Clinical Biomechanics 16 (2001) 213±221

www.elsevier.com/locate/clinbiomech

A dynamical systems investigation of lower extremity coordination

during running over obstacles
Nicholas Stergiou a,*, Jody L. Jensen b, Barry T. Bates c, Shane D. Scholten a,
George Tzetzis d
a
HPER Biomechanics Laboratory, University of Nebraska, 6001 Dodge Street, Omaha, NE 68182-0216, USA
b
Department of Kinesiology and Health Education, The University of Texas, Austin, TX, USA
c
Department of Exercise and Movement Science, University of Oregon, Eugene, OR, USA
d
Department of Physical Education, Aristotle University, Thessaloniki, Greece
Received 20 April 2000; accepted 25 October 2000

Abstract
Objective. To investigate intralimb coordination during running over a level surface and over obstacles of three di€erent heights.
Design. The phasing relationships between the foot and leg motions in the frontal plane, and the shank and thigh motions in the
sagittal plane were used to compare patterns of coordination.
Background. The coordinated actions of lower extremity segments are necessary to absorb the impact forces generated during
running. The behavioral patterns of these segments can be studied under changing task demands using analysis techniques from the
Dynamical Systems Theory.
Methods. Ten subjects ran at their self-selected pace under four conditions: over a level surface and over obstacles of di€erent
heights (5%, 10%, 15% of their standing height). A force platform was used to record impact forces during landing after obstacle
clearance, while kinematics were collected using a two-camera system.
Results. The increases in obstacle height resulted in signi®cant changes in impact forces (34% increase between the two extreme
conditions) and more in-phase relationships between the segments during early stance. No changes were observed in the variability
of the phasing relationships.
Conclusions. The coordination changes observed might be compensatory strategies aimed to reduce forces and potential injury.
However, since the impact forces still increased signi®cantly, it is also possible that the observed changes might be at-risk movement
patterns predisposing runners to injury.

Relevance
Tools from the Dynamical Systems Theory, such as intralimb coordination, can be used as a way to evaluate running mechanics
so that comparisons can be made to various patient populations in subsequent studies. This approach might be a viable alternative
to examine questions in therapeutics. Ó 2001 Elsevier Science Ltd. All rights reserved.

Keywords: Impact forces; Dynamical systems; Lower extremity; Coordination; Obstacle; Running

1. Introduction drome, etc.) are also associated with running [2].

Quantitatively, the frequency of running-induced inju-
Recreational running is a very popular form of ries to the skeletal system is considerable (2.5±12.1/1000
physical activity [1±3]. Bene®ts from running range from h running; [2]). These injuries are classi®ed as overuse
improvement of cardiovascular performance to reducing and involve the knee, leg, ankle, and foot [2,3].
the occurrence of osteoporosis in elderly women [1]. However, as it has been frequently documented in the
Unfortunately, negative side e€ects in terms of muscu- medical literature, there has been little signi®cant pro-
loskeletal injuries (stress fractures, patellar pain syn- gress made in expanding our understanding of running
injury mechanisms [2,4±7]. The medical literature sug-
gests that the lack of single measures to predict speci®c
*
Corresponding author. running injuries may be due to the multifactorial nature
E-mail address: [email protected] (N. Stergiou). of running injuries [2,4,6,8±10]. Running is a complex
0268-0033/01/$ - see front matter Ó 2001 Elsevier Science Ltd. All rights reserved.
PII: S 0 2 6 8 - 0 0 3 3 ( 0 0 ) 0 0 0 9 0 - 5
214 N. Stergiou et al. / Clinical Biomechanics 16 (2001) 213±221

