Lower Limb Spasticity Assessment Using An Inertial Sensor: A Reliability Study

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Home Search Collections Journals About Contact us My IOPscience

Lower limb spasticity assessment using an inertial sensor: a reliability study

This content has been downloaded from IOPscience. Please scroll down to see the full text.

2013 Physiol. Meas. 34 1423

(http://iopscience.iop.org/0967-3334/34/11/1423)

View the table of contents for this issue, or go to the journal homepage for more

Download details:

IP Address: 194.102.58.157
This content was downloaded on 30/10/2015 at 08:58

Please note that terms and conditions apply.


IOP PUBLISHING PHYSIOLOGICAL MEASUREMENT
Physiol. Meas. 34 (2013) 1423–1434 doi:10.1088/0967-3334/34/11/1423

Lower limb spasticity assessment using an inertial


sensor: a reliability study
I Sterpi 1 , A Caroli 2 , E Meazza 2 , G Maggioni 2 , C Pistarini 2
and R Colombo 1
1 Bioengineering Service, ‘Salvatore Maugeri’ Foundation, IRCCS, Rehabilitation Institute of

Pavia, Via Salvatore Maugeri 10, 27100 Pavia, Italy


2 Neurorehabilitation and Coma Units, ‘Salvatore Maugeri’ Foundation, IRCCS, Rehabilitation

Institute of Pavia, Via Salvatore Maugeri 10, 27100 Pavia, Italy

E-mail: irma.sterpi@fsm.it, roberto.colombo@fsm.it, alberto.caroli@fsm.it,


elisa.meazza@fsm.it, giorgio.maggioni@fsm.it and caterina.pistarini@fsm.it

Received 17 June 2013, accepted for publication 9 September 2013


Published 8 October 2013
Online at stacks.iop.org/PM/34/1423

Abstract
Spasticity is a common motor impairment in patients with neurological
disorders that can prevent functional recovery after rehabilitation. In the clinical
setting, its assessment is carried out using standardized clinical scales. The
aim of this study was to verify the applicability of inertial sensors for an
objective measurement of quadriceps spasticity and evaluate its test–retest and
inter-rater reliability during the implementation of the Wartenberg pendulum
test. Ten healthy subjects and 11 patients in vegetative state with severe brain
damage were enrolled in this study. Subjects were evaluated three times on three
consecutive days. The test–retest reliability of measurement was assessed in the
first two days. The third day was devoted to inter-rater reliability assessment.
In addition, the lower limb muscle tone was bilaterally evaluated at the knee
joint by the modified Ashworth scale. The factorial ANOVA analysis showed
that the implemented method allowed us to discriminate between healthy and
pathological conditions. The fairly low SEM and high ICC values obtained for
the pendulum parameters indicated a good test–retest and inter-rater reliability
of measurement. This study shows that an inertial sensor can be reliably used to
characterize leg kinematics during the Wartenberg pendulum test and provide
quantitative evaluation of quadriceps spasticity.

Keywords: spasticity, reliability, Wartenberg test, inertial sensor, neurorehabil-


itation
(Some figures may appear in colour only in the online journal)

0967-3334/13/111423+12$33.00 © 2013 Institute of Physics and Engineering in Medicine Printed in the UK & the USA 1423
1424 I Sterpi et al

1. Introduction

Spasticity is a common motor impairment in patients following stroke (Simpson et al 2009),


