Results from the Upper Limb International Spasticity Study-II (ULIS-II): a large, international, prospective cohort study investigating practice and goal attainment following treatment with botulinum toxin A in real-life clinical management

Abstract

Objective

To describe real-life practice and person-centred outcomes in the treatment of poststroke upper limb spasticity with botulinum toxin A (BoNT-A).

Design

Observational, prospective study.

Setting

84 secondary care centres in 22 countries.

Participants

456 adults (≥18 years) with poststroke upper limb spasticity treated with one cycle of BoNT-A.

Methods/outcomes

Muscle selection, BoNT-A preparation, injection technique and timing of follow-up were conducted according to routine practice for each centre. Primary outcome: achievement of the patient's primary goal for treatment using goal-attainment scaling (GAS). Measurements of spasticity, standardised outcome measures and global benefits were also recorded.

Results

The median number of injected muscles was 5 (range 1–15) and the most frequently injected muscles were the long finger flexors, followed by biceps and brachioradialis. The median (range) follow-up time was 14 (2.6 to 32.3) weeks. The common primary treatment goals were passive function (132 (28.9%)), active function (104 (22.8%)), pain (61 (13.4%)), impairment (105 (23%)), involuntary movement (41 (9%)) and mobility (10 (2.2%)). Overall, 363 (79.6%) (95% CI 75.6% to 83.2%) patients achieved (or overachieved) their primary goal and 355 (75.4%) (95% CI 71.2% to 79.2%) achieved their secondary goal. Mean (SD) change from baseline in GAS T-scores was 17.6 (11.0) (95% CI 16.4 to 18.8; p<0.001). GAS T-scores were strongly correlated with global benefit and other standard measures (correlations of 0.38 and 0.63, respectively; p<0.001).

Conclusions

BoNT-A demonstrated a clinically significant effect on goal attainment for the real-life management of upper-limb spasticity following stroke. The study confirms the feasibility of a common international data set to collect systematic prospective data, and of using GAS to capture person-centred outcomes relating to passive and active functions and to pain.

Registration

ClinicalTrials.gov identifier: NCT01020500

Article summary.

Article focus

• A large international observational cohort study (the Upper Limb International Spasticity (ULIS)-II) to describe the use of botulinum toxin A (BoNT-A) for management of upper-limb spasticity in the context of real-life clinical practice.

• To quantify and characterise the achievement of person centred goals following one BoNT-A injection cycle delivered in the context of routine clinical practice.

• To describe the variations in clinical practice and explore prognostic factors that may impact on outcome.

Key messages

• Despite wide variations in the approach to clinical practice, a large majority (80%) of the patients achieved their treatment goals, mainly in terms of passive and active functions and pain reduction.

• The results provide evidence that BoNT-A injections may contribute to an improvement in the daily life of patients and their carers beyond simple improvement of tone or spasticity.

Strengths and limitations of this study

• The wide geographical distribution of centres across three continents is a strength of this study, but recruitment of only 5–12 patients/site may not adequately reflect the patient population of each centre.

• The study lays the foundation for larger international longitudinal cohort studies to explore further the characteristics and treatment approaches that predict best outcomes in BoNT-A treatment of upper limb spasticity.

Introduction

Spasticity is a common and distressing sequela of stroke, which interferes with upper limb movement and limits the use of the limb for active functional tasks, as well as impacting on mobility and increasing the burden on caregivers.¹

From controlled clinical trials conducted to date,^2–9 it is established that botulinum toxin A (BoNT-A) is safe and effective in reducing spasticity. However, functional gains have been harder to demonstrate,¹⁰ especially as there is wide diversity in the pattern of spasticity and goals for treatment leading to individual variations in response. We also know that clinicians vary in their approach to treatment with respect to selection of muscle, injection technique and follow-up therapy. These variations appear to have more to do with clinician bias and local availability of services than with patient presentation.¹²

It is now time to extend the field of investigation in this area to understand how BoNT-A is used in routine clinical practice around the world and gain better understanding of how to select patients most likely to respond to treatment. Establishing what treatment approaches work best based on clinical presentation would also be of great clinical value. To do this, we will need to build a consistent body of data that captures clinically important change at an individual level and is of sufficient size and generalisability to enable us to answer these critical questions.

The Upper Limb International Spasticity (ULIS) programme consists of a series of large international observational studies to describe current clinical practice in the application of BoNT-A in this context, and to work towards the development of a common international data set for prospective systematic recording of longitudinal outcomes.¹³ Importantly, the programme incorporates elements of a) training in the use of agreed outcome measures, and b) development of electronic data-collection tools suitable for use by the wider international community.