motor skill that involves numerous interacting compo- structures, causing problems associated with excessive
nents or degrees of freedom. It is the mastery of these stress [14,16,19].
degrees of freedom that results in a stable coordinated However, limited research exists in the running liter-
movement. Coordination, then, is de®ned as the process ature where coordination between the interacting seg-
by which the degrees of freedom are organized in time ments has been examined especially under varied
and in sequence to produce a functional movement conditions that can possibly increase impact forces. An
pattern [11,12]. It is perhaps in the patterns of coordi- example of such varied conditions can be the presence of
nation, not individual biomechanical variables that the obstacles in the training path, which is often associated
insight to running injury mechanisms may be found. with increased impact forces. Such perturbation may
In motor control, stable coordination patterns have also produce instability between the actions of the in-
been considered a fundamental feature of consistent, teracting segments or a transition to a new behavioral
functional action [11±13]. An alternative approach to pattern. Either change may reduce the capacity to ab-
understand the construction of, and subsequent change sorb the increased impact forces. Theoretically [16], de-
in, patterns of coordination comes from the Dynamical viations in loading may lead to soft tissue and bone
Systems Theory (DST; 12, 13). Brie¯y, DST proposes pathology if the musculoskeletal system fails to adapt to
that change from one coordinated motor pattern, to a the increased loading demands.
di€erent coordinated pattern is discontinuous and oc- The purpose of this study was to investigate intralimb
curs when a variable to which the neuromotor system is coordination during running over a level surface and
sensitive is scaled up or down through a critical over obstacles of three di€erent heights. To accomplish
threshold. This variable is referred to as a control pa- this purpose, we used DST analysis techniques, and we
rameter and changes in its value cause the neuromotor examined the phasing relationships between the foot and
system to move through di€erent behavioral states. An leg motions in the frontal plane, and the shank and thigh
example comes from the work of Kelso [13]. The task motions in the sagittal plane.
was the alternate ¯exion and extension of the fore®nger
on each hand. The task began with the fore®ngers
pointing in the same direction. Thus, one ®nger was 2. Methods
¯exed while the other was extended. Under slow oscil-
lation speeds, the ®ngers maintained this orientation,
2.1. Subjects
and out-of-phase relationship, with respect to one an-
other. Upon scaling up on the oscillation speed, how-
Ten healthy males (n ˆ 7) and female (n ˆ 3) runners
ever, a behavioral transition occurred such that both
who had been running a minimum of 10 miles per week
®ngers ¯exed at the same time and then went into ex-
for at least one year volunteered as subjects (mean age:
tension at the same time, an in-phase relationship. Just
25.9 years; mean body mass: 74.0 kg; mean height: 177.7
prior to the transition, greater instability was observed
cm). All subjects exhibited a heel±toe footstrike pattern
in the phasing relationship between the ®ngers.
during running at a comfortable self-selected pace. Prior
In running, it is the coordination and phasing rela-
to testing, each subject read and signed an informed
tionships between the actions of the shank and the thigh
consent document approved by the University of Ore-
in the sagittal plane that produce ¯exion and extension
gon Human Subjects Review Board.
at the knee joint. In the frontal plane, the actions of the
leg and the foot have to be coordinated to produce
pronation and supination at the subtalar joint. To un- 2.2. Instrumentation
derstand the adaptation to changing task demands, we
look at the functional patterns of coordination in the A force platform Advanced Mechanical Technology
lower extremity for signs of instability. (AMTI) (Watertown, MA, USA) was used to measure
Furthermore, the actions of knee ¯exion and prona- the vertical ground reaction forces. The force platform
tion occur during the ®rst 50% of the stance period was installed in the middle of a 30 m runway in the
[14,15] and they are important to attenuate impact Biomechanics Laboratory at the University of Oregon.
forces. The large magnitudes of impact forces have been An AMTI signal conditioner/ampli®er was employed in
implicated as a primary cause of running injuries conjunction with the force platform. The signal condi-
[3,6,16±18]. By acting eccentrically, the knee joint mus- tioner/ampli®er was interfaced with an Ariel Perfor-
cles attenuate 70% of the impact forces [17]. Subtalar mance Analysis System (APAS, ARIEL Dynamics,
pronation allows for the impact forces to be absorbed Trabuco Canyon, CA, USA) containing a 32-channel
during a longer period by the supporting structures re- analog to digital sampling module. The APAS was in-
ducing these forces. Without the mechanisms of knee terfaced to an 80386-processor computer. One channel
¯exion and subtalar pronation, these forces would have of the force signal (Fz, vertical component) and one
to be abruptly and directly absorbed by the supporting synchronizing channel, sampled at 1000 Hz.
 N. Stergiou et al. / Clinical Biomechanics 16 (2001) 213±221 215

