Gait & Posture 33 (2011) 715–717

Contents lists available at ScienceDirect

Gait & Posture

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

Short communication

Estimating functional stability boundaries for bipedal stance

Katharine E. Forth a, Matthew J. Fiedler b, William H. Paloski c,d,*
a
Universities Space Research Association, Houston, TX, USA
b
Wyle Integrated Science and Engineering Group, Houston, TX, USA
c
NASA Johnson Space Center, Houston, TX, USA
d
University of Houston, Houston, TX, USA

A R T I C L E I N F O A B S T R A C T

Article history: We propose a technique to estimate functional limits of stability (LOS) during bipedal stance using a
Received 16 March 2010 controlled, low speed, voluntary leaning protocol requiring feet to remain in contact with the ground.
Received in revised form 22 December 2010 LOS are estimated from ellipses fit to center-of-mass position data obtained during the leaning protocol.
Accepted 26 December 2010
The LOS of nine healthy subjects were found to be 20–59% closer to the center of stance than the more
frequently used anatomical boundaries and were reduced by closing the eyes. We conclude that
Keywords: functional stability boundaries should be used when the outcome measure is related to fall risk.
Balance control
Published by Elsevier B.V.
Stability limits
Falls
Time-to-boundary
Stability margin

1. Introduction
Traditional postural stability measures focus on deviations of
the center-of-mass (COM) from the center of the base-of-support.
However, introduction of spatial [1] and temporal [2] stability
margins has increased the need to accurately estimate the limits of
stability (LOS) because the functional accuracy of stability margin
estimates depends on the accuracy of LOS estimates. For example,
studies employing time-to-boundary (TTB), an increasingly popu-
lar temporal stability margin estimator, frequently define the
stability boundary at the outer edges of the feet [2–4]. However, an
individual’s functional stability boundary during bipedal stance is
more likely an ellipse much smaller than the base of support [3,5–
7], so true TTB is often overestimated. While this may be of little
consequence to basic studies of posture control, it should be of
concern to clinical studies addressing fall risks, as TTB over-
estimates translate directly to fall risk underestimates.
Functional LOS have been estimated previously using leaning
protocols [3,6,8,9], but inconsistencies in techniques may have
limited the accuracy and reproducibility of the estimates due to
technique for estimating the functional LOS using a controlled, low
speed, voluntary leaning protocol requiring feet to remain in
continuous contact with the ground.
2. Methods
2.1. Subjects
Nine healthy subjects (four males, five females; ages 25–39 years) participated in
this study, which was approved by the Johnson Space Center CPHS, after providing
written informed consent.
2.2. Procedure
Height, weight, and foot dimensions were measured before testing began (Table
1). Foot lengths and widths were measured using a Brannock device. Forefoot and
hindfoot lengths were measured from the center of the medial malleolus to the
posterior or anterior extent of the foot, respectively. Subjects then stood upon thin
(<1 mm) pressure-sensing material (TekScan Inc., Boston, MA) atop the forceplate
of a posturography system (NeuroCom International, Clackamas, OR) with feet
parallel to the anterior–posterior (AP) body axis and medial malleoli aligned with
the forceplate center axis. Stance width was measured between the medio-lateral
midpoints of each foot. Throughout testing, center-of-pressure (COP) data and foot-
ground contact pressures were recorded at 100 Hz.

fast leaning circles [9], failure to monitor [3,5] or allowing loss of

2.3. Limits of stability testing
foot contact with ground [6,8,9], and/or constructing ellipses from
limited data samples [3,8]. In this note, we propose an alternative Subjects performed eight 35 s LOS trials: four with eyes open (EO) and four with
eyes closed (EC). They began each trial standing quietly upright with arms crossed
at the chest for 3–5 s. They then leaned forward (slowly) as far as possible while
maintaining heel contact with the forceplate, and, after reaching this forward
stability limit, began moving their bodies slowly and evenly to the left (counter-
* Corresponding author at: Center for Neuromotor and Biomechanics Research,
clockwise) or right (clockwise) while leaning as far away from center as possible
University of Houston, 3855 Holman St., Garrison 104, Houston, TX 77204, USA.
without flexing knee or hip joints or raising the toes or heels of either foot (Fig. 1,
Tel.: +1 713 743 9272; fax: +1 713 743 9860.
top). After completing a full 3608 arc, they slowly returned to the upright (center)
E-mail address: whpaloski@uh.edu (W.H. Paloski).

