The measurement of spasticity

3
The measurement of spasticity
Garth R. Johnson and Anand D. Pandyan
Introduction
Even today, although there are a number of validated
techniques for the measurement of associated dis-
ability, the measurement of spasticity at the level of
impairment is probably in its infancy. Because of the
relative lack of treatment or therapy to reduce spas-
ticity, there has been limited development of meth-
ods for its measurement. However, with the relatively
recent advent of treatments for spasticity, such as
botulinum toxin, there is now a considerable incen-
tive to develop new methods.
One particular barrier to valid measurement
relates to the need for a precise definition. The mea-
surement of any physical phenomenon is impossi-
ble in the absence of a definition, and this is equally
true in the case of spasticity. At the clinical level,
there is almost certainly a wide variety of assumed
definitions concerning stiffness and the lack or diffi-
culty of movement. A relativelyprecisestatement has
been provided by Lance (1980), as follows: Spasticity,
which is directly equated with spastic hypertonia, is
a motor disorder that is ‘characterised by a velocity
dependent increase in the tonic stretch reflex (muscle
tone) with exaggerated tendon reflexes, resulting from
the hyper excitability of the stretch reflex, as one
component of the upper motor neurone syndrome’
following a lesion at any level of the corticofu-
gal pathways – cortex, internal capsule, brainstem

or spinal cord (Burke, 1988). Furthermore, spastic
hypertonia has also been described as the exagger-
ation of the spinal proprioceptive reflexes resulting
from a loss of descending inhibitory control (Burke,
1988).
While these definitions would appear to be rea-
sonably precise, there is a need to ask whether cur-
rent clinical testing procedures are consistent with
the model that underlies them and whether the
model itself is sufficiently representative to allow
reliable testing. Essentially, the neural contributions
to increased tone
1
are likely to result from vol-
untary and involuntary (reflex) activation of the
alpha motor neuron. The presence or absence of
reflex activity is likely to be a function of muscle
length, velocity of stretch, load on the tendon and
threshold and gain in the reflex loops. It therefore
appears that, at minimum, there are five variables
that may account for the level of spasticity. This com-
plexity is not adequately addressed by the defini-
tions described above. The measurement challenge,
therefore, is to develop a procedure which is broadly
consistent with the clinical definition and percep-
tion of the impairment, but which is sensitive to the
important variables. For instance, do the assessment
procedures commonly in use always distinguish
between spasticity, contracture or other abnormal
tone such as the rigidity encountered in Parkinson’s

disease?
1
The definition of tone is another moot point. There are two
broad definitions of tone used in the literature: (a) resistance
to an externally imposed movement and (b) the state of
readiness (or background activity) in a resting muscle. In this
chapter the former definition is used.
64
The measurement of spasticity 65
Reflex hyperexcitability
CNS lesion
Altered muscle
function
Altered mechanical
properties
Increased tone
or resistance
Figure 3.1. The major contributions to resistance to passive motion result from changes in both the reflex behaviour and in
the passive mechanical properties of the muscle. It is important to note that, under certain circumstances, reflex activity
can confounded by interactions between the cognitive system and the environment.
Approaches to measurement
Probably because of neurophysiological complex-
ity and the lack of rigid definitions discussed above,
there has been a variety of approaches to the mea-
surement of spasticity. While the majority of clini-
cians probably rely on descriptive scales, there have
been several attempts to use physical or biomechan-
ical approaches. However, the common element of
all these methods is that they are concerned with the
quantification of resistance to passive motion, and it

must be remembered that this can result from a com-
bination of the neurophysiological effects together
with biomechanical changes to the muscle(s), ten-
don(s) and capsule. The situation is summarized in
Figure 3.1.
While the primary theme of this chapter is to con-
sider methods for the measurement of the impair-
ment associated with spasticity, it is important to
note that techniques of both impairment and dis-
ability may be used clinically. While one particular
approach to the measurement of disability, gait anal-
ysis,is discussed later, itis important to stressthat the
relationships between disability and spasticity are
poorly understood and have yet to be fully explored.
Use of scales to measure spasticity
Requirements of measurement scales
Since most measurement of spasticity is performed
using clinical scales, it is useful first to examine the
properties of these instruments. A prerequisite for
the use of any measurement scale is a knowledge of
its performance characteristics and limitations, as
these play a key part in interpreting the data and
determining the appropriate method of statistical
analysis. The key aspects of measurement scales are
considered before going on to examine the attributes
of instruments for the measurement of spasticity.
Level of measurement
There are four distinct levels of measurement that
can be identified hierarchically as follows: nominal
(categorical), ordinal, interval and ratio levels. These

