Introduction

An upper motor neuron (UMN) syndrome refers to a cluster of broad symptoms including both positive and negative features, with spasticity being one of the most common positive features.1 The traditional definition of spasticity describes it as a motor disorder with a velocity-dependent increase in tonic stretch reflexes, with exaggerated tendon jerks that result from a loss of inhibition of the stretch reflex.2 A high prevalence of spasticity has been found in a number of neurological conditions, such as stroke, spinal cord injury, traumatic brain injury and multiple sclerosis.3–5 Patients with spasticity have poorer quality of life and greater burden of care compared with those without it.6 Such effects are particularly true when spasticity is found in the upper limbs, leading to profound disabilities, low levels of independence and high cost of care.7 8 Therefore, an accurate evaluation of upper limb spasticity is crucial in spasticity management potentially leading to better outcomes and care cost saving. The Ashworth Scale (AS) and Modified Ashworth Scale (MAS) are the most commonly used clinical tools to assess spasticity.9 However, the validity of the AS and MAS has been questioned since they assess the resistance to the passive movement at a constant stretch velocity instead of the velocity-dependent feature of spasticity.10

In addition to the AS and MAS, the Tardieu Scale (TS) is another commonly used tool for spasticity assessment.11 Evidence suggests that the TS is significantly better at identifying spasticity than the AS.12 The main strength of the TS is to identify the presence and severity of spasticity by objectively measuring the muscle responses to passive stretches at different speeds.12 This characteristic allows the TS to reflect the velocity-dependent feature of spasticity better than the AS and MAS.11 The TS was first described in French,13 and later developed and translated into English by Held and Pierrot-Deseilligny.14 Boyd and Graham15 standardised the testing speed and procedures, and this version was later known as the Modified Tardieu Scale (MTS). The MTS comprises four elements, including R1, R2, R2-R1 and X score. R1 represents an angle of catch, where a sudden increase of muscle resistance is felt during a fast passive stretch. R2 is an angle indicating the tested muscle length at a slow passive range of motion. R2-R1 was introduced to differentiate spasticity from contracture.15 Contracture is indicated if the value of R2-R1 is small, while a large value indicates spasticity.15 The X score describes types of muscle resistance when passive stretches are applied during the assessment (grade 0–5).

A systematic review was published in 2006 on the validity and reliability of the TS.16 However, the research evidence underpinning it was limited by methodological weaknesses. These included an unclear search date range, a lack of validity studies and samples comprising mostly paediatric participants. Despite two decades passing since the development of the MTS and its common use clinically,17 relatively little published research has examined its features. Moreover, evidence on evaluation of its psychometric properties is scarce. Therefore, this systematic review is warranted to explore existing evidence on the validity and reliability of the MTS to assess spasticity in the upper limbs of adults with neurological conditions. On the basis of the identified research, we will make conclusions on the quality of evidence to date and recommendations for future research by identifying gaps in evidence surrounding its use.

MethodDesign

Systematic review and narrative synthesis.

Information sources

A systematic literature search was conducted using PubMed/MEDLINE, CINAHL PLUS, EMBASE, the Cochrane Library, Web of Science and Physiotherapy Evidence Database from inception to 31 December 2020. After all the duplicates were removed, a hand search of the reference lists of the remaining articles was conducted to identify additional relevant studies.

Search strategy and study selection

A search strategy was developed in PubMed/MEDLINE using key elements of the research question: population, intervention (action) and outcome. It was later adapted for the other databases searched. Part 1 (population) included common neurological conditions and the term ‘disability’ as well as variations of these keywords. Part 2 (intervention/action) included the search term ‘Modified Tardieu Scale’ within text words. The last part (outcome) focused on spasticity and its variations. Medical subject headings were included in searches where possible. Three parts were combined using Boolean operator AND (see online supplemental appendix 1). The search procedure followed the Preferred Reporting Items for Systematic Review and Meta-Analysis reporting guidelines.18 XS and AK consulted a librarian and designed the search strategy together, while XS and CM conducted the search independently.

Eligibility criteria

The inclusion criteria were: (1) adult participants with neurological conditions; (2) upper limb muscles/joints as tested elements; (3) studies testing the MTS and (4) reliability or validity reported. The exclusion criteria were: (1) non-English articles; (2) non-empirical studies and (3) studies testing the TS.

Data extraction, synthesis and analysis

Following completion of the search, collection of articles and application of the inclusion and exclusion criteria, data on reliability and validity were extracted from the articles. These data were recorded by XS and CM independently and the results were verified. The results were summarised under the following headings: (1) Authors (year); (2) Study design; (3) Participants (n=, health conditions, source of recruitment); (4) Test characteristics (Tested muscle, testing position, testing time and interval, testing speed, measuring, rater training, MTS version, any other relevant details) and (5) Summary of results (inter-rater, intrarater and test–retest reliability; validity).

