ARTICLE IN PRESS

Journal of Biomechanics 43 (2010) 19972001

Contents lists available at ScienceDirect

Journal of Biomechanics
journal homepage: www.elsevier.com/locate/jbiomech
www.JBiomech.com

The variability and complexity of sitting postural control are associated

with discomfort
Karen H.E. Sndergaard a, Christian G. Olesen a,b, Eva K. Sndergaard a, Mark de Zee a,
Pascal Madeleine a,n
a
b

Center for SensoryMotor Interaction (SMI), Department of Health Science and Technology, Fredrik Bajers vej 7, 9220 Aalborg East, Denmark
Department of Mechanical and Manufacturing Engineering, Aalborg University, Denmark

a r t i c l e in fo

abstract

Article history:
Accepted 3 March 2010

The present investigation examined the variability of sitting postural movement in relation to the
development of perceived discomfort by means of linear and nonlinear analysis. Nine male subjects
participated in this study. Discomfort ratings, kinetic and kinematics data were recorded during
prolonged sitting. Body part discomfort index, displacement of the center of pressure (COP) in anterior
posterior and mediallateral directions as well as lumbar curvature were calculated. Mean, standard
deviation and sample entropy values were extracted from COP and lumbar curvature signals. Standard
deviation and sample entropy were used to assess the degree of variability and complexity of sitting.
A correlation analysis was performed to determine the correlation of each parameter with discomfort.
There were no correlations between discomfort and any of the mean values. On the contrary, the
standard deviations of the COP displacement in both directions and lumbar curvature were positively
correlated to discomfort, whereas sample entropies were negatively correlated. The present study
suggests that the increase in degree of variability and the decrease in complexity of sitting postural
control are interrelated with the increase in perceived discomfort. Finally, the present study underlined
the importance of quantifying motor variability for understanding the biomechanics of seated posture.
& 2010 Elsevier Ltd. All rights reserved.

Keywords:
Complexity
Sitting posture
Postural control
Discomfort
Variability

1. Introduction
Musculoskeletal discomfort is expressing manifestations like
perceived tension, muscle fatigue or soreness, numbness and
feeling of pain (de Looze et al., 2003). Discomfort from the lumbar
region is reported to be the main cause for an increase in general
discomfort in the seated position (Vergara and Page, 2002).
Moreover, discomfort may reect an early perception of pain
related to the biomechanical load applied to the musculoskeletal
system (Madeleine et al., 1998). However, the underlying
mechanisms of back pain are not clearly understood (Battie
et al., 2009; Videman et al., 2003). A number of studies have
suggested that prolonged sitting could be a risk factor for the
development of low-back pain (Corlett, 2006; Pope et al., 2002).
Thus, the study of discomfort in relation to prolonged sitting may
reveal important aspects of the transition between discomfort
and pain.
Interestingly, discomfort is considered to be related with
sitting postural changes (Fenety and Walker, 2002; Vergara and
Page, 2002; Liao and Drury, 2000). Liao and Drury (2000) have

Corresponding author. Tel.: + 45 99408833; fax: + 45 98154008.

E-mail address: pm@hst.aau.dk (P. Madeleine).

0021-9290/$ - see front matter & 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jbiomech.2010.03.009

reported a positive relationship between discomfort and the

frequency of postural changes during computer work. This has
been further substantiated by an increase in the frequency of
postural shifts over time also reported during computer work
(Fenety and Walker, 2002). Vergara and Page (2002) proposed
that postures sustained for a long period may be harmful and
underlined the necessity of varying posture. Previous studies have
mainly focused on specic postures and postural changes, and as
such, did not take into account the fact that seating is a dynamic
task (Dempster, 1955; Branton and Grayson, 1967). This calls for
further studies aiming at quantifying the variability of postural
control strategies in seated posture.
Standard deviation and/or coefcient of variation are the most
common linear descriptors used to characterize the amount of
motor variability (Stergiou, 2004). Such analysis is often
complemented with nonlinear analysis, as it provides measures
of the pattern of postural movement. These analysis techniques
enable the quantication of subtle changes in the dynamics of
biological systems (Lipsitz and Goldberger, 1992). In the biomechanics of sitting, nonlinear analysis so far has been mostly
applied to delineate the dynamics of sitting in relation to
development and disorders in infants (Deffeyes et al., 2009;
Harbourne and Stergiou, 2003; Harbourne et al., 2004). Changes
in the pattern of sitting postural control can be assessed by