are described in Table 3.1 with examples.
The Ashworth scales
Inthe clinical setting, the most commonly used tech-
nique of measurement is the Ashworth scale (Ash-
worth,1964), developedoriginally for the assessment
of patients with multiple sclerosis. The Ashworth
test is based upon the assessment of the resistance
to passive stretch by the clinician who applies the
movement. However, although this would appear
to be broadly in conformity with the Lance defi-
nition, its reliability might be expected to depend
upon the ability of the observer both to control the
rate of stretch and to assess the resistance. However,
despite its widespread use and further development
(Bohannon & Smith, 1987), there are relatively few
data available on the reliability of this scale. The
66 Garth R. Johnson and Anand D. Pandyan
Table 3.1. The properties of scales
Type of scale
Mutually
exclusive Logical order
Scaled to perceived
quantity
Intervals of
equal length
True zero
point
Nominal (e.g. type of stroke) x
Ordinal (e.g. strength
measured on MRC scale)

xxx
Interval (e.g. range of
motion)
xxx x
Ratio (e.g. absolute strength) xxx xx
Table 3.2. Definitions of the Ashworth and modified Ashworth scales
Score Ashworth scale (Ashworth, 1964) Modified Ashworth scale (Bohannon & Smith, 1987)
0 No increase in tone No increase in muscle tone
1 Slight increase in tone giving a catch when the limb
was moved in flexion or extension
Slight increase in muscle tone, manifested by a catch
and release or by minimal resistance at the end of
the range of motion (ROM) when the affected
part(s) is moved in flexion or extension
1+ Slight increase in muscle tone, manifested by a catch,
followed by minimal resistance throughout the
remainder (less than half) of the ROM
2 More marked increase in tone but limb easily flexed More marked increase in muscle tone through most
of the ROM, but affected part(s) easily moved
3 Considerable increase in tone – passive movement
difficult
Considerable increase in muscle tone passive,
movement difficult
4 Limb rigid in flexion or extension Affected part(s) rigid in flexion or extension
properties of these scales have been reviewed in
detail by Pandyan and colleagues (Pandyan et al.,
1999) and the major points are outlined in Table 3.2.
Ashworth and modified Ashworth scales – level
of measurement
Since the Ashworth scale does not measure the resis-

tance to passive movement objectively, it cannot be
treated as either a ratio or an interval level measure.
The originator has proposed that the scale should
be treated as an ordinal level measure of resistance
to passive movement (Ashworth, 1964). Although it
is not possible to give a clear guideline as to what
would define a ‘passive stretch’, evidence suggests
that velocities of greater than 10 degrees per sec-
ond could trigger reflex activity, which in turn could
contribute to an increase in the resistance to pas-
sive movement (Dewald & Given, 1994; Lamontagne
et al., 1998; Pandyan et al., 2006). However, further
investigation of this is almost certainly required.
The modified Ashworth scale, proposed by
Bohannon and Smith (1987), contains an additional
level of measurement (1+) and a revised definition
of the lower end of the Ashworth scale. However,
this modification may have introduced an ambigu-
ity in the scale that reduces it to a nominal level
The measurement of spasticity 67
measure of resistance to passive movement. The rea-
sons for this are the lack of clear clinical or biome-
chanicaldefinitions for theterms ‘catch’ and‘release’
and an assumption that ‘catch and release’ at end
range of movement is the same as ‘minimal resis-
tance to passive movement’. In particular, the differ-
entiation between grades 1 and 1+ depends upon
the presence or absence of either ‘release’ or ‘min-
imal resistance to passive movement at end range
of movement’, the latter of which is probably influ-

enced by the viscoelastic properties. Since there is
no published evidence supporting either an ordi-
nal relationship between the grades 1 and 1+ or a
relationship between the ‘catch and release’, ‘min-
imal resistance to passive movement’, ‘increased
resistance to passive movement’, and spasticity,
it is not possible to treat the modified Ashworth
scale as an ordinal measure of resistance to passive
movement.
Published data support the use of the original
Ashworth scale as an ordinal level measure of resis-
tance to passive movement. However, the modified
Ashworth scale could be considered to be an ordinal
level measure of resistance to passive movement if
the ambiguity between the 1 and 1+ categories could
be resolved.
Reliability of the Ashworth scales
Original Ashworth scale
Two studies have investigated the reliability of the
original Ashworth scale (Lee et al., 1989; Nuyens
et al., 1994), and a further four have studied the reli-
ability of the modified Ashworth scale (Bohannon &
Smith, 1987; Bodin & Morris, 1991; Sloan et al., 1992;
Allison et al., 1996). One further study has compared
the reliability of the two scales (Hass et al., 1996).
There appears to be conflicting evidence on the reli-
ability of the Ashworth scales.
In the original paper, the Ashworth scale was used
as one of several clinical observations to classify
spasticity (Ashworth, 1964), although, surprisingly,