Quality assessment

The internal validity of individual studies was assessed using the US National Heart, Lung, Blood Institute (USNHLBI) quality assessment tool for observational cohort and cross-sectional studies19 (online supplemental appendix 2). According to recommendations from Sommer et al,20 studies were rated as poor, fair or good with scores of less than 50%, 50%–75% and greater than 75%, respectively.

Patient and public involvement

Patient and public involvement was a part of this study from conception to completion, with author AK being a community physiotherapy service user for spasticity management. The MTS was used by his community physiotherapist, XS, to assess his spasticity. XS and AK discussed the validity and reliability of the MTS during a physiotherapy session. XS and AK commenced a collaboration to undertake this research paper, with an aim to contribute to the spasticity research literature and to inform practice for clinicians.

ResultsOverview

A total of 292 studies were retrieved from the search of six databases. After the duplicates were removed by a software (Endnote) and researchers manually, the remaining 160 articles were retained for screening. One additional article was identified through reference list searching. Nine papers were retrieved for eligibility assessment after title and abstract screening. Finally, seven articles were included for this review after two non-empirical papers were removed. The details of the search process are displayed in figure 1.

Figure 1

Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow chart of the study identification and inclusion process.

Characteristics of included studies

Seven observational studies met all inclusion criteria. Descriptive data of the studies are listed in table 1. Six articles reported data on reliability: four of these examined inter-rater reliability,21–24 two studied intrarater reliability24 25 and three investigated test−retest reliability.21 23 26 Only one article reported data on criterion validity.27 The total number of participants across seven studies was 317, with sample sizes between 20 and 91. Most participants were stroke survivors (n=222). The studies described the test characteristics at varying levels of detail. For instance, raters’ training was described in three studies.21 23 25 Two studies specified testing speed and marked the tested joints.25 26 Elbow flexors were the most commonly tested muscles across the studies.21–26 The statistical methods were also described in all papers except Sonvane and Kumar.26 The most frequently used statistical methods were intraclass correlation coefficients (ICC) (six studies) and t-tests (seven studies).

Table 1

Data extraction table for MTS studies reporting reliability and validity data

Summary of the study results

Table 1 displays reliability estimates of the MTS in the reviewed studies: ICC values range from 0.56 to 0.851 for inter-rater reliability,21–24 from 0.70 to 0.998 for intrarater reliability24 25 and from 0.627 to 0.815 for test–retest reliability.21 23 Most studies listed data of all four items of the MTS (R1, R2, R2-R1 and X scores).22 24–26 Nevertheless, Mehrholz et al21 and Waninge et al23 only reported ICC values of two items in their papers. X scores, a categorical item, were measured using ICC by Ansari et al22 and Singh et al,25 while Mehrholz et al21 and Li et al24 chose Kappa to assess X scores and reported a wide range between 0.33 and 0.87. Instead of ICC or Kappa, Sonvane and Kumar26 selected Pearson’s correlation coefficients and t-tests to assess the overall reliability of the MTS. They reported satisfactory correlation values (r=0.913–0.973) and statistically significant p values (p<0.001) of all four elements.26 No articles were found in the systematic search discussing the content or construct validity of the MTS, which was a notable gap in the research literature. One article investigated the criterion validity of the MTS.27 Naghdi et al27 reported poor correlations between the MTS and electrophysiological measurements of the H-reflex in all four elements when testing the wrist flexors.

Quality of evidence

We modified the USNHLBI quality assessment tool19 because five items of the tool were not applicable to the reviewed studies (items 6, 7, 8, 10 and 13). Accordingly, the quality of the reviewed studies was classified as poor for total scores between 1 and 4, fair for scores between 5 and 7 and good for scores above 7. The quality of most studies included in the review was poor22–24 26 27 with studies by Mehrholz et al21 and Singh et al25 studies being rated good and fair, respectively, (table 2). All the studies clearly defined the purpose of the research (item 1). None of the studies justified the sample size (item 5), with variable details of participant criteria and selection (items 2, 3 and 4). Only Singh et al,25 Mehrholz et al21 and Naghdi et al27 correctly defined the assessment of the reliability or validity of the MTS (item 11), while assessors were blinded in three studies (item 12).21 22 25 Two studies considered confounders in statistical analysis (item 14).21 23

Table 2

Quality assessment tool for observational cohort and cross-sectional studies

Discussion

There is insufficient evidence to either support or reject the use of the MTS to assess the spasticity in the upper limbs of adults with neurological conditions. Although the selected papers showed positive results regarding the reliability of the MTS,21–26 further analysis revealed methodological weaknesses and low study quality affected the credibility of the results. The only identified paper assessing the validity of the MTS contained study design limitations.27 This discussion summarises the main findings and critically appraises them. Then, possible explanations and implications for clinicians and policy-makers are explored. The strengths and limitations of this review are discussed, followed by recommendations for future research.