ARTICLE IN PRESS
1998

K.H.E. Sndergaard et al. / Journal of Biomechanics 43 (2010) 19972001

e.g. entropy measures derived from information theory. Entropy is

the natural logarithm of a conditional probability, interpreted as
the rate of information generation and provides an estimate of the
complexity of the underlying system producing the dynamics in
question (Lipsitz and Goldberger, 1992; Richman and Moorman,
2000). Surprisingly, sitting postural control has only sparsely been
investigated in adults, despite the aforementioned importance of
understanding seated postures. The relationship between seated
center of pressure (COP) dynamics and driver macro movements
has been investigated by simultaneously assessing driver macro
movements and complexity of the COP displacements during long
term driving (Hermann, 2005). However, no studies have, to our
knowledge, measured the development of sitting discomfort and
related this to the dynamics of sitting postural control.
In the present study, we investigated the development of
discomfort during prolonged sitting and the basic relationships
between perceived seated discomfort and sitting postural movement in terms of variations in COP and lumbar curvature. The
variability of these seated postural variables was evaluated by
means of linear and nonlinear analysis techniques in relation to
perceived discomfort during prolonged sitting.

2. Methods and materials

2.1. Subjects
Nine male volunteers participated in the study (mean 7 SD age 25.2 7 1.6
years, height 186.9 7 5.8 cm, body mass 81.6 7 6.5 kg and BMI 23.3 7 1.1). None of
the subjects had any known spinal deformities or history of back pain. All subjects
gave written, informed consent to participate in this study. The study was
approved by the local ethics committee (N-20070004) and conducted in
conformity with the Declaration of Helsinki.

2.2. Experimental procedure

The method for assessing lumbar curvature was modied from the tangential
radiologic assessment of lumbar lordosis technique (Chernukha et al., 1998) to
allow for non-invasive assessment of lumbar curvature (see Fig. 1 for a schematic
illustration of the modied method). The spinous processes of the L1 and S2
vertebrae were palpated during relaxed standing, and each marked with a
reective marker. The point on the lumbar spine with the furthest perpendicular
distance to an imaginary line between L1 and S2, corresponding to point C in Fig. 1,
was found by sliding a ruler along the lumbar spine between L1 and S2, and
marked with a reective marker. A fourth marker was placed on the sternum as a
reference for discerning kyphotic and lordotic lumbar curvature. Lumbar curvature
was measured to assess the local postural movement in the lumbar region.
Once the markers are attached, the subjects were seated on a force platform
(AMTI OR6-7 1000, Watertown, MA, USA) with no back-, foot support or armrest
and without cushioning. As such, the only contact surface was the force platform.
The subjects were allowed to move their upper bodies in response to discomfort,
but were given restrictions with respect to movement of arms and feet, i.e. they
were instructed to leave their hands at rest on their thighs, and not to move their
legs and feet during the recording sessions. The edge of the force platform was
padded with foam to avoid discomfort at the popliteal area from resting the thighs.
The subjects watched a movie during the recording session to minimize the
inuence of the Hawthorne effect due to subject awareness. Pilot studies revealed
90 min of sitting to be sufcient to provoke a moderate level of discomfort. Each
recording session consisted of 18 intervals of 5 min data recording with a break of
20 s between each interval, resulting in a total of 96 min for each session. The
breaks were inserted for discomfort assessment and to allow the subjects to move
their lower legs and feet to ensure blood circulation in the legs.

