Patterns of Spinal Motion During Walking: Jack Crosbie ", Roongtiwa Vachalathitib, Richard Smith"
Patterns of Spinal Motion During Walking: Jack Crosbie ", Roongtiwa Vachalathitib, Richard Smith"
Patterns of Spinal Motion During Walking: Jack Crosbie ", Roongtiwa Vachalathitib, Richard Smith"
Abstract
We present the results of a study in which we investigated the patterns of movement of the lower thoracic and lumbar spinal
segments and the pelvis in subjects walking at a self-selected speed. Our subjects were 108 healthy adults aged from 20 to 82 years
and were equally divided with respect to gender. Measurements were carried out using a video-based system by which the trunk
was partitioned into segments and the movements of these segments about three orthogonal axes recorded. We found consistent
patterns within and between segments and movements, with apparent consequential trunk motion following pelvic displacements.
This suggests that the spinal movements associated with walking are linked to the primary motions of the pelvis and the lower limbs.
the lower thoracic spine has the capability to move, 2.2. Instrumentation
albeit within a limited range, during the gait cycle. Evi- The kinematic analysis was performed on data
dence from planar investigations of the thoracic spine, collected by an automated video-based system (Expert-
between the sixth and twelfth thoracic vertebral levels, vision; Motion Analysis Corporation (MAC), Santa
has suggested that movements with amplitudes of over Rosa, Ca.). Four video cameras (NEC model Tl-23A
30 degrees may be quite common about each of the three CCD; MAC) with 12.5-75 mm zoom lenses were used
axes during full range active motion [ 121. to record images from body markers at a sampling fre-
We propose in this study to describe the motion of the quency of 60 frames per second. The test zone was 2.5
spine during self-selected free-speed walking. To that metres long by 2 metres high by approximately 2 metres
end, we subdivided the spine into lower thoracic, lumbar wide, with the long axis aligned with the plane of pro-
and pelvic segments and compared the movement pat- gression. Simultaneous input from the four cameras was
terns of these segments during walking. The population processed automatically (VP320 video processor; MAC)
tested was equally represented by male and female sub- and the data files were transferred to a dedicated com-
jects and used subjects ranging in age from 20 to 82 puter workstation (SUN Graphics 4/l 10; Sun Micro-
years. This study formed part of a project which includ- systems Inc, Mountain View, Ca.). Processing of the
ed investigation of age, gender and speed effects on spi- data files included lens correction algorithms and utilis-
nal kinematics. Reporting the patterns and ranges of ed commercial and in-house software for data reduction
motion during unconstrained gait was an essential and analysis.
precursor to more specific analyses. Prior to data collection, we calibrated the camera sys-
tem using an accurately constructed and surveyed rigid
2. Procedures steel cube of dimension 1 m3 with 18 spherical
retroreflective markers. The resolution and error of the
2.1. Subjects camera system was evaluated using static and dynamic
We tested a total of 108 subjects (50 males and 58 images to assess the accuracy and reliability of the sys-
females) aged between 20 and 82 years (Table 1). tem under test conditions. Results indicated average
Subjects’ body mass index values (BMI = height/ measurement errors of less than 0.1% for all coor-
weight2) suggested that our sample was normally distri- dinates. These values are comparable with the data
buted across a range consistent with the general popula- reported by Linden et al. [14] which demonstrated high
tion. We decided that subjects whose BMI ratio lay accuracy and reproducibility for the Motion Analysis
outside 1.5 standard deviations (SD.) from the median video system. The theoretical and empirical resolution
value for the population in general [ 131would be exclud- and sensitivity of the system is less than 0.5 degrees
ed from the study. No subjects were excluded as a result based upon a test model of similar dimensions to the
of this criterion. subjects tested in this study. For the purposes of analy-
Subjects were volunteers from the local community. sis, we report angular displacement values of less than
All subjects included were healthy, reporting no inci- one degree but do not consider differences of this
dence of serious neurological or musculoskeletal illness magnitude to represent functionally relevant values
or injury prior to testing. None of the subjects had suf- irrespective of statistical significance.
