34 cm deepUpstairs Ascending stairs of 30 cm height 34 cm deepCircle Level walking around a circle of 2 m diameter with the

prosthetic leg outside

circle. Detailed descriptions of each activity are shown inTable 2. Load data were measured for at least ve stepsof the prosthetic limb for each activity. The amputees wererequired to walk and manage the stairs and slope at a self-selected, comfortable speed. The order of each activity wasrandomized. Finally, the prosthesis was detached from theresiduum to enable a one-minute recording without loadapplied on the transducer for calibration purposes.

2.4. Data processing

The raw force and moment data were imported and pro-cessed by a customized Matlab software program (TheMathWorks Inc., MA, USA). The load data were osetaccording to the magnitude of the load recorded duringunloaded conditions. The rst and last strides recordedfor each trial were also removed in order to avoid the ini-tiation and termination of walking. The patterns of thethree-dimensional forces and moments for each gait cyclewere analysed. The heel contact and toe-o time weredetermined according to the curve of the long-axis force.

300

400

500

600

700

800

Forc

e (N

)

FAP

FML

FL

Resultant

FL1

FL1

FAP+

Toe off

10

t (N

m ML+

a

668 W.C.C. Lee et al. / Clinical Biom0 10 20 30 40

-30

-20

-10

0

% gait cycle

Mom

en

MML-

50 60 70 80 90 100

Fig. 2. The local extrema and typical patterns of (a) forces and (b)-100

0

100

200

0 10 20 30 40 50 60 70 80 90 100

% gait cycleFAP-

FML+

20

30

40

)

MAP

MML

ML

MAP+MML+

Toe off bmoments along the antero-posterior (AP), medio-lateral (ML), and long-axis (L) axes of subject 1 performing straight-line level walking.A gait cycle was dened as the period between two consec-utive heel contacts.

The magnitude of local extrema of the three componentsof forces and moments presented in Fig. 2 were determinedfor each step of the prosthetic limb. Resultant forces (Fr)were calculated by the vector sum of FAP, FML and FL.The time of occurrence (expressed in percentage of stancephase time) as well as the magnitude (expressed in percent-age of body weight) of the local extrema of Fr were identi-ed. Impulse (IAP, IML, IL and IR) for each step of theprosthetic limb were calculated by using the conventionaltrapezoid method to integrate the area under the forcetime curves (FAP, FML, FL and Fr). Each parameter (thelocal extrema, time of occurrence and impulse) was aver-aged across steps for each subject. To discuss the dier-ences in loading strategies among various activities ofdaily living, the means and standard deviations of eachparameter across subjects were computed for each activity.Statistical analyses were performed in SPSS statistical soft-ware (LEAD technologies, Inc.). Dierences among theactivities were determined by repeated measures analysisof variance (ANOVA). A post-hoc Tukeys test was usedto identify the signicant dierence. A signicance levelof P < 0.05 was used.

3. Results

3.1. Patterns of forces and moments

Fig. 2 shows the typical patterns of forces and momentsin straight-line level walking. It can be seen that althoughthe three loading axes (AP, ML, L) were xed relative tothe limb during ambulation, the three components offorces followed a pattern that was similar to the groundreaction forces obtained with a xed force-plate (Perry,1992; Zahedi et al., 1987). As expected, FL was the largestin magnitude among the three components of forces andpresented two peaks and a valley. Small plateaus werefound immediately before the rst peak of the curve in afew subjects, which may be explained by a sense of insecu-rity or discomfort. During level walking, the abutmentexperienced some posterior braking forces at the earlystance phase, and anterior propulsive forces at the latestance phase, and consistently experienced some lateralforces throughout the entire stance phase of the gait.Lateral rotational moment was consistently experiencedduring the stance phase of the gait. Anterior rotationalmoment was experienced during the mid-stance phase,and posterior rotational moment at the late-stance phaserelated to unlocking of the prosthetic knee joint was expe-rienced when performing level walking. Due to the tractioncreated by gravity acting on the mass of the prosthesiswhich was located below the transducer, the forces andmoments had small magnitudes during the swing phase.All subjects demonstrated similar patterns of forces and

echanics 22 (2007) 665673moments for level walking, except for ML which showedinconsistent patterns among participants (Fig. 3).