Kinematic data were collected using two NEC (NEC change their stride length when clearing the obstacle.
USA, Nashville, TN, USA) high speed video cameras The subjects were also instructed to run over the ob-
(200 Hz) interfaced to a real time automated video based stacles and avoid jumping over them, in order to
tracking system (Motion Analysis Corporation, Santa maintain a normal heel±toe running pattern. The ob-
Rosa, CA, USA). The cameras were positioned to ob- stacles were made of light weight wood so that if a
tain a right sagittal and rear (frontal) view of the right subject stepped on or hit the obstacle by mistake while
lower extremity during stance. Camera distances were 14 running, the obstacle was destroyed. This minimized
and 11 m, respectively and each was used in conjunction subjects' fear for tripping and falling. Each condition
with a 10  12 A zoom lens to optimize image size while consisted of 10 trials for a total of 40 trials.
minimizing perspective error. Prior to recording the The order of the presentation of conditions was pre-
movement, re¯ective markers were placed on the sub- determined starting with the no obstacle condition fol-
ject's right lower extremity. Speci®cally, the sagittal view lowed by the obstacle conditions presented from the
markers were placed as follows: (a) lateral malleolus, lowest to the highest obstacle. The rationale for using
(b) knee joint center, and c) greater trochanter. Rear this predetermined order was based on DST, where
(frontal) view markers were placed as described by Ed- scaling up in a continuous fashion of the control pa-
ington et al. [15]. The retrore¯ective images from each rameter (obstacle height) will result in changes re¯ected
camera were obtained and translated to cartesian coor- on the phasing relationships. Therefore, an order e€ect
dinates using a Motion Analysis VP320 video processor is actually desirable [20±22]. In addition, the obstacle
interfaced to an 80486-processor computer. Data col- heights were established based upon the related litera-
lection by the APAS and the video tracking system was ture [23,24].
triggered by a manual transistor/transistor/logic (TTL)
switch to synchronize the video and force data.
3. Data reduction
2.3. Procedures
A typical vertical ground reaction force (Fz) plot
Running speed was monitored over a 3 m interval from a heel-strike runner exhibits three distinct points
using a photocell timing system. Subjects were given [18]: the ®rst maximum value which is the ground re-
time to accommodate to the experimental setup and action impact force (IF), the second maximum value or
to adequately warm-up prior to testing. Warm-up ground reaction active force (AF) and the minimum
consisted of running through the testing area without value between the two maximums (Fmin). These three
concern for stepping on the force platform. During points were identi®ed for each trial by the same inves-
warm-up the subject established a comfortable running tigator using laboratory software. This software allows
pace which was recorded. This speed 5%† was used as identi®cation of both the values and the corresponding
a baseline speed for subsequent testing. Following this times. The IF values and the times from contact to Fmin
procedure a foot placement marker was located ap- (TFmin), to AF (TAF), and to toeo€ (To€) were re-
proximately 10 m before the timed interval to allow for a tained for further analysis. The IF values were normal-
normal right foot contact (FC) on the force platform. ized to body mass, and the mean value was calculated
Each trial consisted of a run of approximately 40 m. for each subject-condition. Group means were also
Data transfer from the cameras to the computer and the calculated for each condition. The TFmin and TAF
qualitative inspection of the force curves allowed for a values were normalized to percent of stance by dividing
1 min inter-trial rest interval. them by To€ values and multiplying the result by 100.
All subjects were asked to run at their previously The normalized TFmin and TAF values were averaged
established baseline pace under four di€erent condi- for each subject-condition and across conditions.
tions. The ®rst condition was running on a level surface The normalized TFmin and TAF values were used to
while the other three conditions were running over ob- identify two distinct periods from the stance phase: the
stacles of three di€erent heights: 5%, 10%, and 15% of impact period which is from contact to Fmin, and the
their standing height. The obstacles were placed directly active period which is from Fmin to AF. Separate ex-
before the force platform so that the subject had to clear amination of each period was utilized since measure-
the obstacle with the right leg and land on the force ments over the entire stance can mask di€erences for a
platform. While the subjects were performing at their single period. Functionally, TAF is synchronous with
self-selected pace, a piece of athletic tape was positioned maximum knee ¯exion [14,15] and it is the transition
one step before the force platform to identify left foot point from braking of the forward motion to propul-
position. When the obstacle was placed on the runway, sion. At this point, the subtalar joint is changing from
the subjects were instructed to hit the tape with their left pronation to supination. TFmin divides the braking of
foot prior to clearing the obstacle with the right leg. the forward motion into two periods. The impact period
Using this procedure insured that the subjects did not is time matched with the occurrence of the impact
216 N. Stergiou et al. / Clinical Biomechanics 16 (2001) 213±221