fered back, hip or leg pain during the six months Following transfer of the video files, the coordinate
preceding the study. None of the subjects was using data were analysed using Fourier analysis. Coordinate
prescribed medication at the time of testing. We inter- data were filtered using a Butterworth-type digital filter
viewed the subjects prior to inclusion in the study to with a low pass value of 5 Hz.
categorise their activity levels. On the basis of questions
derived from the Australian Heart Association’s activity 2.3. Protocol
questionnaire, all subjects were graded as physically ac- Subjects wore a backless swimsuit or similar clothing
tive and lived independently. The study was approved for the test. We applied lightweight adhesive body mark-
by the University of Sydney Human Ethics Committee. ers, consisting of 2.5 cm hollow spherical reflective tar-
Table 1
Subject characteristics
Females 45.24 18.55 20-80 1.61 0.07 1.43-1.8 59.6 9.83 41-95
Males 46.34 18.25 20-82 1.72 0.08 1.57-1.9 73.7 10.5 55-100
J. Crosbie et al. I Gait & Posture 5 (1997) 4-12
Fig. I. Location of reflective markers on subject and conventions used to define axes and motions.
gets. Thirteen body markers were attached to the dorsal perpendicular to the X-axis and parallel to the ground,
surface of the trunk and three each to the right and left with the positive direction towards the left side, the Z-
thighs. A further marker was used to define each foot axis was vertical and mutually perpendicular to the
for identification of the temporal and spatial parameters other axes, being positive upwards.
of gait (Fig. 1). In order to minimise and check for rela- We divided the trunk and lower limbs into discrete
tive motion between the markers and the skin surface, segments using the reflective marker coordinates to
we examined each marker between every test walk and define a series of rigid bodies. An upper trunk segment
at the conclusion of the test. Markers were applied to was defined using markers located over the spinous pro-
the skin over the bony landmarks as indicated in the fig- cess of the first thoracic vertebra (Tl), the spinous pro-
ure. The thigh markers were located according to a cess of the sixth thoracic vertebra (T6) and the two
predetetmined orientation which permitted transforma- paravertebral markers at the T6 level located over the
tion of the loci to define the three dimensional motion angles of the ribs. The direction of the principal local
of the thigh segment relative to the pelvis and thus rec- embedded axis (caudo-cephalic) was defined by the ver-
ord the hip joint motion. tebral markers, the provisional mediolateral axis by the
We required the subjects to walk three times over a lateral markers and the anteroposterior axis as the cross
clear, level floor area without restricting them to a product of the first and second axes. The corrected sec-
marked walkway. Subjects walked barefoot across a 10 ond axis was then computed as the cross product of the
metre area towards a target marker at the far end of the first and third axes [15].
room. The central zone of the walkway acted as the test Similar procedures were used to define the lower
area, but subjects were not aware of the ‘active’ portion trunk segment, the lumbar segment, the pelvic segment
of the laboratory. We tested each subject at a walking and the two thigh segments. The model used in the study
speed which was self-selected by the subject but followed treated the defined segments as rigid bodies and we
simple instructions from the investigator, that is, that defined movement within spinal regions in terms of the
subjects should walk at the speed which they considered relative motion between the rigid body above and the
most comfortable. rigid body below the region of interest. Thus, motion of
the lumbar spine was defined as the relative motion of
2.4. Coordinate systems the lower trunk segment with respect to the pelvic
The global reference frame was defined with respect segment and the lower thoracic region was defined by
to the laboratory. The X-axis corresponded to the line the relative motion of the lumbar and upper trunk
of progression of the subject across the room and was segments.