0966-6362/$ – see front matter . Published by Elsevier B.V.

doi:10.1016/j.gaitpost.2010.12.024
716 K.E. Forth et al. / Gait & Posture 33 (2011) 715–717

Table 1
Subject characteristics.

Subjects N Weight (Kg) Height (cm) Foot length (cm) Forefoot (cm) Hindfoot (cm) Foot width (cm) Stance width (cm)

Males 4 77.6  3.3 178.9  4.8 27.5  1.1 20.1  0.9 7.4  1.0 10.3  0.5 25.1  1.5
Females 5 55.7  8.4 161.9  4.0 24.3  0.5 18.2  0.8 6.1  0.4 9.4  0.4 25.5  1.4
Total 9 65.4  13.1 169.5  9.8 25.7  1.9 19.1  1.3 6.6  1.0 9.8  0.6 25.3  1.4

Values are mean  SD.

position. For each vision condition, movement direction was randomly assigned on
the first trial, and then alternated between left and right.

2.4. Data reduction and analysis

COM position was estimated by low-pass filtering the COP time series data for
each trial using a phaseless 2nd order Butterworth FIR filter with a cutoff frequency
at 0.85 Hz (Matlab 7, the MathWorks, Natick, MA). Data were then removed from
the initial and final lean periods (from/to center) and from each period when the
subject’s feet lost full contact with the support surface (pressure <0.07 Pa under
heel or ball of either foot; Fig. 1, middle). Average LOS boundaries were estimated by
fitting an unconstrained ellipse to the concatenated COM position data for each
visual condition using a numerically stable least squares fitting technique [10], and
functional stability boundaries were defined by the mean + 2SD LOS ellipse (Fig. 1,
bottom). SD was determined from differences between the individual data points
and the best-fit ellipse, and 2SD was added to the major and minor ellipse axes to
create the mean + 2SD ellipse. Repeated measures analysis was used to examine
differences in LOS between vision conditions. Data are expressed as mean  SD and
statistical significance was accepted at p < 0.05.

3. Results

The LOS ellipse center locations and axes orientations were

unchanged between EO and EC, and the axes were closely aligned
with the cardinal body axes (rotation angle = 1.4  3.78 with EO and
1.5  3.38 with EC). The functional stability boundaries defined by
these axes were smaller than the corresponding anatomical
boundaries with EO, particularly in the posterior direction: anterior
boundary = 80.0  7.8% of forefoot length, posterior
boundary = 41.0  10.6% of the hindfoot length, and ML
boundary = 73.5  7.3% of the stance width plus foot width. Each
of these boundaries was reduced significantly by closing the eyes:
anterior boundary = 77.1  7.6% of forefoot length, posterior
boundary = 31.2  10.9% of the hindfoot length, and ML
boundary = 69.0  9.7% of the stance width plus foot width, as was
the area of the functional stability ellipse (363  54 cm2 with EO vs.
321  65 cm2 with EC).

