ARTICLE IN PRESS

Journal of Biomechanics 42 (2009) 736–742

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/jbiomech

Journal of Biomechanics

0021-92

doi:10.1

� Corr

E-m1 Te

www.JBiomech.com

Lower limb dynamics change for children while walking with backpack loadsto modulate shock transmission to the head

Tarkeshwar Singh a,�, Michael Koh b,1

a Physical Education and Sports Science, National Institute of Education, Nanyang Technological University, Singaporeb School of Sports, Health and Leisure, Republic Polytechnic, Singapore

a r t i c l e i n f o

Article history:

Accepted 9 January 2009Backpack load carriage increases ground reaction forces and increases the stiffness in the upper

extremity that can cause transmission of higher amount of forces from the lower extremity to the head.

Keywords:

Backpack

Gait

Pelvic rotation

90/$ - see front matter & 2009 Elsevier Ltd. A

016/j.jbiomech.2009.01.035

esponding author.

ail addresses: tarkesh@psu.edu (T. Singh), koh

l.: 94593107.

a b s t r a c t

This study investigated the effect of load carriage and placement of load on the back on the shock

transmission mechanisms amongst children. Fifteen primary school boys with mean age 10.01 (71.31)

years, mean height 136.40 (710.08) cm and mean mass 31.83 (77.13) kg completed the study. Subjects

carried 10%, 15% and 20% bodyweight (BW) loads on two locations on the back, namely upper and lower.

Results showed a significant reduction in pelvic and trunk rotation in the transverse plane and an

increase in the upper body stiffness for loads exceeding 15% of BW. The lower limb results showed a

reduction in the first peak force and cadence and a significant change in the walking velocity and time to

the first peak force for 20% load. No significant differences were found for the load configuration but the

upper configuration showed slightly higher shock transmission. The changes in the lower limb

dynamics indicated that there are locomotion mechanisms in place amongst children to modulate shock

transmission to the head.

& 2009 Elsevier Ltd. All rights reserved.

1. Introduction

Large forces are produced during gait at the lower extremityjoints due to muscle contractions and impacts with the ground(Ratcliffe and Holt, 1997). Backpack load carriage induces stiffnessin the overall system (Holt et al., 2003) and increases the force byincreasing the ground reaction force (GRF) (Kinoshita, 1985;Tilbury-Davis and Hooper, 1999) in the vertical and anterior–posterior direction. Previous backpack studies have shown thatincreases in muscle-mediated stiffness maintains a constantvertical excursion of the centre of mass (COM) under loadedconditions and this stiffness thereby limits increases in metaboliccost that would otherwise have occurred if the COM would travelthrough a greater vertical range of motion (Holt et al., 2003).However, these stiffer systems facilitate the transmission of forcesbetween segments through elastic energy storage between thedistal and proximal segment (LaFiandra et al., 2002). This stiffnessis influenced by co-contraction of the antagonist pair of trunkrotators (the internal and external obliques abdominus). Thehigher GRF and the increased stiffness due to backpacks mayresult in a higher transmission of impulsive shock from the

ll rights reserved.

_teik_hin@rp.sg (M. Koh).

inferior to the superior segments during the loading stage of thestance phase of the gait cycle (0–15% of the gait cycle approx.) andthis could possibly result in musculoskeletal injuries. Unless somecompensatory locomotion mechanism is adopted by the centralnervous system (CNS) to regulate shock transmission to thesuperior segments, there is a possibility that the higher GRF willtravel in the form of shock from the inferior to the superiorsegments. An inability of the CNS to modulate the increased GRFand increased stiffness during load carriage could potentially leadto musculoskeletal injuries.

At preferred walking speeds and frequencies much of the forceat heel-strike is discharged as the shock wave passes through thelower extremity joints (Ratcliffe and Holt, 1997). A study on activeadults has shown that load carriage does not induce significantincrement in the transmission of shock from the ankle to the headduring the loading stage of the stance phase (Holt et al., 2005).However, backpack studies have shown that backpack loadsinduce a higher change in gait of children compared to adults(Pascoe et al., 1997; Charteris, 1998; Hong and Brueggemann,2000; Wang et al., 2001; Hong and Cheung, 2003).

