Res-h Report

Neuromuscular Coordination of Squat Lifting, 11:

Individual Differences

Backgmund and Purpose. 7;bisarticle reports individual drferences in the John P &hok
coordination (ie, the relative timing of joint movements and muscle activity) of Amy G McMillan
squat lrjling identzjkd by extended analysis of data reported in the authors'
companion article in this issue. Subjects. Two post hoc groups of 6 subjects
each were identijied from the original sample of 15 subjects based on qualita-
tive drferences in knee-lumbar spine relative motion plots during load acceler-
ation. Metbods. Subjects lifted a crate containing 15% to 75%of their maxi-
mum lzj2ing capacity using a symmetrical squat-lift technique. Movement
kinematic data were obtained with videography, and the electromyographic
(EMG) activity of the vastus lateralis and erector spinae muscles was recorded
with surface EMG. Measurements of coordination derived both kinematically
and via EMG and the kinematic data were examined for group dzferences.
Resu&s. Subjects in group 2 limited lumbar spine motion during load acceler-
ation for all loads lifted, whereas those in group 1 limited lumbar spine motion
more when lifting the heaviest loads. 7;bese dzyerences were obvious both quali-
tatiuely, via knee-lumbar spine relative motion plots, and quantitatively, via
measures of the relative timing of joint motions early in the lift. Z5e efect of
load on the coordination of these joints was the same for both post hoc groups
afier initial load acceleration. Signijicant dzyerences in other kinematic mea-
suremenrs were alsofound between these groups. Conclusion and Discus-
sion Despite specijic instructions about how to lzj2 the load, individual subjects
coordinated theirjoints d w e n t l y during the initial, accelerativephase of
squat lzfling. Individual dr$erences in coordination in response to load in-
creases could be categorized into two patterns, although the data of 2 subjects
were dtficult to categorize and thus not included in these analyses. Whether
the two dominant patterns have consequencesfor stress to the joints during
lzfling remains to be determined. [ScholzJP, McMillan AG. Neuromuscular
coordination of squat liJting,II: individual dzymnces. Phys mer.
1995;75:13.3- 144.1

Key Wotds: Coordination, Electmmyography, Lijling, Mataials handling, Motlement

kinematics, Spine.

An implicit assumption in much of the

literature on manual lifting appears to
JP Scholz, PhD, PT, is Associate Professor, Physical Therapy Department and Interdisciplinary be that the lifting "technique"used can
Neuroscience Program, University of Delaware. Newark, DE 19716 (USA). Address all correspon-
dence to Dr Scholz. be defined in terms of the body's
starting posture alone. Despite the fact
AG McMillan, PT, is a doctoral candidate in the Interdisciplinary Neuroscience Program, School of
Life and Health Sciences, University of Delaware.
that successful completion of this task
requires the coordination of a large
This study was approved by the Human Subjects Committee of the University of Delaware Re- number of movement components
search Office.
and the solution of a number of
This work was supported by a research grant to Dr Scholz from the Foundation for Physical Ther- potentially conflicting requirements
apy Inc. (eg, maintaining balance, producing a
7hb attick was submitted October 5, 1993, and was accepted September 8, 1994. precise load trajectory, minimizing