4. Discussion

The functional stability boundary region during bipedal stance

Fig. 1. Example COM position data from four sequential EC LOS trials showing
orientation with respect to feet. Top: Raw data tracing from the 1st trial. Arrows show
direction of motion, beginning at the solid triangle and ending at the solid circle.
Middle: Trimmed data from 1st trial. Same as top, but with data removed from initial
and final leans (to/from center) as well as any time when the heel or ball of either foot
lost contact with the support surface. Bottom: Concatenated trimmed data from all
four trials, with best-fit COP ellipse (solid line) and proposed functional stability limit
(+2SD ellipse; dotted line). Note: the grid lines in each panel are spaced 5 cm apart.
appears to be well described by an ellipse [3,5–7]. Our results
suggest that reproducible, individual functional stability bound-
aries can be easily obtained in about 3–5 min using our proposed
leaning protocol (Fig. 1). While it remains unclear what (if any)
limits the posture control systems use during dynamic activities,
and it seems clear that most people can maintain balance (at least
momentarily) when one or more heels or toes lose contact with the
support surface, we suggest that requiring the feet to be always in
contact with the ground improves reliability of the estimates and
sets a minimal boundary limit that even frail or movement
disordered people can achieve.
The relatively low variability (10–20%) across subjects in the
boundary locations as a percentage of foot/stance dimensions
suggests that it might be possible to define normalized stability
limits that would obviate the need to perform leans testing on
every future subject. However, the current sample was limited to
healthy individuals having relatively narrow ranges of age and
anthropometrics, so these results would not be fully adequate for
that purpose. Also, the observed alignment of the axes of the LOS
ellipses with the cardinal body axes may not hold for all
 K.E. Forth et al. / Gait & Posture 33 (2011) 715–717
populations. Misalignment of these axes might be expected in
those having biomechanical or sensorimotor control problems.
The reduction in functional stability limits during EC trials,
which has been reported previously [3], suggests that these limits
are not purely biomechanical, but are affected by available sensory
information and perhaps by uncertainties or anxieties regarding
the current COM location, biomechanical stability limits, or the
spatio-temporal difference between them. Such context-depen-
dence has been observed in other aspects of sensorimotor control
(e.g., VOR gain), and would likely be affected by other factors (e.g.,
environmental affordances, previous experiences, psychological
factors, etc.). If so, then characteristics of the ellipse (e.g., area)
might be useful indicators of changes in those underlying
uncertainties or anxieties. Indeed, Slobounov et al. [3] and King
et al. [7] have shown that they decrease with age, which is
associated with increases in some of these factors.
We conclude that fall risk assessment by spatial or temporal
stability margin estimates (e.g., TTB) will be more accurate when
estimated using functional (elliptical) boundaries than anatomical
boundaries, particularly for general (oblique) COM motions not
aligned with AP or ML axes. We therefore recommend that
functional stability boundaries be used when the outcome
measure is related to fall risk.
Acknowledgements
Thanks to Laura Taylor, Lance Dean, and Jared Cammon for their
technical help in data collection.
717
Conflict of interest
None of the authors has any financial or personal conflicts of
interest in relation to the submission, other people, or any
organizations.
References
[1] Koozekanani S, Stockwell CW, McGhee RB, Firoozmand F. On the role of
dynamic models in quantitative posturography. IEEE Trans Biomed Eng
1980;27:605–9.
[2] Slobounov SM, Slobounova ES, Newell KM. Virtual time-to-collision and
human postural control. J Motor Behav 1997;29(3):263–81.
[3] Slobounov SM, Moss SA, Slobounova ES, Newell KM. Aging and time to
instability in posture. J Gerontol A Biol Sci Med Sci 1998;53(1):B71–8.
[4] Forth KE, Metter EJ, Paloski WH. Age associated differences in postural
equilibrium control: a comparison between EQscore and minimum time to
contact (TTCmin). Gait Posture 2007;25(1):56–62.
[5] Duarte M, Zatsiorsky VM. Effects of body lean and visual information on the
equilibrium maintenance during stance. Exp Brain Res 2002;146(1):60–9.
[6] Holbein-Jenny MA, McDermott K, Shaw C, Demchak J. Validity of functional
stability limits as a measure of balance in adults aged 23–73 years. Ergonomics
2007;50(5):631–46.
[7] King MB, Judge JO, Wolfson L. Functional base of support decreases with age. J
Gerontol 1994;49(6):M258–63.
[8] Hirvonen TP, Aalto H, Pyykko I. Stability limits for visual feedback posturo-
graphy in vestibular rehabilitation. Acta Otolaryngol Suppl 1997;529:104–7.
[9] Haibach PS, Slobounov SM, Slobounova ES, Newell KM. Virtual time-to-contact
of postural stability boundaries as a function of support surface compliance.
Exp Brain Res 2007;177(4):471–82.
[10] Halir R, Flusser J. Numerically stable direct least squares fitting of ellipses. In:
Proceedings of 6th international conference central Europe Comp Graph
Visualiz; 1998.