2.3. Data recordings and analysis

Every 5 min, body part discomfort (BPD) ratings were collected during the 20 s
break using a 6 level scale from 0 to 5. Zero representing no discomfort and 5
worst imaginable discomfort (Corlett and Bishop, 1976). Reaction forces were
sampled at 100 Hz after amplication (gain of 4000) and analogue low pass
ltering (Fcut-off: 10.5 Hz). Similarly, kinematics data were sampled at 100 Hz using
eight Qualisys Pro-Reex 240 cameras (Qualisys, Gothenburg, Sweden).
BPD scores for each subject were summed for each time-interval, providing a
BPD index (Helander and Zhang, 1997). COP displacement over time was

Fig. 1. Experimental setup: Subject was seated on a force platform with no

support for the back, feet or arms. The arrows in the coordinate system indicate
the positive axes for the x, y and z force components. Reective markers
were placed at the spinous processes of L1 and S2. A third marker was placed at
the farthest perpendicular distance to a line between L1 and S2 markers. a denotes
the angle of lumbar curvature. A fourth marker was placed at the sternum as a
reference point to discern lordotic and kyphotic lumbar curvatures. Negative
values denoted kyphotic curvatures and positive lordotic curvatures.
computed in the anteriorposterior (AP) and mediallateral (ML) directions
(Winter, 1990) in Matlab (Mathworks, Natick, MA). Kinematics data were
processed using Qualisys track manager software, exported to Matlab and
transformed to lumbar curvature angle. Negative values denoted kyphotic
curvatures and positive lordotic curvatures. The mean, standard deviation (SD)
and sample entropy (SaEn) of the COP displacement in AP and ML directions and
of the lumbar curvature were computed. SD and SaEn were used to characterize
respectively the amount of motor variability and the complexity of the sitting
postural control. SaEn quanties regularity in a data series by assessing
the probability that sequences of length m that are similar will remain similar
when incrementing the length of the sequences to m+ 1. The similarity condition is
determined by the tolerance, r. Output is a unit-less, non-negative number, where
higher values indicate more complex data series. For more details, see Richman
and Moorman (2000). The embedding dimension, m, was chosen as 2, and the
tolerance, r, was chosen to 0.1  SD of the time series (Pincus, 1991).

2.4. Statistical analysis

Pearson correlation coefcients were calculated using SPSS version 16.0
(Chicago, IL, USA) to assess the relationship between perceived discomfort and
dynamics of sitting postural control, using BPD sum as the dependent variable and
mean/SD/SaEn of COP displacement in AP and ML directions and of lumbar
curvature as predictor variables. p o 0.05 Was considered as signicant.

3. Results
Figs. 2 and 3 illustrate changes in BPD and in mean/SD/SaEn of
COP displacement in AP and ML directions and of lumbar
curvature over time. BPD increased signicantly over time (Fig. 2).
The mean COP displacement in AP direction (Fig. 3a) and mean
lumbar curvature (Fig. 3c) increased signicantly over time,
revealing a shift towards more lordotic curvatures. Meanwhile,
mean COP displacement in ML direction did not show any
signicant changes over time (Fig. 3b). The SD of the COP
displacement in both AP and ML directions (Fig. 3d and e)
and of lumbar curvature (Fig. 3f) signicantly increased over time,
while SaEn signicantly decreased over time (Fig. 3gi).

ARTICLE IN PRESS
K.H.E. Sndergaard et al. / Journal of Biomechanics 43 (2010) 19972001

Sum of regions

Discomfort Index

1999

prolonged sitting by means of linear and non-linear analysis of

center of pressure displacement and variations in lumbar
curvature.
4.1. Methodological considerations

Lower
Back

Upper
Back

Shoulders

Neck

In previous studies on seated discomfort, experimental setups

have typically been directed towards specic applications, e.g.
discomfort during driving (Hermann, 2005), chair design/assessment (Vergara and Page, 2002; Kingma and van Dieen, 2009) or
work tasks (Starr et al., 1985; Fenety and Walker, 2002; Ziee,
2003). The main drawback of such approach is the difculty to
discern between factors (e.g. work task, sitting task) interacting
with the perception of discomfort. Furthermore, the presence or
absence of backrest or armrests complicates comparison between
studies. In the present study, we omitted backrest, armrests and
cushion. The sitting constraints (no arms/legs movement) during
the recording sessions may explain the lack of changes in the
displacement of the COP in the ML direction over time. The front
edge of the force platform was padded to minimize discomfort in
the popliteal area. This resulted in perceived discomfort mainly
originating from the buttocks and the low back regions. The
experimental setup did not reect realistic seating, but it enabled
to assess changes over time in sitting postural dynamics in
relation to perceived seated discomfort. Thus, the present study
could be used as a baseline for future studies investigating more
realistic setup, including for instance armrests, footrests or seats.