Goal-attainment scaling (GAS) was chosen as the primary person-centred outcome measure to capture the diversity of treatment intentions that are important to the individual patient and their family/carers. First described by Kiresuk and Sherman¹⁴ in the 1960s, GAS is increasingly used in the context of spasticity management^{15 16} and is shown to be sensitive to changes following focal intervention that are not detected by more global measures.¹¹ Although GAS provides a useful measure of achievement of treatment intention, it does not measure outcome per se, and therefore does not stand alone but is used alongside other standardised outcome measures.¹⁷ It is therefore pertinent to understand the relationship between GAS and other outcome tools.

This second stage in the ULIS programme (ULIS-II) is a large international observational cohort study to describe the real-life practice and outcomes in the treatment of poststroke upper limb spasticity with BoNT-A. This is the first large international cohort study to use GAS as a primary outcome measure for the person-centred evaluation of treatment with BoNT-A for spasticity. In a separate paper, we have presented the rationale and methodology for ULIS-II in detail, and described the steps taken to ensure the validity of GAS as a measure of functional gains in this context.¹³ In this study, we describe the baseline clinical characteristics, details of interventions and the primary results in accordance with the STROBE guidelines for presentation of cohort studies.¹⁸

Aims and objectives

The primary objective of the ULIS-II study was to assess the responder rate (as defined by the achievement of the primary goal from GAS) following one BoNT-A injection cycle delivered in the context of routine clinical practice.

Secondary objectives were to

• Describe the baseline characteristics, including demographics, duration and pattern of spasticity, concomitant therapies/medication, etc;

• Describe injection practices (muscle identification, dosage, dilution, etc);

• Assess achievement of secondary goals and evaluate the overall attainment of treatment goals using the GAS T-score as a patient-centred measure of outcome;

• Document the use of standardised outcome measures and their results;

• Assess the global benefits as perceived by the investigator and the patient.

Additional exploratory objectives, addressed through the analysis plan, were to

• Describe the common goal areas for treatment and to identify those in which goals were most often achieved;

• Examine the relationship between GAS and other standardised outcome measures;

• Identify any prognostic factors for response.

Methods

Full details of the methodology are described elsewhere.¹³ In brief, ULIS-II was an 18-month, postmarketing, international, multicentre, observational, prospective, before-and-after study, conducted in 84 centres in 22 countries spanning Europe, Pacific Asia and South America.

The study was conducted in compliance with Guidelines for Good Pharmacoepidemiology Practices. Marketing authorisation for the use of BoNT-A in this context was ensured for each participating country prior to the start of the study. Ethical approval and written informed consent to the recording of anonymous data were obtained in countries where this was required.

Recruitment took place between January 2010 and May 2011. To limit the potential bias from over-recruiting sites, the number of patients was limited to 5–12 patients/treatment centre. All centres recruited at least five patients, but 25 of the more experienced centres (which were usually also larger) could recruit up to 12 patients. This was allowed to try to ensure a representative sample from clinicians with experience in this area of practice. It also offers an opportunity for future subanalysis of the differences between experienced and less experienced injectors.

Centres included consecutive patients—or spaced inclusions in a predefined manner (eg, one for every 2–3 patients) if necessary for pragmatic reasons—until their recruitment target was achieved.

Study population

The main inclusion criteria required patients to be consenting adults ≥18 years with poststroke upper limb spasticity in whom a decision had already been made to inject BoNT-A, and who had had no previous treatment with BoNT-A or BoNT-B within the last 12-weeks. Agreement on an achievable goal set and ability to comply with the prescribed treatment were also required. The efficacy population analysed here included all participants who received one BoNT-A injection and who underwent a postinjection visit including an assessment of GAS.

Study schedule

Baseline evaluation at Time 1 included

• Demography and history of the stroke including type, location and time since onset.

• The pattern of impairment in the affected upper limb (modified Neurological Impairment Scale).^{19 20}

• Previous/concomitant treatments for upper limb spasticity.

• Clinical examination, including measurements of spasticity and other standardised outcome measures as routinely performed in that centre.

• Goal setting and GAS applied using the ‘GAS-light’ method,²¹ as detailed in the rationale and methodology paper,¹³ with emphasis on setting SMART (specific, measurable, achievable, realistic and timed) function-related goals agreed on between the investigator, the patient and the treating team.

• One primary and up to three secondary goals were set and assigned to one of seven goal categories.

Injection of BoNT-A

To reflect real-life practice in this observational study, physicians were free to choose targeted muscles, BoNT-A preparation, injected doses, number of points and volume for each point and use of EMG/electrical stimulation in accordance with their usual practice, and with their local Summary of Product Characteristics and therapeutic guidelines. The timing of follow-up was at the discretion of the investigator, based on their usual practice and the nature of the goals set, usually between months 3 and 5.

Follow-up evaluation at Time 2 included

• Achievement of primary and secondary GAS goals rated on a six-point verbal rating scale, and transcribed within the computer software to the five-point numerical scale (range −2 to +2), and the GAS T-score.