spinal cord injury (Kirshblum 1999), traumatic brain injury (Sheean 2002), multiple sclerosis
(Mayer 1997) and cerebral palsy (Fowler et al 2000). It is described as a motor disorder
characterized by a velocity-dependent increase of the tonic stretch reflexes (muscle tone) with
exaggerated tendon jerks, resulting from hyperexcitability of the muscle spindle (Lance 1980).
Spasticity hinders voluntary movements, and prolonged spasticity may lead to pain and severe
reduction of joint range of motion. Physicians and therapists in their clinical setting come up
against the problem of spasticity daily and although it is generally agreed that it is easy to
recognize, it is not easy to quantify (Biering-Sørensen et al 2006). It is, therefore, important
to be able to measure spasticity accurately and objectively in order to plan, deliver and
monitor appropriate rehabilitation strategies. Often, the measurement of spasticity influences
clinical decision making for the application of pharmacological treatments such as baclofen
and botulinum toxin. Although the literature offers a variety of alternatives for measuring
spasticity (Biering-Sørensen et al 2006), none of these appear to be widely used in clinical
practice, and the Ashworth scale (AS) (Ashworth 1964) and its modified version (MAS)
(Bohannon and Smith 1987) remain the most diffused evaluation methods (Kumar et al 2006).
Both scales determine the amount of resistance felt during the passive displacement of a
limb in a specific joint. Whilst they are simple to apply, the evaluations obtained are neither
quantitative nor objective and can vary greatly from examiner to examiner. Moreover, both
AS and MAS have poor sensitivity. Fleuren et al (2010) recently showed that the AS is not
reliable nor valid and even recommended against its further use. In spite of this, clinicians
are unwilling to abandon the use of these subjective measures in the absence of a reasonable
alternative.
The Tardieu scale has been suggested as a more suitable alternative to AS and MAS
as it assesses and compares the response of the muscle to passive movement at both slow
and fast speeds (Boyd and Graham 1999). However, a recent literature review pointed out
that at present, publications regarding the validity and reliability of the Tardieu scale are
scarce, with none relating to the adult population (Haugh et al 2006). An alternative and
more practical method to evaluate quadriceps spasticity is to generate movement by gravity, as
Wartenberg did with his pendulum test for quantifying lower limb hypertonia in neurological
diseases (Wartenberg 1951). The pendulum test is a simple biomechanical method useful for
quadriceps spasticity evaluation. It quantifies the reduction of swing due to spasticity during
the gravity induced pendulum-like movement of the lower limb (see details in section 2). It
is non-invasive, quick and easy to apply. During the test, the shank of the leg is horizontally
extended and subsequently allowed to hang freely.
In the literature, oscillations of knee angle during the pendulum test have typically been
captured using electrogoniometers (Brown et al 1988, Fowler et al 2000) or 3D motion
analysis systems (ultrasonic devices (Valle et al 2006); optical systems (Syczewska et al
2009, Stillman and McMeeken 1995); electromagnetic systems (Bohannon et al 2009)).
Electrogoniometers are low-cost sensors that can be easily applied for the execution of the
pendulum test and for other biomechanical evaluation tests. They have moderate test–retest
reliability (Isacson et al 1986, Piriyaprasarth et al 2008, Rome and Cowieson 1996), but the
difficulty in repositioning the sensor requires trained personnel and a strictly standardized
protocol to obtain a reproducible measurement. Likely due to this fact, published studies that
use electrogoniometers for measurements of body motions and angles report poor or moderate
intra-rater reliability (Piriyaprasarth et al 2008, Rome and Cowieson 1996). Motion analysis
systems are more reliable, but they have a high cost and require a specific laboratory setup. As a
Lower limb spasticity assessment using an inertial sensor 1425

result, their application is labor-intensive and the diffusion of the pendulum test as an evaluation
method of spasticity is not so high. In addition, most of the published studies reported the
reliability analysis of the pendulum test on the same day but with contrasting results (reliability
varied from good to bad) (Fowler et al 2000, Syczewska et al 2009, Bohannon 1987, Bohannon
et al 2009, Nordmark and Anderson 2002), and apparently only two studies reported the inter-
day repeatability of the test in pathological subjects (El-Gohary and McNames 2012, Paulis
et al 2011). Moreover, we could not find any study in the literature evaluating the inter-
rater reliability of the pendulum test. However, even if in the literature we can find several
applications of the pendulum test, the most diffused method to evaluate spasticity in clinical
practice remains the MAS.
Inertial measurement units (IMU) are recently developed devices that can be used in a
wide range of applications and, being highly accurate, small and lightweight, they can be
considered an ideal 3D human motion tracker. Recently, Van den Noort et al (2009, 2010)
validated the use of an IMU for the quantitative evaluation of spasticity in children with cerebral
palsy using the spasticity test (Scholtes et al 2007). The authors concluded that inertial sensors
yield a superior performance to goniometers mainly due to the repositioning problems of the
latter. El-Gohary and McNames (2012) compared shoulder and elbow joint angles estimated
by an inertial system to those estimated by an optical tracking system. They found that, during
large movements, errors between optical and inertial angle are minimal (less than 8◦ ) with an
excellent average correlation coefficient (r  0.95) (El-Gohary and McNames 2012). Paulis
et al (2011) evaluated their reliability for the Tardieu scale with electrogoniometers and with
IMU in elbow flexors of stroke patients. They showed that the inter-rater reliability was better
for IMU than for goniometry.
The aim of this study is to verify the applicability of inertial sensors for measuring
quadriceps spasticity and evaluate their test–retest and inter-rater reliability during the
implementation of the Wartenberg pendulum test. We hypothesized that IMU can offer a
reliable measure of spasticity and, in combination with the pendulum test, can provide an easy
and economic method suitable for extended clinical application.

2. Methods

2.1. Subjects
Ten healthy subjects (five male, five female) and 11 patients in the vegetative state (six male,
five female) with severe brain damage secondary to traumatic lesions (n = 3), hypoxic lesion
(n = 2) or vascular stroke (n = 6) were enrolled in this study (table 1). All subjects were
examined at the Neurorehabilitation and Coma Unit.
Inclusion criteria were the presence of a severe acquired brain injury, no musculoskeletal
impairment, no sedation within the 24 h prior to testing and no use of anti-spasticity drugs.
Exclusion criteria were the appearance of tremors and other patient reactions disturbing test
execution. The study and the measurement protocol were approved by the local scientific and
ethics committees. Informed consent to participate in the study was obtained from the family
members of patients and from the healthy subjects in person.

