- Email: [email protected]
Biomechanical examination of a commonly used measure of spasticity
Clinical Biomechanics 16 (2001) 859±865
www.elsevier.com/locate/clinbiomech
Biomechanical examination of a commonly used measure of spasticity A.D. Pandyan
a,*
, C.I.M. Price b, H. Rodgers c, M.P. Barnes a, G.R. Johnson
a
a
c
Centre for Rehabilitation and Engineering Studies (CREST), University of Newcastle, Stephenson Building, Newcastle upon Tyne NE1 7RU, UK b Freeman Hospital, Newcastle upon Tyne, UK Department of Medicine (Geriatrics) and Department of Epidemiology and Public Health, University of Newcastle, Newcastle upon Tyne, UK Received 3 May 2001; accepted 29 August 2001
Abstract Background. An increase in the prevalence of neurological disability puts pressure on service providers to restrict costs associated with rehabilitation. Spasticity is an important neurological impairment for which many novel and expensive treatment options now exist. The antispastic eects of these techniques remain unexplored due to a paucity of valid outcome measures. Aim. To develop a biomechanical measure of resistance to passive movement, which could be used in routine clinical practice, and to examine the validity of the modi®ed Ashworth scale. Study design. Repeated measure cross-section study on 16 subjects who had a unilateral stroke one-week previously and had no elbow contractures. Outcome measures. Simultaneous measurement of resistance to passive movement using a custom built measuring device and the modi®ed Ashworth scale. Passive range of movement and velocity were also measured. The ``catch'', a phenomenon associated with the modi®ed Ashworth scale, was identi®ed by the assessor using a horizontal visual analogue scale and biomechanically quanti®ed using the residual calculated from a linear regression technique. Results. Half the study population had a modi®ed Ashworth score greater than zero. The association between the two measures was poor (j 0:366). The speed and range of passive movement were greater in subjects with modi®ed Ashworth score ``0'' (P < 0:05). Resistance to passive movement was higher in the impaired arm (P < 0:05) and tended to decrease with repeated measures and increasing speeds. Conclusions. A device to measure resistance to passive movement at the elbow was developed. The modi®ed Ashworth scale may not provide a valid measure of spasticity but a measure of resistance to passive movement in an acute stroke population. Relevance Spasticity is an important neurological impairment for which many novel and expensive treatment options are being made available. There is a paucity of clinically usable outcomes to measure spasticity. A device to measure resistance to passive movement at the elbow, which was more reliable than the modi®ed Ashworth scale was developed. This device may provide a much needed objective clinical measure to evaluate the ecacy of antispasticity treatment. Ó 2001 Elsevier Science Ltd. All rights reserved.
1. Introduction Demographic changes in the Western world are placing an increasing burden on social and health service providers, as the number of people over 70 years of age grows rapidly [1]. The consequent increase in the incidence of stroke and other neurological disorders, com-
*
Corresponding author. E-mail address: [email protected] (A.D. Pandyan).
bined with the increase in survival brought about by improvements in acute medical care, is leading to a dramatic increase in the prevalence of people with neurological impairment and associated disabilities. These factors, combined with the pressure to restrict the costs of healthcare, will demand more ecient rehabilitation techniques. Spasticity is one of many impairments that can develop following an injury to the central nervous system [2]. It is de®ned as ``. . . a velocity dependent increase in the tonic stretch re¯ex (muscle tone) with exaggerated
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 1 ) 0 0 0 8 4 - 5
860
A.D. Pandyan et al. / Clinical Biomechanics 16 (2001) 859±865
tendon re¯exes, resulting from the hyper excitability of the stretch re¯ex, as one component of the upper motor neurone syndrome'' [3]. Excessive and uncontrolled spasticity limits functional recovery, causes pain, and is associated with contractures [4]. Thus, many therapeutic paradigms in neurological rehabilitation aim to reduce disability, and so improve function, by reducing spasticity e.g., botulinum toxin injections, intrathecal baclofen pumps and novel surgical techniques [4±6]. Even though such techniques can bene®t patients, their eect on spasticity is unclear as there is a paucity of valid measures that can be used in clinical practice [7,8]. Further the relationship between spasticity (an impairment) and disability also remains unexplored. A variety of electrophysiological, biomechanical and manual techniques have been developed to quantify spasticity [9±19]. Few of the electrophysiological techniques are used in routine clinical practice, as they are complex, invasive or unreliable [11]. Although biomechanical techniques are considered most promising [11] they are also not routinely used due to a combination of factors, i.e., the need for specialised equipment and testing procedures (e.g., controlled displacement techniques [11,12]), poor sensitivity and repeatability in measuring severe spasticity (e.g., Wartenberg's pendulum test [13,15,16]), safety factors (e.g., using powered systems in a cognitively impaired population [11,12,18,19]) and cost. Therefore the clinician currently relies upon clinical scales to measure spasticity in routine practice [8±10]. The Ashworth and modi®ed Ashworth scales [9,10] are the most commonly used clinical measures of spasticity. When using these scales the assessor is required to move the limb passively about a joint in one second and then grade the resistance encountered using a set of rules (Table 1) to quantify the re¯ex activity elicited in the muscle groups that oppose the passive movement [10]. Re¯ex activity at the lower end of the scale is additionally identi®ed by a phenomenon described as a ``catch'' [10]. It is not clear from descriptions in the source article if this is similar to another clinical phenomenon described in the literature, i.e., the clasp knife
phenomenon ± when the assessor measures resistance to passive movement (RTPM) about a single joint, the RTPM initially increases and then suddenly reduces [14,20]. Although considerable research has been done to test the reliability of these clinical scales their validity as measures of either RTPM or spasticity remains unproven [8]. The aim of this project was, therefore, to develop a biomechanical non-invasive measure of RTPM, which could be used in routine clinical practice, as a ®rst step towards developing a quantitative measure of spasticity and to examine the validity and reliability of the modi®ed Ashworth scale (MAS). The elbow joint was selected for this study, as this a joint in which the MAS can be used most reliably [8], and biomechanically this is probably the best understood of upper limb joints. 2. Methods 2.1. The measurement device The system developed was required to be portable, non-invasive, simple and safe to use in a clinical setting on dierent patient populations (e.g., stroke, traumatic brain injury, multiple sclerosis, etc.), whilst providing a quantitative and clinically relevant measure. The ®nal system Fig. 1(a) consists of a force transducer and a ¯exible electrogoniometer (Biometrics Ltd., Gwent, UK), allowing simultaneous measurement of applied force and passive range of movement. As the point of application of the force with respect to the elbow joint was not standardised the moments were not calculated. After attaching the measurement device to the subject's arm, the assessor abducted the humerus to 90° (or within a pain free range of shoulder movement), then fully ¯exed and rapidly extended the elbow within a pain free range of movement Fig. 1(b). This procedure closely mimics the Ashworth test [10]. The outputs from the transducers were ampli®ed, sampled at 100 Hz, and stored and/or displayed on a portable computer.