• Any concomitant treatments for upper limb spasticity given since baseline.

• Clinical examination including measurements of spasticity as normally routinely performed.

• Global assessments of benefits were rated by the investigator and patient as: ‘great benefit (+2)’, ‘some benefit (+1)’, ‘same (0)’, ‘worse (−1)’ or ‘much worse (−2)’.

• Change on any standardised measures performed was recorded on the same five-point scale.

• The next therapeutic strategy—including any planned re-injection with BoNT-A—whether using the same agent and protocol or an adjusted one.

As this was an observational study, reporting of related adverse events followed the standard regulations related to spontaneous adverse event reporting for marketed products.

Study size

The sample size calculation was based on an estimate that 60% of patients would achieve their primary goal following their first BoNT-A injection cycle. Using a 5% two-sided significance level, with a power of 80%, 450 patients were needed to allow estimation of this proportion with a precision of 4.5%. This sample size also allowed the detection of potential prognostic factors to response (based on the detection of odd ratio (OR) larger than or equal to 2).

Statistical analysis

Data were entered by the treating clinicians into an electronic case report form (eCRF). After cleaning and validation of the data set, statistical evaluations were performed using Statistical Analysis System (V.9).

Analyses were conducted on the efficacy population. For the primary statistical analysis, ‘Responders’ were those who achieved their primary goal (GAS score 0, 1 or 2).

• Baseline characteristics and efficacy evaluations are presented as descriptive statistics, including 95% CI where relevant.

• Mean and SD are reported for interval quality data, including long-ordinal data that fulfilled the criteria for normal distribution (eg, Modified Ashworth Scale (MAS) total and GAS T-scores).

• 95% CI for percentage were calculated as p±1.96×SE. SEs were calculated as √(pq/n), where p is the rate, q=1–rate and n=sample size.

• Short ordinal data are described by median and IQR and analysed using non-parametric statistical techniques.

As originally described by Kiresuk and Sherman, the GAS T-score provides a composite score (the sum of the attainment levels×the relative weights (optional) for each goal) transformed into a standardised measure with a mean of 50 and an SD of 10. Several different scoring methods are currently used in the literature to account for partial achievement of goals in GAS.²² In this study, baseline scores were rated as −1=‘some function’ and −2=‘no function’ with respect to the goal. T-scores were calculated with weighting (importance), and partial achievement was rated as −1 to conserve the normal distribution of scores.²²

MAS scores were recorded for the shoulder, elbow, wrist, fingers and thumb joints. Scores of ‘1+’ were entered as 1.5 in the calculation of the total MAS score and combined to composite scores as follows: MAS-Proximal=shoulder+elbow scores; MAS-Distal=wrist+fingers+thumb scores; MAS-total=composite sum of all five joints.

Relationships between GAS T-scores and other measures of outcome (eg, measures of spasticity, global benefit and other standardised measures) were examined using Spearman's rank correlation coefficients.

Stepwise logistic regression modelling was used to identify prognostic factors for achievement of the primary goal. Potential covariates included stroke aetiology, primary goal area, duration and severity of spasticity, time interval to follow-up, presence of confounding factors (including the presence of contractures; impaired motor, sensory, cognitive, emotional and cortical function (ie, neglect, dyspraxia, visuospatial deficits, etc). A backward elimination analysis was followed using a significance level of 0.2 to retain variables in the model. The Hosmer and Lemeshow goodness-of-fit test¹⁸ was used and 95% CI for OR estimated by the logistic model was calculated.

Results

Recruitment and participation

A total of 468 participants were enrolled in this study of which 12 were excluded from the efficacy population (n=456). Eleven participants were excluded because they did not attend their follow-up visit for assessment of GAS (n=5 lost to follow-up, n=3 subject death unrelated to study medication and n=3 for other reasons). An additional participant was excluded as treatment was given in the lower (rather than the upper) limb. No participants withdrew because of withdrawal of consent, lack of efficacy or adverse effect related to the BoNT-A treatment.

Demographics and disease characteristics

The geographic distribution and demographics of the efficacy population are described in detail elsewhere.¹³ The mean (SD) age was 57 (13.5) years and mean (SD) time since onset of stroke was 61.4 (69.1) months. Fifty-eight per cent of the population was men; 70% had had infarcts and 30% had haemorrhagic stroke. Left and right hemisphere localisation was approximately equal (47.1 : 51.1%, respectively) and 3% had posterior circulation strokes.

Baseline characteristics are detailed in table 1. Distal patterns of spasticity predominated. A quarter had evidence of soft tissue shortening and more than half had severe motor weakness. Over half of the participants (57%) had no useful hand function and 46% had sensory impairment. These findings confirm that the majority of participants had chronic spasticity with severe impairment and therefore little potential for recovery of useful motor function. In contrast, the cognitive and communication impairments were relatively uncommon and were mostly mild in nature.