The transverse plane moment of inertia (TPMI) of the trunkcould be different when backpack loads are placed on the thoraxor lumbar region. Since the inertial characteristics of thorax andlumbar region are different because of difference in segmentcompositions, it is likely that the location of the backpack on thelumbar and thorax rotations could affect a different amount of

ARTICLE IN PRESS

T. Singh, M. Koh / Journal of Biomechanics 42 (2009) 736–742 737

trunk rotation in the transverse plane and thereby transmitdifferent amounts of impulsive shock from the ankle to the headduring the loading stage of the stance phase. The upper loadconfiguration could specifically increase the TPMI of the thoracicregion and thereby reduce the mean relative phase between thethorax and lumbar region by interlocking the thorax–lumbarsystem. On the other hand when load is placed in the lowerconfiguration, it could increase the TPMI of the lumbar region andrender the lumbar–pelvic system stiffer. The purpose of this studywas to investigate shock absorption capabilities amongst childrenwhen they carry backpack loads and to measure the differences inshock transmission due to location of the load on the back.

2. Methods

2.1. Participants

Informed consent and participation assent was obtained from the parents and

participants as mandated by the Research and Graduate Review Committee of the

Ank

Kn

Pelv

Trun

Hea

-18-13

-8-327

12

-6-4-20246

-3-2-101234

-3-2-101234

-3-2-101234

1 41 81

1 41 81

1 41 81

1 41 81

1 41 81

Fig. 1. Vertical acceleration time series immediately following heel-strike showing peak

Academic Group of Physical Education and Sports Science (PESS), Nanyang

Technological University, Singapore. All experimental procedures were done

indoors at the Sports Biomechanics Laboratory. Seventeen primary school boys

aged 8–12 years completed the study. The data of two participants were not

considered because of technical problems. The mean age, height and weight of the

remaining 15 participants were 10.01 (71.31) years, 136.40 (710.08) cm and 31.83

(77.13) kg, respectively.

2.2. Instruments

A belt-driven instrumented H-P Cosmoss Gaitway treadmill with Kistler force

plates was used for the study. The belt surface of the treadmill was 1500 mm long,

500 mm wide and 170 mm above the floor. Data was collected at 100 Hz. The

incline of the treadmill was set to 01. A synchronized Hawk digital six-camera,

Motion Analysis Corporation (MAC), Santa Rosa, California, USA, optical motion

analysis system was used to capture motion. The two systems were synchronized

with an external trigger. The body segments were marked according to a modified

Helen Hayes marker configuration. Two markers on the left and right posterior

superior iliac spine (PSIS) were used instead of the marker at L5-S1. This was done

for better visibility of the markers upon the introduction of backpacks on the

subjects’ back.

Kinematic data was collected at 60 Hz at a shutter speed of 0.001 s. The

collected data was interpolated with the cubic spline technique and smoothing

le

ee

is

k

d

121 161 201

121 161 201

121 161 201

121 161 201

121 161 201

accelerations (m s�2) is the temporal occurrence of the peak vertical acceleration.

ARTICLE IN PRESS

T. Singh, M. Koh / Journal of Biomechanics 42 (2009) 736–742738

was done with a Butterworth low pass band filter at 6 Hz. EVaRTs, version 4.6,

software was used for motion capture. Unframed AirShadow backpacks manu-

factured by AirPacks Systems with dimensions 29.2�45.7�14.0 cm3 made of

600D polyester were used for the study.