Physical Therapy/Volume 75, Number 2 /February 1995

musculoskeletal stress),l few studies
have directly addressed the coordina-
tion (ie, relative timing of joint move-
ments and muscle activity) of this task.
One may ask then whether lifts per-
formed from the same initial posture
can be assumed to evolve to the final
posture in the same fashion under
different task conditions? It has been
suggested that our limited understand-
ing of the coordination of lifting signif-
icantly limits our understanding of the
mechanisms of back inj~ry.~23
In the earliest known report that ad-
dressed the coordination of manual
I h g , Davis and colleagues4showed
that subjects who were instructed to
llft using :3 squat lift tended to extend
their legs faster than their back when
the load to be lifted was made
heavier. Measures of coordination
were not used, however, and all sub-
jects lifted the same loads despite
possible (runreported) dlfferences in
individual strength or lifting capacity.
Only one article3 could be found in
the 20 years following this report that
addressed the coordination of t h ~ s
task, albeit indirectly.
Several recent studies have considered
the coordination of lifting, although
many of these studies were limited in
scope and used somewhat indirect
methods for assessing coordination.58
Schipplein et a19 recently provided
indirect evidence that subjects convert
their lifting technique from one s i d a r
to a squat lift to more of a back lift as
the load to be lifted is made heavier.
Interpretation of their results is compli-
cated, however, by the fact that sub-
jects were allowed to lift using a tech-
nique of their choice (ie, freestyle),
and information was not provided on
the initial posture from which the lift
was initiated.
Thus, indications have appeared in the
literature that the coordination of lift-
ing is affected by manipulating task
variables such as the amount of load
being lifted. Our own experiments
were designed to explore the coordi-
nation of manual I h g more system-
atically than was the case in previous
work (see our companion article, part
I, in this issue).lJOSubjects were asked
Physical Therapy /Volume 75, Number 2 /February 1995
to llft loads that were scaled to their
estimated maximum l i n g capacity
(MLC). Specific measures of coordina-
tion (eg, relative phasing between
joint movements and the relative tim-
ing between electromyogmphic [EMG]
events) were used. Prelimnary evi-
dence was provided that the coordina-
tion of this task can be adequately
described and studied by measuring a
small number of the many possible
coordination variables.1°
To date, our studies have concentrated
on the effect of changes in load mag-
nitude on the coordination of a spe-
cific lifting technique (ie, squat l h g ) .
This "technique" is generally the one
most recommended by clinicians for
minimizing stress to the back, al-
though controversy exists about
whether this is always the case.596
Despite the provision of specfic and
consistent instructions to subjects, our
results show that the coordination of
squat lifting changes sigmficantly in
the face of changes in the load to be
lifted (see our other article in this
i~sue).~JO The greatest change in coor-
dination has been found to occur
between motion of the knee and
motion of the lumbar spine. In an-
other article in this issue, we report
the extension of our earlier findings
on college students to industrial work-
ers. Changes in the timing of EMG
events of the vastus lateralis muscle
(VL) and the erector spinae muscle
(ES) were shown to be generally con-
sistent with changes in knee-lumbar
spine coordination defined kinemati-
cally, using relative phase (&) vari-
ables. Moreover, the body's coordina-
tion during the lowering phase of the
task was shown to be relatively unaf-
fected by changes in the load.
Despite the overall consistency of our
results across subjects, we noted indi-
vidual differences in how subjects
performed certain aspects of the 1A.
Specifically, a qualitative analysis of
knee-lumbar spine coordination using
angle-angle or relative motion plots
(see Fig. 4 in our other article in this
issue) revealed individual differences
in the effect of load on the coordina-
tion of the initial, accelerative phase of
the lift. Moreover, these differences
appeared to be limited to a few pat-
terns used to accomplish this early
part of the Ilft. The purpose of the
analyses presented in this article was
to further explore the nature of these
apparent individual differences. We
were interested in knowing whether
individual differences observed quali-
tatively could also be identified quanti-
tatively and whether subjects who
differed in their pattern of knee-
lumbar spine coordination during load
acceleration also differed on other
measures related to task performance.
Although individual differences are a
common feature of all research, we
believe that exploring these dlfferences
formally is important in clinically rele-
vant studies such as this. In therapy,
we do not evaluate or treat the mean
patient, but the individual. Thus, sensi-
tivity to individual differences in pre-
sentation of symptoms and response
to treatment is a hallmark of effective
therapeutic intervention. The fact that
most of our subjects exhibited patterns
of coordination that could be catego-
rized into two groups may also have
future significance for diagnostic clas-
sification of patients with back pain.
Methodological details about the ex-
periments are presented in our com-
panion article in this issue. Only the
most pertinent information is pre-
sented here.
Fifteen male subjects, ranging in age
from 26 to 52 years (%?SDI=
35.127.61, were recruited as the sam-
ple of convenience for the study.
Physical characteristics of the sample
are presented in the part I article.
Techmcal diaculties resulted in some
lost data for 1 subject. Based on quali-
tative analyses of knee-lumbar spine
relative motion plots, an attempt was
made to categorize the coordination
patterns of the 14 remaining subjects.
Two dominant patterns of coordina-
tion during load acceleration were
apparent (see the "Results" section).
The data of 2 subjects could not be
easily categorized into either pattern.
Therefore, the analyses presented in
this article are for the remaining 12
subjects, whose patterns could be
ready categorized.
Reflective markers were attached to
major landmarks of the right extremi-
ties and to the trunk using adhesive
Velcro@.*The markers were placed at
the base of the fifth metatarsal, lateral
rnalleolus, lateral femoral condyle,
lateral humeral condyle, greater tro-
chanter, posterior superior iliac spine,
and immediately inferior to the tip of
the acrornial process laterally. A
marker was positioned on the crate to
be aligned with the styloid process of
the ulna when the subject grasped the
crate's handles. For the purpose of
these experiments, the hand was con-
sidered to be rigidly connected with
the forearm. Markers mounted on
balsa wood fins were placed at the
midline of the spinous process of the
1st sacral 6 - 0 , 3rd lumbar (L-3), 12th
thoracic (T-12), and 7th cervical (C-7)
vertebrae. The two-dimensional mo-
tion of these markers during each task
repetition was videotaped and digi-
tized using a Peak Performance Tech-
nologies Inc (PPTI) motion analysis
~ystem.~
Electromyographic records were taken
from the left ES and from the left VL
using the Therapeutics Unlimited Inc
Model-544 EMG system.* Preamplifier
electrode assemblies were used to
detect the EMG signal at the skin
surface. The EMG signals were ampli-
fied (10K gain factor), high-pass fil-
tered (75-Hz cutom, and digitized
on-line at a rate of 360 Hz.
Prior to each experiment, each sub-
ject's MLC was determined by having
him perform a maximumeffort lifting
motion against a load cell from a
squatting position, with the knees
flexed and the back relatively straight.
- - - - - -
'Velcro USA Inc, 406 Brown Ave, Manchester, NH 03108.
+peak Performance Technologies Inc, 7388 Revere Pkwy, Suite 601, Englewood, CO 80112-9765.
%erapeutics Unlimited Inc, 2835 Fnendship St, Iowa City, IA 52245.
60 / 135
Subjects were instructed to l~ftthe
weighted crate to the height of their
waist from a starting posture with the
knees bent, back relatively straight
(squat lift), and feet positioned sym-
metrically. Subjects were carefully
monitored to ensure that they began
each trial in the generally defined
position and were reminded verbally.
The exact amount of initial knee flex-
ion and back extension at lift onset,
however, was not controlled because
of a desire to maintain some degree of
external validity. Each subject's pre-
ferred foot position with respect to the
fixed location of the crate was deter-
mined from a test trial at 60% MLC.
This position was then marked on the
floor and maintained as the starting
position for all experimental trials.
Data Reduction