phenomena [18]. The active period is associated with the aged across trials and mean ensemble curves were gen-
aftermath of impact, the active loading, and the con- erated for all subject-conditions. To statistically test
clusion of the absorption period of the ground reaction di€erences between CRP curves, it was necessary to
forces by the musculoskeletal structures [18]. The focus characterize the curves by single numbers, therefore, two
of this study is on changes in intralimb coordination and additional parameters were calculated using the ensem-
impact forces. Thus, the analysis is limited to the de- ble curves.
pendent variables associated with the time of occurrence The ®rst parameter was the mean absolute value of
of the impact phenomena ± the impact and the active the ensemble CRP curve values (MARP). It was calcu-
period. lated by averaging the absolute values of the ensemble
All kinematic coordinates were scaled and smoothed curve points for the designated periods (impact and
using a Butterworth Low-Pass Filter with a selective cut- active).
o€ algorithm based on Jackson [25]. The cut-o€ values Pp
jU REL:PHASE ji
used were 13±16 and 16±20 Hz for the sagittal and the MARP ˆ iˆ1 ; 1†
p
frontal view coordinates, respectively. Subsequently,
from the frontal plane coordinates, the foot and leg where p is the number of points in each of the two
absolute (regarding the left horizontal) angular positions periods.
and velocities were calculated. From the sagittal plane Functionally, a low MARP value indicates a more in-
coordinates, the shank and thigh absolute (regarding the phase relationship between the two segments' actions for
left horizontal) angular positions and velocities were this condition and for this given subject. The second
calculated. All kinematic parameters were normalized to parameter was the deviation phase (DP) and was cal-
100 points for the stance period using a cubic spline culated by averaging the standard deviations of the en-
routine to enable mean ensemble curves to be derived semble CRP curve points for the designated periods
for each subject-condition. (impact and active).
To examine intralimb coordination the phase por- Pp
SDi
traits for the foot, leg, shank and thigh were generated. DP ˆ iˆ1 ; 2†
p
The phase portrait is a plot of each segment's position
versus its velocity. The phase portrait analysis follows where p is the number of points in each of the two pe-
Rosen's [26] suggestion that the behavior of a dynamical riods.
system may be captured by a variable and its ®rst de- Functionally, a low DP value indicates a less variable
rivative with respect to time. After the phase portraits relationship between the two segments' actions for this
were constructed, the resulting phase plane trajectories condition and for this given subject. The normalized
were used to calculate the phase angles u ˆ tan 1 ‰x0 =xŠ times of TFmin and TGRAF identi®ed from the Fz
[12,13,26,27]. To allow for the calculation of the phase plots were used to calculate the MARP and DP pa-
angles, the phase plots were normalized according to Li rameters for each of the two periods. Group means were
et al. [27]. also calculated for MARP and DP for each segmental
Subsequently, the normalized phase angles of the relationship, for each period, and for each condition.
segments' trajectories were used to examine phasing
relationships. From the frontal plane, the foot and leg 3.1. Statistical analysis
can be viewed as rotating clockwise and counterclock-
wise around the subtalar joint axis, while for the sagittal One-way repeated measures A N O V A s (obstacle height
plane, the shank and the thigh can be viewed as rotating with subjects as the repeated factor) were performed on
clockwise and counterclockwise around the knee joint the subject means for IF, MARP, and DP. For MARP
axis. Continuous relative phase (CRP) represents the and DP, statistical analysis was performed for each co-
phasing relationships or coordination between the ac- ordinative relationship (foot±leg for frontal and shank±
tions of the two interacting segments at every point thigh for sagittal) and for each period (impact and
during a speci®c time period; i.e., it depicts how the two active). In tests that resulted in a signi®cant F -ratio
segments are coupled in their movements while per- (P < 0:05), a Tukey multiple comparison test was used
forming the task. CRP was calculated throughout stance to identify the signi®cant di€erences.
by subtracting the phase angles of the corresponding
segments: uFRONTAL REL: PHASE ˆ uFOOT uLEG and
uSAGITTAL REL: PHASE ˆ uSHANK uTHIGH . Values close 4. Results
to 0° indicate that the two segments are moving in a
similar fashion or in-phase, while values close to 180° The group analysis results are presented in Table 1.
indicate that the two segments are moving in opposite The IF group results were statistically signi®cant, with
directions or out-of-phase. The CRP curves for each the post-hoc analysis revealing statistical di€erences
segmental relationship (frontal and sagittal) were aver- among all possible comparisons. It can be observed that
 N. Stergiou et al. / Clinical Biomechanics 16 (2001) 213±221 217