Legs

Popliteal

Thighs

Buttocks

4.2. Postural variability measures as indicators of seated discomfort

Time (min)
Fig. 2. Mean+ SD of the sum of body part discomfort, neck, shoulders, upper back,
lower back, buttocks, thighs, popliteal region and legs over 90 min seating.

Table 1 lists the results of the statistical analysis. There were

statistically signicant correlations between BPD and all predictor
variables, except mean COP displacement in AP and ML
directions and lumbar curvature. The SDs of the COP
displacement and lumbar curvature were positively correlated
with BPD, while SaEn were negatively correlated.

4. Discussion
In the present study, we investigated the correlations between
perceived discomfort and sitting postural control during

The present study depicted correlations between perceived

seated discomfort and mean, amount of variability and complexity of the COP trajectories as well as lumbar curvature. The
predominance of discomfort reported from the low back and
the buttocks conrmed that the variables under investigation can
be considered as indicators of seated discomfort. Regarding mean
posture, we computed the mean curvatures of the 18 intervals
constituting the recording session. This procedure was repeated
for the displacement of the COP in AP and ML directions to
further explore relationships between seated discomfort and
sitting postural changes. There were no correlations between
discomfort and mean values of postural movement and mean
lumbar curvature, suggesting that mean sitting posture variables
do not seem to provide an objective way to assess the
development of seated discomfort at a group level in the present
experimental setup.
On the other hand, the present analysis conrmed the
importance of variability in relation to sitting posture and
discomfort (Vergara and Page, 2002). The amount of variability
for both lumbar curvature and COP displacement increased over
time as well as the discomfort, indicating a relationship. Gross
mediallateral displacements of the COP can be interpreted as a
means of pressure relief of the gluteal region, as the peak pressure
is lifted from either side. The increase in SD with increased
discomfort probably indicates a progressively larger need for
greater or more effective pressure relief of the soft tissue under
the buttocks and suggests an association between prolonged
tissue pressure under the buttocks and seated discomfort. Given
the lumbar-pelvic anatomy, with the shape of the spine closely
intertwining with the tilt of the pelvis, the COP displacements in
AP direction are closely related to the variations in lumbar
curvature. This is substantiated by the positive correlation
between standard deviation of the lumbar curvature and standard
deviation of COP displacement in AP direction. Gross displacements of the lumbar curvature are likewise interpreted as a

ARTICLE IN PRESS
2000

K.H.E. Sndergaard et al. / Journal of Biomechanics 43 (2010) 19972001

Lumbar curvature (deg.)

CoP displacement in M-Ldir. ( m)

CoP displacement in A-P dir. (m)

0.03

0.03

0.02

0.02

0.01

0.01

-0.01

-0.01

-0.02

-0.02

-0.03

-0.03

6
4

Mean

2
0
-2
-4
-6

Standard Deviation

0 5 1015202530354045505560657075808590

-8
-10
0 5 1015202530354045505560657075808590

0 5 1015202530354045505560657075808590

0.03

0.03

0.025

0.025

0.02

0.02

0.015

0.015

0.01

0.01

0.005

0.005

5
4
3

Sample Entropy

0 5 1015202530354045505560657075808590
1.8

1.8

1.6

1.6

1.4

1.4

1.2

1.2

0.8

0.8

0.6

0.6

0.4

0.4

0.2

0.2

0 5 1015202530354045505560657075808590

2
1
0
0 5 1015202530354045505560657075808590

0 5 1015202530354045505560657075808590
2.5
2
1.5
1
0.5
0

0 5 1015202530354045505560657075808590

0 5 1015202530354045505560657075808590

Time (min)

Fig. 3. Mean+ SD of the mean (top row), standard deviation (middle row) and sample entropy (bottom row) of the center of pressure (CoP) displacement (m) in the
anteriorposterior (AP, left column) and mediallateral (ML, middle column) and lumbar curvature (deg., right column) over 90 min seating. Negative values denoted
kyphotic curvatures and positive lordotic curvatures.