When performing the other activities of daily living, thepatterns of forces and moments were close to those of levelwalking as described above. However, there were excep-tions for FAP, ML, and MML. Fig. 4a and b show FAPalong a gait cycle when the majority of the subjects walkedupstairs and downstairs. Unlike level walking with someposterior forces exerted at the abutment at the early stancephase and anterior forces at the late stance phase, the

for walking inclines and around a circle follow those forstraight-line level walking. All subjects demonstratedinconsistent ML patterns among the six activities depictedin Fig. 5. Fig. 6ad display the dierent patterns of MMLfor all subjects among the four activities: ascending anddescending stairs and a ramp. Walking upslope andupstairs produced an average anterior rotational momentof 17 Nm and 10 Nm, respectively, while walking down-slope and downstairs did not produce a prominent peakanterior rotation moment (Fig. 6). Walking around a circleproduced similar MML patterns to level walking along astraight-line.

3.2. Local extrema

A total of 10 local extrema for the three components offorces and moments were studied, which represented thekey features of the curve plotting force/moment dataagainst time as presented in Fig. 2. The local extrema werethe turning points of the curves which presented the highestabsolute magnitude of loads. Two local extrema were iden-tied for the maximum anterior (FAP+) and posterior

-8

-6

-4

-2

0

2

4

6

8

0 10 20 30 40 50 60 70 80 90 100

% gait cycle

ML

(Nm

)

External rotation

Internal rotation

Fig. 3. External/internal rotational moment (ML) for all subjectsperforming straight-line level walking.

c

W.C.C. Lee et al. / Clinical Biomechanics 22 (2007) 665673 669abutments of seven subjects experienced posterior forcesmost of the time at the stance phase when they walkeddownstairs. When walking upstairs, the abutment of sixsubjects experienced anterior forces during the entirestance phase. The FAP patterns of the remaining subjectsare displayed in Fig. 4c and d, which show entirely dier-ent curve patterns. In calculating the local extrema of FAP,those remaining subjects were excluded. Patterns of FAP

-150

-100

-50

0

50

100

0 10 20 30 40 50 60 70 80 90 100

F AP (

N)

AnterioraDownstairs, n=7

Upstairs, n=6

-300

-250

-200

% gait cyclePosterior

-250Posterior

-50

-30

-10

10

30

50

70

90

% gaitcycle

F AP

(N)

Anterior

Posterior

1000 10 20 30 40 50 60 70 80 90

b d

Fig. 4. Antero-posterior force for (a) seven subjects walking downstairs, and(c) walking downstairs, and (d) walking upstairs, which showed dierent curv Downstairs, n=2

Upstairs, n=3

F AP (

N)

-300 % gait cycle

-50

-30

-10

10

30

50

70

90

% gait cycle

Anterior

Posterior

1000 10 20 30 40 50 60 70 80 90(FAP) forces, one for the maximum lateral force (FML+),and two peaks for the axial force (FL1 and FL2). One localextrema was identied for the maximum lateral rotationalmoment (MAP+), two for the maximum anterior (MML+)and posterior (MML) rotational moments and two localextrema were identied for the peak external (ML+) andinternal (ML) rotational moments. In addition, the mag-nitudes of the two peaks (Fr1 and Fr2) and the time of

F AP (

N)

-200

-150

-100

-50

0

50

100

100

Anterior

0 10 20 30 40 50 60 70 80 90(b) six subjects walking upstairs. The remaining subjects are shown ine patterns from the majority of the amputees.

occurrence of the two peaks (TFR1 and TFR2) wereidentied.

Table 3 shows the mean and standard deviation acrosssubjects of each local extrema in six dierent activities.There was no statistical dierence among activities in eachlocal extrema of the three components of forces andmoments. As far as the body weight normalized resultantforces are concerned as presented in Table 4, it was foundthat the Fr1 in walking upstairs [101% (SD 14%) bodyweight] was statistically higher than in walking downstairs[78% (SD 12%) body weight]. Meanwhile, the TFR2 inwalking downstairs [56% (SD 11%) stance phase] was sta-tistically earlier than other activities except walking down-slope. The TFR1 in walking downstairs [26% (SD 33%)stance phase], in addition, appeared statistically earlierthan walking upstairs [39% (SD 8%) stance phase]. Timeof occurrence of local extrema in each particular compo-nent of force and moment was not computed because ofthe highly inconsistent patterns in some activities.