Table 1
Group means and (SD) for IF, MARP and DPa

Variables No obstacle 5% Obstacle 10% Obstacle 15% Obstacle

IF (N/kg) 17:735%O; 10%O; 15%O 20:4210%O; 15%O

23:2015%O 26.47
(2.24) (2.94) (3.88) (4.97)
Sagittal MARP impact (°) 50:4010%O 45.75 43:8315%O 49.56
(7.27) (9.39) (10.24) (9.24)
Sagittal MARP active (°) 54.03 55.46 56.54 54.31
(7.16) (9.18) (8.42) (7.91)
Frontal MARP impact (°) 77:7510%O; 15%O 68.66 58.41 60.72
(11.94) (9.17) (9.40) (7.98)
Frontal MARP active (°) 20.21 16.61 21.20 24.23
(9.11) (7.84) (9.60) (14.48)
Sagittal DP impact (°) 7.25 7.81 8.33 9.83
(1.79) (2.61) (2.99) (2.68)
Sagittal DP active (°) 7.74 8.39 8.35 9.49
(2.47) (1.97) (3.03) (3.28)
Frontal DP impact (°) 24.53 21.68 23.05 26.69
(7.37) (6.29) (11.51) (12.27)
Frontal DP active (°) 23.12 23.68 20.90 23.89
(8.12) (14.61) (9.94) (12.43)
a
For MARP and DP the values presented are for each coordinative relationship (foot±leg for frontal and shank±thigh for sagittal) and for each
period (impact and active). Condition means that are signi®cantly di€erent P < 0:05† are shown in superscripts.

the higher the obstacle, the greater the IF. The signi®- jectories are more similar geometrically, and the foot
cant increases in IF allowed for the evaluation of the segment changed its direction only once during stance.
other dependent variables over a wide range of impact The frontal CRP ensemble curves of the same subject
force increases. are displayed in Fig. 2. For all conditions, CRP begins
The DP group results were not statistically signi®cant around +100° (Fig. 2). A positive value indicates that
for either the frontal or the sagittal segmental relation- the foot is leading the leg. Toward midstance the two
ships (Table 1). The MARP group results were statisti- segments are in-phase (0°), while during late stance the
cally signi®cant for both frontal and sagittal segmental relationship is reversed with the leg leading. This is in-
relationships during the impact period (Table 1). No dicated by the negative values. The e€ect of the obstacle
statistical di€erences were found for the active period. In in the ®rst portion of stance is quite interesting. In all
the impact period and for the frontal segmental rela- obstacle conditions, CRP goes through zero more than
tionship, the no obstacle condition resulted in the once (Fig. 2). That means that the obstacle caused the
highest value and was statistically di€erent from both two segments to move in-phase in early stance, followed
the 10% and 15% obstacle conditions. For the sagittal, with the foot regaining the lead for a while, before
the lowest value was produced by the 10% obstacle eventually the two segments go through in-phase again
condition and it was statistically di€erent from the 15% and the leg obtaining the lead. These phenomena are
and no obstacle conditions. The greatest MARP value consistent with the emergence of the additional cycles
was produced for the no obstacle condition indicating identi®ed in the leg phase portraits.
that the introduction of the obstacle decreased the Comparing the phase portraits of the sagittal shank
MARP values. and thigh motions, the thigh trajectory shows greater
To better understand the above signi®cant ®ndings change in its geometric form between conditions (Fig. 3).
we looked at the phase portraits of the individual seg- It can be seen that the introduction of the obstacle
ments and the ensemble CRP curves. Comparing the caused an additional cycle to be developed within the
phase portraits of the frontal foot and leg motions, the original cyclic pattern during early stance. This addi-
leg trajectory shows greater change in its geometric form tional cycle increased in size as obstacle height increased
between conditions (Fig. 1). It can also be observed that and showed that the thigh reversed its oscillation twice
an additional cycle is emerging within the original leg during stance. The increasing height of this vertical loop
cyclic pattern. This additional cycle can be clearly ob- just after FC indicates rapid and abrupt changes in the
served in the 5% and 10% obstacle conditions. Every thigh's segmental behavior. The shank trajectories are
time that the trajectory goes through zero a segmental more similar geometrically and without reversals, indi-
reversal is observed. Thus, the leg segment changed its cating a backward only rotation around the knee joint
oscillatory direction twice during stance. The foot tra- during stance.
218 N. Stergiou et al. / Clinical Biomechanics 16 (2001) 213±221