Table 1 Grading RTPM based on rules described in the MAS [10] Grade
MAS (Bohannon and Smith [10])
0 1
No increase in muscle tone Slight increase in muscle tone, manifested by a catch and release or by minimal resistance at the end of the range of motion when the aected part(s) is moved in ¯exion or extension Slight increase in muscle tone, manifested by a catch, followed by minimal resistance throughout the remainder (less than half) of the range of movement (ROM) More marked increase in muscle tone through most of the ROM, but aected part(s) easily moved Considerable increase in muscle tone, passive movement dicult Aected part(s) rigid in ¯exion or extension
1+ 2 3 4
A.D. Pandyan et al. / Clinical Biomechanics 16 (2001) 859±865
861
measuring system in people who had a unilateral stroke, causing upper limb weakness, one week previously. Subjects were included if they were able to comply with the study protocol, had no previous history of orthopaedic problems that might cause changes in RTPM about the elbow, and were able to provide informed consent directly or via a care giver. This study was approved by the Joint Ethics Committee of the Newcastle Health Authority and University of Newcastle upon Tyne. (a)
2.3. Measuring resistance to passive movement A single assessor experienced at using the modi®ed Ashworth scale simultaneously measured spasticity and RTPM about the elbow joint. Measurements were taken ®rst in the non-impaired and then in the impaired arm. Three repeated measures, separated by an interval of approximately 5±15 s, were taken on each arm. The assessor was blinded to all the biomechanical outcome measures. The primary measures were MAS, RTPM, passive range of movement (PROM), and speed (ratio of PROM to time taken to complete the movement). For this study a elbow ¯exion angle of 90° was de®ned as neutral, therefore, ¯exion angles were negative and extension angles were positive. 2.4. Quantifying the catch
(b)
Fig. 1. (a) The device used to measure RTPM about the elbow consists of a force transducer and a ¯exible electrogoniometer (Biometrics Ltd., Gwent, UK), allowing simultaneous measurement of the applied force and passive range of movement. (b) The assessor abducted the humerus to 90° and fully ¯exed the forearm. The forearm was then rapidly extended.
RTPM was quanti®ed by calculating the slope of the graph of applied force versus passive range of movement (using linear regression techniques). The R2 values (a measure of the goodness of ®t; ``0'' indicates a poor ®t and ``1'' a perfect ®t) were used to select the curve shapes that deviated from a straight line ®t. In addition, speed (deg/s) was calculated from displacement and time data. 2.2. Study design A cross-section study with a single assessor was used to investigate the clinical applicability of the
When quantifying spasticity using the MAS, the assessor has to identify a phenomenon loosely de®ned as a ``catch'' and/or ``minimal resistance at end range of movement'' to distinguish between grades ``1'' and ``1+''. In order to identify speci®c biomechanical correlates associated with this phenomenon, the assessor estimated the angle at which the catch was perceived (as a percentage of each individual's PROM) on a horizontal visual analogue scale that represented full range of movement. The clinical catch angle (CCA) was calculated from these data. From a biomechanical perspective, a catch was de®ned as a transient increase in the force that opposed passive extension. At the position of the catch the residual, i.e., the dierence between the actual value of force at a given angle and the value of force predicted by the linear regression equation, was expected to be a maximum. The errors associated with perceiving the position of the catch were measured by calculating the oset between the CCA and the mid-point in the range of movement (MPO) and the RMS error (RMSE). The MPO was de®ned as the dierence between the CCA and mid-point in the PROM. The RMSE was de®ned as the square root of the squared dierence between the CCA and the angle at which the residual was a maximum (Fig. 2).
862
A.D. Pandyan et al. / Clinical Biomechanics 16 (2001) 859±865
The time from set up to taking measurements on both the impaired and non-impaired arms and then removing the system from the subject's arm was less than seven minutes. 4.2. Primary measures
Fig. 2. Identifying biomechanical correlates associated with the catch: the angle of catch identi®ed by the clinician (CCA) was calculated from the horizontal visual analogue scale. The angle at which the residual (i.e., the dierence between the actual value of force and the value of force predicted by the linear regression equation at a given angle) was a maximum Ares . The RMSE was de®ned as the root of the squared dierence between CCA and Ares . The Mid-point Oset was the difference between CCA and mid-point of the passive range of movement (in this example mid-point was 55°).
3. Statistical analysis The ``paired t-test'' was used to test whether signi®cant dierences existed between the impaired and nonimpaired arms. Analysis of variance and the Tukey's post hoc test were used to determine whether signi®cant dierences existed between the three repeated measures of RTPM and between the dierent MAS scores. The RTPM (in the impaired arm) was then classi®ed into three groups (as only three MAS grades were identi®ed in this study) using the k-means cluster procedure. The Cohen's kappa was used to study the association between classi®cation of RTPM using the MAS and classi®cation of RTPM using the clustering procedure. All procedures were carried out using SPSS for windows V10.0. 4. Results
In the impaired elbow the MAS grade was ``0'' in 24 measures, ``1'' in 18 and ``1+'' in 6 (Table 2). In the ``non-impaired elbow'' MAS was ``0'' in 43 and ``1'' in 5. In the impaired arm, subjects with an MAS of ``1+'' had a signi®cantly higher resistance to passive movement than subjects with a MAS of ``0'' or ``1'' (P < 0:05) (Table 2). (RTPM) was not signi®cantly dierent in subjects with MAS of ``1'' and ``0'' (P > 0:1). Speed and PROM were higher in subjects with a MAS of ``0'' (P < 0:05) when compared to subjects with a MAS of ``1'' and ``1+'' (Table 2). Speed and PROM were not signi®cantly dierent between subjects with MAS of ``1'' and ``1+''. The association between the MAS and the RTPM classi®cation was poor (j 0:366; Standard error 0.119). In subjects with a MAS of ``1'', catches were observed in 11 of the 18 measurements and the mean catch angle was 47° (range: 28±67°). The catches were always observed after the mid-point in the passive range of movement; mean MPO was 12° (range: 4±24°); mean RMSE was 9° (range: 0±37°). In subjects with a MAS of ``1+'', catches were observed in three of the six measurements and the mean catch angle was )2° (range: )4° to 1°). The negative values signify that the catch occurred before neutral. The catches were always observed before mid-point in the PROM was reached; mean MPO was )17° (range: )19° to )16°); mean RMSE was 25° (range: 18±32°). The speed of passive movement was not signi®cantly dierent between the impaired (mean, (standard error): 54, (3) deg/s) and non-impaired arms (59, (4) deg/s; P > 0:10), however, the RTPM was higher in the impaired arm (0.23, (0.2) N/deg) than in the non-impaired arm (0.14, (0.1) N/deg; P < 0:01). Although the mean RTPM of the ®rst measures was higher than the second and third, there was no signi®cant dierence between them in either the impaired or the non-impaired arm (P > 0:10) (Table 3). In both
4.1. Subject details Sixteen subjects, six female and 10 male, diagnosed as having unilateral stroke and who ful®lled all inclusion criteria were recruited. Mean age was 67.3 years (range: 54±84). The left arm was aected in seven subjects and the right in nine. All measurements were taken one-week after stroke. As the assessor was blind to the biomechanical measures, the three repeated measures were treated as independent measures for analysis.