The motion of the reflective markers
during each task repetition was digi-
tized, calibrated, and filtered (3 Hz),
yielding two-dimensional data for 12
lifts at each percentage of MLC for
each subject. A link-segment model
was used to calculate sagittal-plane (ie,
flexion-extension) angles for the an-
kle, knee, hip, lumbar spine, shoulder,
and elbow joints from the digitized,
calibrated coordinates of the reflective
markers.
Electromyographic signals were recti-
fied using software. Cumulative sum
histograms of each EMG signal then
were obtained using the procedure of
Ellaway. The histograms were then
used to determine the onset and peak
of EMG activity on each lifting-
lowering cycle.
Determination of Post Hoc
Groups
The basis for identifying individual
diierences in performance was the
- - -- -- - - --
Physical Therapy /Volume 75, Number 2 / February 1995
qualitative analysis of normalized
relative motion plots of diferent joint
pairs. The normalization procedure for
these plots is presented in detail in
part I. The amplitude of each joint was
normalized to [- 1,1]in the process of
calculating the relative phase of the
joint's motion. Because of intertrial
diferences in task period, all trials
were then normalized to 1000h using
linear interpolation. AU 12 normalized
trials of a given load condition were
then averaged, resulting in an average,
normalized movement trajectory for
each joint at each load. The averaged
knee trajectory for each load was then
plotted against that of the lumbar
spine and examined for changes
across loads.
Independent Variables
The independent variables of interest
for these analyses were the percentage
of MLC lifted and the post hoc group,
defined based on knee-lumbar spine
coordination during load acceleration
(see "Results" section).
Dependent Measures
Relative phase (a. The relative
phase of motion between the knee
and lumbar spine (&) and between
the knee and hip (&) were calcu-
lated as continuous measures of coor-
dination.' Technical difficulties limited
the number of subjects for whom the
relative phase of motion between the
knee and shoulder (&.J could be
calculated to five for post hoc group 1
and three for post hoc group 2. This
measure, therefore, was not examined
for group diferences. The within-lift
mean relative phase (eg, &)across
all samples was obtained for each lift,
and the mean of this value across all
trials of a load condition per subject
was calculated for statistical analyses.
Relative timing. To help distinguish
coordination daerences between the
period of initial load acceleration and
later in the lift cycle, discrete relative
timing measurements for knee-lumbar
spine and knee-hip coordination were
calculated at four diferent points in
the lift cycle.1° For purposes of this
calculation, the knee was defined as
the reference joint. The four relative
timing measures per lift cycle are
characterized as the relative time in
the knee extension cycle (onset of
knee extension to maximum exten-
sion) during a lift at which the target
joint (lumbar spine or hip) reaches
I@?,30%, 50%, and 75% of full exten-
sion. That is, if T,, and T,, repre-
sent the times at which the knee be-
gins and c.:ompletes extending,
respectively, and T,, is the time at
which the hip joint reaches 50% of full
extension during the lift, then
where (TGtop- TKstart)is the period
of knee extension.
Electromyogaphic relative timing.
The relative time of ES and VL onsets
and peaks of EMG activity within a lift
cycle was determined from the cumu-
lative sum histograms as, for example,
(2) RT, = (T-, - TJ/T
where RTn is the relative time of VL
onset of activity in a lifting cycle, T-,
is absolute time of VL onset of EMG
activity, T, is the time of initiation of
vertical crate motion, and T is the
period of' the lifting phase of the task
(ie, the time from initiation of vertical
crate motion to completion of the
l h g phase). All other EMG timing
measurements were determined
similarly.
Other measures. A number of addi-
tional variables were compared be-
tween the post hoc groups to deter-
mine whether physical characteristics
and other kinematic factors related to
task pedbrmance also differed for
these groups. These variables included
the following:
1. Physical characteristics. Subject
height and weight.
2. Initialposture. The angle of the
knee and the angle of the lumbar
spine at the onset of the lift were used
"YSTAT Inc, 1800 Sherman Ave, Evanston, IL 60201.
Physical Therapy / Volume 75, Number 2 / February 1995
to index the body's posture at lift
onset.
3. Task period. The period (in milli-
seconds) of the lifting phase of the
task was assessed for each task cycle.
4. Location of crate and body center
of muss (COM) at lif2off The location
of the crate and the body's COM with
respect to the head of the fifth meta-
tarsal (5MTH) was determined for
each trial (1) immediately before liftoff
and (2) at 15% of the period of vertical
crate movement after liftoff. At the
latter point in the lift cycle, the crate
had achieved, on average, 5 cm of
vertical excursion, or less than 8%of
full excursion. The relative crate loca-
tion was with respect to the ctate
marker, which was centered on the
upper edge of the side of the crate.
The starting crate location with respect
to the feet was self-selected and fixed
for each subject for all lifts. All subjects
chose a crate location that placed the
near edge of the crate slightly between
the distal phalanx of the great toes,
with the feet angled outward slightly.
Thus, subjects had no room to slide
the crate further backward prior to
liftoff and reduce the clockwise torque
produced by the load about the base
of support.
With a relatively fixed lever arm at
liftoff,this clockwise torque was deter-
mined largely by the magnitude of the
load. As load became heavier, there-
fore, the challenge posed to a subject's
balance necessarily increased. One
way to meet this challenge, and pre-
sumably reduce the required muscular
effort, would be to move the body's
COM further backward with respect to
the base of support prior to lifting
heavier loads. To determine whether
this was done and the extent to which
this "strategy" d8ered among subjects,
the body's COM location with respect
to the 5MlR was analyzed. The loca-
tion of the crate with respect to the
5MTH at liftoff was also analyzed to
determine whether differences in crate
position existed between post hoc
groups.
5. Horizontal velocity of the COM.
Indirect evidence has been provided
that, when rising from a chair, subjects
convert initial horizontal momentum
into vertical momentum to assist with
rising.12If this represents a general
strategy, then one could expect it to
be used to assist with rising in the
lifting task as well. If so, greater poste-
rior horizontal momentum would be
expected prior to lifting heavier loads.
The peak horizontal velocity of the
body's COM prior to liftoff of the crate
was examined to determine whether
peak velocity differed across loads and
between the two post hoc groups.
Statistical Analyses
The effects of load, post hoc group,
and their interaction were tested by
repeated-measures analyses of vari-
ance (ANOVAs) using the SYSTAT
statistical package.s Only the lifting
phase of the task was examined in
these analyses. The test of subject
characteristics was done with a
between-groups ANOVA, with group
as the only independent variable.
Determination of Post Hoc
Groups
An examination of averaged, normal-
ized relative motion plots from the
data presented in part I led to the
identification of individual dlaerences
in the way subjects coordinated knee-
lumbar spine motion during the accel-
erative portion of lifting. The patterns
exhibited by 12 of the subjects tended
to fall into two categories, whereas the
patterns exhibited by the other 2 sub-
jects could not be easily categorized
and are not examined further here.
Fortuitously, an equal number of sub-
jects fit into the two diferent patterns
exhibited. These patterns are illus-
trated in Figure 1.The relative motion
plots in Figure 1A show that as the
load was made heavy, lumbar spine
extension lagged further behind knee
extension. This was true at all stages
of the lii, including the initial, acceler-
 accelerative portion of the lift, lumbar
spine motion was severely limited
dhile the knee extended when lifting
all loads. In many trials, the spine
A actually flexed slightly during this
initial period. The percentage of the
EXT 1.0 lift period over which lumbar spine
motion was limited varied somewhat
a across subjects in this group, although
d
all subjects clearly showed the effect.
f OS
Based on this finding, we sought to
2
.d
determine whether there were quanti-
,P 0.0 tative indicators of the differences
illustrated in Figure 1. We then asked
s
& whether these differences in knee-
lumbar spine coordination were re-
6 -03 lated to other differences in task per-
3 formance between these groups.