Fig. 1. Phase portraits of the frontal foot (top panel) and leg (bottom panel) motions from a representative subject for all conditions. The ®rst 80% of
stance is plotted. The occurrence of FC is also identi®ed.

relationship is reversed with the shank leading the thigh.

The introduction of the obstacle resulted in changes
mainly in early stance. For all obstacle conditions, rel-
ative phase started at 0° indicating an in-phase rela-
tionship at FC. This result indicates that when the
obstacle was present, both segments moved backwards
at FC, while later the thigh reversed its motion forward
and then backwards again. The formation of the addi-
tional cycles observed for the thigh-phase portraits, also
supports this explanation.

5. Discussion

The purpose of this study was to investigate intralimb

coordination during running over a level surface and
Fig. 2. Frontal (foot±leg) relative phase from the same representative over obstacles of three di€erent heights. To accomplish
subject for all conditions. Each curve is an ensemble average over all
trials. The no obstacle condition is represented by a solid line, while the
this purpose, we used DST analysis techniques.This
obstacle conditions with markers (5% circles; 10% squares; 15% tri- approach requires one to follow a speci®c methodology
angles). to give the DST ideas a concrete meaning [12,13,20±22].
The ®rst step is to characterize the movement patterns
using the appropriate variables, the order parameters.
The sagittal CRP ensemble curves of the same subject Then, it is important to identify control parameters that
are displayed in Fig. 4. For the no obstacle condition, move the system through its behavioral states, so when a
CRP begins around )40° indicating that the thigh is control parameter reaches a critical threshold a transi-
leading the shank (Fig. 4). Toward midstance the two tion to a new coordinative behavior will occur. Changes
segments are in-phase (0°), while during late stance the of the control parameter are re¯ected upon the order
 N. Stergiou et al. / Clinical Biomechanics 16 (2001) 213±221 219

Fig. 3. Phase portraits of the sagittal shank (top panel) and thigh (bottom panel) motions from the same representative subject for all conditions. The
®rst 80% of stance is plotted. The occurrence of FC is also identi®ed.

order parameter. Based upon the literature [12,21,22],

the use of continuous relative phase was the logical
choice since it incorporates both the periodic and the
coupling motion of the segments involved. The height
of an obstacle that a runner had to avoid was used as a
control parameter. This proposal was driven by the
premise that IF would increase as obstacle height was
increased. The results supported our idea and showed
that IF increased signi®cantly with increases in obstacle
height (Table 1).
Subsequently, we examined the order parameters for
any changes caused by the scaling up of the suggested
control parameter. The results revealed changes both
statistically and graphically. The partitioning of the
stance period also assisted in locating the statistical
changes. Both frontal and sagittal relative phase, de-
Fig. 4. Sagittal (shank±thigh) relative phase from the same represen- scribed by MARP, signi®cantly decreased during the
tative subject for all conditions. Each curve is an ensemble average impact period. These decreases indicate that the inter-
over all trials. The no obstacle condition is represented by a solid line,
acting segments became more in-phase, reducing the
while the obstacle conditions with markers (5% circles; 10% squares;
15% triangles). independent action of each segment. Since these changes
occurred during the impact period, they may be related
to the increased IF. Graphically, the appearance of ad-
parameter and therefore reveal the dynamics of the ditional cycles for the leg (frontal) and the thigh (sagit-
system. tal) trajectories might indicate the emergence of new
In the present study, this procedure was used. Thus, coordinated patterns. The CRP values and curves found
the phasing relationship of limb segments (continuous in this study for level running were similar with those
relative phase) within the same leg was used as the presented by other authors [22,27].
220 N. Stergiou et al. / Clinical Biomechanics 16 (2001) 213±221