Table 1.

Baseline clinical characteristics of the efficacy population

Parameter	Values (%)	Range	n/missing or untestable
Distribution of spasticity (MAS ≥2), n (%)
Shoulder	235 (55.7)		422/34
Elbow	335 (75.1)		446/10
Wrist	344 (77.8)		442/14
Fingers	368 (82.9)		444/12
Thumb	292 (66.8)		437/19
Soft tissue shortening (limiting ≥half range), n (%)			456/0
Shoulder	156 (34.2)
Elbow	150 (32.9)
Wrist	179 (39.3)
Hand	206 (45.2)
Motor paralysis (no useful function), n (%)			456
Proximal	95 (20.8)
Distal	260 (57.0)
Sensory, n (%)			447/9
Partial	208 (45.6)
Complete	20 (4.4)
Cognitive/communicative impairment, n (%)
Cognitive	98 (21.5)		455/1
Speech and language	159 (34.9)		446/10
Neglect, dyspraxia or visuo-perceptual difficulties	60 (13.2)		456/0
Other symptoms, which may impact on functional outcome, n (%)			456/0
Emotional/behavioural impairment	149 (32.7)
Pain*	78 (17.1)
Fatigue	48 (10.5)
Impairment scores (modified NIS) †	Median (Q1–Q3)	Range	n/untestable
Proximal upper limb motor score (arm raising/reaching)	2 (1–2)	0–3	456/0
Distal upper limb motor score (hand function)	3 (2–3)	0–3	456/0
Sensation	1 (0–2)	0–3	447/9
Communication impairment	0 (0–1)	0–3	455/1
Cognitive impairment	0 (0–0)	0–3	446/10
Spasticity (MAS)‡			n/missing
Distal composite MAS score	7.0 (6–9)	0–12	431/25
Proximal composite MAS score	4.0 (3–5)	0–8	422/34
Baseline outcome measures
Total MAS score, mean (SD)	11.0 (3.3)	1–20	414/42
GAS weighted score, mean (SD)	36.4 (7.7)	21.2–43.8	456/0

As multiple answers were possible, the sum of some percentages in this table may exceed 100%.

*Data extracted from injected segments at baseline, when at least one muscle of the segment was injected.

†Modified NIS: the score range for each of the five domains is 0=none; 1=mild impairment affecting high-level function only; 2=significant impairment; 3=severe impairment, effectively preventing function. Further details are available from the corresponding author.

‡MAS composite scores: proximal MAS score=shoulder+elbow scores; MAS-distal=wrist+fingers+thumb scores; MAS-total=composite sum of all five joints (total composite scores were only calculated if individual MAS scores were recorded at all five joints).

MAS, Modified Ashworth Scale; NIS, Neurological Impairment Scale.

BoNT-A injection history

Approximately two-thirds of the participants (n=307; 67.3%) had received a previous injection of BoNT-A in the upper limb. The mean (SD) time since the last injection was 8.0 (11.5) months and the median time was 5 months (IQR 3–5, range 1–102 months). Some had had treatment spanning several years.

The median time since the first injection was 24 months (range 3–168 months), but two-thirds of the participants had had BoNT-A treatment for over 1 year. The median number of BoNT-A injections previously received by the participants was 4 (IQR 1–8; range 1–45).

BoNT-A treatment

In this cohort, abobotulinumtoxinA (Dysport) was the most commonly used agent (70.4%), followed by onabotulinumtoxinA (Botox; 21.5%) and incobotulinumtoxinA (Xeomin; 7.7%); two patients received another local BoNT-A preparation. The median number of injected muscles was 5.0 (range 1–15). There was very wide variation in the total dose of BoNT-A (see table 2).

Table 2.

BoNT-A treatment

Current injection
Agent*	Dysport  n=321 (70.4%)	Botox  n=98 (21.5%)	Xeomin  n=35 (7.7%)
No. of injected muscles
Median (IQR)	5 (2)	5 (2)	5 (2)
Range	1–11	1–15	3–9
Dose
Total dose range (units)	40–1900	50–500	100–600
Localisation of injection (used for at least one muscle), n (%)
EMG	91 (28.3)	34 (34.7)	6 (17.1)
Electrical stimulation	145 (45.2)	53 (54.1)	9 (25.7)

*Two participants received other BoNT-A preparations.

BoNT-A, botulinum toxin A; EMG, electromyography.

The most commonly injected muscles and doses, by treatment, are shown in table 3. The most frequently injected were the long-finger flexors, followed by biceps and brachioradialis. With the exception of the pectoralis major (which was injected in 19.3% of patients), the shoulder muscles were relatively rarely injected. Multiple injection points were most commonly used in the larger and more proximal muscles, such as the biceps and the pectoralis major. Electrical stimulation was more commonly used to locate muscles than EMG (45.8% vs 29.2%), especially for the smaller muscles such as the flexor pollicis longus.