2.2. Testing protocol


Subjects were comfortably positioned supine on the padded examination table with the trunk
placed at approximately 40◦ from the horizontal plane and knees poised on the edge of the
table so that the shanks were hanging freely. The examiner grasped the patient’s foot and
1426 I Sterpi et al

Table 1. Characteristics of 11 patients in the vegetative state.


Weight Height MAS MAS
Patient ID Age Sex (kg) (cm) Injury (right leg) (left leg)

P1 61 F 55 168 Vascular stroke 3 2


P2 39 M 55 187 Traumatic lesions 3 2
P3 28 M 45 170 Traumatic lesions 2 2
P4 23 M 55 184 Traumatic lesions 3 3
P5 54 F 58 168 Vascular stroke 2 2
P6 18 M 58 169 Vascular stroke 0 0
P7 65 F 69 148 Vascular stroke 2 3
P8 61 M 60 176 Hypoxic lesion 0 0
P9 66 M 66 165 Vascular stroke 2 3
P10 55 F 40 170 Vascular stroke 3 4
P11 45 F 51 178 Hypoxic lesion 3 3

extended the shank under evaluation to the horizontal position or alternatively to the maximal
extension position allowed by the subject’s knee joint. The shank was suddenly released
and allowed to oscillate freely under the action of gravity until it stopped. For each subject,
spasticity was evaluated in both lower limbs. During the evaluation procedure, three successful
measurements were performed for each leg. There was a resting period of 90 s between two
consecutive measurements in order to allow complete muscle recovery and relaxation.
The test–retest reliability of measurement was assessed by performing two test sessions
on consecutive days under the supervision of the same therapist. The third day, a second
therapist performed an additional test session to allow the evaluation of inter-rater reliability.
For each patient, the two physiotherapists were randomly selected out of a team of five trained
therapists. Acquisitions were always performed at the same time of day. In addition, quadriceps
spasticity was bilaterally evaluated by the MAS prior to performing the pendulum test on the
first evaluation day. All patients were clinically assessed by the same therapist.

2.3. Recording system


Shank movements during the pendulum test were recorded using an inertial sensor (MTx,
Xsens Technologies, Enschede, The Netherlands). It is a compact wearable device that contains
a triaxial accelerometer, gyroscope and magnetometer. The sensor was placed on the external
side of the shank proximal to the knee joint, at the head of the fibula. It was fixed to the shank
through a soft VelcroR
strap provided with a dedicated pocket to prevent change of orientation
of the sensor. Thanks to this pocket, we could minimize the misalignment of the sensor’s x axis
with the long axis of the fibula, thereby allowing the approximation of the knee flexion angle
in the sagittal plane with the x axis sensor angle in the same plane. In any case, the examiner
carefully verified this alignment so as to minimize variations in sensor placement caused by
differences in anatomy. Data were acquired with a sample frequency of 100 Hz. Figure 1(a)
shows the inertial sensor, the physiotherapist and one subject during the execution of the test.
The angle of the knee joint was measured by integrating the angular velocity signal data in the
sagittal plane.

2.4. Parameters
In accordance with Bohannon et al (2009) and White et al (2007), in order to precisely quantify
the pendulum kinematics, we computed the following set of six parameters during the swing
of the knee angle.
Lower limb spasticity assessment using an inertial sensor 1427

(a) (b)

Figure 1. (a) Laboratory setting during the execution of the Wartenberg pendulum test. At the
bottom-right is shown a detail of the inertial sensor applied in our tests. (b) Theoretical shank
oscillation in degrees during the Wartenberg pendulum test with quantitative parameters identified.
The x axis of the sensor is aligned with the long axis of the fibula.

(i) Initial angle (IA) (deg): knee angle at the beginning of the test during maximal limb
extension.
(ii) Resting angle (RA) (deg): knee angle at the end of oscillations.
(iii) First angle of reversal (FAR) (deg): knee angle when shank motion first switched from
flexion to extension.
(iv) Area under the curve (AUC) (deg × s): area between the knee angle during oscillations
and the resting angle; it is the integral of the absolute value of the knee angle ϕ(t) in the
sagittal plane during motion:
 T
AUC = |ϕ(t ) − RA| dt
0
where T = 12 s.
(v) Velocity to first reversal (VFR) (deg s−1): the change in knee angle between the starting
position and the first reversal divided by the time to first reversal:
FAR
VFR = .
TFR
(vi) Time to first reversal (TFR) (s): time interval between the start of shank motion and the
first reversal from flexion to extension.
Figure 1(b) shows a typical sagittal knee angle trace (ϕ(t)) obtained during the pendulum
test, with the measured parameters highlighted.