this paper does not describe the exact testing pro-
tocol. Based on the Ashworth scale guidelines, Lee
et al. (1989) investigated the inter- and intrarater
reliability of spasticity measurement using a recoded
and summated spasticity score. While it was not pos-
sible to draw any conclusions on the reliability of the
Ashworth scale as a measure of spasticity in individ-
ual joints, there are important data analysis issues
that need to be highlighted. If it is accepted that the
Ashworth scale is not an interval or ratio level mea-
surement of spasticity, then the use of parametric
measures of intrarater reliability may be questioned.
Similarly, the summing of individual joint scores to
producea summated Ashworth scoreis methodolog-
ically flawed.
Nuyenset al. (1994) investigated the interrater reli-
ability of the Ashworth scale to measure spasticity
in selected muscles of the lower limb, although it
is not entirely clear how the authors differentiated
between some muscle groups tested (e.g. m. soleus
and m. gastrocnemius). Based on an initial assump-
tion that it was an ordinal measure of spasticity, the
authors supported the continued use of the Ash-
worth score as a clinical measure of spasticity. They
also suggested that the inter-rater reliability of the
scale when measuring spasticity in the lower limb
may vary according to the muscle group being tested
and concluded that the inter-raterreliability was bet-
ter for the distal than the proximal muscle groups. In
the same study, they summed the (nonparametric)

Ashworth scores obtained from individual muscles
to obtain a total score and showed that the median
of these totals was similar for both assessors, even
though the two raters often assessed spasticity dif-
ferently. This latter finding highlights how the use
of a summated score in intervention and reliability
studies may mask any unreliability arising with the
use of individual joint scores.
Modified Ashworth scale
Bohannonand Smith(1987),as wellas beingthe orig-
inators, were the first to test the inter-rater reliability
of the modified Ashworth scale. They concluded that
the inter-rater reliability at the elbow was accept-
able, but noted the possibility that the high degree of
agreement may have been attributable to the inter-
actions (mutual testing and discussions) between
assessors. Bodin and Morris (1991) investigated the
68 Garth R. Johnson and Anand D. Pandyan
inter-rater reliability of the scale for measuring wrist
flexor spasticity and concluded that it was a reliable
measure of wrist flexor spasticity when used by two
trained testers. The authors were of the view that
the good agreement was independent of interactions
between assessors during the study period. Sloan
et al. (1992) investigated the reliability of the scale in
measuring spasticity of the elbow flexors and exten-
sors and the knee flexors. Assuming an ordinal level
of measurement, they concluded that the modified
Ashworth scale was a reliable measure of spasticity
at the elbow but not at the knee. The results from

this study were similar in some respects to that of
Bohannon and Smith (1987) and supported the con-
clusions that the modified Ashworth scale may have
sufficient reliability to classify resistance to passive
motion at the elbow.
Allison et al. (1996) investigated the intra- and
inter-rater reliability of the modified Ashworth scale
when measuring ankle plantar flexor spasticity and
concluded, despite reservations, that it had suffi-
cient reliability in measuring spasticity at the ankle
in the clinical setting. The authors also highlighted
some practical difficulties experienced when using
the scale to classify spasticity in the ankle plantar
flexors.
Comparison of the Ashworth and modified
Ashworth scales
Hass et al. (1996) compared the inter-rater reliability
of the Ashworth and the modified Ashworth scales
achieved by two assessors grading spasticity in the
lower limbs of 30 subjects with spinal cord injury.
Using the Cohen’s ␬ to test for the inter-rater relia-
bility, they concluded that both scales should be used
with extreme caution since the inter-rater reliability
in classifying spasticity in the lower limb was poor.
They also showed that inter-rater reliability was bet-
ter for the original Ashworth scale.
It could be argued that by adding an extra level of
classification to increase the sensitivity, Bohannon
and Smith (1987) had also increased the probability
of errors occurring in the modified Ashworth scale.