Table 3.

Most commonly injected muscles and technique within each upper limb segment

Group/muscle	Total number injected (%)*	Number of units: median (IQR)			Range of units (min, max)			Multiple points (%)	Use of EMG (%)	Electrical stimulation (%)
Dysport		Dysport	Botox	Xeomin	Dysport	Botox	Xeomin
Shoulder	n=147 (32.2)
Pectoralis major	88 (19.3)	200.0 (150.0)	30.0 (40.0)	70.0 (30.0)	30, 750	10, 100	20, 140	55 (62.5)	11 (12.5)	25 (28.4)
Teres major	17 (3.7)	75.0 (50.0)	10.0 (10.0)	50.0 (60.0)	50, 200	10, 50	40, 100	2 (11.8)	8 (47.1)	8 (47.1)
Deltoideus	15 (3.3)	100.0 (100.0)	50.0 (0.0)	N/A	50, 300	50, 50	N/A	1 (6.7)	3 (20.0)	7 (46.7)
Subscapularis	14 (3.1)	200.0 (100.0)	30.0 (0.0)	50.0 (0.0)	75, 320	20, 40	50, 50	3 (21.4)	1 (7.1)	11 (78.6)
Latissimus dorsi	9 (2.0)	120.0 (125.0)	55.0 (90.0)	40.0 (0.0)	75, 200	10, 100	40, 40	6 (66.7)	4 (44.4)	2 (22.2)
Upper arm	n=336 (73.7)
Biceps brachii	270 (59.2)	200.0 (150.0)	50.0 (10.0)	55.0 (40.0)	50, 750	20, 100	20, 100	208 (77.0)	52 (19.3)	70 (25.9)
Brachialis	130 (28.5)	150.0 (100.0)	50.0 (20.0)	60.0 (27.5)	20, 400	10, 100	30, 80	76 (58.5)	30 (23.1)	62 (47.7)
Triceps brachii	18 (3.9)	175.0 (100.0)	30.0 (25.0)	N/A	60, 300	20, 100	N/A	15 (83.3)	8 (44.4)	8 (44.4)
Lower arm	n=434 (95.2)
Flexor digit. superficialis	325 (71.3)	150.0 (100.0)	50.0 (35.0)	60.0 (37.5)	20, 500	15, 150	30, 100	195 (60.0)	91 (28.0)	151 (46.5)
Flexor digit. profondus	265 (58.1)	150.0 (100.0)	50.0 (30.0)	50.0 (20.0)	50, 600	15, 150	20, 200	142 (53.6)	66 (24.9)	114 (43.0)
Flexor carpi radialis	262 (57.5)	125.0 (100.0)	33.3 (25.0)	45.0 (40.0)	20, 350	5, 100	20, 80	83 (31.7)	73 (27.9)	99 (37.8)
Brachioradialis	156 (34.2)	112.5 (100.0)	40.0 (35.0)	50.0 (20.0)	25, 300	10, 75	20, 75	55 (35.3)	29 (18.6)	46 (29.5)
Pronator teres	138 (30.3)	100.0 (75.0)	40.0 (25.0)	40.0 (30.0)	25, 500	10, 100	10, 50	44 (31.9)	26 (18.8)	53 (38.4)
Flexor pollicis longus	136 (29.8)	100.0 (50.0)	25.0 (12.5)	35.0 (30.0)	20, 250	5, 50	10, 60	29 (21.3)	31 (22.8)	79 (58.1)
Hand/fingers	n=204 (44.7)
Flexor pollicis brevis	47 (10.3)	50.0 (50.0)	15.0 (12.5)	20.0 (0.0)	10, 200	5, 30	20, 40	7 (14.9)	6 (12.8)	19 (40.4)
Adductor pollicis	37 (8.1)	50.0 (22.5)	25.0 (15.0)	20.0 (20.0)	20, 125	10, 30	10, 30	0	8 (21.6)	10 (27.0)
Lumbricales	32 (7.0)	100.0 (120.0)	25.0 (10.0)	40.0 (47.5)	50, 400	15, 40	20, 75	29 (90.6)	8 (25.0)	11 (34.4)
Interossei dorsales	22 (4.8)	150.0 (150.0)	25.0 (30.0)	N/A	50, 375	10, 80	N/A	20 (90.9)	6 (27.3)	9 (40.9)
Opponens pollicis	18 (3.9)	50.0 (50.0)	30.0 (20.0)	5.0 (0.0)	10, 120	20, 40	5, 5	0	3 (16.7)	9 (50.0)

Percentages are based on the number of participants injected in the muscle, except * for which percentages are based on the number of participants in the efficacy population.

EMG, electromyography; IQR, interquartile range.