-8

-6

-4

-2

0

2

4

0 10 20 30 40 50 60 70 80 90 100

% gait cycle

ML

(Nm

)

DownslopeUpslopeDownstairsUpstairsWalkingCircle

External rotation

Internal rotation

Fig. 5. Typical external/internal rotational moment (ML) for the sixdierent activities

-30-20-10

010203040

0 10 20 30 40 50 60 70 80 90 100

MM

L(N

m)

Anterior rotation

-70-60-50-40-30-20-10

010203040

0 10 20 30 40 50 60 70 80 90 100

% gait cycle

MM

L(N

m)

Anterior rotation

Posterior rotation

-70-60-50-40-30-20-10

010203040

0 10 20 30 40 50 60 70 80 90 100

% gait cycle

MM

L(N

m)

Anterior rotation

Posterior rotation My-

My+

My+

My-

-70-60-50-40-30-20-10

010203040

0 10 20 30 40 50 60 70 80 90 100

% gait cycle

MM

L(N

m)

Anterior rotation

Posterior rotation

a

b

c

d

Fig. 6. Moment about the medio-lateral axis (My) for all subjects man

Table 3Mean and standard deviation (in bracket) across subjects of the nine local ext

Walking Downslope Upslo

FAP (N) 74 (36) 93 (44) 53 (3FAP+ (N) 101 (19) 87 (29) 90 (25FML+ (N) 89 (35) 79 (22) 93 (39FL1 (N) 671 (139) 699 (149) 697 (1FL2 (N) 675 (138) 660 (146) 704 (1MAP+ (Nm) 21 (10) 25 (9) 22 (8)

(120 (9(1.

670 W.C.C. Lee et al. / Clinical Biomechanics 22 (2007) 665673MML+ (Nm) 9 (10) 17MML (Nm) 20 (9) 30 (20) 2ML+ (Nm) 3.7 (1.2) 5.3 (2.7) 3.2

ML (Nm) 5.0 (2.0) 3.8 (1.3) 6.3 (No statistical dierences were found in those local extrema. (A indicates-70-60-50-40 % gait cycle

Posterior rotation

aging (a) downslope, (b) downstairs, (c) upslope, and (d) upstairs.

rema

pe Downstairs Upstairs Circle

4) 137 (98) 69 (29)) 74 (20) 84 (27)) 53 (14) 76 (30) 93 (34)53) 587 (157) 769 (171) 706 (165)44) 649 (112) 715 (170) 703 (148)

18 (8) 19 (8) 27 (9)) 10 (14) 11 (11)) 18 (6)7) 5.3 (3.6) 3.0 (1.1) 3.8 (1.7)

2.5) 3.5 (1.0) 3.7 (1.2) 5.4 (1.2)that there were no consistent peaks/valleys in that activity)

ltan

slop

91 (94 (39 (72 (

e

(39)(47)(79)(85)

iom3.3. Impulses

The overall loading of the prosthesis over the supportphase represented by the impulse is provided in Table 5.As expected, the impulse produced in long-axis (IL) wasthe largest in magnitude among the three axes. IAP in walk-ing upslope was statistically higher than that of walkingdownslope and downstairs, while IL and IR in walkingdownstairs was statistically lower than other activitiesexcept walking down the slope.

4. Discussion

Conventionally, the load experienced by prosthetic com-ponents, including implants in the lower-limbs, can be esti-

Table 4Mean and standard deviation (in bracket) across subjects of the peak resu

Walking Downslope Up

Fr1 (%BW) 89 (6) 89 (8)Fr2 (%BW) 93 (9) 86 (7)TFR1 (%SP) 32 (8) 30 (6) *

TFR2 (%SP) *70 (3)&,$63 (8) #,&

A pair of *,#,+,&,$ in each parameter represents a statistical dierence.