2.3. Backpack load configuration

A lower and upper region was defined on the back as the region inferior and

superior to the eighth and ninth thorax (T8–T9) vertebrae, respectively. The T8–T9

vertebrae were chosen as the midpoint because they are approximately at the

geometrical midpoint of the back. The regions were divided on the back using

erasable ink by drawing a medial–lateral line segment. Three backpack loads of

10%, 15% and 20% of body weights (BW) were used for the study. The centre of the

backpack was marked by measuring the end-to-end length of the cavity in which

the load was placed. The midpoint of this cavity was marked and considered as the

midpoint of the bag. The upper load configuration had load placed in the upper

half of the backpack and foam was placed in the bottom. The process was reversed

for the lower load configuration. For each trial, after the backpack was put on, the

length of the shoulder straps were adjusted so that the load was located in the

desired location. The adjustment was made so that the centre of the backpack was

aligned either superior or inferior to the medial–lateral line segment drawn at the

T8–T9 vertebrae for the upper and lower load conditions, respectively.

2.4. Experimental procedure

Sufficient walking time was given to participants for familiarization on

treadmill. The walking protocol involved walking at self-selected speeds under

different load conditions. The first trial for all participants was the unloaded

condition, henceforth, referred to as B0. The participants walked bare feet to

minimize the effects due to shoe rigidity (Forner et al., 1995) on the treadmill for

6 min. at the end of which kinematic and spatiotemporal parameters were

collected during the last 10 s (Matsas et al., 2000). The order of load conditions was

randomized. Sufficient rest was provided between two trials to minimize fatigue.

Participants carried the three different loads of 10%, 15% and 20% of their

bodyweight in the upper (U) and lower (L) load configurations. The six loaded

conditions will be referred to as L10, L15, L20, U10, U15 and U20. For example, L10

refers to the lower load configuration with 10% BW load while U10 refers to

the upper load configuration with 10% BW load and so on. The values for the

spatiotemporal parameters were averaged over 10 s of collected data and the

means were reported. One stride cycle was chosen at random from 10 s of collected

data to obtain the kinematic and kinetic parameters. The right leg was considered

the ipsilateral leg and gait symmetry was assumed as all participants were healthy

with normal gait. The normalized first peak force (NFPF) was normalized to the

body weight of the participant and backpack. The time to the NFPF was measured

from the instance of heel-strike to the maximum first peak force. Heel-strike was

marked as the point where the vertical GRF exceeded the threshold value of 5 N.

Pelvic and trunk rotation in the transverse plane was measured in global

coordinates. Phase difference between the pelvis and trunk was computed as

described by LaFiandra et al. (2003).

2.5. Shock transmission

To calculate the shock transmission ratio (TR), the time series of vertical

acceleration (ai) of the joints including ankle, knee, pelvis, trunk and head were

0.8

0.9

1

1.1

1.2

1.3

Load Condition

Velo

city

(m/s

ec) *

*

B0 10% 15% 20%

Fig. 2. The means of walking velocity and cadence for different load conditions. * indicat

one SD. The lower load configuration induced a lower gait velocity and cadence compa

calculated using the following equation (Ratcliffe and Holt, 1997):

ai ¼ ðZiþ1 � 2Zi þ Zi�1Þ=Dt2, (1)

where Zi is the vertical displacement of the virtual marker on the body segment at

the timeframe i. Dt was 0.0167 s. i is an instant in time during the gait cycle and

varies from 1 to 100.

The ankle and knee joint centres were calculated as the midpoint of the medial

and lateral ankle and knee markers, respectively. The virtual pelvis centre was

defined as the circumcentre of the triangle formed by the two markers on anterior

superior iliac spine (ASIS) and the virtual marker at the L5–S1 joint. The virtual

trunk centre was defined as the circumcentre of the triangle formed by the two

markers on the acromion process and the circumcentre of the pelvic plane. The

virtual head centre was defined as the circumcentre of the three markers placed on

the head. Five peak accelerations corresponding to markers on the five virtual

joints and centres (ankle, knee, pelvis, trunk and head) were measured for one

walking cycle for all seven conditions for all participants (Fig. 1). The acceleration

due to heel contact travels sequentially from the lower body joints to upper body

joints (Ratcliffe and Holt, 1997). The temporal sequential order of the occurrence of

the acceleration peaks of the five segments were recorded following heel-strike.