At Visit 2, the median (range) follow-up time was 14 (2.6–32.3) weeks, and further injection was planned in 361 (79.2%) participants. Clinicians planned to inject the same muscles in 254 (70.4%) with the same dose in 227 (62.9%). In 134 (37.1%), a different dose was planned (increased in 26.3% and decreased in 10.8%), but only 10 (2.8%) planned to use a different agent.

Concomitant treatments

Nearly two-thirds (61.6%) of the patients were receiving physiotherapy in association with the BoNT-A treatment at follow-up, and over a third (39.5%) of participants also received occupational therapy. However, there was wide variation in the number of sessions received (see table 4).

Table 4.

Concomitant treatments at baseline and follow- up

n (%)	Baseline	Visit 2
Physiotherapy
1–4 sessions		33 (11.7%)
5–10 sessions		50 (17.8%)
11–20 sessions		74 (26.3%)
>20 sessions		120 (42.7%)
Unknown		4 (1.4%)
Occupational therapy
1–4 sessions		34 (18.9%)
5–10 sessions		33 (18.3%)
11–20 sessions		46 (25.6%)
>20 sessions		63 (35.0%)
Unknown		4 (2.2%)
Splinting	188 (41.2%)	148 (32.5%)
Orthotics	114 (25.0%)	92 (20.2%)
Exercise	393 (86.2%)	360 (78.9%)
Passive stretching	423 (92.8%)	410 (89.9%)
Functional electrical stimulation	83 (18.2%)	58 (12.7%)
Positioning	266 (58.3%)	237 (52.0%)
Oral antispastic medication	210 (46.1%)	130 (28.5%)
No concomitant treatment	3 (0.7%)	12 (2.6%)

The types of concomitant treatments at baseline and follow-up are shown in table 4. At baseline, the frequency of concomitant treatments ranged from 18% (functional electrical stimulation) to 93% (passive stretching). By follow-up, the overall frequency of concomitant therapies had diminished, although the range of modalities remained similar. Notably, the proportion of patients on antispasmodic medication fell from 46% to 28.5%.

Primary and secondary goal areas

Primary and secondary goal areas set at baseline are shown in table 5. Goals were most commonly set in the areas of passive function, impairment and active function followed by pain, and involuntary movement. Less commonly in this data set, goals focused on mobility (ie, improvement in balance or gait quality by restoring freedom of upper limb movement) or other areas, such as cosmesis or supporting therapy interventions.

Table 5.

Primary and secondary goal areas

Goal area	Primary goals by area (n=456)			Secondary goals by area (n=471)
	Goal set n (%)	Goal achieved n (%) (95% CI)	Partially achieved n (%)	Goal set n (%)	Goal achieved n (%) (95% CI)	Partially achieved n (%)
Pain	61 (13.4)	51 (83.6) (71.9 to 91.8)	10 (16.4)	84 (17.8)	70 (83.3) (73.6 to 90.6)	6 (7.1)
Passive function (ease of care)	132 (29.0)	113 (85.6) (78.4 to 91.1)	11 (8.3)	109 (23.1)	84 (77.1) (68.0 to 84.6)	15 (13.9)
Active function (active use of limb)	104 (22.8)	75 (72.1) (62.5 to 80.5)	13 (12.5)	78 (16.5)	47 (60.3) (48.5 to 71.2)	18 (23.1)
Mobility (balance, gait)	10 (2.2)	7 (70.0) (34.8 to 93.3)	3 (30.0)	19 (4.0)	14 (73.7) (48.8 to 90.9)	2 (10.5)
Involuntary movement (associated reaction)	41 (9.0)	32 (78.0) (62.4 to 89.4)	5 (12.2)	56 (11.9)	45 (80.4) (67.6 to 89.8)	6 (10.7)
Impairment (eg, range of movement)	105 (23.0)	82 (78.1) (69.0 to 85.6)	11 (10.5)	117 (24.8)	91 (77.8) (69.2 to 84.9)	13 (11.2)
Other	3 (0.7)	3 (100) (29.2 to 100)	0	5 (1.1)	3 (60.0) (14.7 to 94.7)	2 (40.0)
Total	456	363 (79.6) (75.6 to 83.2)	53 (11.6)	468	354 (75.6) (71.2 to 79.2)	52 (11.1)

Overall, 363 (79.6%; 95% CI 75.6 to 83.2) patients achieved (or overachieved) their primary goal with 355 (75.4%; 95% CI 71.2% to 79.2%) achieving their secondary goal (see table 5). Although the rate of achievement was lower (but not statistically significant) for active function goals in comparison with passive function and impairment goals, in this series, a total of 182 (primary and secondary) goals were set in relation to active function of which 122 (67%) were achieved, either as expected (73 (40.1%)) or beyond expectation (49 (26.9%)). Pain reduction was a goal for treatment in nearly one-third of the patients (145 (31.8%)), and was achieved in 83.5%.