2.5. Data analysis and statistics


The intrasession coefficient of variation (CVi = standard deviation divided by mean) was
calculated across the three repetitions performed during the same examination session, to test
the repeatability of measures. Subsequently, the three measures of each day were averaged
and the analysis of reliability was performed using these mean values. In order to verify if the
method we implemented was able to discriminate between healthy and pathological subjects,
factorial analysis of variance (ANOVA) was conducted by pooling data of all subjects.
Then, the variability of measurements obtained on three different days was assessed by
repeated measures ANOVA. For each measurement condition, the null hypothesis was no
significant difference. A significance level of 0.05 was adopted for the statistical tests. Then,
the Bonferroni correction was applied to adjust the significance level for multiple comparisons.
1428 I Sterpi et al

(a) (b)

Figure 2. Knee angle during the pendulum test. Typical example of the pendulum test traces for
one healthy subject (panel (a)) and one patient (panel (b)). The dashed and solid lines correspond
respectively to the right and left leg swings. Panel (b) clearly illustrates the reduced oscillatory
pattern due to the patient’s increased leg stiffness.

Subsequently, the test–retest and inter-rater reliability for the six parameters were evaluated by
computing the intraclass correlation coefficients (ICC) and the standard error of measurement
(SEM) (Fleiss 1986, Weir 2005). For each parameter, the CV of the SEM (CV = SEM divided
by mean) (Brodie et al 2009) was also computed by expressing the SEM value as a percentage
of the overall mean of the parameter.√ From the SEM, the minimal detectable change (MDC)
was calculated as MDC = 1.96 × 2 × SEM. Finally, in order to allow comparison of our
results with those of other systems reported in the literature, the results from the pendulum test
of the patients’ group were correlated with the MAS values of the lower limb using Spearman’s
rank correlation coefficient. Additionally the correlation coefficients between the six indices
were calculated using Pearson’s correlation matrix.
Statistical analysis was performed using the StatView R
statistical package (SAS Institute,

R
Cary, NC, USA) and MATLAB 7.5 (R2007b, The MathWorks Inc., Natick, MA, USA)
software.

3. Results

Two out of ten healthy subjects were not able to relax one of their legs, as assessed with muscle
palpation by the examiner, and consequently the leg stopped autonomously during the test
execution. For this reason, they were excluded from analysis. One patient was enrolled only
in the test–retest reliability protocol due to the appearance of clinical complications. Another
patient experienced tremor and discomfort during the second evaluation session and therefore
only the inter-rater reliability could be tested.
For each subject and examination session, we computed the coefficient of variation for
the six parameters; the mean values of CVi were in the worst case lower than 10%. Based on
this result, we decided to average the values of the same session for further analysis.
The results of the factorial ANOVA showed that the Wartenberg pendulum test was able
to distinguish spasticity between healthy and pathological subjects; all six parameters differed
significantly between the two groups (p < 0.001). Figure 2 presents an example of the typical
pendulum test traces for one healthy subject and one patient.
The shank movement in the healthy subject (figure 2(a)) is characterized by a smooth
swing with a low damping factor, so a number of oscillations can be seen in the traces. Of note,
Lower limb spasticity assessment using an inertial sensor 1429

Table 2. Test–retest reliability results for the six Wartenberg pendulum test parameters.
Test–retest
Test Retest p-val SEM MDC ICC (95% CI)

IA (deg) 25.97 ± 8.60 28.16 ± 9.07 0.12 5.78 16.02 0.71(0.44–0.85)


RA (deg) 63.57 ± 11.04 59.22 ± 10.57 0.001 – – –
FAR (deg) 90.17 ± 22.12 84.66 ± 20.25 0.01 8.48 23.51 0.90(0.80–0.95)
AUC (deg × s) 84.72 ± 37.43 88.11 ± 49.19 0.48 19.94 55.27 0.89(0.78–0.94)
VFR (deg s−1) 163.22 ± 29.51 154.46 ± 28.24 0.03 16.93 46.93 0.77(0.55–0.88)
TFR (s) 0.549 ± 0.064 0.545 ± 0.072 0.62 0.032 0.09 0.91(0.83–0.96)

Table 3. Inter-rater reliability results for the six Wartenberg pendulum test parameters.
Inter-rater
Rater 1 Rater 2 p-val SEM MDC ICC (95% CI)
IA (deg) 25.95 ± 8.61 23.19 ± 8.40 0.11 7.11 19.71 0.44(0.10–0.71)
RA (deg) 63.10 ± 11.97 63.83 ± 11.60 0.64 6.45 17.88 0.83(0.66–0.91)
FAR (deg) 90.31 ± 22.71 91.73 ± 22.03 0.48 8.49 23.53 0.92(0.85–0.96)
AUC (deg × s) 79.92 ± 35.03 80.87 ± 33.89 0.74 12.03 33.35 0.94(0.88–0.97)
VFR (deg s−1) 162.17 ± 33.55 168.63 ± 33.18 0.22 21.95 60.84 0.72(0.45–0.86)
TFR (s) 0.563 ± 0.078 0.543 ± 0.074 0.21 0.063 0.17 0.37(0.23–0.68)

the amplitude and number of shank oscillations in the patient (figure 2(b)) were lower than
those in the healthy subject. It is worth noting that patients generally had a different level of
impairment in each leg, as illustrated in figure 2(b). For the reliability study, the data acquired
from both healthy and pathological subjects were pooled together.