Relative Phase (+)

-1.0
-1.0 -0.5 0.0 0.5 1.0
FLEX Mean & decreased (increased phase
Knee Angle EXT lag of lumbar spine motion) with
B increasing load for both post h a -
groups (F4,,,=56.0, P<.001), as de-
scribed in part I. Ovelall, the post hoc
groups did not differ on this measure
(P=.920). Load and group interacted
to affect & (F4,,=2.6, P<.05). This
effect was due to slight differences in
the order of the group means across
loads. For example, at 15% MLC, &
was -21.3 and -23.9 for groups 1
and 2, respectively, whereas at 75%
MLC, & was -41.6 and -38.5 (Fig.
2).

Increasing the load sigruficantly af-

fected & (F4,40=15.5, P<.001), lead-
ing to a slight increase in the lag of
hlp motion behind knee motion, as
FLEX reported in part I. Although relative
Knee Angle EXT motion plots suggested a possible
dgerence between many subjects in
Figure 1. Awrage, normalized relatiw motion plo& of knee-lumbar spine motion the two post hoc groups at the heavi-
during the l$ingphase of the squatting task for a rpesentatiw subect of ( A l p s t ha- est loads (Fig. 31, neither the group
group 1 (CF) and ( 2 ) p s t hoc gmup 2 UR). Normalized W m u mjointJ2exion and factor (P=,397) nor its interaction with
extension equals - 1.0and 1.O,e@ectitx?Iy. (EXT= extension, FLEX=$'exion, load (P=. 158) had a significant effect
MLC= &mum lifling capacity.) on &. None of the measures of
relative phase variability (eg, standard
ative phase. This pattern is similar to hocgroup 2. Note that during the last deviation of &) differed significantly
that exhibited by a different subject half of the Ilft, the effect of load on the between the groups.
(see Fig. 4 in part I) and constitutes coordination of the knee and spine
the pattern exhibited by all subjects in was similar to that observed for sub- Relative Timing
post hoc group 1. jects in post hoc group 1. That is, lum-
bar spine motion lagged behind knee The effect of load on all four relative
Figure 1B presents a pattern of results motion to a greater extent when llfting timing measures was highly sigruficant
for one representative subject of post heavier loads. In contrast, during the (F4,40>15.0,P<.001). This finding was

62 / 137 Physical Therapy / Volume 75, Number 2 / February 1995


0-
-
-O
-10-
Group 1
-
8 the load (all P>. 19).
2 -20- T
PI
i? 30-
m
CP
t -40-
-50
15 30
Percentage of MLC
Figure 2. m e average relative phase of knee-lumbar spine motion (4& (2 1 sEM)
for subjects in post hoc groups 1 and 2 for each load condition (pwcentage of maxi-
mum lifting capacity [MLCI).
expected and consistent with the rela-
tive phase results (see part I). Of
greatest interest here was the effect of
group and its interaction with the load
on the relative timing measures at
different points in the lifting cycle.
The relative time in the cycle of knee
extension.at which the lumbar spine
achieved 10% of its motion (ie, KL,J
differed between the two post hoc
groups (F,,,,= 14.1, P<.01). The effect
is illustrated in Figure 4, which shows
that for group 2 the knee extension
cycle was, on average, more than 45%
complete by the time the lumbar spine
achieved 10% of full extension. Knee
extension led lumbar spine extension
less for subjects in group 1. The differ-
ence between the groups appeared to
decrease with increasing load, which
might be expected from the relative
motion plots (Fig. 1). That is, the
phase lag of lumbar spine motion
during load acceleration was similar
between the two groups at the heavi-
est loads. Planned comparisons be-
tween the group means for KL,, at
each load condition indicated signifi-
cant differences between the groups
Physical Therapy /Volume 75, Number 2 / February 1995
45 60 75
(all P<.05) except at 75% MLC
(P= .147).
Both load and group interacted to
affect the relative time in the knee
extension cycle where spinal exten-
sion reached 30% of maximum
(F4 ,=3.1, P<.05). Planned compari-
soh showed that the groups differed
only for the 15% MLC (group 1: 0.530,
group 2: 0.636) and 30% MLC (group
1: 0.575, group 2: 0.674) conditions.
Again, lumbar spine motion lagged
behind knee motion more at this point
in the lift cycle for group 2 subjects.
Neither the group (D.650) nor its
interaction with load ( D . 1 0 ) had a
s i m c a n t effect on relative timing
measures KL, and KL,,.
The relative time in the knee exten-
sion cycle at which the hip achieved
lo?? of full extension was unaffected
by the load (P=.221) and did not
differ between groups (P=.301). The
three measures of knee-hip timing
assessed at later points in the lift cycle
(ie, KH,, KH,, and KH,,) were also
unaffected by the group (all D.320),
but they were affected by an increase
in the load (all P<.001). In all trials,
the lag of hip extension behind knee
extension increased with increasing
load. For example, 58.5%, 59.4%,
60.8%,62.7%, and 67.3% of the knee
extension cycle had been completed
when the hip achieved 50% of full
extension (ie, KH,J, when lifting 15%
to 75% MLC, respectively. There was
no interaction between the group and
Electromyographk Relative
Timing
Other than the significant effect of
load on the relative timing of onsets
(F4,40=6.0,P<.01) and peaks of ES
activity (F4,,=23.9, P<.001), as re-
ported in the companion article in this
issue, there were no sigmiicant effects
of post hoc group (all D . 1 9 ) on any
of the EMG timing measures.
The relative timing of VL peak activity
was affected by an interaction of
group with load (F4,,=4.8, PC.01).
The mean relative times of occurrence
of VL peak activity were 36.5% and
31% of the lift cycle at 15% and 75%
MLC, respectively, for group 1. The
slight advance in VL peak activity with
increasing load in this group was not
sigmiicant. For subjects in group 2,
however, VL peak activity occurred
s i m c a n t l y later in the lift cycle when
lifting heavy loads (52.8% at 75% MLC)
compared with light loads (46.2% at
15% MLC).
Physic~lCharacteristics
Although subjects were slightly taller,
on average, in group 2 (1.803 m)
compared with group 1 (1.789 m),
these differences did not approach
sigmiicance (P= .725). Differences in
weight between the two groups
(group 1: 920.3 N, group 2: 812.1 N)
were also not significant (P=.W).
The size of the mean weight difference
was primarily due to one subject in
group 1 who weighed 1,147.6 N.
Initial Posturn
The knee angle at lift onset was aF-
feaed significantly by an interaction
between load and post hoc group
(F4,40=7.4,P<.001). As shown in
 than did those in group I (192.0" and
185.5O,respectively) at the onset of the
lift (195°=neutral)l Both groups were
slightly more flexed at the onset of
lifting the heaviest loads (Fig. 6).