Summary of findings

Koo and Li28 suggested an ICC value above 0.5 indicated moderate-to-excellent reliability. Across all the papers reviewed, even though all ICC values were found above 0.5, four studies reported moderate-to-excellent levels of inter-rater, intrarater and test–retest reliabilities.21 24–26 This is due to a lack of consensus on acceptable levels of agreement among these studies. For instance, in spite of high ICC values for most tested items, Waninge et al23 reported insufficient inter-rater and test–retest reliabilities because the results of the reliability tests failed to meet their own statistical criteria (acceptable ICC>0.75). Similarly, using different assessment criteria, Ansari et al22 concluded that the inter-rater reliability was in fact unacceptable due to great variabilities of the R2-R1 values and the X scores between raters.

Evidence on the validity of the MTS is lacking, and further research in this area is required. The only validity study returned by our search suggested low confidence in using the MTS clinically as a validated tool.27 However, the authors acknowledged there was no gold standard to assess spasticity against the MTS,27 thereby the internal validity of this study is questionable.

Furthermore, the quality of study assessment suggests five out of seven reviewed studies were poor quality.22–24 26 27 The only study rated as good quality,21 testing several upper limb muscle groups, but only R1 and X scores data for the elbow flexors were presented in the paper. Therefore, the available data may not be sufficient to lend credibility to the findings. In summary, the seven studies included in this review do not provide adequate evidence to support the use of the MTS to assess spasticity in the upper limbs of adults with neurological conditions. However, this may be due to the low overall quality of the reviewed studies.

Critical discussion of review findings

Closer examination of the seven studies shows all studies contained methodological limitations, such as non-standardised use of the MTS and small samples without power calculations.21–27 The largest sample was 91 participants,25 while 3 studies had sample sizes below 35.21–23 Small sample sizes are known to cause a wide 95% Confidenc Interval (CI) range.29 For instance, the 95% CI values of R2 reported by Ansari et al22 and Li et al24 were 0.26 to 0.76 and 0.36 to 0.73, respectively, making it difficult to have confidence in the reported means. Moreover, none of the studies acknowledged the fact that there were two versions of X scores. Boyd and Graham15 added an extra item on the X score category of the TS, extending it from five to six scores. Mehrholz et al21 used categories 0–5, while Ansari et al22 and Singh et al25 selected the TS version (0–4). Neither Waninge et al23 nor Sonvane and Kumar26 specified which scale was investigated. Li et al24 claimed to study a 5-score scale, but their results suggested they used a 6-score scale instead. Non-standardised use of the scale made it hard to compare the X score results across the studies.

Misuses of statistical methods can be found in the reviewed papers, leading to misinterpretation of results and incorrect conclusions being drawn. For example, ICC and 95% CI are parametric statistical procedures that should only be used on continuous data.30 However, X scores (discrete data) were analysed using ICC and 95% CI in two studies.22 25 Similarly, Sonvane and Kumar26 analysed X scores choosing Pearson’s correlation coefficient, a measure of the association between two continuous variables.31 Furthermore, reporting biases were noted in some studies. For instance, Waninge et al23 only listed satisfactory ICC values of R1 and R2, but measurements all four elements were described in the method. The positive conclusion may have been overstated if based on the results of only two elements. The results of the X score and R2-R1 could affect the overall estimates of the MTS. In summary, despite reported moderate-to-excellent reliability, close examination of the research studies cited in this review suggests that many factors affect the credibility of the research findings, including reporting biases and methodological weaknesses.

Implications

The present results and analysis lead to some practical implications for clinicians, researchers and policy-makers. Reviewing the procedures employed across the studies suggests that further standardisation of the MTS may be warranted. For instance, sitting, supine and prone MTS testing positions have been reported.32 Researchers found significantly lower tone in the upper limb muscles in the supine position than in sitting or standing positions.33 In addition, testing speed was not standardised during the tool development. Mackey et al34 found ‘considerably variable’ speed between participants when the biceps muscles were assessed. Moreover, the fact that there are two versions of the X score scale may cause confusion, resulting in miscommunication between researchers and clinical users.