3.1. Test–retest reliability

Table 2 shows the test–retest reliability results for the six pendulum parameters of all subjects.
The repeated measures ANOVA, performed for each variable to rule out a possible systematic
error between the two consecutive days of trial, revealed significant systematic errors for
RA, thereby nullifying the computation of the ICC and SEM; therefore, their values were
not included in the table for this parameter (Weir 2005). The remaining five parameters were
characterized by high and significant (p < 0.001) ICC values ranging from 0.71 to 0.91; the
VFR was the parameter with the best performance. Overall, for the six parameters, the CV of
SEM ranged from 5% to 20%. The SEM values of angular parameters (IA, RA, FAR) were
quite small (<9◦ ).

3.2. Inter-rater reliability

Table 3 summarizes the inter-rater reliability results for the pendulum parameters evaluated.
Repeated measures ANOVA revealed no significant systematic error for all parameters. ICC
values were all statistically significant but only RA, FAR, AUC and VFR could reach a
consistent value (> 0.7). The highest reliability index (ICC) was found for the AUC parameter
(ICC = 0.94). Inter-rater SEM values of the angular parameters were once again lower than
10◦ . The CV of SEM of the other three parameters ranged from 7% to 18%.
1430 I Sterpi et al

Table 4. Pearson’s correlation coefficients matrix for the six Wartenberg pendulum test parameters
(first six rows) and Spearman’s rank correlation coefficients between the six pendulum test
parameters and the modified Ashworth scale (MAS) (last row) in 11 patients in the vegetative
state.
IA (deg) RA (deg) FAR (deg) AUC (deg × s) VFR (deg s−1) TFR (s) MAS

IA (deg) 1 – – – – – –
RA (deg) −0.26 1 – – – – –

FAR (deg) −0.27 0.85 1 – – – –
∗ ∗
AUC (deg × s) 0.28 −0.76 −0.67 1 – – –
−1 ∗ ∗ ∗
VFR (deg s ) 0.10 −0.74 −0.93 0.53 1 – –
∗ ∗ ∗ ∗
TFR (s) 0.503 −0.61 −0.70 0.55 0.45 1 –
∗ ∗
MAS 0.01 0.47 0.43 −0.18 −0.36 0.01 1

p-value < 0.05.

3.3. Correlation between the MAS and quantitative parameters


None of the healthy subjects demonstrated increased quadriceps muscle tone during the MAS
testing. Table 4 reports the correlation matrix between the modified Ashworth clinical scale
and the six Wartenberg pendulum test parameters evaluated in the patients’ group. Spearman’s
correlation between the six pendulum test parameters and the MAS showed a significant
but poor correlation with a maximum correlation of 0.47 obtained for the RA parameter.
In contrast, the correlations between the diverse kinematic parameters apart from IA were
statistically significant and ranged from moderate to high values.

4. Discussion

The purpose of this study was to examine the reliability of the Wartenberg pendulum test
when the swings of the knee joint are captured by means of an inertial sensor (IMU-based
pendulum test). The findings demonstrate the ability of this method to discriminate between
healthy and pathological conditions. In addition, high test–retest and inter-rater reliability of
measure were found for most of the computed parameters, thereby demonstrating the validity
of this measurement approach.
Our study confirmed the test–retest reliability of measure with good to excellent ICC
values and quite small SEM values, which are not usual for a clinical setting application.
We measured the sagittal knee angle, since movement will mainly be around the medio-
lateral knee joint axis. White et al (2007) evaluated the three-dimensional pendulum test in
able-bodied and cerebral palsy children and concluded that, because of the relatively small
magnitude of frontal and transverse plane motions, three-dimensional motion analysis may
not be required to perform the pendulum test.
The IA exhibit an ICC of 0.71 probably because of low between-subjects variability.
Conversely, the IA SEM was quite small (<6◦ ) and highly acceptable from a clinical point
of view. This error might be due to an objective difficulty on the part of the therapist in
maximally extending the patient’s leg in a reproducible way and obtaining the same level of
muscle relaxation across different sessions. In order to compensate for this error and obtain the
same initial conditions, the inclusion of the electromyography (EMG) signal to monitor the
baseline activity of the quadriceps muscles could have been beneficial. It could also be a useful
tool to better standardize the test when the evaluation is carried out by different therapists. The
good inter-rater reliability was confirmed for RA, FAR and AUC; in particular, ICC and SEM
values were comparable with those obtained in the test–retest protocol. It is worth recalling the
Lower limb spasticity assessment using an inertial sensor 1431