Table 5Mean and standard deviation (in bracket) across subjects of impulses

Walking Downslope Upslop

IAP (Ns) 108 (37) *75 (38) *,#130

IML (Ns) 58 (41) 51 (36) 65IL (Ns) *363 (76) 332 (56)

#385IR (Ns) *386 (81) 348 (59)

#415

A pair of *,#,+, in each parameter represents a statistical dierence.

W.C.C. Lee et al. / Clinical Bmated using inverse dynamics methods which are based onthe motion of the limb and the ground reaction forces(DiAngelo et al., 1989; Stephenson and Seedhom, 2002).The drawbacks of this method are that only one or twosteps of walking can usually be measured, force-plate tar-geting can produce un-natural gait (Wearing et al.,2001), and accurate determination of inertia of naturaland prosthetic limb segments is needed. In addition, loadscannot be measured easily using force plates when walkingon uneven terrain. This study used a portable recordingsystem based on a load transducer and a wireless modem.This allowed direct measurement of load applied to theabutment. In addition, the wireless system allowed the trueloading to be measured in dierent environments, and sub-jects could walk unimpeded when they performed the var-ious activities.

The main objective of this study was to examine the dif-ferences in load applied on osseointegrated xation duringvarious activities of daily living. If level straight-line walk-ing is considered to be the baseline, it was found thatthe other ve activities which were believed to be morephysically demanding did not produce any statisticallysignicant increase in loading when compared to the base-line. This may imply that these ve activities would notinduce higher risk to the structural integrity of the xationsystem. Impulses represented the utilisation of the prosthe-sis. IL in walking downstairs was statistically lower than instraight-line walking, but it was also statistically lower thanwalking upslope, upstairs and around a circle.

There was no statistical dierence in local extrema iden-tied in the three components of forces and momentsbetween every pair of activities. Although there were largedierences in some cases, for instance, FAP in walkingdownstairs was 47% higher than the average of all the otheractivities, a statistical dierence was not reached because ofthe large standard deviations across subjects. Some statisti-cal dierences were found in the magnitude and the time of

t forces and the time of occurrence of the peaks

e Downstairs Upstairs Circle

5) *78 (20) *101 (14) 91 (6)8) 85 (16) 94 (7) 93 (6)8) *26 (11) 30 (6) 34 (7)4) *,#,+,56 (11) +,$79 (3) 70 (4)

Downstairs Upstairs Circle

#67 (33) 104 (35) 113 (33)42 (27) 54 (38) 62 (42)

*,#,+,256 (31) +367 (69) 388 (79)*,#,+,271 (35) +388 (72) 411 (82)

echanics 22 (2007) 665673 671occurrence of peak resultant forces. Fr1 in walking down-stairs was on average the lowest among all activities andwas statistically lower than in walking upstairs. In addi-tion, TFR2 in walking downstairs and downslope werestatistically earlier than some other activities, while TFR1in walking downstairs was statistically earlier than in walk-ing upslope. The earlier peaks as well as lower magnitudeof Fr1 are likely to be due to the rapid forward progressionof the prosthetic limb, resulting from the lack of plantarexion and active knee exion of the prosthetic joints,which are important in maintaining stability when descend-ing from a step or a ramp. This also explains the signicantreduction of impulses in walking downstairs. In addition,because of the lack of active motion of the joints, amputeestend to roll the prosthetic foot over the step edges whenmanaging stair descent as reported for conventional trans-femoral amputees (Schmalz et al., 2007).

Similar patterns of forces and moments were seen inthe six dierent activities, except for FAP, ML and MML.Rotational moment (ML) was inconsistent among subjectsand activities. The maximum absolute value of externaland internal rotational moment fell within a narrow range

I impulse of F (integrated the area under the

Berme, N., Lawes, P., Solomonidis, S., Paul, J.P., 1975. A shorter pylon

iombetween 3.0 Nm and 6.3 Nm among the six activities. Thiscould have implications in the minimal torque required totighten the abutment to the implant, as well as thresholdvalue for the Rotasafe rational protective devices. The vari-ations in the patterns of FAP and MML could suggest thatamputees use dierent strategies in prosthetic knee jointcontrol when they manage dierent activities. Walking ups-lope and upstairs produced some anterior rotationalmoment during stance phase. Walking downslope anddownstairs, on the other hand, did not produce a promi-nent peak anterior rotation moment (Fig. 5), which indi-cates the line of action of the ground reaction force wasalways kept behind the knee joint. There was a peak pos-terior rotational moment at the late stance phase of the gaitwhen the subjects performed level walking, managingslope, and walking around a circle. This could be explainedby the eort of initiating knee exion in order to providefoot clearance during the swing phase.