Table 2 A summary of the results (mean; standard error) from the impaired arm for measures of RTPM, speed and PROM for each Ashworth group MAS (Frequency)
0 (24)
1 (18)
1+ (6)
RTPM (N/deg) Speed (deg/s) PROM (deg)
0:17; 0:02 66; 5 91; 3
0:23; 0:02 44; 3 78; 2
0:46; 0:08 28; 5 70; 4
A.D. Pandyan et al. / Clinical Biomechanics 16 (2001) 859±865
863
Table 3 A summary of the results (mean; standard error) showing the relationship in RTPM and speed between the three repeated measures (RMn repeated measure n; n 1, 2, or 3) taken on both the impaired and non-impaired arm Non-impaired RTPM (N/deg) Speed (deg/s)
RM1 0:17; 0:02 45; 7
Impaired RM2 0:12; 0:01 65; 5
RM3 0:13; 0:02 69; 5
arms, the speed of the ®rst of the three measures was lower than the speed of movement used for subsequent two measures (P < 0:05) (Table 3). 4.3. Non-linear curve shapes The mean R2 value (goodness of ®t measure) was 0.844 (range: 0.114±0.993). It was greater than 0.6 in 87 of the measures (90.6%) and the curve shapes were classi®ed as linear (mean R2 0:887; range: 0.655± 0.993). The R2 was less than 0.6 in seven measures and was classi®ed as non-linear (mean R2 0:433; range: 0.114±0.591). Of the nine non-linear curve shapes, two were seen in the non-impaired arm and seven in the impaired arm. The non-linear curve shapes in the non-impaired arm showed no speci®c characteristics (Fig. 3(a)) and one non-linear curve shape in the impaired arm was shown to have a biomechanically identi®able catch (Fig. 3(b)). Of the six remaining non-linear curves in the impaired arm two distinct curve shapes, as described below, were observed. · Yield (Fig. 3(c)): In this curve shape there was an initial build up of resistance that gave way (melted/ yielded) after crossing the mid-point in the range of movement. Three of the subjects who demonstrated yield were graded as MAS ``1''. · Initial catch (Fig. 3(d)): In this curve shape there was an initial build up of resistance that reduced before the mid range of movement was reached. 5. Discussion The prototype system developed in this study has been successfully used to quantify RTPM in a clinical environment under various patient testing conditions (e.g., in an acute ward, in patients with poor sitting balance etc.). The data presented here show that the technique can provide reliable data that relate to clinically important variables when used with subjects one week after stroke. 5.1. Biomechanical correlates associated with the MAS Spasticity is a velocity dependent phenomenon [3]. Based on the neurophysiological de®nition, RTPM
RM1 0:25; 0:04 36; 5
RM2 0:22; 0:04 61; 5
RM3 0:21; 0:03 64; 6
should increase when the velocity of passive movement is increased [2,3]. However, the relationship between velocity and the RTPM in this study population was not consistent with this neurophysiological de®nition (Table 3), i.e., as the velocity increased the resistance to passive movement tended to decrease. In half of the study population the MAS grade was greater than 0. However, the association between the MAS and RTPM even at the elbow was poor. Considering that most of the subjects in this trial had MAS of less than ``1+'', it is possible that this poor association could have been related to the uncertain validity and poor reliability of the MAS at the lower end of the scale [8]. The catch and its association with PROM is key in dierentiating between MAS grades ``1'' and ``1+''. From theoretical considerations one would expect greater spasticity to be associated with the catch occurring earlier in the range of movement [8]. However, it was not possible to identify any speci®c biomechanical correlates associated with the catch because the descriptions provided in the scale are ambiguous. The non-linear curve shapes seen in the non-impaired arms (Fig. 3(a)) could not be speci®cally classi®ed and could have resulted from poor subject compliance, i.e., subject actively assisting in the movement. Although the catch seen in Fig. 3(b) was consistent with the biomechanical de®nition, a clinical catch was not identi®ed in this subject, which raises doubts about the sensitivity of the MAS. The phenomena of ``yield'' and ``initial catch'' were both consistent with the description of the claspknife phenomenon [20]. However, inertial eects encountered at the start of movement could also contribute to the initial catch. In the absence of any emg data it was not possible to comment on the exact nature of the pathophysiological mechanism that resulted in these curve shapes. In summary, the MAS has limited association with the measure of resistance to passive motion and may not exclusively measure spasticity. No speci®c biomechanical correlates of the catch could be identi®ed. It was also not possible to demonstrate an ordinal relationship between the grades ``1'' and ``1+'' based on the catch. Hence, the validity of using the MAS as an ordinal level measure of spasticity will need to be questioned. It may, however, provide a crude clinical measure of resistance to passive movement.
864
A.D. Pandyan et al. / Clinical Biomechanics 16 (2001) 859±865
5.2. Standardisation when using the MAS The speed of passive movement showed signi®cant variation between the MAS grades and between repeated measures. This variation could have resulted from limitations in the range of movement and pain (about the shoulder and elbow) in individual subjects. Equally they may have also resulted from the assessor anticipating/perceiving an increase in RTPM prior to carrying out the actual test. Evidence from the literature would suggest that re¯ex activity in the elbow ¯exors would have been triggered by the speeds measured in this study [21]. However, the in¯uence of this activity on RTPM could not be assessed in the absence of emg data. In this pilot study involving a single assessor we have demonstrated the practical limitations in following the instructions associated with the use of the MAS in routine clinical practice. This further supports the conclusion that there may be limitations in the construct validity of this scale [8]. 5.3. Factors that confound MAS
Fig. 3. (a) A plot demonstrating the non-linear relationship between the angle and force in the non-impaired arm of a study subject (RTPM 0.047; R2 0:582). (b) A plot demonstrating the non-linear relationship with a catch between the angle and force in the impaired arm of a study subject. However, no clinical catch was identi®ed in this subject (RTPM 0.367; R2 0:591). (c) A plot demonstrating a nonlinear curve shape demonstrating yield in the impaired arm of a subject (RTPM 0.290; R2 0:532). This phenomenon closely resembles a clinical phenomenon de®ned as the clasp-knife phenomenon in the literature [20]. (d) A plot demonstrating a non-linear curve shape in the impaired arm of a subject (RTPM 0.044; R2 0:419). There was an initial increase in the RTPM that may have been confounded by inertial eects.