The validity of the MTS remains unclear, partially due to a lack of consensus on a spasticity definition and a full understanding of its pathophysiology. A systematic review of 250 papers revealed inconsistency in defining spasticity.1 Only about 30% of reviewed studies explained spasticity according to Lance’s definition, a third simply described it as increased muscle tone, and the rest used their own definitions or none at all. Lacking an agreement on spasticity definition has made it difficult to decide on the most valid measurement. Furthermore, the classic definition of spasticity2 is believed only to reflect its motor aspects.35 The TS and MTS adhere to this definition by addressing the velocity-dependent feature of spasticity.11 A more recent definition describes spasticity as a disorder of sensorimotor control, resulting from an UMN lesion, presenting as intermittent or sustained involuntary muscle activation.35 This broad description includes other positive signs of an UMN syndrome, such as clonus and spasm, which are often used interchangeably with spasticity.36 Moreover, an abnormal sensory process was also acknowledged as a factor to induce spasticity.37 Nevertheless, no further recommendations have been made either to measure the sensory aspect of spasticity or to differentiate spasticity from other positive signs of an UMN syndrome. The MAS, which measures spasticity as resistance to passive movements,38 may be more likely to address the ‘intermittent or sustained involuntary muscle activation’ feature described by Pandyan et al.35 Therefore, to support research on the reliability and validity of the MTS, we recommend further work to establish a consensual definition of spasticity.

Despite all the pitfalls discussed above, the MTS may still be clinically useful to assess the velocity-dependent feature of the spasticity based on the current knowledge of spasticity pathophysiology and previous studies of the TS. First, losing central inhibition following UMN lesions, spasticity is caused by alpha motor neuron hyperexcitability leading to an exaggerated stretch reflex.37 Power et al39 reported that by applying the same stretch velocity, the stretch response threshold reduced and the amplitude was increased in people with spasticity compared with unaffected subjects. Therefore, the neuropathophysiology supports that using the MTS by performing a fast speed stretch could provoke spasticity as a result of an overactive stretch reflex.

Furthermore, the content validity of the TS was investigated by comparing the manual identification of spasticity with laboratory measurements.40 The TS successfully detected 88.9% of spasticity (p<0.05). Patrick and Ada12 reported 100% exact agreement between the TS and laboratory measurements. In both studies,12 40 spasticity was identified by performing stretches at a fast speed. The MTS assesses spasticity in the same way, meaning the MTS may also have excellent content validity in assessing the motor features of the spasticity. So, the MTS could still be considered a useful clinical tool to assess the presence of spasticity.

Other strengths of the MTS should not be underestimated. For example, the costs associated with conducting the MTS are very low, including just a goniometer and a standard plinth. The MTS instructions and assessment forms are available online free of charge, and training to use the tool can be completed in as little as 45 min.21 22 Moreover, rater experience does not seem to have a significant impact on the results. Studies recruiting experienced physiotherapists21 25 and students22 all reported consistent results. Feasibility of the MTS was also assessed by comparing the successful measurements to total measurement numbers.23 The MTS has been reported to have good feasibility, with a 94% success rate.23

Strengths and limitations

To our knowledge, this study is the first to investigate the validity and reliability of the MTS to assess upper limb spasticity for the adults with neurological conditions. Moreover, a comprehensive search of multiple databases and hand searches of references ensured studies selected for the review were up to date. Furthermore, an in-depth analysis of results and characteristics has allowed us to understand the state of current evidence on the MTS. We acknowledge that our search was biased toward English publications and may have missed relevant studies written in other languages. Furthermore, this search only identified seven studies in total, six of which tested the elbow flexors. Therefore, the reported findings may not be applicable to other untested upper limb muscle groups.

The only similar study to ours in the literature was a systematic review of the TS conducted in 2006.16 Of the research reviewed, most studies recruited children as subjects. Our review has added evidence of the MTS in adult population. Although neither Haugh et al16 nor the present review found sufficient evidence to support the validity and reliability of the TS and MTS. A direct comparison of the two papers is problematic due to the heterogenous samples.

Recommendations

Based on a critical review of the study results, we suggest that future research studying psychometric properties of the MTS should focus on improving study designs, including test standardisation, statistical power and sample size calculation, and using appropriate statistical analysis. Second, the internal validity of the MTS study could be compromised by an inconsistent definition of spasticity.1 Therefore, further efforts could be made to develop the definition of spasticity and subsequently develop the MTS congruent to that definition. Finally, according to the International Classification of Functioning, Disability and Health (ICF) framework,41 the MTS only reflects the motor aspect of the spasticity on the body functions and structures level. Multidimensional assessments of disabilities, including those on the activity and participation levels, were recommended to form a holistic understanding of the impacts’ spasticity could make on individuals’ lives.1 17 For instance, the Action Research Arm Test, and the Nottingham Extended ADL can be applied as measurements of functions (activity level) and quality of life (participation level), respectively.17 42