positive in that direction. The Y-axis was defined as We defined motion about the three orthogonal local
J. Crosbie et al. I Gait & Posture 5 (1997) 61.2 9
2.5. Analysis
We established the temporal events of the gait cycle
with reference to heel strike on each side. Heel strike was
Fig. 2. Patterns of lateral flexion in the lower thoracic and lumbar defined as the time at which the heel marker ceased
regions and in the pelvis (pelvic list) during free-speed walking. Thick moving vertically downwards after the swing phase. The
lines represent mean values and tine lines represent f standard error data were time normalised with respect to one complete
of the mean (S.E.M.). Time values are referred to the complete gait
gait cycle (right heel strike to the next right heel strike).
cycle from right heel strike (0%) to the subsequent right heel strike
(100%). Toe off times are indicated (LTO and RTO). Positive values We preceded this normalisation by performing all analy-
(up on the vertical axis) refer to rotation about the local anteropos- ses to which time variables were pertinent. The three
terior axis towards the right side (see text). cycles were averaged for each subject. Statistical analysis
was performed using the SPSSx program (SPSS Inc.,
Chicago).
axes using the right hand convention; thus, counter-
clockwise rotation about the axis, as viewed looking 3.Redts
back along the axis towards the body, was expressed as
a positive value. Using conventional descriptors of ana- We have followed a conventional description of the
tomical motion, we describe rotation about the X-axis temporal divisions of the gait cycle [ 161. Based upon our
as lateral flexion, with bending towards the right side of data, the double support phase extended from 0% to
the subject’s body of the cephalad segment with respect 10% and from 50% to 60% of the cycle, single sup-
to the caudal defined as positive. Rotation about the port/swing phase from 10% to 50% and from 60% to
Y-axis corresponds to forward flexion and extension. 1Ooo! of the cycle and mid-stance at 30% and 80% of the
Positive motion is forward flexion. Similarly, rotation gait cycle. The cycle was taken from right heel strike to
about the Z-axis is termed axial rotation, with the posi- the next right heel strike. We observed a high level of
tive motion involving rotation of the cephalad segment inter-subject consistency in all the data obtained during
towards the left with respect to the caudal segment testing (Figs. 2-7). Subjects walked at their self-
(Fig. 1). determined speed, which averaged 1.3 m.s-’ (S.D.
Because the movements within the trunk may be 0.19). When normal&d for height, this value equated to
directed towards providing the upper trunk segment 0.78 statures.s-’ (S.D. 0.1). Subjects walked with a
with a relative stability with respect to the global frame cadence of 57.1 stepsmin-’ (S.D. 5.2) and their average
of reference [5], we also investigated the relative attitude height normal&d step length was 0.69 statures (S.D.
of the rigid bodies representing the trunk segments with 0.09).
respect to the ground. We used similar conventions to
describe global motion of the segments as we employed 3.1. Lateral jlexion
for the relative segmental motion. In these cases, we Peak lateral flexion occurred at 15% and 65% of the
refer to the lower trunk rigid body (LT), defined by the gait cycle, which coincides with early swing phase (Fig.
twelfth thoracic vertebra (T12) as its caudal extremity, 2). In the lower thoracic and lumbar regions, the dis-
and the upper trunk rigid body (UT) with T6 as its placement was towards the weight bearing limb, while
caudal extremity. the pelvic segment listed towards the swinging side. The
The hip joint motion was defined in terms of the rela- segmental motions demonstrated close time correla-
tive motion between the relevant thigh rigid body and tions. The pelvis and trunk ‘stabilise’ near the neutral
the pelvis. The movements of the pelvis were considered position for around 15% to 20% of the cycle, correspon-
with respect to the global reference system. Particular ding to the period of late swing and early double
conventions have evolved for description of the move- support.