For each trial, the first peak acceleration following heel-strike of the most inferior

segment was recorded. The next temporal occurrence of the peak acceleration of a

more superior segment was recorded. Only positive peak acceleration values

were considered. The shock TR between the ankle and the head for each trial

were computed by cumulatively multiplying the shock transmission values

from the ankle to head. A MATLAB program was written to determine the

temporal sequential peak acceleration values for the different segments (Ratcliffe

and Holt, 1997).

The shock TR was calculated using the formula:

TR ¼ Umax=Lmax, (2)

where Lmax is the positive peak acceleration of the inferior joint and Umax is the

next temporal positive peak acceleration of the more superior joint.

2.6. Statistics

Descriptive statistics were reported as mean7SD. Alpha (a) level was set at

0.05. Repeated measures ANOVA analyses (with Bonferroni confidence interval

adjustment) were used and conducted for each of the dependent parameters.

Tukey’s Post-Hoc test was used whenever significant differences were found.

3. Results

3.1. Spatiotemporal and kinetic parameters

Walking velocity decreased for all loaded conditions ascompared to the B0 condition and decreased significantly forthe L20 and U20 condition (Fig. 2). The lower load configurationinduced a higher reduction in velocity compared to the upperconfiguration. No significant differences were found for thewalking cadence though it decreased consistently for the loadedconditions (Fig. 2). The lower load configuration induced a higher

110

120

130

Load Condition

Cad

ence

(Num

ber o

f Ste

ps/M

in)

B0 10% 15% 20%

es a significant difference (po0.05) with the B0 condition. The vertical bars indicate

red to the upper configuration. B0 is the unloaded condition.

ARTICLE IN PRESS

T. Singh, M. Koh / Journal of Biomechanics 42 (2009) 736–742 739

reduction in cadence compared to the upper load. No significantchanges were found for the NFPF (Table 1) although the differencebetween the L20 and B0 condition approached significance(p ¼ 0.06). The time to the NFPF was significantly greater for theL20 and U20 conditions compared to the B0 and U10 condition(Table 1).

4.0

3.2. Kinematic parameters

The range of pelvic and trunk rotation reduced significantly(po0.05) for almost all the loaded conditions (Fig. 3) compared tothe B0 condition (except trunk rotation between U10 and B0). Nosignificant differences were found between any of the loadedconditions. The pelvis and trunk transverse plane rotation wereout-of-phase at heel-strike and during the early part of theloading stage of the stance phase for B0 and 10% conditions andalmost in-phase for the 15% and 20% load conditions (Fig. 4).

No significant differences were found between the loaded andB0 condition for the knee flexion angle. However, the knee flexionangle during the loading stage of the stance phase of the gait cyclewas higher for the loaded conditions compared to the B0condition (Fig. 5). The approximate average difference betweenthe loaded and B0 condition was 31.

0

4

8

12

16

Load Condition

Pelv

is (D

egre

es)

* * *

* **

B0 10% 15% 20%

Fig. 3. The means of range of pelvic rotation and trunk rotation in the transverse plane.

indicate one SD. B0 is the unloaded condition.

Table 1The means of the normalized first peak force (NFPF) and the time to the

normalized first peak force for the different load conditions.

Load conditions Normalized first peak force Time to first peak force (s)

B0 1.17� (70.06) 0.13fg (70.02)

L10 1.15 (70.06) 0.15 (70.03)

U10 1.15 (70.08) 0.14fg (70.02)

L15 1.14 (70.09) 0.15 (70.03)

U15 1.14 (70.08) 0.14 (70.02)

L20 1.10� (70.06) 0.16ac (70.02)

U20 1.13 (70.07) 0.16ac (70.02)

The values in brackets indicate one SD. B0 is the unloaded condition. L10 and U10

are the 10% loads in the lower and upper configuration, respectively, and so on.