Although all measurements were made one-week post stroke, the RTPM in the impaired arm was higher than that in the non-impaired arm and more than half the population had an MAS greater than 0. The rationale for selecting subjects who were one-week post stroke was to ensure that contractures were not a confounding factor when studying the association between RTPM and the MAS. Evidence from the literature suggests that this increase in RTPM could have resulted from decreased soft tissue compliance associated with reduced use [22]. Even though spasticity and other tone related abnormalities are not expected to develop one week after a stroke [23] the presence of curve shapes consistent with the clasp-knife phenomenon would suggest that these factors could have also in¯uenced RTPM. RTPM was in¯uenced by the immediate past history of movement. This would suggest that the increase, in RTPM, observed in the impaired arm might have been predominantly associated with changes in the viscoelastic properties of the soft tissues and not spasticity. However, it could also have resulted from poor subject compliance associated with an inability to relax or apprehension of the test procedure. 5.4. Limitations in the study In this small sample pilot study simultaneous emg measurements were not taken from the elbow ¯exors and extensors and so it was not possible to associate changes in RTPM and MAS with changes in neuronal activity. Despite initial assumptions that the repeated
A.D. Pandyan et al. / Clinical Biomechanics 16 (2001) 859±865
measures could be treated as mutually exclusive, serial dependency in the measures of RTPM was observed. The non-linearities, when quantifying RTPM, could have resulted from poor subject compliance, or abnormal re¯ex activity but a detailed analysis was beyond the scope of this initial study. All subjects had a MAS less than ``2'', therefore the reliability and sensitivity of this scale was not completely investigated.
[4]
[5] [6]
6. Conclusion
[7]
The need to quantify neurological impairment is increasing rapidly with a perceived need to justify clinical procedures used routinely. The technical challenge of doing this relates to the need to provide reliable and sensitive measurement systems that can be used within the short time available in the clinic. A device to measure resistance to passive movement at the elbow, which is more reliable than the MAS, has been presented in this paper. Evidence from this study supports previous ®ndings suggesting that MAS may not have sucient construct validity or reliability in assessing spasticity [8], however it may still provide a crude measure of resistance to passive movement. Further work is now required to investigate the sensitivity and reliability of the MAS in quantifying RTPM in a general stroke population. The clinical relevance of this measuring device may be improved by adding emg measurement capabilities. With the development of such a device it will also be possible to quantitatively investigate the eects of neuronal activity on RTPM and the MAS.
[8]
Acknowledgements This study was supported by Action Research, UK (Grant ± AP0603). We would like to thank all patients from the Northumbria Healthcare NHS Trust who participated in this trial. We would also like to thank Allergan, UK for an unrestricted educational grant and Biometrics, UK for equipment support.
[9] [10] [11] [12]
[13] [14] [15] [16] [17] [18] [19] [20]
[21]
References [1] Mayo NE. Stroke: epidemiology and recovery. Physical Medicine and Rehabilitation: State of the Art Reviews 1993;7:1±25. [2] Burke D. Spasticity as an adaptation to pyramidal tract injury. Advances in Neurology 1988;47:401±22. [3] Lance JW. Pathophysiology of spasticity and clinical experience with Baclofen. In: Lance JW, Feldman RG, Young RR, Koella
[22]
[23]
865
WP, editors. Spasticity: disordered motor control. Year Book. Chicago: 1980. p. 185±204. Barnes MP. An overview of the clinical management of spasticity. In: Barnes MP, Johnson GR, editors. Upper motor neurone syndrome and spasticity: clinical management and neurophysiology. Cambridge: Cambridge University Press; 2001. p. 1±11. Bobath B. Adult hemiplegia: evaluation and treatment. London: Heinemann Medical Books; 1990. Harburn KL, Potter PJ. Spasticity and contractures. Physical Medicine and Rehabilitation: State of the Art Reviews 1993;7:113±32. Hass BM, Crow JL. Towards a clinical measurement of spasticity?. Physiotherapy 1995;81:474±9. Pandyan et al. A review of the properties and limitations of the Ashworth and modi®ed Ashworth scales as measures of spasticity. Clinical Rehabilitation 1999;13:373±83. Ashworth B. Preliminary trial of carisoprodal in multiple sclerosis. Practitioner 1964;192:540±2. Bohannon RW, Smith MB. Inter rater reliability of a modi®ed Ashworth scale of muscle spasticity. Physical Therapy 1987;67:206±7. Katz RT et al. Objective quanti®cation of spastic hypertonia. Correlation with clinical ®ndings. Archives of Physical Medicine and Rehabilitation 1992;73:339±47. Given JD et al. Joint dependent passive stiness in paretic and contralateral limbs of spastic patients with hemiparetic stroke. Journal of Neurology, Neurosurgery, and Psychiatry 1995;59:271±9. Wartenberg R. Pendulousness of the legs as a diagnostic test. Neurology 1951;1:18±24. Rothwell JC. Control of human voluntary movement. 2nd ed. London: Chapman & Hall; 1994. Bajd T, Vodovnik L. Pendulum testing of spasticity. Journal of Biomedical Engineering 1984;6:9±16. He J et al. A dynamic neuromuscular model for describing the pendulum test for spasticity. IEEE Transactions on Biomedical Engineering 1997;44:175±83. He J. Stretch re¯ex sensitivity: eects of postural and muscle length changes. IEEE Transactions on Rehabilitation Engineering 1998;6:182±9. Lakie M et al. Assessment of human hemiplegic spasticity by a resonant frequency method. Clinical Biomechanics 1988;3:173±8. Lehmann JF et al. Spasticity: quantitative measurement as a basis for assessing eectiveness of therapeutic intervention. Archives of Physical Medicine and Rehabilitation 1989;70:6±15. Sheean G. Neurophysiology of spasticity. In: Barnes MP, Johnson GR, editors. Upper motor neurone syndrome and spasticity: clinical management and neurophysiology. Cambridge: Cambridge University Press; 2001. p. 12±78. Katz RT, Rovai GP, Brait C, Rymer WZ. Objective quanti®cation of spastic hypertonia. Correlation with clinical ®ndings. Archives of Physical Medicine and Rehabilitation 1992;73:339± 47. Goldspink G, Williams S. Muscle ®bre and connective tissue changes associated with use and disuse. In: Ada L, Canning C, editors. Key issues in neurological physiotherapy. Oxford: Butterworth & Heinemann; 1990. p. 127±218. Speach DP, Dombovy ML. Recovery from stroke: rehabilitation. Bailliere's Clinical Neurology 1994;4:317±38.