Task M o d

There was no sigtllficant difference in

the period of llfting the load between
the two post hoc groups (P=.725), nor
was there an interaction of the group
with load (P=.N6). As noted in our
other article in this issue, the lift pe-
riod increased significantly with in-
creasing load after 45% MLC
(F4,,= 10.0, P<.OOl).

Location of Crate at Liftoff

There was no difference between post

hoc groups in the location of the crate
with respect to the 5MTH, either just
prior to liftoff (P=.580) or immediately
after liftoff (P=.986). The position of
the crate immediately after liftoff, how-
ever, did differ across loads
(F4,,,= 26.9, P<.001). All subjects
moved the crate closer to the 5MTH
when the load was heavier (eg, 15%
MLC=0.264 m, 75% MLC=0.228 m).
Load also interacted with group
(F4,40=8.4,P<.001), but this interac-
tion was due to a significant ddference
in relative crate locations only at 15%
MLC (F,,,,=8.5, PC.05; group
1=0.306 m, group 2=0.256 m).

Location of Body Center of Mass

at Liftoff

The COM was always slightly anterior

to the 5 m immediately prior to the
crate leaving the ground, and more so
for subjects in group 1 (F1,,,=7.2,
Figure 3. Average, normalized relative motion plots of knee-hip motion during the P<.05; group 1=0.052 m, group
lifting phase of the task for a representative subject of ( I ) post hoc group 1 (0and
) 2=0.022 m). Both groups moved the
(2) post hoc group 2 (m).Normalized maximum joint flexion and extension equals
- 1.0 and 1.0, respectively. ( m T = extension, FLEX=flexion, MLC= maximum lifting
COM hrther posterior prior to llftoff as
capacity.) the load was made heavier (F4,40=9.5,
P<.001; Fig. 7A).
Figure 5, knee flexion at lift onset was assumed a more extended knee p s i -
about 75 degrees (180°=full exten- tion (90° at 75% MLC). This same pattern of results held just
sion) for both groups when lifting 15% after Ilftoff (Fig. 7B), the body's COM
MLC. With heavier loads, particularly The initial lumbar spine angle still tending to be located anterior to
above 300h MLC, the knee was slightly changed with the load (F4,,,=11.0, the 5MTH. On average, the COM was
more flexed for subjects in group 2 at P<.001) and ddfered between the located 0.037 m and 0.007 m anterior
llft onset (70°, on average, at 75% groups (F1,,,=9.7, Pc.05). Subjects in to the 5MTH for subjects in groups 1
MLC), whereas subjects in group 1 group I1 had more extended spines and 2, respectively (F1,,,=6.8, Pc.05).
For subjects in group 2, however, the

64 / 139 Physical Therapy / Volume 75, Number 2 / February 1995

 to liftoff increased as the magnitude of
the load was increased (F,,,= 13.4,
P<.Wl) and tended to be higher for
subjects in group 1 than for those in
0.75 group 2 for all but the heaviest loads
t (Table). The apparent group differ-
ences, however, failed to reach signifi-
Q)
I Group 2 cance (P= ,075).
U
h
u 0.45 Discussion
Q)
Our companion article in this issue
presents evidence that increasing the
cw load a subject lifts can lead to changes
0
1 in the coordination of squat lifting
gb 0.15- G"'P despite specific instructions and fre-
a
C, quent reminders about how to llft. The
3 data presented in this article show that
z 0.001
0 15 30 45 60 75
1

90 individual differences in performance

a" Percentage of MLC also exist under the same llfting
conditions.

Reanalysis of the relative phase of

Figure 4. 7he average relative time in the he extension cycle during lifing at which
the lumbar spine completes 10%offull atenqion (21 SEM) for subjec.5 in post hoc g m u p 1 knee-lumbar spine motion, including
and 2for each load condition ( p e n t a g e of mazimum lging capacit3, [MLCI). the group factor, yielded a quantitative
result apparently inconsistent with the
mean COM location was posterior to Horizontal Velocity of the Center qualitative analysis of relative motion
the 5MTH at 600h to 75% MLC (Fig. of Mass plots used to categorize the groups
7B). The body's COM was closer to (Fig. 1). It must be remembered, how-
the SMTH when lifting heavier loads Peak posterior horizontal velocity (in ever, that this continuous measure of
(F4,40=50.3,P<.Wl). meters per second) of the COM prior coordination (ie, &) is the average
across all samples of a lift. The identi-
fied categories of knee-lumbar spine
coordination were based on d8er-
ences in relative motion of these joints
during the initial, accelerative phase of
the lift only. Because the effect of load
on knee-lumbar spine coordination
did not appear to dfier substantially
between the groups after this early
phase (Fig. 11, & probably obscures
these dfierences early in the llft.