Note: All times were calculated from the instant the ipsilateral leg made contact

with the treadmill walking surface.a po0.05 vs. B0.b po0.05 vs. L10.c po0.05 vs. U10.d po0.05 vs. L15.e po0.05 vs. U15.f po0.05 vs. L20.g po0.05 vs. U20.� p ¼ 0.06.

3.3. Peak vertical acceleration following heel-strike

The means and the standard deviations of the peak accelera-tion of the ankle, knee, pelvis, trunk and head for the differentloaded and B0 conditions are shown in Table 2. There were nosignificant differences between any of the loaded conditions andB0 condition. There were also no significant differences betweenany two loaded conditions. The acceleration values show asubstantial decrease from the ankle to the knee joint and a slightdecrease from the knee joint to the pelvis. The acceleration valueswere fairly similar between the pelvis, trunk and head.

4. Discussion

Our results showed a significant reduction in the transverseplane range of pelvic and trunk rotation for the load conditionscompared to the B0 condition (Fig. 3). No significant changes inthe range of knee and ankle rotation were observed indicating thatthe lower limb stiffness was not altered for loaded walking.

0

4

8

12

16

Load Condition

Trun

k (D

egre

es)

** *

**

B0 10% 15% 20%

* indicates a significant difference (po0.05) with the B0 condition. The vertical bars

-4.0

-2.0

0.0

2.0

6141211

Rad

ians

Percent Gait CycleB0 L10 U10 L15 U15 L20 U20

Fig. 4. Mean relative phase difference between the pelvis and trunk for the stance

phase of the ipsilateral leg. For 15% and 20% load conditions, the pelvis and trunk

are almost in-phase at heel-strike. During the loading stage of the stance phase of

the gait cycle (0–15% of the gait cycle approx.) the 15% and U20 condition shows a

significant in-phase movement. The L20 condition showed in-phase movement at

the starting of the gait cycle, but showed a steep increase during the loading stage.

B0 is the unloaded condition. L10 and U10 are the 10% loads in the lower and upper

configuration, respectively, and so on.

ARTICLE IN PRESS

0

20

40

60

Percent Gait Cycle

Kne

e Fl

exio

n A

ngle

B0 L10 U10 L15 U15 L20 U20

0 2040 60 80 100

Fig. 5. Mean knee flexion angle over one gait cycle for the different load conditions. The gait cycle was normalized to 100 data points and the mean for the 15 subjects was

computed at each data point. The highlighted section shows the higher knee flexion angle for the loaded conditions during the loading stage of the stance phase. During this

stage, the approximate average difference between the loaded and B0 condition was 31. B0 is the unloaded condition. L10 and U10 are the 10% loads in the lower and upper

configuration, respectively, and so on.

Table 2The means of the peak vertical acceleration (following heel-strike) of the ankle, knee, pelvis, trunk and head.

Ankle (m/s2) Knee (m/s2) Pelvis (m/s2) Trunk (m/s2) Head (m/s2)

B0 10.32 (72.32) 4.32 (72.57) 3.08 (70.92) 3.26 (70.93) 2.63 (70.72)

L10 11.03 (72.21) 5.21 (71.99) 2.91 (71.03) 3.41 (71.26) 3.16 (71.12)

U10 9.92 (71.91) 5.44 (72.15) 3.12 (70.56) 3.49 (71.17) 3.49 (71.17)

L15 10.68 (71.52) 3.61 (71.49) 3.08 (70.83) 3.41 (71.00) 2.96 (70.83)

U15 11.11 (71.66) 4.54 (71.45) 3.17 (70.93) 3.55 (70.62) 3.18 (70.64)

L20 10.75 (71.93) 3.89 (72.25) 2.98 (71.03) 3.27 (71.51) 2.64 (71.48)

U20 10.54 (73.61) 3.68 (72.42) 2.67 (70.85) 3.04 (70.86) 2.73 (70.84)

The values in brackets indicate one SD. B0 is the unloaded condition. L10 and U10 are the 10% loads in the lower and upper configuration, respectively, and so on.