Secondary outcomes

At follow-up, the mean (SD) weighted GAS T-score was 52.0 (10.1; median 50.0, IQR 13.8), giving a mean (SD) change from baseline of 17.6 (11.0; 95% CI 16.6 to 18.6; p<0.001). Baseline and mean change from baseline in GAS T-scores were similar between BoNT-A preparations.

The mean (SD) MAS total score at follow-up was 8.4 (3.4), giving a mean (SD) change from baseline of −2.6 (2.6; 95% CI −2.9 to −2.4; p<0.0001). Overall, 90.1% of investigators and 85.8% of patients considered BoNT-A treatment to be of benefit.

GAS T-scores correlated, albeit rather weakly, with a reduction in total MAS at follow-up (Spearman r 0.28; p<0.0001). They correlated more strongly, however, with global assessment of benefit (Spearman r 0.38; p<0.0001 for investigator assessment and 0.45; p<0.0001 for patient assessment).

Standardised measures of upper limb spasticity

Reflecting common impairment, the ranges of active (56.1%) and passive (54.4%) motion were the most commonly recorded standardised measures of upper limb spasticity used at baseline (table 6). Approximately one-third (34.4%) of the patients also used a visual analogue rating to reflect symptoms such as pain or carer burden. However, only a very small minority (7.2%) had a standardised measure of functional outcome recorded, such as the Arm Activity Scale (ArMA—a self-report measure of active and passive functions; 5.9%) or the Leeds Adult Spasticity Impact Scale (LASIS—an investigator-reported measure of passive function; 1.3%). As shown in table 6, the GAS T-score correlated strongly with a change in these standardised measures wherever they were measured.

Table 6.

Standardised measures to assess upper limb spasticity at baseline

Standardised measure	Recorded at baseline	Recorded at follow-up	No. showing change	Correlation with GAS T-Score
	n (%)*	n (%)*	n (%)	r	p value
Impairment
Tardieu	77 (16.9)	62 (13.6)	56 (90.3)	0.43	<0.001
Active range of motion	256 (56.1)	231 (50.7)	137 (59.3)	0.41	<0.001
Passive range of motion	248 (54.4)	249 (54.6)	172 (69.1)	0.43	<0.001
Associated Reaction Rating Scale23	24 (5.3)	26 (5.7)	14 (53.8)	0.76	<0.001
Symptoms/carer report
Visual Analogue Scale†	157 (34.4)	139 (30.5)	109 (78.4)	0.46	<0.001
Function
Leeds Adult Spasticity Impact Scale3	6 (1.3)‡	5 (1.1)‡		—	—
Arm Activity Scale24	27 (5.9)	27 (5.9)	14 (51.9)	0.63	<0.001

*Percentage of efficacy population.

†Parameters measured with VAS were not specified.

‡Numbers too small to compute.

GAS, goal attainment scaling.

Identification of potential prognostic factors

Potential predictors with a p<0.20 were the primary goal area, primary goal score at baseline, first administration of BoNT-A and cortical function. These were entered into the multivariate logistic model. Results indicated that patients with impaired cortical function were half as likely to respond to treatment (OR 0.48 (95% CI 0.26 to 0.89)). The most likely explanation for this is that patients with dyspraxia, neglect or poor visuospatial perception are more likely to have difficulty complying with and carrying through any treatment programme.

A non-significant trend was also seen for poorer goal achievement in patients receiving their first administration of BoNT-A, compared with those who had had previous injections (OR 0.69 (95% CI 0.43 to 1.12)). Possible explanations for this may be that dose-ranging and muscle selection had been optimised through previous injections, or that patients benefited from their prior experience to better define their treatment goals.

Discussion

The findings from this large prospective international cohort study showed good response rates overall to BoNT-A injection delivered in the context of routine clinical practice for the management of upper limb spasticity. The study demonstrated wide variation in clinical practice with respect to the selection of muscles and approach to injection, highlighting the need for further systematic research into which approaches are likely to be most effective for which patients. Nevertheless, almost 80% of patients achieved their primary goal, as defined by the patients—together with their clinical team—at the start of treatment. The study also confirms the feasibility of collecting data across a large international community, using an eCRF.

Patient-reported outcomes are increasingly recognised as important indicators of quality of daily life for patients and/or their carers, and in this study GAS was selected as the primary endpoint in order to evaluate the benefits of treatment in terms of the attainment of individual person-centred goals. It also provides important insight into the nature of goals that are chosen as priorities for treatment, and also those that are most likely to be achieved.