www.elsevier.com/locate/clinbiomech
Biomechanical examination of a commonly used measure of spasticity A.D. Pandyan
a,*
, C.I.M. Price b, H. Rodgers c, M.P. Barnes a, G.R. Johnson
a
a
c
Centre for Rehabilitation and Engineering Studies (CREST), University of Newcastle, Stephenson Building, Newcastle upon Tyne NE1 7RU, UK b Freeman Hospital, Newcastle upon Tyne, UK Department of Medicine (Geriatrics) and Department of Epidemiology and Public Health, University of Newcastle, Newcastle upon Tyne, UK Received 3 May 2001; accepted 29 August 2001
Abstract Background. An increase in the prevalence of neurological disability puts pressure on service providers to restrict costs associated with rehabilitation. Spasticity is an important neurological impairment for which many novel and expensive treatment options now exist. The antispastic eects of these techniques remain unexplored due to a paucity of valid outcome measures. Aim. To develop a biomechanical measure of resistance to passive movement, which could be used in routine clinical practice, and to examine the validity of the modi®ed Ashworth scale. Study design. Repeated measure cross-section study on 16 subjects who had a unilateral stroke one-week previously and had no elbow contractures. Outcome measures. Simultaneous measurement of resistance to passive movement using a custom built measuring device and the modi®ed Ashworth scale. Passive range of movement and velocity were also measured. The ``catch'', a phenomenon associated with the modi®ed Ashworth scale, was identi®ed by the assessor using a horizontal visual analogue scale and biomechanically quanti®ed using the residual calculated from a linear regression technique. Results. Half the study population had a modi®ed Ashworth score greater than zero. The association between the two measures was poor (j 0:366). The speed and range of passive movement were greater in subjects with modi®ed Ashworth score ``0'' (P < 0:05). Resistance to passive movement was higher in the impaired arm (P < 0:05) and tended to decrease with repeated measures and increasing speeds. Conclusions. A device to measure resistance to passive movement at the elbow was developed. The modi®ed Ashworth scale may not provide a valid measure of spasticity but a measure of resistance to passive movement in an acute stroke population. Relevance Spasticity is an important neurological impairment for which many novel and expensive treatment options are being made available. There is a paucity of clinically usable outcomes to measure spasticity. A device to measure resistance to passive movement at the elbow, which was more reliable than the modi®ed Ashworth scale was developed. This device may provide a much needed objective clinical measure to evaluate the ecacy of antispasticity treatment. Ó 2001 Elsevier Science Ltd. All rights reserved.
1. Introduction Demographic changes in the Western world are placing an increasing burden on social and health service providers, as the number of people over 70 years of age grows rapidly [1]. The consequent increase in the incidence of stroke and other neurological disorders, com-
*
Corresponding author. E-mail address: [email protected] (A.D. Pandyan).
bined with the increase in survival brought about by improvements in acute medical care, is leading to a dramatic increase in the prevalence of people with neurological impairment and associated disabilities. These factors, combined with the pressure to restrict the costs of healthcare, will demand more ecient rehabilitation techniques. Spasticity is one of many impairments that can develop following an injury to the central nervous system [2]. It is de®ned as ``. . . a velocity dependent increase in the tonic stretch re¯ex (muscle tone) with exaggerated
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 1 ) 0 0 0 8 4 - 5
860
A.D. Pandyan et al. / Clinical Biomechanics 16 (2001) 859±865
tendon re¯exes, resulting from the hyper excitability of the stretch re¯ex, as one component of the upper motor neurone syndrome'' [3]. Excessive and uncontrolled spasticity limits functional recovery, causes pain, and is associated with contractures [4]. Thus, many therapeutic paradigms in neurological rehabilitation aim to reduce disability, and so improve function, by reducing spasticity e.g., botulinum toxin injections, intrathecal baclofen pumps and novel surgical techniques [4±6]. Even though such techniques can bene®t patients, their eect on spasticity is unclear as there is a paucity of valid measures that can be used in clinical practice [7,8]. Further the relationship between spasticity (an impairment) and disability also remains unexplored. A variety of electrophysiological, biomechanical and manual techniques have been developed to quantify spasticity [9±19]. Few of the electrophysiological techniques are used in routine clinical practice, as they are complex, invasive or unreliable [11]. Although biomechanical techniques are considered most promising [11] they are also not routinely used due to a combination of factors, i.e., the need for specialised equipment and testing procedures (e.g., controlled displacement techniques [11,12]), poor sensitivity and repeatability in measuring severe spasticity (e.g., Wartenberg's pendulum test [13,15,16]), safety factors (e.g., using powered systems in a cognitively impaired population [11,12,18,19]) and cost. Therefore the clinician currently relies upon clinical scales to measure spasticity in routine practice [8±10]. The Ashworth and modi®ed Ashworth scales [9,10] are the most commonly used clinical measures of spasticity. When using these scales the assessor is required to move the limb passively about a joint in one second and then grade the resistance encountered using a set of rules (Table 1) to quantify the re¯ex activity elicited in the muscle groups that oppose the passive movement [10]. Re¯ex activity at the lower end of the scale is additionally identi®ed by a phenomenon described as a ``catch'' [10]. It is not clear from descriptions in the source article if this is similar to another clinical phenomenon described in the literature, i.e., the clasp knife
phenomenon ± when the assessor measures resistance to passive movement (RTPM) about a single joint, the RTPM initially increases and then suddenly reduces [14,20]. Although considerable research has been done to test the reliability of these clinical scales their validity as measures of either RTPM or spasticity remains unproven [8]. The aim of this project was, therefore, to develop a biomechanical non-invasive measure of RTPM, which could be used in routine clinical practice, as a ®rst step towards developing a quantitative measure of spasticity and to examine the validity and reliability of the modi®ed Ashworth scale (MAS). The elbow joint was selected for this study, as this a joint in which the MAS can be used most reliably [8], and biomechanically this is probably the best understood of upper limb joints. 2. Methods 2.1. The measurement device The system developed was required to be portable, non-invasive, simple and safe to use in a clinical setting on dierent patient populations (e.g., stroke, traumatic brain injury, multiple sclerosis, etc.), whilst providing a quantitative and clinically relevant measure. The ®nal system Fig. 1(a) consists of a force transducer and a ¯exible electrogoniometer (Biometrics Ltd., Gwent, UK), allowing simultaneous measurement of applied force and passive range of movement. As the point of application of the force with respect to the elbow joint was not standardised the moments were not calculated. After attaching the measurement device to the subject's arm, the assessor abducted the humerus to 90° (or within a pain free range of shoulder movement), then fully ¯exed and rapidly extended the elbow within a pain free range of movement Fig. 1(b). This procedure closely mimics the Ashworth test [10]. The outputs from the transducers were ampli®ed, sampled at 100 Hz, and stored and/or displayed on a portable computer.