T. Singh, M. Koh / Journal of Biomechanics 42 (2009) 736–742740

Therefore, these data have not been reported. The reduction inpelvic rotation is consistent with previous research (Smith et al.,2006). At heel-strike and during the early loading stage of thestance phase (0–5% of the gait cycle) the pelvis and trunktransverse plane rotation were out-of-phase for B0 and 10%conditions and were in-phase for the 15% and 20% load conditionsfor most of the stance phase (Fig. 4). This shows that load carriageabove 15% BW induced a high amount of stiffness in thepelvic–trunk system. This is consistent with another study byLaFiandra et al., (2003). They showed that backpack loadplacement on thorax reduces the mean relative phase betweenpelvic and thoracic rotations. Since the inertial characteristics ofthorax and lumbar region are different, we had hypothesized thatthe location of the backpack on the lumbar and thorax rotationscould affect a different amount of transverse plane rotation of thetrunk and pelvis. However, no significant differences were foundbetween the two load locations. It is possible that the TPMI of thebackpack and trunk system did not change sufficiently enoughwith respect to the location of the load on the axis (in this case thetrunk) because the inertial characteristic of the trunk did notchange enough along the length of the axis to be affected by loadsas high as 20% BW.

The increase in stiffness is a locomotion strategy adopted byCNS during backpack load carriage that minimizes torqueproduction in the upper body (LaFiandra et al., 2002). If thisstrategy was not employed, unwarranted increases in torquegeneration about the upper body would require large muscularforces to control the resulting increase in angular momentum, at apotentially higher metabolic cost and this may contribute to theincrease in low back injuries reported in the literature (Knapik etal., 1996). The reduction in transverse rotation of the pelvis andtrunk could also be a mechanism for head stability (Stokes et al.,1989; Ratcliffe and Holt, 1997). Since addition of backpacks wouldcause greater inertia and lesser locomotive control on thecombined COM of the backpack and body system, it is possiblethat these mechanisms facilitate a stable forward propagation ofthe head and COM trajectory. Therefore, reduction in the upperbody rotation creates two opposing effects of (a) reducing theupper body torque to minimize injury occurrences and (b)increase the global body stiffness that could allow a higherpropagation of shock from the inferior to the superior segments.

The means of vertical acceleration (Table 2) and the shock TR(Table 3) indicate that majority of the shock absorption took placeat the lower extremity between the ankle–knee and the

ARTICLE IN PRESS

Table 3The means of the shock transmission ratio (TR) between the ankle-knee, knee-pelvis, pelvis-trunk and trunk-head, for the different load conditions.

Load conditions Ankle–knee Knee–pelvis Pelvis–trunk Trunk–head Ankle–head

B0 0.44 (70.28) 0.82 (70.3) 1.03 (70.21) 0.82 (70.11) 0.27 (70.10)

L10 0.44 (70.12) 0.65 (70.29) 1.10 (70.27) 0.94 (70.15) 0.30 (70.11)

U10 0.54 (70.29) 0.72 (70.3) 1.12 (70.26) 0.94 (70.18) 0.36 (70.14)

L15 0.41 (70.11) 0.8 (70.26) 1.0 (70.28) 0.8 (70.2) 0.26 (70.10)

U15 0.41 (70.14) 0.81 (70.37) 1.13 (70.34) 0.90 (70.13) 0.28 (70.07)

L20 0.41 (70.20) 0.76 (70.33) 1.09 (70.27) 0.78 (70.15) 0.21 (70.10)

U20 0.37 (70.13) 0.68 (70.29) 1.16 (70.26) 0.94 (70.22) 0.25 (70.09)

The values in brackets indicate one SD. B0 is the unloaded condition. L10 and U10 are the 10% loads in the lower and upper configuration, respectively, and so on.