information that the ICC and SEM convey about the reliability of a measure. ICC is a measure
of relative reliability and reflects the ability of a measurement to discriminate between different
subjects. SEM represents the absolute reliability and provides information about the random
variation within an individual. If we look at tables 2 and 3, we can see that the maximum ICC
values were obtained by the AUC and FAR parameters. In addition, the AUC (deg × s) SEM
values observed in our study seem lower compared with the AUC variability reported by Bajd
and Vodovnik (1984) and Stillman and McMeeken (1995).
This would suggest that the AUC and the first swing excursion (FAR) are the most
appropriate parameters for evaluating spasticity among subjects. This claim is in agreement
with other findings in the literature (Fowler et al 2000, Johnson 2002, Le Cavorzin et al 2002).
The MDC was relatively small for all parameters except IA and AUC. In the case of
IA, this is probably because the parameter actually has poor reliability. Conversely, AUC has
excellent reliability but an MDC that seems a bit high. It represents the smallest change that
can be considered above the measurement error with a given level of confidence (95% in our
case). In practice, this high value could depend on the time window selected (12 s) for AUC
calculation, thus suggesting the need for refinement of its definition.
In summary, based on our results we can say that the pendulum test executed by an IMU
sensor is reliable both across time and across raters. The comparison between the six pendulum
test parameters and the MAS showed a significant but poor to moderate correlation (table 4).
This result is in agreement with the findings of Nordmark and Anderson (2002). It might be
explained by the clustering effect on the MAS, due to the fact that the therapists usually tend
to group the majority of evaluations in grades 2 and 3 of the clinical scale (table 1). Although
many studies have shown that metrics characteristics of the MAS are unsatisfactory for the
assessment of spasticity, we compared the six pendulum test parameters and the MAS, on
account of its wide diffusion in the clinical practice for evaluation of spasticity (Pandyan et al
1999, Platz et al 2005). However, our results confirm that the MAS might not be sensitive
enough to detect small changes in spasticity thereby eliminating its application to highlight
intervention-related effects (Damiano et al 2002, Leonard et al 2001, Katz et al 1992).
Similar to the findings of Bohannon et al (2009), the correlation coefficients (table 4)
between the diverse measured kinematic parameters were statistically significant and ranged
from moderate to high. This finding suggests that these parameters should reflect the
measurement of a unique underlying variable. In contrast, the IA parameter exhibited poor
correlation with the remaining parameters. This was likely due to the fact that it is a static
measure and therefore it evaluates only the limitation in the joint’s range of motion due to
spasticity and not its speed-related component. Based on these results, we hypothesize that in
subsequent studies, it could be useful to identify a reduced set of parameters to quantify the
level of spasticity. In accordance with Bajd and Vodovnik (1984), Le Cavorzin et al (2002)
and White et al (2007), we confirm that AUC is a good and sensitive indicator of change
in quadriceps spasticity. Furthermore, in consideration of the findings of Fowler et al (2000)
showing that the FAR was the most sensitive measure of spasticity in children with cerebral
palsy, we suggest that AUC and FAR parameters are the best candidates to evaluate quadriceps
spasticity because of their reliability properties.
The availability of a reliable assessment method opens the way to its application for the
evaluation of changes in spasticity during pharmacological treatments (baclofen, botulinum
toxin), in particular for dosage adjustment purposes (Bianchi et al 1999), during surgical
intervention (selective dorsal rhizotomy) (Nordmark and Anderson 2002) and during local
treatment of the quadriceps musculature. In addition, the simplicity of use, the portability and
the low cost of the IMU sensor should play a key role in promoting its extended application
in clinical practice. Wartenberg’s test was proposed for the spasticity assessment of the knee
1432 I Sterpi et al

joint, but it would be cumbersome to apply it to other joints such as the elbow joint. The main
difficulties include the smallness of forearm inertia and the uncomfortable posture. However,
with appropriate modifications it could also be applied to other joints, as reported in the recent
literature (Vodovnik et al 1984, Lin et al 2003, Huang et al 2009). The major limitation of the
pendulum test is its dependence on the subject’s sitting position and relaxation capacity. For
this reason, a well-standardized protocol is necessary to ensure minimal posture errors and to
avoid voluntary activation of the investigated muscles. Additional precautions could include
the advisability of monitoring EMG activity of the involved muscles in order to verify the
subject’s relaxation (Bajd and Vodovnik 1984, Fowler et al 2000, Lin and Rymer 1991) even
if this should not be considered a priority in rapid routine tests (Bajd and Vodovnik 1984).
Additionally, to better validate the measure of the IMU-based pendulum test, future studies
should investigate the correlation between this test (biomechanical spasticity measurement)
and electrophysiological spasticity measurements such as the H-reflex.
Finally, modern cell phones include sensors similar to the IMU sensor and have been shown
to provide a reliable measurement of knee joint goniometry in static conditions (Ferriero et al
2013). Therefore, it is plausible that in the future a simple application could be used to perform
the pendulum test at even lower cost with readily available hardware.