Table 1 Grading RTPM based on rules described in the MAS [10] Grade
MAS (Bohannon and Smith [10])
0 1
No increase in muscle tone Slight increase in muscle tone, manifested by a catch and release or by minimal resistance at the end of the range of motion when the aected part(s) is moved in ¯exion or extension Slight increase in muscle tone, manifested by a catch, followed by minimal resistance throughout the remainder (less than half) of the range of movement (ROM) More marked increase in muscle tone through most of the ROM, but aected part(s) easily moved Considerable increase in muscle tone, passive movement dicult Aected part(s) rigid in ¯exion or extension
1+ 2 3 4
A.D. Pandyan et al. / Clinical Biomechanics 16 (2001) 859±865
861
measuring system in people who had a unilateral stroke, causing upper limb weakness, one week previously. Subjects were included if they were able to comply with the study protocol, had no previous history of orthopaedic problems that might cause changes in RTPM about the elbow, and were able to provide informed consent directly or via a care giver. This study was approved by the Joint Ethics Committee of the Newcastle Health Authority and University of Newcastle upon Tyne. (a)
2.3. Measuring resistance to passive movement A single assessor experienced at using the modi®ed Ashworth scale simultaneously measured spasticity and RTPM about the elbow joint. Measurements were taken ®rst in the non-impaired and then in the impaired arm. Three repeated measures, separated by an interval of approximately 5±15 s, were taken on each arm. The assessor was blinded to all the biomechanical outcome measures. The primary measures were MAS, RTPM, passive range of movement (PROM), and speed (ratio of PROM to time taken to complete the movement). For this study a elbow ¯exion angle of 90° was de®ned as neutral, therefore, ¯exion angles were negative and extension angles were positive. 2.4. Quantifying the catch
(b)
Fig. 1. (a) The device used to measure RTPM about the elbow consists of a force transducer and a ¯exible electrogoniometer (Biometrics Ltd., Gwent, UK), allowing simultaneous measurement of the applied force and passive range of movement. (b) The assessor abducted the humerus to 90° and fully ¯exed the forearm. The forearm was then rapidly extended.
RTPM was quanti®ed by calculating the slope of the graph of applied force versus passive range of movement (using linear regression techniques). The R2 values (a measure of the goodness of ®t; ``0'' indicates a poor ®t and ``1'' a perfect ®t) were used to select the curve shapes that deviated from a straight line ®t. In addition, speed (deg/s) was calculated from displacement and time data. 2.2. Study design A cross-section study with a single assessor was used to investigate the clinical applicability of the
When quantifying spasticity using the MAS, the assessor has to identify a phenomenon loosely de®ned as a ``catch'' and/or ``minimal resistance at end range of movement'' to distinguish between grades ``1'' and ``1+''. In order to identify speci®c biomechanical correlates associated with this phenomenon, the assessor estimated the angle at which the catch was perceived (as a percentage of each individual's PROM) on a horizontal visual analogue scale that represented full range of movement. The clinical catch angle (CCA) was calculated from these data. From a biomechanical perspective, a catch was de®ned as a transient increase in the force that opposed passive extension. At the position of the catch the residual, i.e., the dierence between the actual value of force at a given angle and the value of force predicted by the linear regression equation, was expected to be a maximum. The errors associated with perceiving the position of the catch were measured by calculating the oset between the CCA and the mid-point in the range of movement (MPO) and the RMS error (RMSE). The MPO was de®ned as the dierence between the CCA and mid-point in the PROM. The RMSE was de®ned as the square root of the squared dierence between the CCA and the angle at which the residual was a maximum (Fig. 2).
862
A.D. Pandyan et al. / Clinical Biomechanics 16 (2001) 859±865
The time from set up to taking measurements on both the impaired and non-impaired arms and then removing the system from the subject's arm was less than seven minutes. 4.2. Primary measures
Fig. 2. Identifying biomechanical correlates associated with the catch: the angle of catch identi®ed by the clinician (CCA) was calculated from the horizontal visual analogue scale. The angle at which the residual (i.e., the dierence between the actual value of force and the value of force predicted by the linear regression equation at a given angle) was a maximum Ares . The RMSE was de®ned as the root of the squared dierence between CCA and Ares . The Mid-point Oset was the difference between CCA and mid-point of the passive range of movement (in this example mid-point was 55°).
3. Statistical analysis The ``paired t-test'' was used to test whether signi®cant dierences existed between the impaired and nonimpaired arms. Analysis of variance and the Tukey's post hoc test were used to determine whether signi®cant dierences existed between the three repeated measures of RTPM and between the dierent MAS scores. The RTPM (in the impaired arm) was then classi®ed into three groups (as only three MAS grades were identi®ed in this study) using the k-means cluster procedure. The Cohen's kappa was used to study the association between classi®cation of RTPM using the MAS and classi®cation of RTPM using the clustering procedure. All procedures were carried out using SPSS for windows V10.0. 4. Results
In the impaired elbow the MAS grade was ``0'' in 24 measures, ``1'' in 18 and ``1+'' in 6 (Table 2). In the ``non-impaired elbow'' MAS was ``0'' in 43 and ``1'' in 5. In the impaired arm, subjects with an MAS of ``1+'' had a signi®cantly higher resistance to passive movement than subjects with a MAS of ``0'' or ``1'' (P < 0:05) (Table 2). (RTPM) was not signi®cantly dierent in subjects with MAS of ``1'' and ``0'' (P > 0:1). Speed and PROM were higher in subjects with a MAS of ``0'' (P < 0:05) when compared to subjects with a MAS of ``1'' and ``1+'' (Table 2). Speed and PROM were not signi®cantly dierent between subjects with MAS of ``1'' and ``1+''. The association between the MAS and the RTPM classi®cation was poor (j 0:366; Standard error 0.119). In subjects with a MAS of ``1'', catches were observed in 11 of the 18 measurements and the mean catch angle was 47° (range: 28±67°). The catches were always observed after the mid-point in the passive range of movement; mean MPO was 12° (range: 4±24°); mean RMSE was 9° (range: 0±37°). In subjects with a MAS of ``1+'', catches were observed in three of the six measurements and the mean catch angle was )2° (range: )4° to 1°). The negative values signify that the catch occurred before neutral. The catches were always observed before mid-point in the PROM was reached; mean MPO was )17° (range: )19° to )16°); mean RMSE was 25° (range: 18±32°). The speed of passive movement was not signi®cantly dierent between the impaired (mean, (standard error): 54, (3) deg/s) and non-impaired arms (59, (4) deg/s; P > 0:10), however, the RTPM was higher in the impaired arm (0.23, (0.2) N/deg) than in the non-impaired arm (0.14, (0.1) N/deg; P < 0:01). Although the mean RTPM of the ®rst measures was higher than the second and third, there was no signi®cant dierence between them in either the impaired or the non-impaired arm (P > 0:10) (Table 3). In both
4.1. Subject details Sixteen subjects, six female and 10 male, diagnosed as having unilateral stroke and who ful®lled all inclusion criteria were recruited. Mean age was 67.3 years (range: 54±84). The left arm was aected in seven subjects and the right in nine. All measurements were taken one-week after stroke. As the assessor was blind to the biomechanical measures, the three repeated measures were treated as independent measures for analysis.