In keeping with findings from other studies,^{11 25} improvement in passive function was the most frequently selected primary goal area for treatment (29%), and primary goals in this area were achieved in around 86% of cases. Goals for active function were also commonly set in nearly a quarter of the patients (22%). Perhaps unsurprisingly in this population with chronic spasticity and severe motor impairment, the achievement rate was somewhat lower for active function. Nevertheless, 72% of patients achieved their primary active function goal. The second most commonly achieved primary goal area was pain (set in 13% of patients), where again achievement was around 84%. Functional goals relating to involuntary movements and mobility were less commonly set in this population but were nevertheless achieved at broadly similar rates overall. The wide diversity in treatment goals between patients, however, highlights the importance of defining the primary treatment intentions clearly and then evaluating outcome specifically in relation to those.

Over and above the achievement of the primary goal, the GAS T-score provides an overall assimilation of attainment of primary and secondary goals regardless of the number of goals set. Some authors have cast doubt on the value of calculating a GAS T-score.²⁶ In this study, we went to considerable lengths to ensure that GAS was applied rigorously and goals were focused on functional gains,¹³ so it is worth reflecting on the added value of the GAS T-score in this context.

• If goals are set in an unbiased fashion, and are neither overambitious or overcautious, the mean GAS T-score should be around 50 (±SD 10).²¹ Our mean (SD) GAS T-score at follow-up of 52 (10.1) provides a useful quality check of the team’s ability to set and negotiate achievable goals, neither overestimating nor underestimating the expected outcome.

• Previous authors have recorded that GAS change scores >10 represent clinically meaningful change.^{11 17 21} In this study, the mean improvement in the GAS T-score from baseline to follow-up was 17.6. This confirms the findings from one RCT,¹⁶ and provides supportive evidence that BoNT-A produces clinically meaningful change at a functional level in the treatment of spasticity in real life practice.

• Importantly, GAS T-scores provide a single numerical evaluation of overall goal achievement for comparison with other outcome measures. If the gains occur as a result of reduction in spasticity, some correlation with change in the MAS score would be expected, even though the relationship may not necessarily be very strong^{11 16} and indeed this was found. The significant correlations with other standardised measures (especially the Arm Activity Scale) provide further support for GAS as a meaningful person-centred measure of outcome in this context.

The findings of this study also give insight into the longer term treatment of upper limb spasticity. Many of the patients enrolled in this study had received several previous BoNT-A injections, often over several years, suggesting that patients continue to receive benefit from repeat treatments, as indeed was shown by our findings. Additionally, the mean duration since the last BoNT-A treatment was 8 months, suggesting that patients with upper limb spasticity may not require retreatment as frequently as in other conditions (such as cervical dystonia). The reduction in use of other antispasmodic medication suggests that successful treatment may possibly have allowed the reduction in or withdrawal of other systemic agents. Many patients were receiving other concomitant therapies at the time of their injection (eg, therapy, splinting home exercise, etc). It is possible that these interventions play a role in reducing the frequency of BoNT-A injections, but insufficient detail regarding the frequency, duration and content of therapies was collected in this study to examine this possibility in great detail. Further longitudinal studies with a systematic recording of concomitant interventions are needed to confirm these observations and to ‘open the black box’ of a holistic approach to spasticity management.

The authors recognise a number of limitations to this study:

• Although there was a wide geographical distribution of centres across three continents, the numerical representation in each region was by no means representative. The recruitment of just 5–12 patients/site may have been insufficient to ensure adequate representation of practice in each centre, especially given the wide diversity of patients and goals for treatment.

• The relatively low frequency of reported impairments in cognitive and communicative function suggests either that there is selection bias (patients with these problems are less likely to be referred for treatment) or that there is under-reporting (clinicians focused on treating spasticity are not good at identifying associated impairments which may potentially impact on outcome).

• For pragmatic reasons, change on standardised outcome measures was recorded only subjectively on a standard scale of −2 to +2 and should therefore be interpreted with caution.

• The study was not sufficiently powered to perform a detailed investigation of prognostic factors for outcome, but it has given some preliminary insights into potential prognostic factors within the baseline data set.

Despite these limitations, the study provides useful information about the way BoNT-A is used in clinical practice around the world, and demonstrates that its effectiveness in the management of poststroke spasticity can be documented using individual person-centred goals. More importantly, a large majority of the patients achieved their treatment goals, mainly in terms of passive and active functions, demonstrating that BoNT-A injections contribute to an improvement in the daily life of the patients beyond improvement of tone or spasticity. Further secondary analyses will be presented separately to explore the impact of different treatment strategies including injection technique, early versus late treatment and the role of concomitant therapies. Further refinement of the tools and data set are now in train to produce a concise Upper Limb Spasticity Index that combines GAS with selected standardised measures targeted on the key goals for intervention (including the GAS evaluation of outcome in upper-limb spasticity (GAS-EOUS) tool).²⁷ Tools have also been developed for recording systematically the nature and content of any concomitant treatments provided in routine clinical practice. These tools will be used in the next phase of the ULIS programme (ULIS-III) to expand the cohort and to capture the benefits of integrated spasticity management in a fully generalisable sample recorded longitudinally over several cycles of BoNT-A treatment.