5. Conclusion

This study shows that the implementation of the Wartenberg pendulum test using an inertial
sensor can provide quantitative evaluation of spasticity of the quadriceps’ femoris muscles,
allowing us to discriminate between healthy and pathological conditions. The fairly low SEM
and high ICC values obtained for some parameters describing knee joint kinematics indicated
a good test–retest and inter-rater reliability of measurement. The proposed experimental
setup allows easy application, portability and reliability of this measurement method. These
properties, in combination with the low cost of the system, should make the method suitable
for extended clinical application.

Acknowledgments

We would like to thank Pt Mattia Malagni for his contribution to the data collection. This
research was partly supported by the project ‘5 × 1000–2009’ funded by the Italian Ministry
of Health.
The authors do not have any actual or potential conflict of interests to declare in relation
to the content of this paper.

References

Ashworth B 1964 Preliminary trial of carisoprodol in multiple sclerosis Practitioner 192 540–2
Bajd T and Vodovnik L 1984 Pendulum testing of spasticity J. Biomed. Eng. 6 9–16
Bianchi L, Monaldi F, Paolucci S, Iani C and Lacquaniti F 1999 Quantitative analysis of the pendulum test: application
to multiple sclerosis patients treated with botulinum toxin Funct. Neurol. 14 79–92
Biering-Sørensen F, Nielsen J B and Klinge K 2006 Spasticity-assessment: a review Spinal Cord 44 708–22
Bohannon R W 1987 Variability and reliability of the pendulum test for spasticity using a Cybex II isokinetic
dynamometer Phys. Ther. 67 659–61
Bohannon R W, Harrison S and Kinsella-Shaw J 2009 Reliability and validity of pendulum test measures of spasticity
obtained with the Polhemus tracking system from patients with chronic stroke J. Neuroeng. Rehabil. 6 30
Bohannon R W and Smith M B 1987 Interrater reliability of a modified Ashworth scale of muscle spasticity Phys.
Ther. 67 206–7
Lower limb spasticity assessment using an inertial sensor 1433

Boyd R N and Graham H K 1999 Objective measurement of clinical findings in the use of botulinum toxin type A
for the management of children with cerebral palsy Eur. J. Neurol. 6 s23–35
Brodie F G, Atkins E R, Robinson T G and Panerai R B 2009 Reliability of dynamic cerebral autoregulation
measurement using spontaneous fluctuations in blood pressure Clin. Sci. 116 513–20
Brown R A, Lawson D A, Leslie G C and Part N J 1988 Observations on the applicability of the Wartenberg pendulum
test to healthy, elderly subjects J. Neurol. Neurosurg. Psychiatry 51 1171–7
Damiano D L, Quinlivan J M, Owen B F, Payne P, Nelson K C and Abel M F 2002 What does the Ashworth scale
really measure and are instrumented measures more valid and precise? Dev. Med. Child Neurol. 44 112–8
El-Gohary M and McNames J 2012 Shoulder and elbow joint angle tracking with inertial sensors IEEE Trans. Biomed.
Eng. 59 2635–41
Ferriero G, Vercelli S, Sartorio F, Muñoz Lasa S, Ilieva E, Brigatti E, Ruella C and Foti C 2013 Reliability of a
smartphone-based goniometer for knee joint goniometry Int. J. Rehabil. Res. 36 146–51
Fleiss J L 1986 The Design and Analysis of Clinical Experiments (Wiley Series in Probability and Statistics)
(New York: Wiley)
Fleuren J F M, Voerman G E, Erren-Wolters C V, Snoek G J, Rietman J S, Hermens H J and Nene A V 2010 Stop
using the Ashworth scale for the assessment of spasticity J. Neurol. Neurosurg. Psychiatry 81 46–52
Fowler E G, Nwigwe A I and Ho T W 2000 Sensitivity of the pendulum test for assessing spasticity in persons with
cerebral palsy Dev. Med. Child Neurol. 42 182–9
Haugh A B, Pandyan A D and Johnson G R 2006 A systematic review of the Tardieu scale for the measurement of
spasticity Disabil. Rehabil. 28 899–907
Huang H-W, Ju M-S, Wang W-C and Lin C-C K 2009 Muscle tone of upper limbs evaluated by quantitative pendulum
test in patients with acute cerebellar stroke Acta Neurol. Taiwan 18 250–4
Isacson J, Gransberg L and Knutsson E 1986 Three-dimensional electrogoniometric gait recording
J. Biomech. 19 627–35
Johnson G R 2002 Outcome measures of spasticity Eur. J. Neurol. Suppl. 1 9 53–61 10–16 (see also ‘A panel
discussion: guidelines of best practice’ (chaired by Professor M Barnes))
Katz R T, Rovai G P, Brait C and Rymer W Z 1992 Objective quantification of spastic hypertonia: correlation with
clinical findings Arch. Phys. Med. Rehabil. 73 339–47
Kirshblum S 1999 Treatment alternatives for spinal cord injury related spasticity J. Spinal Cord Med. 22 199–217
Kumar R T S, Pandyan A D and Sharma A K 2006 Biomechanical measurement of post-stroke spasticity Age
Ageing 35 371–5
Lance J W 1980 The control of muscle tone, reflexes, and movement: Robert Wartenberg lecture
Neurology 30 1303–13
Le Cavorzin P, Hernot X, Bartier O, Carrault G, Chagneau F, Gallien P, Allain H and Rochcongar P 2002 Evaluation
of pendulum testing of spasticity Ann. Readapt. Med. Phys. 45 510–6
Leonard C T, Stephens J U and Stroppel S L 2001 Assessing the spastic condition of individuals with upper
motoneuron involvement: validity of the myotonometer Arch. Phys. Med. Rehabil. 82 1416–20
Lin C-C, Ju M-S and Lin C-W 2003 The pendulum test for evaluating spasticity of the elbow joint Arch. Phys. Med.
Rehabil. 84 69–74
Lin D C and Rymer W Z 1991 A quantitative analysis of pendular motion of the lower leg in spastic human subjects
IEEE Trans. Biomed. Eng. 38 906–18
Mayer N H 1997 Clinicophysiologic concepts of spasticity and motor dysfunction in adults with an upper motoneuron
lesion Muscle Nerve Suppl. 6 S1–13
Nordmark E and Anderson G 2002 Wartenberg pendulum test: objective quantification of muscle tone in children
with spastic diplegia undergoing selective dorsal rhizotomy Dev. Med. Child Neurol. 44 26–33
Pandyan A D, Johnson G R, Price C I, Curless R H, Barnes M P and Rodgers H 1999 A review of the properties
and limitations of the Ashworth and modified Ashworth scales as measures of spasticity Clin. Rehabil.
13 373–83
Paulis W D, Horemans H L D, Brouwer B S and Stam H J 2011 Excellent test–retest and inter-rater reliability for
Tardieu scale measurements with inertial sensors in elbow flexors of stroke patients Gait Posture 33 185–9
Piriyaprasarth P, Morris M E, Winter A and Bialocerkowski A E 2008 The reliability of knee joint position testing
using electrogoniometry BMC Musculoskeletal Disord. 9 6
Platz T, Eickhof C, Nuyens G and Vuadens P 2005 Clinical scales for the assessment of spasticity, associated
phenomena, and function: a systematic review of the literature Disabil. Rehabil. 27 7–18
Rome K and Cowieson F 1996 A reliability study of the universal goniometer, fluid goniometer, and electrogoniometer
for the measurement of ankle dorsiflexion Foot Ankle Int. 17 28–32
Scholtes V A, Dallmeijer A J and Becher J G 2007 The spasticity test: a clinical instrument to measure spasticity
in children with cerebral palsy The Effectiveness of Multilevel Botulinum Toxin Type A and Comprehensive
Rehabilitation in Children with Cerebral Palsy (Amsterdam: Ponsen & Looijen BV) pp 65–86
1434 I Sterpi et al