T. Singh, M. Koh / Journal of Biomechanics 42 (2009) 736–742 741

knee–pelvis couple with a significant contribution by theankle–knee couple. The overall shock transmission from the ankleto the head was approximately 25–36% which is consistent withother research (Holt et al., 2005). However, there were nosignificant differences between the loaded conditions and B0(Table 3) indicating that there are locomotion mechanisms inplace even for children to regulate inferior to superior shocktransmission. Our study showed that the lower limb dynamicschanged for loaded walking to compensate for stiffer superiorsegments. There was an effort to land more softly on the walkingsurface as was evident by a lowering of the NFPF for the loadedconditions (Table 1). The difference between the L20 and B0condition approached significance (p ¼ 0.06). The lowering of theNFPF could be caused by a higher knee flexion for loadedconditions during the loading stage (Fig. 5). The approximateaverage difference during this stage between the loaded and B0condition was 31. Knee flexion during the loading stage of thestance phase assists in lowering the vertical GRF as the knee flexormuscles absorb part of the forces (Kinoshita, 1985) and this couldhave also assisted in lowering the shock transmission at the lowerlimb. The time to the NFPF was also significantly longer for the20% load conditions compared to the B0 and U10 condition. Lowerwalking speeds result in lower GRF (Chiu and Wang, 2007). Areduction in walking speed (Fig. 2) besides serving the purpose ofincreasing gait stability (England and Granata, 2007; Dingwell andMarin, 2006) could have also assisted in lowering the NFPF. Thereduction in impulsive shock due to a lowering of NFPF andvelocity and increase in time to NFPF (Table 1) would result in ageneration and propagation of a lesser amount of shock from thelower limb towards the superior segments.

However, after heel-strike, the L20 condition showed a quickertransition from in-phase to out-of-phase movement between thepelvis and trunk compared to the U20 condition (Fig. 4). No suchdifferences were observed for the 10% and 15% load conditions.This indicates that despite a high amount of induced upper bodystiffness through the stance phase, the L20 condition entailed aquicker transition to out-of-phase movement between the pelvisand trunk during the loading stage of the stance phase to absorbthe shock at the lower trunk. This was done along with theadoption of compensatory mechanisms at the lower limbindicating that there is a possibility that for the L20 configuration,lower limb compensations may not be enough to attenuate shock.

Between the two configurations, the upper configurationshowed slightly higher shock transmission compared to the lowerconfiguration and the shock transmission decreased as the loadswere increased (Table 3). However, these differences were notsignificant. A slightly greater shock transmission for the upper loadconfiguration indicates that there is a possibility that an inter-locking of the thorax–lumbar system and a freer lumbar–pelvicsystem transmits higher shock towards the head than a systemwhere lumbar–pelvic system is interlocked and the thorax–lumbarsystem is freer to rotate in the transverse plane. A stiff

thorax–lumbar system may cause a higher TPMI of the trunk thana stiff lumbar–pelvic system and may therefore propagate a higheramount of shock. Alternatively, for a freer thorax–lumbar systemand for walking speeds above 1 m/s, a possible out-of-phasecounter-rotation between the thorax and lumbar region couldhave reduced trunk stiffness and assisted in a higher amount ofshock absorption.

5. Conclusions

Our results show that gait changes occurred at the lower limbto probably counter an increase in stiffness at the upper extremity.This shows that like adults, there are locomotor mechanisms inplace for 10-year-old children to attenuate the transmission ofshock through the musculoskeletal system (Ratcliffe and Holt,1997). Further investigation is required to understand the steeptransition from in-phase to out-of-phase rotation between thetrunk and the pelvis for the L20 condition during the loading stageof the stance phase. No significant differences were found in theshock TR for the two load configurations, but the upperconfiguration showed slightly higher shock transmission indicat-ing differences in TPMI for the two load configurations. To furtherinvestigate this in a more controlled setting, studies with morecontrolled backpack load configurations along with placement ofmarkers at the upper and lower trunks need to be done.

Conflict of interest statement

We confirm that for this study there is no conflict of interestwith any internal or external party.