ments of the pelvis in space. We employed a convention The motion of the LT segment with respect to the
J. Crosbie et al. I Gait & Posture 5 (1997) 612
0 20 10 60 80 100
Time (percenla,oegail CJ&)
global reference system appeared reduced when com- of the cycle to a maximum positive pelvic tilt in the early
pared to the intra-segmental motions. The segment UT single support phase. The pelvis then progressively
moved in a complementary manner to the pelvis with re- reversed its tilt to the next heel strike. The lower thoracic
spect to the ground (Fig. 3). segment extended maximally at heel strike, returned to
The peak-to-peak range of motion for lateral flexion a neutral position at mid-stance, then extended again
(Table 2) indicated that the lumbar segment displaced through late stance.
more than the other segments, reaching nine degrees The patterns of movement in the lumbar segment
(S.D. 3.5 degrees). This range was significantly greater complemented those of the pelvis. Maximum lumbar
than that found in the pelvis or lower thoracic segments spine flexion occurred at heel strike. This was followed
(F2,21= 16.55; P < 0.001). There was no significant dif- by a relatively rapid extension back to neutral until the
ference between values for the pelvis and lower thoracic beginning of single support. A relatively slow flexion
segment. Lateral flexion range was consistently and then followed, reaching maximum at heel strike.
significantly greater in all segments than the other move- The upper trunk demonstrated a greater oscillation
ments (F2.r1 = 56.46; P < 0.001). with respect to the ground than did the lower trunk (Fig.
5). Both segments were in a near neutral orientation at
3.2. Forward Jlexion/extension ‘toe off, being displaced into a few degrees of extension
The patterns of flexion and extension demonstrated at heel strike.
biphasic movements through the gait cycle correspon-
ding to one flexionextension cycle per step (Fig. 4). The
pelvis rotated into negative pelvic tilt at heel strike. This 0
was followed by a counter-motion during the first 10%
Table 2
Peak-to-peak range of motion of spinal segments, pelvis and hip
during free-speed gait cycle (standard deviation)
Lumbar 9.0 (3.5) 3.5 (2.0) 4.5 (2.0) Time $ercrntqegoir cycle)
subjects with clinical problems affecting the lower limbs 1101Bastian A, Thompson D M, Bums S B, Schlegel J H. Correla-
tions between lumbar and pelvic movement during normal gait.
and also in subjects with spinal disorders. The influence Phys Therap 1991; 71: S92.
of walking speed, and the changes in spinal motion 1111Nottrodt J W, Charteris J, Wall J C. The effects of speed on pel-
associated with advancing age, provide further insights vic oscillations in the horizontal plane during level walking. J
Hum Move Stud 1982; 8: 27-40.
into the nature of the association between spinal kine-
matics and gait. We propose to report the results of WI Vachalathiti R. An Investigation of Age-Related Changes in
Three-Dimensional Kinematics of the Spine. Unpublished PhD
these investigations in a separate paper. thesis, University of Sydney, Australia, 1994.
1131 Med J Ausrral 1985; 182 (supplement): S3.
5. Conclusion 1141 Linden D W V, Carlson S J, Hubbard R L. Reproducibility and
accuracy of angle measurement obtained under static conditions
with the Motion AnalysisTM video system. Phys Therap 1992;
We believe that the spinal segments demonstrate com- 72: 300-305.
plementary movements to the motion of the pelvis, in IlSl Pearcy M J, Gill J M, Whittle M W, Johnson G R. Dynamic
particular with respect to flexion/extension and lateral back movement measured using a three dimensional television
flexion. Pelvic motion responds to the needs of the sub- system. J Biomech 1987; 20: 943-949.
ject to advance the lower limb and to transfer the body [I61 Inman V T, Ralston H J, Todd F. Human Walking. Baltimore:
Williams and Wilkins, 198 1.
weight from one supporting side to the other during nor- I171 Sutherland D H. Gait Disorders in Childhood and Adolescence.
mal locomotion. Baltimore: Williams and Wilkins, 1984.
1181Townsend M A. Dynamics and coordination of torso motions
in human locomotion. J Biomech 1981; 14, 727-738.