Based on the patterns exhibited in

Figure 1, we expected that group
diferences on knee-lumbar spine
relative timing would be greatest at
early points in the lift, especially when
llfting light loads, and that they would
disappear at later points in the lift
cycle. The results of the relative timing
I I Group 2
I I analysis confinned this prediction.

For KL,, , knee extension led spine

I Percentage of MLC
I extension by a greater amount for
subjects in group 2 (Fig. 41, consistent
with dfierences in the relative motion
Figure 5. Average fnitiul knee angle( 2 1 SEMI for both post hoc groups plots (Fig. 1). Although the interaction
(EXT= extension, FLFX=Pexion. MLC= maximum lifiing capacity.) between group and load was not

Physical Therapy / Volume 75, Number 2 / February 1995


EXT 200-
n
e peared to be affected more by in-
Q)
Q)
a ent that knee-hip coordination is much
h load lifted than is knee-lumbar spine
sEl 185-
3 work to date (see part I).lJO
FLEX 180 ij 30
I
Percentage of MLC
Figure 6. Average initial lumbar spine angle (2 I SEM) for both post hoc groups.
Neutml spine is 195 degrees. (EXT= extension, FLEX=Jexion, MLC= maximum lifting
capacity.)
sighcant, planned comparisons
showed that the group diierence was
s i m c a n t at all loads except 75%
MLC. This finding also appears consis-
tent with the relative motion plots. At
a slightly later point in the lift cycle,
load and group interacted to affect
knee-spine timing (KL3J. Here, the lag
of lumbar spine extension behind
knee extension was greater for group
2 only at 15% to 30% MLC. Relative
timing measurements calculated later
in the lift cycle, after load acceleration
(ie, KLWand KL,,), did not differ be-
tween the two post hoc groups.
The results of these analyses provide
quantitative support for conclusions
drawn from the relative motion plots
(Fig. 1). During the accelerative phase
of lifting, the inertia of the load has to
be overcome, leading to the greatest
likelihood of joint stress." Subjects in
group 2 restricted lumbar spine mo-
tion regardless of the load magnitude,
possibly to minimize stress to the back
during this period of high stress. In
contrast, subjects in group 1 appeared
to use this "strategy" primarily when
lifting the heaviest loads.
66 / 141
Group 2
45 60 75
Schipplein et a19 reported results sirni-
lar to those exhibited by group 1, that
is, an increasing lag between knee
and lumbar spine extension as the
load became heavy. Their subjects,
however, were allowed to lift freestyle.
The authors concluded that the ob-
served change in lifting pattern was
due to the inability of knee extensors
to generate sufficient torque when
lifting the heaviest loads. Thus, the
burden was assumed to be shifted to
the back after the knee had already
extended. The data suppolting this
conclusion, however, were weak. If
this interpretation were accurate, it
would suggest that our group 2 sub-
jects had weaker knee extensors be-
cause they showed a substantial lum-
bar spine extension lag behind knee
extension at all loads during load
acceleration. This conclusion seems
unlikely. For example, our earlier
experiments on squat liftinglJOin-
cluded one subject who was a triple-
jumper. This athlete was very strong
and yet exhibited an increased phase
lag between knee and back motion
even across only moderately heavy
loads (similar to pattern exhibited by
Physical Therapy / Volume 75, Number 2 / February 1995
group 1). It is difficult to imagine that
this result was due to the inability of
the knee extensors to generate suffi-
cient force in this subject.
Group 1's knee-hip coordination ap-
creases in the load as compared with
group 2 (Fig. 3). Neither & nor the
relevant relative timing measures,
however, differed sigmficantly be-
tween the groups. It is certainly appar-
less affected by the magnitude of the
coordination (Figs. 1 and 3). This has
been a consistent finding in all of our
Although we controlled the starting
posture for the lift by specrFyig loose
boundaries (ie, providing general
instructions for starting position, such
as with knees bent, rather than spe-
cific angles of flexion), this posture
was not fixed. Significant group differ-
ences in the starting posture of the
body were associated with group
diaerences in knee-lumbar spine coor-
dination measured during the acceler-
ative phase of lifting. Although both
groups assumed, on average, the same
initial knee flexion position when
lifting light loads, this position became
more divergent when lifting heavy
loads (Fig. 5). Subjects in group 1
showed the greatest change, starting
more extended when lifting all but the
lightest load. Moreover, these subjects
had more initial lumbar spine flexion
at lift onset at all loads than did sub-
jects in group 2 (Fig. 6). Both groups
assumed a more flexed spine to start
the lift when the load became heavy.
The exact starting position has not
been reported for most studies of
manual liftiig techniques, even when
the subjects were allowed to lift free-
style.9 Thus, the findings of those
studies for which the investigators
failed to report the starting position
are especially diff~cultto interpret.
Despite these significant effects of
group and load on the initial posture
of the body at lift onset, the actual
values (Figs. 5 and 6) indicate that the
subjects in both groups began the lift
in a squatting posture. That is, the
 -
Table. Peak Horizot~talPosterior
Velocity (5 Standard Deviation) of the
Body's Center of Mass Pn'or to Crate
Liftoff

Peak Velocity (rnls)

Group 1 Percentage
of MLCe Group 1 Group 2
0.08

p p p p p

%LC =maximum lifting capacity.

cance. Moreover, at loads at which the

starting posture of the knee was most
-0.02 different between the groups (eg, 75%
0 15 30 45 60 75 90
MLC), differences in knee-lumbar
Percentage of MLC spine coordination were smallest (Fig.
1).

We could have mechanically fixed the

subjects' starting posture, as has been
done in other recent studies.14J5Al-
though such a procedure would help
eliminate the starting posture as a
possible intervening variable, it is
inconsistent with how squat llfting is
taught clinically and implemented in
the real world. In our experience,
clinicians do not necessarily speclfy
precisely the initial posture of clients
when teaching them how to lift, al-
though control of lumbar spine pos-

2
-0.05 15 30 45

Percentage of MLC
60
Group 2
75
ture has generated much recent inter-
est.'6z17 If future work shows that small
differences in lifting posture make a
meaningful daerence, then more
precise training procedures than are
currently used will be essential.