References

Charteris, J., 1998. Comparison of the effects of backpack loading and of walkingspeed on foot-floor contact patterns. Ergonomics 41 (12), 1792–1809.

Chiu, M.-C., Wang, M.-J., 2007. The effect of gait speed and gender on perceivedexertion, muscle activity, joint motion of lower extremity, ground reactionforce and heart rate during normal walking. Gait and Posture 25 (3), 385–392.

Dingwell, J.B., Marin, L.C., 2006. Kinematic variability and local dynamic stability ofupper body motions when walking at different speeds. Journal of Biomecha-nics 39 (3), 444–452.

England, S.A., Granata, K.P., 2007. The influence of gait speed on local dynamicstability of walking. Gait and Posture 25 (2), 172–178.

Forner, A., Garcia, A.C., Alcantara, E., Ramiro, J., Hoyos, J.V., Vera, P., 1995. Propertiesof shoe insert materials related to shock wave transmission during gait. FootAnkle International 16 (12), 778–786.

Holt, K.G., Wagenaar, R.C., LaFiandra, M.E., Kubo, M., Obusek, J.P., 2003. Increasedmusculoskeletal stiffness during load carriage at increasing walking speedsmaintains constant vertical excursion of the body center of mass. Journal ofBiomechanics 36 (4), 465–471.

Holt, K.G., Wagenaar, R.C., Kubo, M., LaFiandra, M.E., Obusek, J.P., 2005. Modulationof force transmission to the head while carrying a backpack load at differentwalking speeds. Journal of Biomechanics 38 (8), 1621–1628.

Hong, Y., Brueggemann, G.P., 2000. Changes in gait patterns in 10-year-old boyswith increasing loads when walking on a treadmill. Gait and Posture 11 (3),254–259.

ARTICLE IN PRESS

T. Singh, M. Koh / Journal of Biomechanics 42 (2009) 736–742742

Hong, Y., Cheung, C.K., 2003. Gait and posture responses to backpack load duringlevel walking in children. Gait and Posture 17 (1), 28–33.

Kinoshita, H., 1985. Effects of different loads and carrying systems on selectedbiomechanical parameters describing walking gait. Ergonomics 28 (9), 1347–1362.

Knapik, J.J., Harman, E., Reynolds, K.L., 1996. Load carriage using packs: a review ofphysiological, biomechanical and medical aspects. Applied Ergonomics 27 (3),207–216.

LaFiandra, M., Holt, K.G., Wagenaar, R.C., Obusek, J.P., 2002. Transverse planekinetics during treadmill walking with and without load. Clinical Biomecha-nics 17 (2), 116–122.

LaFiandra, M., Wagenaar, R.C., Holt, K.G., Obusek, J.P., 2003. How do load carriageand walking speed influence trunk coordination and stride parameters?Journal of Biomechanics 36 (1), 87–95.

Matsas, A., Taylor, N., McBurney, H., 2000. Knee joint kinematics from familiarisedtreadmill walking can be generalised to overground walking in youngunimpaired subjects. Gait and Posture 11 (1), 46–53.

Ratcliffe, R.J., Holt, K.G., 1997. Low frequency shock absorption in human walking.Gait and Posture 5 (2), 93–100.

Smith, B., Ashton, K.M., Bohl, D., Clark, R.C., Metheny, J.B., Klassen, S., 2006.Influence of carrying a backpack on pelvic tilt, rotation, and obliquity in femalecollege students. Gait and Posture 23 (3), 263–267.

Stokes, V.P., Andersson, C., Forssberg, H., 1989. Rotational and translationalmovement features of the pelvis and thorax during adult human locomotion.Journal of Biomechanics 22 (1), 43–50.

Tilbury-Davis, D.C., Hooper, R.H., 1999. The kinetic and kinematic effects ofincreasing load carriage upon the lower limb. Human Movement Science 18(5), 693–700.

Wang, Y.T., Pascoe, D.D., Weimar, W., 2001. Evaluation of book backpack loadduring walking. Ergonomics 44 (9), 858–869.