Sheean G 2002 The pathophysiology of spasticity Eur. J. Neurol. Suppl. 1 9 53–61 3–9; (see also ‘A panel discussion:
guidelines of best practice’ (chaired by Professor M Barnes))
Simpson D M, Gracies J M, Yablon S A, Barbano R and Brashear A 2009 Botulinum neurotoxin versus tizanidine in
upper limb spasticity: a placebo-controlled study J. Neurol. Neurosurg. Psychiatry 80 380–5
Stillman B and McMeeken J 1995 A video-based version of the pendulum test: technique and normal response Arch.
Phys. Med. Rehabil. 76 166–76
Syczewska M, Lebiedowska M K and Pandyan A D 2009 Quantifying repeatability of the Wartenberg pendulum test
parameters in children with spasticity J. Neurosci. Methods 178 340–4
Valle M S, Casabona A, Sgarlata R, Garozzo R, Vinci M and Cioni M 2006 The pendulum test as a tool to evaluate
passive knee stiffness and viscosity of patients with rheumatoid arthritis BMC Musculoskeletal Disord. 7 89
Van den Noort J C, Scholtes V A, Becher J G and Harlaar J 2010 Evaluation of the catch in spasticity assessment in
children with cerebral palsy Arch. Phys. Med. Rehabil. 91 615–23
Van den Noort J C, Scholtes V A and Harlaar J 2009 Evaluation of clinical spasticity assessment in cerebral palsy
using inertial sensors Gait Posture 30 138–43
Vodovnik L, Bowman B R and Bajd T 1984 Dynamics of spastic knee joint Med. Biol. Eng. Comput. 22 63–69
Wartenberg R 1951 Pendulousness of the legs as a diagnostic test Neurology 1 18–24
Weir J P 2005 Quantifying test–retest reliability using the intraclass correlation coefficient and the SEM J. Strength
Cond. Res. 19 231–40
White H, Uhl T L, Augsburger S and Tylkowski C 2007 Reliability of the three-dimensional pendulum test for
able-bodied children and children diagnosed with cerebral palsy Gait Posture 26 97–105

You might also like