Table 1
Pearson correlation coefcients between sum of body part discomfort (BPD) and mean, standard deviation (SD) and sample entropy (SaEn) of the center of pressure
displacement in anteriorposterior and mediallateral directions (COPA P and COPM L) as well as lumbar curvature (LC).
BPD
BPD
Mean COPA P
Mean COPM L
Mean LC
SD COPA P
SD COPM L
SD LC
SaEn COPA P
SaEn COPML
SaEn LC
n

1
0.125
0.029
0.002
0.273nn
0.329nn
0.140n
0.271nn
0.278nn
0.193nn

Mean COPA

Mean COPM

0.541nn
0.065
0.158n
0.052
0.139n
0.235nn
0.120
0.252nn

1
0.091
0.003
0.313nn
0.139n
0.070
0.091
0.113

Mean LC

SD COPA

1
0.171n
0.361n
0.480nn
0.128
0.200nn
0.392nn

1
0.163n
0.675nn
0.097
0.151n
0.565nn

SD COPM

1
0.357nn
0.134n
0.696nn
0.368nn

SD LC

SaEn COPA

1
0.098
0.116

1
0.033
0.162n
0.823nn

SaEn COPM

1
0.166n

SaEn LC

p o 0.05.
po 0.01.

nn

means of pressure relief, as changes in lumbar curvature rotate

the pelvis, and thus shift the location of the ischial tuberosities
under the buttocks. Moreover, it may provide muscle- and
ligament tension relief of the lumbar, sacral and gluteal body
regions.
In parallel to the observed changes in the amount of
variability, sample entropy as a measure of complexity decreased

over time for COP postural movement in AP and ML direction

and lumbar curvature, along with signicant negative correlations
of sample entropy to discomfort. The changes in sample entropy
values suggest that the intrinsic dynamics of seated postural
control are similar to those of standing postural control and
generally perceived as uctuating around an equilibrium point
(Collins and DeLuca, 1993). Finally, the present study showed that

ARTICLE IN PRESS
K.H.E. Sndergaard et al. / Journal of Biomechanics 43 (2010) 19972001

the complexity of sitting postural control is affected by increasing

discomfort due to prolonged sitting.
In conclusion, the present study revealed for the rst time that
the degree of variability of COP displacements and lumbar
curvature increased, while its complexity decreased in relation
to increased perceived discomfort. Quantitative measurement of
spinal posture behavior over an extended period of time
contributed to a better understanding of the biomechanics of
seating. As discomfort increased, sitting movement patterns
became larger and more regular. Further studies investigating
sitting postural in working conditions are warranted.

Conict of interest statement

All authors hereby declare that there are no conicts of interest.

Acknowledgements
The authors are grateful to Rene Lindstrm (Center for
Sensory-Motor Interaction (SMI), Aalborg University) for identifying anatomical landmarks. This study was nancially supported
by the European Regional Development Fund (Project Seating
Position and Functional Ability).
References
Battie, M.C., Videman, T., Kaprio, J., Gibbons, L.E., Gill, K., Manninen, H., Saarela, J.,
Peltonen, L., 2009. The twin spine study: contributions to a changing view of
disc degeneration. Spine Journal 9, 4759.
Branton, P., Grayson, G., 1967. An evaluation of train seats by observation of sitting
behaviour. Ergonomics 10, 35 &.
Chernukha, K.V., Daffner, R.H., Reigel, D.H., 1998. Lumbar lordosis measurement: a
new method versus Cobb technique. Spine 23, 7479.
Collins, J.J., DeLuca, C.J., 1993. Open-loop and closed-loop control of posturea
random-walk analysis of center-of-pressure trajectories. Experimental Brain
Research 95, 308318.
Corlett, E.N., 2006. Background to sitting at work: research-based requirements for
the design of work seats. Ergonomics 49, 15381546.
Corlett, E.N., Bishop, R.P., 1976. Technique for assessing postural discomfort.
Ergonomics 19, 175182.
J., 2003. Sitting comfort and discomfort
de Looze, M.P., Kuijt-Evers, L.F.M., Van Dieen,
and the relationships with objective measures. Ergonomics 46, 985997.