The measured differences in coordina-

Figure 7. Location of the body's center of mass (COM) with respect to theJfth tion also do not appear to be related
metatarsal head (A) immediateL$prior to crate [@off and (B) at 15% of the period of to differences in the physical character-
vertical crate motion. Positive values indicate that the COM is anterior to tbeJflh istics of the subjec'ts. There was no
metatarsal head. (MLC=maximum ltJti7g capacity.) si@cant difference in either the
height or the weight of the subjects in
knee was flexed 90 degrees or more posture directly determined the ob- the two post hoc groups. Weight dif-
when llfting all loads and the lumbar served group diferences in coordina- ferences came close to si@cance, as
spine was flexed, on average, no tion. Correlation matrices (unreported) the mean weight of group 1 was
more than 10 degrees from neutral. revealed that starting angles accounted higher than that of group 2. As noted
for no more than 15% of the variance earlier this mean difference was pri-
Although clearly related, it does not in any coordination measure, and only marily due to one subject in group 1
appear that the dltferences in starting a few correlations reached signifi- who was extremely heavy.

Physical Therapy /Volume 75, Number 2 /February 1995 142 / 67

Based on the kinematic results (Fig.
11, we expected onset and peak of ES
activity to occur later for subjects in
group 2 than for those in group 1, and
especially so when lifting the lighter
loads. The results of our reanalysis of
EMG timing data, however, were
generally inconsistent with the group
differences in knee-lumbar spine coor-
dination, defined hematically, during
the initial period of the lift. The only
significant group effect on EMG timing
was that peaks of VL activity occurred
later in the lift cycle for subjects in
group 2, especially when lifting the
heaviest loads. Moreover, there was a
tendency, although nonsigmficant, for
ES peaks of activity to occur later in
the lift cycle for group 1 compared
with group 2.
The different-than-expected timing of
peak EMG activity may be because VL
and ES peaks of activity both occur
relatively late in the lift. Therefore,
group diierences on these measures
are likely to be unrelated to coordina-
tion differences occurring early in the
lift cycle (ie, during load acceleration).
The lack of a group effect for timing of
onset of ES activity is somewhat more
puzzling. This result as well as the
smaller, albeit significant, effect of load
on EMG timing compared with kine-
matic relative timing measures (see
part I) may be due to our limited
sampling of EMG activity. For exam-
ple, ES activity was recorded at only
one spinal segment. Similarly, VL was
the only knee extensor recorded
directly.
Many additional muscles are likely to
participate in coordinating knee-
lumbar spine motion during perfor-
mance of this task. The data of Ek-
holm et a17 suggest that although
quadriceps femoris muscle activity
initiates squat lifting, hamstring muscle
activity occurs as early as during the
first 1W/o to 20% of the lift cycle. Their
results are consistent with those of a
recent analysis of a more constrained
squat lift.14Toussaint et all4 suggested
that coordination of the monoarticular
quadriceps fernoris muscle and the
biarticular hamstring muscle helps to
preseme balance during squat ldting
by keeping the vertical ground reac-
tion force vector close to the knee
axis, and that this phenomenon re-
flects a general principle of antagonis-
tic muscle coordination.
Likewise, many spinal extensors are
likely to be involved in controlling
spinal motion during lifting. The role
of the abdominal muscles also cannot
be ruled out, although recent data
suggest that these muscles are rela-
tively quiescent during load accelera-
tion.ls Moreover, understanding kine-
matically defined changes in
coordmation is complicated by contri-
butions to joint motion from reactive
forces from adjacent segments. Analy-
ses of the task's kinetics, which are
currently under way in our laboratory,
as well as future experiments that
record from multiple muscles will be
necessary to resolve this issue.
Differences between the post hoc
groups on other variables suggest that
the measured differences in knee-
lumbar spine coordination may be
related to individual differences in
meeting task requirements such as the
control of balance. Although the posi-
tion of the crate with respect to the
foot did not differ between the groups
either prior to or shortly after liftoff,
the position of the body's COM did
differ. The COM was positioned more
anterior to the foot by subjects in
group 1 compared with those in
group 2 for all loads lifted.
Surprisingly, the COM was in mast
trials still anterior to the 5MTH just
after the load left the ground. A stan-
dard method was used to calculate the
body's COM.lWalculations of the
overall COM, however, are affected by
errors in estimating the location of the
COM of individual segments. Thus, it
is possible that the COM was not
actually t h s far forward. If the COM
had been this far forward, more dy-
namic muscular action would have
been required to counteract the clock-
wise torque generated by the crate
and to maintain balance than if the
COM had been located more posteri-
orly. Because subjects in group 2
positioned their COM more posteriorly
at this point in the lift cycle and, on
average, well behind the 5MTH for the
Physical Therapy / Volume 75, Number 2 / Februarv 1995
heaviest loads, they presumably re-
quired less muscular effort to maintain
balance. If so, this "strategy" might
also reduce joint stress, although a
kinetic analysis will be required to
determine this for certain. Certainly,
the limited spine extension during
load acceleration in group 2 is consis-
tent with lower stress to the back.l3
Nonetheless, the results represent a
difference in degree, as both groups
positioned the COM more posteriorly
when lifting heavier loads (Fig. 7). It is
not clear why the group differences
did not disappear for the heaviest
loads lifted, because coordination
during load acceleration was more
similar between the groups at these
loads (Fig. 1).
Increasing the horizontal velocity of
the COM prior to liftoff would gener-
ate greater horizontal momentum. If
this momentum could be converted to
vertical momentum, as has been sug-
gested for a sit-to-stand task,l2 this