Various strategies of prosthetic knee control mightexplain the dierent loading recorded in dierent activities.A good control of the prosthetic knee is critical for asmooth ambulation and there are dierent prosthetic kneesin the market that employ various control mechanisms ingiving motion and stability to the knee. The dierences inloading may also be explained by the dierent walking pat-terns. For example, rolling the prosthetic foot over the edgeof the step may explain the reduction of loads. Prostheticalignment, which refers to the special position of the pros-thetic foot relative to the residual limb, is also important.Variation in alignment can change the load transfer biome-chanics as well as walking ability. In order to aid in preciseexplanation of the load, control of the prosthetic knee andalignment, future studies can collect simultaneous kineticand kinematic data to determine body motion, knee jointangle and loading. Studies can also be performed to inves-tigate the eect of dierent knee mechanisms and align-ment on gait.

This study focuses on the loading applied on the xationsystem which will be helpful in the future mechanicaldesign of the system and rening of the rehabilitation pro-cess. Future studies will perform experimental structuraltests and computational stress analyses to investigate thestress/strain distribution at the bone-implant interfaceand the entire xation system and estimate the fatigue lifeusing the existing load data.

5. Conclusions

This study measured and compared the load acting onthe xation of nine transfemoral amputees tted withosseointegrated xation performing six dierent activitiesof daily living. The magnitudes of local extrema as wellas the curve patterns of each component of forces andmoments were revealed. Results suggested that managingramp and stairs, and walking around a circle did not pro-

672 W.C.C. Lee et al. / Clinical Bduce a signicant increase in load compared to straight-line level walking. Results also suggested that dierenttransducer for measurement of prosthetic forces and moments duringamputee gait. Engineering in Medicine 4, 68.

Branemark, R., Branemark, P.I., Rydevik, B., Myers, R., 2001. Osseo-integration in skeletal reconstruction and rehabilitation: a review.Journal of Rehabilitation Research and Development 38, 175181.

DiAngelo, D.J., Winter, D.A., Ghista, D.N., Newcombe, W.R., 1989.AP AP

forcetime curve)IML impulse of FML (integrated the area under the

forcetime curve)IL impulse of FL (integrated the area under the force

time curve)IR impulse of FR (integrated the area under the force

time curve)

Referencesstrategies were used to control the prosthetic knee jointfor dierent activities. Future studies will attempt to utilizethis load data to optimize the xation design as well as torene the rehabilitation process, and to collect kinematicdata to produce full explanation of the load data.

Acknowledgements

The authors would like to express their gratitude tomembers of the School of Engineering Systems and Insti-tute of Health and Biomedical Innovation, particularlyDr James Smeathers, for their valuable contribution andfeedback during the writing of this manuscript. This studywas partially funded by ARC Discovery Project(DP0345667), ARC Linkage (LP0455481), QUT StrategicLink with Industry and IHBI Medical Device DomainGrants.

Appendix.

FAP antero-posterior forceFML medio-lateral forceFL long-axis (of residual femur) forceFr resultant forceMAP moment about the antero-posterior axisMML moment about the medio-lateral axisML moment about the long-axisFAP, FAP+ the most positive and negative values of FAPFML+ the most positive value of FMLFL1, FL2 the two peaks of FLMAP+ the most positive value of the MAPMML+, MML the most positive and negative value of the

MMLML+,ML the most positive and negative value of theMLFr1, Fr2 the two peaks of FrTFR1, TFR2 time of occurrence of Fr1 and Fr2

echanics 22 (2007) 665673Performance assessment of the Terry Fox jogging prosthesis for above-knee amputees. Journal of Biomechanics 22, 543548.