In addition, as pointed out earlier, there is a certain
degree of ambiguity between the grades 1 and 1+
in the modified Ashworth (Kumar et al., 2006). The
lower reliability observed when using the modified
Ashworth scaleto gradespasticity could be explained
by the above two factors.
Ashworth scales – conclusions and
recommendations
Based on the published evidence,the Ashworth scale
and the modified Ashworth scale can be regarded
as ordinal and nominal level measures of resistance
to passive movement, respectively. These scales are
unable to reliably differentiate changes in resistance
to passive movement between the grades 0, 1, 1+ and
2. However, they may only be regarded as measures
of spasticity if the velocity of passive joint move-
ment is consistent, the joint range of movement is
not compromised and in the absence of patholo-
gies which may cause other forms of increased
tone such as rigidity. The use of parametric pro-
cedures such as a recoded and/or summated Ash-
worth score in the place of individual joint (or mus-
cle) scores is not recommended, since two indi-
viduals who rate resistance to passive movement
quiet differently can produce similar summated
scores.
Some further key points which arise are as follows:
1. Although the use of the frequency distributions,
median and interquartile ranges (mean and stan-
darddeviation/confidence intervals) may be used

in descriptive studies, it is appropriate only
to use categorical/nonparametric data analysis
techniques in reliability and intervention studies
(Chatfield & Collins, 1980; Bland, 1995; Agresti,
1996).
2. In any clinical trials, it is essential that the investi-
gators apply the scales as described in the source
publications (Ashworth, 1964; Bohannon &
Smith, 1987) and are not tempted to introduce
intermediate levels (e.g. spasticity grades of 2.5)
(Agresti, 1996).
3. Given the uncertainty surrounding the inter-rater
reliability of these scales, it is advisable that a
The measurement of spasticity 69
single assessor is used in all clinical trials. If this
is not possible (e.g. multicentre studies), then it is
suggested that the consistency between assessors
be tested before the actual trial.
4. While an implicit assumption in the original
scales is that the resistance to passive movement
is tested through the full range of passive move-
ment (except grade 4), this may not always be
possible in clinical practice (Kumar et al., 2006).
Although many investigators provide information
related to passive range of movement, few pro-
vide a measure of the starting position of the limb
or an indication of whether the subject experi-
enced pain during the assessment of spasticity.
It should be remembered that reflex excitability
may be influenced by the resting length of the

limb and pain (Burka, 1988; Rymer & Katz, 1994;
Rothwell, 1994). Thus, it is recommended that in
future studies, information on the passive range
of movement, the resting limb posture before
stretch, and pain during the stretch be recorded.
5. Many authors use repeated cycles of passive
stretching prior to grading spasticity. It is also
important to realize that the viscoelastic contri-
butions to the resistance to passive movement are
likely to decrease with repeated cycles of stretch-
ing (Pandyan, 1997) while the changes in the
tone-related components will need to be consid-
ered indeterministic (i.e. it could either increase,
reduce or remain unchanged and will depend on
many extraneous factors). It is therefore essen-
tial that repeated movements be kept to a min-
imum and the guidelines described by Nuyens
et al. (1994) would be recommended in future
clinical trials.
6. Environmental and postural considerations are
also likely to be important. For instance, measure-
ments should always be carried out in a room of
the same temperature on each occasion, and the
posture of the subject should be kept the same at
each measurement occasion.
7. It would appear that the modified Ashworth
scale, when compared with the original Ashworth
scale, has lower reliability when used to classify
resistance to passive movement at the lower limb
(Sloan et al., 1992; Nuyens et al., 1994; Hass et al.,

1996). It is possible that the difference arises
from the modified Ashworth scale having an addi-
tional level classification (Kumar et al., 2006). In
addition, the loss of reliability in the lower limb
may be attributable to difficulties in perceiving
reflex-mediated stiffness when moving the heav-
ier shank and foot segments.
Further work is now required to examine both the
validity and the reliability of both the Ashworth and
modified Ashworth scales thoroughly, particularly as
there may be an increase in their clinical use with the
adventof more therapeutic interventions focussed at
reducing spasticity.
The Tardieu method of assessment
Following the original research of Tardieu and col-
leagues (1954) in the early 1950s, a new scale for
classifying spasticity based was developed by Held
and Pierrot-Deseilligny (1969). This scale has since
been translated to English and undergone substan-
tial modifications. Under currently published guide-
lines, for classifying spasticity using the Tardieu
method (Heldet al., 1969; Gracies,2001), the assessor
is required initially to apply two sequential stretches
to the limb segment, as follows:
r
A slow stretch using a velocity below which the
stretch reflex cannot be elicited.
r
A fast stretch, which, depending on the limb seg-
ment under test, could either be (1) the natural

drop of the limb segment under gravity (in a way
similar to the Wartenberg approach described in
the following section) or (2) passively stretched at
a rate faster than the rate of the natural drop of the
limb segment under gravity.
Spasticity is then classified using the quality of the
muscle reaction (X) (Table 3.3) and the angle at which
this muscle reaction occurred (Y).
The use of two velocities for quantifying the mus-
cle reaction makes this method of measurement
consistent with the Lance definition (Lance, 1980).
Although the original methods described by Tardieu