Table 2 A summary of the results (mean; standard error) from the impaired arm for measures of RTPM, speed and PROM for each Ashworth group MAS (Frequency)
0 (24)
1 (18)
1+ (6)
RTPM (N/deg) Speed (deg/s) PROM (deg)
0:17; 0:02 66; 5 91; 3
0:23; 0:02 44; 3 78; 2
0:46; 0:08 28; 5 70; 4
A.D. Pandyan et al. / Clinical Biomechanics 16 (2001) 859±865
863
Table 3 A summary of the results (mean; standard error) showing the relationship in RTPM and speed between the three repeated measures (RMn repeated measure n; n 1, 2, or 3) taken on both the impaired and non-impaired arm Non-impaired RTPM (N/deg) Speed (deg/s)
RM1 0:17; 0:02 45; 7
Impaired RM2 0:12; 0:01 65; 5
RM3 0:13; 0:02 69; 5
arms, the speed of the ®rst of the three measures was lower than the speed of movement used for subsequent two measures (P < 0:05) (Table 3). 4.3. Non-linear curve shapes The mean R2 value (goodness of ®t measure) was 0.844 (range: 0.114±0.993). It was greater than 0.6 in 87 of the measures (90.6%) and the curve shapes were classi®ed as linear (mean R2 0:887; range: 0.655± 0.993). The R2 was less than 0.6 in seven measures and was classi®ed as non-linear (mean R2 0:433; range: 0.114±0.591). Of the nine non-linear curve shapes, two were seen in the non-impaired arm and seven in the impaired arm. The non-linear curve shapes in the non-impaired arm showed no speci®c characteristics (Fig. 3(a)) and one non-linear curve shape in the impaired arm was shown to have a biomechanically identi®able catch (Fig. 3(b)). Of the six remaining non-linear curves in the impaired arm two distinct curve shapes, as described below, were observed. · Yield (Fig. 3(c)): In this curve shape there was an initial build up of resistance that gave way (melted/ yielded) after crossing the mid-point in the range of movement. Three of the subjects who demonstrated yield were graded as MAS ``1''. · Initial catch (Fig. 3(d)): In this curve shape there was an initial build up of resistance that reduced before the mid range of movement was reached. 5. Discussion The prototype system developed in this study has been successfully used to quantify RTPM in a clinical environment under various patient testing conditions (e.g., in an acute ward, in patients with poor sitting balance etc.). The data presented here show that the technique can provide reliable data that relate to clinically important variables when used with subjects one week after stroke. 5.1. Biomechanical correlates associated with the MAS Spasticity is a velocity dependent phenomenon [3]. Based on the neurophysiological de®nition, RTPM
RM1 0:25; 0:04 36; 5
RM2 0:22; 0:04 61; 5
RM3 0:21; 0:03 64; 6
should increase when the velocity of passive movement is increased [2,3]. However, the relationship between velocity and the RTPM in this study population was not consistent with this neurophysiological de®nition (Table 3), i.e., as the velocity increased the resistance to passive movement tended to decrease. In half of the study population the MAS grade was greater than 0. However, the association between the MAS and RTPM even at the elbow was poor. Considering that most of the subjects in this trial had MAS of less than ``1+'', it is possible that this poor association could have been related to the uncertain validity and poor reliability of the MAS at the lower end of the scale [8]. The catch and its association with PROM is key in dierentiating between MAS grades ``1'' and ``1+''. From theoretical considerations one would expect greater spasticity to be associated with the catch occurring earlier in the range of movement [8]. However, it was not possible to identify any speci®c biomechanical correlates associated with the catch because the descriptions provided in the scale are ambiguous. The non-linear curve shapes seen in the non-impaired arms (Fig. 3(a)) could not be speci®cally classi®ed and could have resulted from poor subject compliance, i.e., subject actively assisting in the movement. Although the catch seen in Fig. 3(b) was consistent with the biomechanical de®nition, a clinical catch was not identi®ed in this subject, which raises doubts about the sensitivity of the MAS. The phenomena of ``yield'' and ``initial catch'' were both consistent with the description of the claspknife phenomenon [20]. However, inertial eects encountered at the start of movement could also contribute to the initial catch. In the absence of any emg data it was not possible to comment on the exact nature of the pathophysiological mechanism that resulted in these curve shapes. In summary, the MAS has limited association with the measure of resistance to passive motion and may not exclusively measure spasticity. No speci®c biomechanical correlates of the catch could be identi®ed. It was also not possible to demonstrate an ordinal relationship between the grades ``1'' and ``1+'' based on the catch. Hence, the validity of using the MAS as an ordinal level measure of spasticity will need to be questioned. It may, however, provide a crude clinical measure of resistance to passive movement.
864
A.D. Pandyan et al. / Clinical Biomechanics 16 (2001) 859±865
5.2. Standardisation when using the MAS The speed of passive movement showed signi®cant variation between the MAS grades and between repeated measures. This variation could have resulted from limitations in the range of movement and pain (about the shoulder and elbow) in individual subjects. Equally they may have also resulted from the assessor anticipating/perceiving an increase in RTPM prior to carrying out the actual test. Evidence from the literature would suggest that re¯ex activity in the elbow ¯exors would have been triggered by the speeds measured in this study [21]. However, the in¯uence of this activity on RTPM could not be assessed in the absence of emg data. In this pilot study involving a single assessor we have demonstrated the practical limitations in following the instructions associated with the use of the MAS in routine clinical practice. This further supports the conclusion that there may be limitations in the construct validity of this scale [8]. 5.3. Factors that confound MAS
Fig. 3. (a) A plot demonstrating the non-linear relationship between the angle and force in the non-impaired arm of a study subject (RTPM 0.047; R2 0:582). (b) A plot demonstrating the non-linear relationship with a catch between the angle and force in the impaired arm of a study subject. However, no clinical catch was identi®ed in this subject (RTPM 0.367; R2 0:591). (c) A plot demonstrating a nonlinear curve shape demonstrating yield in the impaired arm of a subject (RTPM 0.290; R2 0:532). This phenomenon closely resembles a clinical phenomenon de®ned as the clasp-knife phenomenon in the literature [20]. (d) A plot demonstrating a non-linear curve shape in the impaired arm of a subject (RTPM 0.044; R2 0:419). There was an initial increase in the RTPM that may have been confounded by inertial eects.