An interesting observation is that the changes of the window into control processes. Traditional time series
control parameter were not re¯ected statistically on the analysis may not be able to reveal such information. By
order parameter in terms of changes in variability. DP examining the phase portraits of the interacting seg-
was used as a measure to describe the variability of the ments changes in coordination can be observed. The
phasing relationship. The fact that DP did not change usage of phase portraits and subsequently of continuous
but MARP decreased during impact, it suggests that the relative phase, allows the incorporation of both angular
variability of the system remained constant. The system displacement and velocity to examine coordination and
was maintaining itself by maintaining the variability of movement [12,32]. This paper utilized this approach and
the segmental couplings. examined intralimb coordination during running. It was
These changes in the system's behavior can be ac- found that IF increased with increases in obstacle
commodative in nature. Since the IF increased, the heights. The increased IF a€ected the phasing relation-
system has to use some compensatory strategies aimed ships of both the shank±thigh (sagittal) and foot±leg
to reduce forces and potential injury. However, such (frontal) segmental relationships during early stance.
adaptations were probably not sucient in the present
study because IF still increased signi®cantly (17.73±
26.47 N/kg; Table 1). It is well established from the References
footwear related research [18,28,29], that adaptations
are usually the reason for the lack of signi®cant di€er- [1] US Department of Health and Human Services. Physical activity
and health: a report of the surgeon general. Atlanta, GA: US
ences in IF when shoes of various hardness are com- Department of Health and Human Services, 1996.
pared. In such experiments the individuals tested, [2] Van Mechelen W. Running injuries: a review of the epidemio-
usually changed their kinematics to maintain small IF logical literature. Sports Med 1992;14:320±35.
when they run with harder shoes. However, in the pre- [3] Walter SD, Hart LE, McIntosh JM, Sutton JR. The Ontario
sent study the adaptive mechanisms used were not en- cohort study of running-related injuries. Arch Int Med
1989;149:2561±4.
ough because the IF increased drastically. An alternative [4] James SL, Jones DC. Biomechanical aspects of distance running
explanation is that the observed changes of coordination injuries. In: Cavanagh PR, editor. Biomechanics of distance
could be at-risk movement patterns predisposing run- running. Champaign, IL: Human Kinetics; 1990. p. 249±70.
ners to injury. However, it should be mentioned that [5] Lombardo JA. Practice guidelines: which ones to follow? Physi-
comparisons to patient populations should be made to cian Sports Med 1996;24:51±4.
[6] McKenzie DC, Clement DB, Taunton JE. Running shoes,
draw such conclusions and gain insight into possible orthotics, and injuries. Sports Med 1985;2:334±47.
injury mechanisms. [7] Smith R. Where is the wisdom? The poverty of medical evidence.
A limitation of this study concerning the frontal re- Br J Med 1991;303:798±9.
lationships is how representative a two-dimensional [8] Meeuwisse WH. Assessing causation in sport injury: a multifac-
biomechanical evaluation of the subtalar joint is since it torial model. Clin J Sports Med 1994;4:166±70.
[9] Warren BL, Jones CJ. Anatomical factors associated with
is a three-dimensional phenomenon. The literature predicting plantar fasciitis in long-distance runners. Med Sci
[30,31] indicates that the di€erences between the two Sports Exerc 1984;16:60±3.
types of analyses are minimal after FC and through [10] Wen DY, Pu€er JC, Schmalzried TP. Lower extremity alignment
approximately 80% of stance. Di€erences increase as the and risk of overuse injuries in runners. Med Sci Sports Exerc
foot moves out of plane with maximum di€erences oc- 1997;29:1291±8.
[11] Bernstein N. The co-ordination and regulation of movement.
curring during toeo€. Since the focus of this study is Oxford, England: Pergamon Press; 1967.
from early- to mid-stance, a two-dimensional analysis [12] Scholtz JP. Dynamic pattern theory: some implications for
was considered adequate. However, future studies therapeutics. Phys Ther 1990;70:827±43.
should consider validating our results with a three-di- [13] Kelso JAS. Dynamic patterns. Boston, MA: MIT Press; 1995.
mensional analysis. [14] Bates BT, James SL, Osternig LR. Foot function during the
support phase of running. Running 1978;3:24±30.
Both subtalar pronation and knee ¯exion have been [15] Edington CJ, Frederick EC, Cavanagh PR. Rearfoot motion in
presented as mechanisms to decrease IF during running distance running. In: Cavanagh PR, editor. Biomechanics of
[14,16,17,19]. Since IF have been implicated as a major distance running. Champaign, IL: Human Kinetics; 1990. p. 135±
cause of running injuries, it is of great importance to 64.
optimally execute these motions. Improper coordination [16] Chu ML, Yazdani-Ardakani S, Gradisar IA, Askew MJ. An in
vitro simulation study of impulsive force transmission along the
between the actions of the two joints might limit the lower skeletal extremity. J Biomech 1986;19:979±87.
ability of the lower extremity as a shock absorbing [17] Kim W, Voloshin AS, Johnson SH. Modeling of heel strike
system. However, limited research examining the coor- transients during running. Hum Mov Sci 1994;13:221±44.
dinative actions of these two joints has been accom- [18] Nigg BM. Biomechanical aspects of running. In: Nigg BM,
plished. DST has been proposed as an alternative editor. Biomechanics of running shoes. Champaign, IL: Human
Kinetics; 1986. p. 1±25.
approach to therapeutics [12,32,33]. As Winstein and [19] Sangeorzan BJ. Biomechanics of the subtalar joint. In: Stiehl JB,
Gar®nkel [34] suggested a phase plane analysis cannot editor. Inman's joints of the ankle. Baltimore, MD: Williams and
only be used to describe movement, but can provide a Wilkins; 1991. p. 65±73.
 N. Stergiou et al. / Clinical Biomechanics 16 (2001) 213±221 221