Frossard, L., Beck, J., Dillon, M., Evans, J., 2001. Comparison betweenthe forces and moments applied on the residium of above-kneeamputees during daily life activities and walking. In: Proceedings ofXth International Society of Prosthetics and Orthotics World Con-gress. Glasgow, Scottland.

Frossard, L., Beck, J., Dillon, M., Chappell, M., Evans, J.H., 2003.Development and preliminary testing of a device for the directmeasurement of forces and moments in the prosthetic limb oftransfemoral amputees during activities of daily living. Journal ofProsthetics and Orthotics 15, 135142.

Gallagher, P., Allen, D., MacLachlan, M.I., 2001. Phantom limb pain andresidual limb pain following lower limb amputation: a descriptiveanalysis. Disability and Rehabilitation 26, 522530.

Hagberg, K., Branemark, R., 2001. Consequences of non-vascular trans-femoral amputation: a survey of quality of life, prosthetic use andproblems. Prosthetics and Orthotics International 25, 186194.

Hagberg,K.,Haggstrom,E.,Uden,M., Branemark,R., 2005. Socket versusbone-anchored trans-femoral prostheses: hip rangeofmotionand sittingcomfort. Prosthetics and Orthotics International 25, 186194.

Jaegers, S.M.H.J., Arendzen, J.H., de Jongh, H.J., 1995. Prosthetic gait ofunilateral transfemoral amputees: a kinematic study. Archives ofPhysical Medicine and Rehabilitation 76, 736743.

James, U., Oberg, K., 1973. Prosthetic gait pattern in unilateral above-knee amputees. Scandinavian Journal of Rehabilitation Medicine 5,3550.

Mak, A., Zhang, M., Boone, D., 2001. State-of-the-art research in lower-limb prosthetic biomechanics-socket interface: a review. Journal ofRehabilitation Research and Development 28, 161174.

Murray, M.P., Mollinger, L.A., Sepic, S.B., Gardner, G.M., 1983. Gaitpatterns in above-knee amputee patients: hydraulic swing control vs

constant-friction knee components. Archives of Physical Medicine andRehabilitation 64, 339345.

Nietert, M., Englisch, N., Kreil, P., Alba-Lopez, G., 1998. Loads in hipdisarticulation prostheses during normal daily use. Prosthetics andOrthotics International 22, 199215.

Perry, J., 1992. Gait Analysis: Normal and Pathological Function. Slack,Thorofare, NJ.

Robinson, K.P., Branemark, R., Ward, D., 2004. Future developments:osseointegration in transfemoral amputees. In: Smith, D.G., Michael,J.w., Bowker, J.H. (Eds.), Atlas of Amputations and Limb Decien-cies: Surgical, Prosthetic and Rehabilitation Principles. AmericaAcademy of Orthopaedic Surgeons.

Schmalz, T., Blumentritt, S., Marx, B., 2007. Biomechanical analysis ofstair ambulation in lower limb amputees. Gait & Posture 25, 267278.

Stephenson, P., Seedhom, B.B., 2002. Estimation of forces at the interfacebetween an articial limb and an implant directly xed into the femurin above-knee amputees. Journal of Orthopaedic Science 7, 192297.

Sullivan, J., Uden, M., Robinson, K., Sooriakumaran, S., 2003. Rehabil-itation of the trans-femoral amputee with an osseointegrated prosthe-sis: the United Kingdom experience. Prosthetics and OrthoticsInternational 27, 114120.

Wearing, S.c., Urry, S.r., Smeathers, J.E., 2001. Ground reaction forces atdiscrete sites of the foot derived from pressure plate measurements.Foot & Ankle International 22, 653661.

Zahedi, M.S., Spence, W.D., Solomonidis, S.E., Paul, J.P., 1987.Repeatability of kinetic measurements in gait studies in gait studiesof the lower limb amputee. Prosthetics and Orthotics International 11,5564.

W.C.C. Lee et al. / Clinical Biomechanics 22 (2007) 665673 673

Kinetics of transfemoral amputees with osseointegrated fixation performing common activities of daily livingIntroductionMethodsParticipants and prosthesesApparatusProtocolData processing

ResultsPatterns of forces and momentsLocal extremaImpulses

DiscussionConclusionsAcknowledgementsAppendixReferences