Although all measurements were made one-week post stroke, the RTPM in the impaired arm was higher than that in the non-impaired arm and more than half the population had an MAS greater than 0. The rationale for selecting subjects who were one-week post stroke was to ensure that contractures were not a confounding factor when studying the association between RTPM and the MAS. Evidence from the literature suggests that this increase in RTPM could have resulted from decreased soft tissue compliance associated with reduced use [22]. Even though spasticity and other tone related abnormalities are not expected to develop one week after a stroke [23] the presence of curve shapes consistent with the clasp-knife phenomenon would suggest that these factors could have also in¯uenced RTPM. RTPM was in¯uenced by the immediate past history of movement. This would suggest that the increase, in RTPM, observed in the impaired arm might have been predominantly associated with changes in the viscoelastic properties of the soft tissues and not spasticity. However, it could also have resulted from poor subject compliance associated with an inability to relax or apprehension of the test procedure. 5.4. Limitations in the study In this small sample pilot study simultaneous emg measurements were not taken from the elbow ¯exors and extensors and so it was not possible to associate changes in RTPM and MAS with changes in neuronal activity. Despite initial assumptions that the repeated
A.D. Pandyan et al. / Clinical Biomechanics 16 (2001) 859±865
measures could be treated as mutually exclusive, serial dependency in the measures of RTPM was observed. The non-linearities, when quantifying RTPM, could have resulted from poor subject compliance, or abnormal re¯ex activity but a detailed analysis was beyond the scope of this initial study. All subjects had a MAS less than ``2'', therefore the reliability and sensitivity of this scale was not completely investigated.
[4]
[5] [6]
6. Conclusion
[7]
The need to quantify neurological impairment is increasing rapidly with a perceived need to justify clinical procedures used routinely. The technical challenge of doing this relates to the need to provide reliable and sensitive measurement systems that can be used within the short time available in the clinic. A device to measure resistance to passive movement at the elbow, which is more reliable than the MAS, has been presented in this paper. Evidence from this study supports previous ®ndings suggesting that MAS may not have sucient construct validity or reliability in assessing spasticity [8], however it may still provide a crude measure of resistance to passive movement. Further work is now required to investigate the sensitivity and reliability of the MAS in quantifying RTPM in a general stroke population. The clinical relevance of this measuring device may be improved by adding emg measurement capabilities. With the development of such a device it will also be possible to quantitatively investigate the eects of neuronal activity on RTPM and the MAS.
[8]
Acknowledgements This study was supported by Action Research, UK (Grant ± AP0603). We would like to thank all patients from the Northumbria Healthcare NHS Trust who participated in this trial. We would also like to thank Allergan, UK for an unrestricted educational grant and Biometrics, UK for equipment support.
[9] [10] [11] [12]
[13] [14] [15] [16] [17] [18] [19] [20]
[21]
References [1] Mayo NE. Stroke: epidemiology and recovery. Physical Medicine and Rehabilitation: State of the Art Reviews 1993;7:1±25. [2] Burke D. Spasticity as an adaptation to pyramidal tract injury. Advances in Neurology 1988;47:401±22. [3] Lance JW. Pathophysiology of spasticity and clinical experience with Baclofen. In: Lance JW, Feldman RG, Young RR, Koella
[22]
[23]
865
WP, editors. Spasticity: disordered motor control. Year Book. Chicago: 1980. p. 185±204. Barnes MP. An overview of the clinical management of spasticity. In: Barnes MP, Johnson GR, editors. Upper motor neurone syndrome and spasticity: clinical management and neurophysiology. Cambridge: Cambridge University Press; 2001. p. 1±11. Bobath B. Adult hemiplegia: evaluation and treatment. London: Heinemann Medical Books; 1990. Harburn KL, Potter PJ. Spasticity and contractures. Physical Medicine and Rehabilitation: State of the Art Reviews 1993;7:113±32. Hass BM, Crow JL. Towards a clinical measurement of spasticity?. Physiotherapy 1995;81:474±9. Pandyan et al. A review of the properties and limitations of the Ashworth and modi®ed Ashworth scales as measures of spasticity. Clinical Rehabilitation 1999;13:373±83. Ashworth B. Preliminary trial of carisoprodal in multiple sclerosis. Practitioner 1964;192:540±2. Bohannon RW, Smith MB. Inter rater reliability of a modi®ed Ashworth scale of muscle spasticity. Physical Therapy 1987;67:206±7. Katz RT et al. Objective quanti®cation of spastic hypertonia. Correlation with clinical ®ndings. Archives of Physical Medicine and Rehabilitation 1992;73:339±47. Given JD et al. Joint dependent passive stiness in paretic and contralateral limbs of spastic patients with hemiparetic stroke. Journal of Neurology, Neurosurgery, and Psychiatry 1995;59:271±9. Wartenberg R. Pendulousness of the legs as a diagnostic test. Neurology 1951;1:18±24. Rothwell JC. Control of human voluntary movement. 2nd ed. London: Chapman & Hall; 1994. Bajd T, Vodovnik L. Pendulum testing of spasticity. Journal of Biomedical Engineering 1984;6:9±16. He J et al. A dynamic neuromuscular model for describing the pendulum test for spasticity. IEEE Transactions on Biomedical Engineering 1997;44:175±83. He J. Stretch re¯ex sensitivity: eects of postural and muscle length changes. IEEE Transactions on Rehabilitation Engineering 1998;6:182±9. Lakie M et al. Assessment of human hemiplegic spasticity by a resonant frequency method. Clinical Biomechanics 1988;3:173±8. Lehmann JF et al. Spasticity: quantitative measurement as a basis for assessing eectiveness of therapeutic intervention. Archives of Physical Medicine and Rehabilitation 1989;70:6±15. Sheean G. Neurophysiology of spasticity. In: Barnes MP, Johnson GR, editors. Upper motor neurone syndrome and spasticity: clinical management and neurophysiology. Cambridge: Cambridge University Press; 2001. p. 12±78. Katz RT, Rovai GP, Brait C, Rymer WZ. Objective quanti®cation of spastic hypertonia. Correlation with clinical ®ndings. Archives of Physical Medicine and Rehabilitation 1992;73:339± 47. Goldspink G, Williams S. Muscle ®bre and connective tissue changes associated with use and disuse. In: Ada L, Canning C, editors. Key issues in neurological physiotherapy. Oxford: Butterworth & Heinemann; 1990. p. 127±218. Speach DP, Dombovy ML. Recovery from stroke: rehabilitation. Bailliere's Clinical Neurology 1994;4:317±38.