[20] Clark JE. On becoming skillful: patterns and constraints. Res Q [29] Bates BT, James SL, Osternig LR, Sawhill JA. An assessment of
Exerc Sport 1995;66:173±83. subject variability, subject-shoe interaction, and the evaluation of
[21] Clark JE, Whitall J. Changing patterns of locomotion. From running shoes using ground reaction force data. J Biomech
walking to skipping. In: Woollacott MH, Shumway-Cook A, 1983;16:181±91.
editors. Development of posture and gait across the life span. [30] Areblad M, Nigg BM, Ekstrand J, Olsson KO, Ekstrom H. Three
Columbia, SC: University Press; 1989. p. 25±47. dimensional measurement of rearfoot motion during running.
[22] Diedrich FJ, Warren Jr WH. The dynamics of gait transitions: J Biomech 1990;23:933±40.
e€ects of grade and load. J Motor Behav 1998;30:60±78. [31] Hamill J, Milliron MJ, Healy JA. Stability and rearfoot motion
[23] Patla AE, Rietdyk S. Visual control of limb trajectory over testing: a proposed standard. In: Proceedings of the Eighth CSB
obstacles during locomotion: e€ect of obstacle height and width. Conference. Calgary, Canada: Canadian Society of Biomechan-
Gait Posture 1993;1:45±60. ics; 1994. p. 324±25.
[24] Warren WH, Young DS, Lee DN. Visual control of step length [32] Hamill J, Van Emmerick REA, Heiderscheit BC, Li L. A
during running over irregular terrain. J Exp Psychol Hum Percept dynamical systems approach to lower extremity running injuries.
Perform 1986;12:259±66. Clin Biomech 1999;14:297±308.
[25] Jackson KM. Fitting of mathematical functions to biomechanical [33] Kamm K, Thelen E, Jensen JL. A dynamical systems approach to
data. IEEE Trans Biomed Eng 1979;26:122±4. motor development. Phys Ther 1990;70:763±75.
[26] Rosen R. Dynamical system theory in biology: vol. 1. Stability [34] Winstein CJ, Gar®nkel A. Qualitative dynamics of disordered
theory and its application. NY: Wiley; 1970. human locomotion: a preliminary investigation. J Motor Behav
[27] Li L, Van Den Bogert ECH, Caldwell GE, Van Emmerick REA, 1989;21:373±91.
Hamill J. Hum Mov Sci 1999;18:67±85.
[28] Nigg BM. Biomechanics, load analysis and sports injuries in the
lower extremities. Sports Med 1985;2:367±79.