conversion could make lifting the load
easier. Horizontal velocity increased
significantly for both groups as the
load was made heavier, suggesting
that all subjects used this "strategy"
(Table). The greater mean horizontal
velocities generated by group 1 at all
MLC conditions except the 75% MLC
condition would appear to reflect a
greater use of this mechanism com-
pared with subjects in group 2. The
group difference approached but did
not reach sigruficance. Nevertheless,
the results of this analysis and the
results reported in our companion
article in this issue hint that individual
diierences in coordination strategy are
related to differences in the primacy of
different task constraints among
subjects.
More research is needed to confirm
the existence of these individual differ-
ences in performance before their
clinical significance can be fully as-
sessed. Nonetheless, our results sug-
gest that minimizing spinal motion
during the initial accelerative phase of
a lift may minimize threats to balance.
As noted, if individual differences in
how the neuromuscular system is
coordinated during squat b g have
consequences for the magnitude of
joint stress or maintaining balance
during lifting, then it will be important
to consider these subtle differences
when training workers to lift properly.
Certamly, the best way to prevent
injury to the back is to prevent work-
ers fmm lifting in situations in which
the risk of injury is increased substan-
tially, including minimizing the load
that they are required to lift and its
height with respect to the floor. Given
that workers are often required to llft
heavy loads from the floor, deterrnin-
ing and teaching the optimum coordi-
nation strategy would be beneficial.
Our results suggest that we may not
have an adequate understanding at
present of what is an optimal lifting
technique. The results presented in
this article and in the companion
article in this issue indicate that a
lifting "technique" such as squat lifting
cannot be defined in terms of the
body's starting posture alone. The
magnitude of load lifted and perhaps
other factors such as the speed of
lifting can have significant effects on
how the body is coordinated when
performing this task.
Individual differences occur in how
subjects lift a load from an initial
squatting posture despite identical and
consistent instructions. Differences in
coordination among subjects appear to
be limited to the initial, accelerative
phase of lifting, and the coordinative
patterns are limited in number. More-
over, these differences appear to be
related to how subjec- manage differ-
ent task constraints such as maintain-
Physical Therapy / Volume 75, Number 2 / February 1995
ing balance or generating the vertical
load trajectory. A full understanding of
the neuromuscular basis of the kine-
matically measured coordination dif-
ferences is complicated by the possi-
ble action of multijoint muscles and
the contribution of reactive forces
fmm adjacent motion segments. More
detailed studies of muscular contribu-
tions as well as kinetic analyses are
needed to better understand the influ-
ence of changing task variables on the
coordination of this task and the clini-
cal sigruficance of individual differ-
ences in performance. Nonetheless,
the measured coordination differences
suggest that clinicians may need to
pay more attention to subtle differ-
ences in coordination when training
clients to lift objects.
Acknowledgments
We are grateful to the Foundation for
Physical Therapy Inc for its support of
this project and to Jeffrey P Millford,
PT, for his assistance with data pro-
cessing. We give special thanks to Dr
Julie Moyer and Dr Irene McClay for
their many helpful suggestions.
References
1 Scholz JP. Organizational principles for the
coordination of lifting. Human Movemat Sci-
a c e . 1993;12:537-576.
2 Grieve DW. The dynamics of lifting. k e n
Sport Sci Rw.1977;5:157-179.
3 Grieve DW. Dynamic characteristics of man
during crouch- and stoop-lifting. In: Nelson
RC, Morehouse CA, eds. Biomechanics N.
Baltimore, Md: University Park Press; 1974:9-
29.
4 Davis PR, Troup JDG, Burnard JH. Move-
ments of the thoracic and lumbar spine when
lifting: a chronocyclo photographic study.
J Anat. 1965;991526.
5 Bejjani FJ, Gross CM, Pugh JW. Model for
static lifting: relationship of loads on the spine
and the knee. J Biomech. 1984;17:281-286.
6 Pamianpour M, Bejjani FJ, Pavlidis L.
Worker training: the fallacy of a single, correct
lifting technique. Ergonomics. 1987:30:331-
334.
7 Ekholm J, Arborelius UP, Nemeth G. The
load on the lumbosacral joint and trunk mus-
cle activity during lifting. Ergonomics. 1982;25:
145-161.
8 Patterson P, Congleton J, Koppa R, Huch-
ingson RD. The effects of load knowledge on
stresses at the lower back during lifting. Eqo-
nomics. 1987;30:539-549.
9 Schipplein OD, Trafimow JH, Andersson
GBJ, Andriacchi TP. Relationship between
moments at the L5/S1 level, hip and knee
joint when lifting. J Biomech. 1%0;23:307-912.
10 Scholz JP. The effect of load scaling in the
coordination of manual lifting. Human Move-
ment Science. 19!33;12:427-459.
11 Ellaway PH. Cumulative sum technique
and its application to the analysis of peris-
timulus time histograms. Flectmcephalogr
CIin Natrophysiol. 1978;45:302-304.
12 Riley PO, Shenkrnan ML, Mann RW,
Hodge WA. Mechanics of a constrained chair-
rise. JBiomech. 1991;24:77-85.
13 Lindh M. Biomechanics of the lumbar
spine. In: Nordin M, Frank1 VH, eds. Basic
Biomechanics of the Musculoskeletal System.
Philadelphia, Pa: Lea & Febiger; 1989:195-202.
14 Toussaint HM, van Baar CE, van Langen
PP, et al. Coordination of the leg muscles in
leglift and backlift. J Biomech. 1992;25:1279-
1289.
15 Toussaint HM, Commissaris DACM, Dieen
JH van, et al. Controlling the ground reaction
force during lifting. journal of Motor Behav-
ior. In press.
16 Delitto RS, Rose SJ, Apts DW. Electromyo
graphic analysis of two techniques of squat
lifting. Phys 7her. 1987;67:1329-1334.
17 Holmes JK, Damaser MS, Lehman SL. Erec-
tor spinae activation and movement dynamics
about the lumbar spine in lordotic and ky-
photic squat-lifting. Spine. 1992;17:327-334.
18 McGill SM, Norman RW. Reassessment of
the role of intra-abdominal pressure in spinal
compression. Ergonomics. 1987;30:15651588.
19 Winter DA. Biomechanics and Motor Con-
tml of Human Movement. New York, NY:
Wiley-Interscience; 1990.