2001

Deffeyes, J.E., Harbourne, R.T., Kyvelidou, A., Stuberg, W.A., Stergiou, N., 2009.
Nonlinear analysis of sitting postural sway indicates developmental delay in
infants. Clinical Biomechanics 24, 564570.
Dempster, W.T., 1955. The anthropometry of body action. Annals of the New York
Academy of Sciences 63, 559585.
Fenety, A., Walker, J.M., 2002. Shortterm effects of workstation exercises on
musculoskeletal discomfort and postural changes in seated video display unit
workers. Physical Therapy 82, 578589.
Harbourne, R., Stergiou, N., Tscharnuter, I., Dejong, S., 2004. Nonlinear analysis of
sitting postural control in infants with and without movement disorders.
Journal of Sport and Exercise Psychology 26, S88.
Harbourne, R.T., Stergiou, N., 2003. Nonlinear analysis of the development of
sitting postural control. Developmental Psychobiology 42, 368377.
Helander, M.G., Zhang, L.J., 1997. Field studies of comfort and discomfort in sitting.
Ergonomics 40, 895915.
Hermann, S., 2005. Exploring sitting posture and discomfort using nonlinear
analysis methods. IEEE Transactions on Information Technology in Biomedicine 9, 392401.
Kingma, I., van Dieen, J.H., 2009. Static and dynamic postural loadings during
computer work in females: sitting on an ofce chair versus sitting on an
exercise ball. Applied Ergonomics 40, 199205.
Liao, M.H., Drury, C.G., 2000. Posture, discomfort and performance in a VDT task.
Ergonomics 43, 345359.
Lipsitz, L.A., Goldberger, A.L., 1992. Loss of complexity and agingpotential
applications of fractals and chaos theory to senescence. JamaJournal of the
American Medical Association 267, 18061809.
Madeleine, P., Voigt, M., Arendt-Nielsen, L., 1998. Subjective, physiological and
biomechanical responses to prolonged manual work performed standing on
hard and soft surfaces. European Journal of Applied Physiology 77, 19.
Pincus, S.M., 1991. Approximate entropy as a measure of system-complexity. In:
Proceedings of the National Academy of Sciences of the United States of
America 88, 22972301.
Pope, M.H., Goh, K.L., Magnusson, M.L., 2002. Spine ergonomics. Annual Review of
Biomedical Engineering 4, 4968.
Richman, J.S., Moorman, J.R., 2000. Physiological time-series analysis using
approximate entropy and sample entropy. American Journal of PhysiologyHeart and Circulatory Physiology 278, 20392049.
Starr, S.J., Shute, S.J., Thompson, C.R., 1985. Relating posture to discomfort in VDT
use. Journal of Occupational and Environmental Medicine 27, 269271.
Stergiou, N., (2004). Innovative analysis of human movement. Human Kinetics,
Champaign, IL.
Vergara, M., Page, A., 2002. Relationship between comfort and back posture and
mobility in sitting-posture. Applied Ergonomics 33, 18.
Videman, T., Battie, M.C., Gibbons, L.E., Maravilla, K., Manninen, H., Kaprio, J., 2003.
Associations between back pain history and lumbar MRI ndings. Spine 28,
582588.
Winter, D.A., 1990. Biomechanics and Motor Control of Human Movement. Wiley
Interscience Publication, New York.
Ziee, M., 2003. Sitting posture, postural discomfort, and visual performance: a
critical view on the interdependence of cognitive and anthropometric factors
in the VDU workplace. International Journal of Occupational Safety and
Ergonomics 9, 503514.