Added-value of Early Post-stroke Spasticity Reduction during Arm-hand Rehabilitation in Improving Functional Arm-hand Skill Performance: A Multiple Baseline Single Case Experimental Design Study

Background: Focal spasticity management in patients in the sub-acute phase after a stroke is mainly based on expert opinion. Evidence for the optimal type and intensity of multidisciplinary rehabilitation programs is scarce. This study will investigate the added-value of reduction of early signs of spasticity in the sub-acute phase after stroke on arm-hand rehabilitation outcome involving a well-described rehabilitation intervention. Reduction of early signs of spasticity will be done using abobotulinum toxin A.


Introduction
In stroke survivors, the presence or absence of voluntary motor activity in the affected arm and hand is the most important predictor of dexterity outcome and the level of performance regarding daily activities. In order to select the potentially most effective treatment, it is advocated to stratify persons with an impaired arm and hand into a limited number of arm-hand function strata [1,2]. at the mildly impaired AHF subgroup (UAT 4-7). Program 1 is titled "taking care and prevention". It is designed for stroke survivors who are not able to use their affected arm and hand for skill performance in daily life situations (non-functional arm-hand). Programs 2 and 3 are high intensity, task-oriented arm-hand performance training programs in which patients learn to integrate their affected arm and hand in daily occupations to optimize their overall functional abilities in daily situations. In this part a distinction is made between persons who have a moderately affected arm and hand, i.e. those who are able to use their affected arm and hand for passive and active stabilization tasks, like fixating bread while making a sandwich; and persons with a mildly affected arm and hand, who are able to use their affected arm and hand instantaneously in daily situations. Currently, CARAS has been implemented in everyday stroke rehabilitation practice in a number of rehabilitation centers throughout the Netherlands.
In order to monitor the development, i.e. either progression or deterioration, of both AHF and AHSP during and after rehabilitation, a single-arm prospective cohort study has been performed, involving 89 patients who participated in CARAS. This study, called AMUSE (Activity Monitoring of Upper extremity use in Stroke patients during and after rehabilitation (CCMO code: NL35681.068.11)), involved data collection at all levels of the ICF (International Classification of Function, disability and health [8]) at the start of rehabilitation (baseline), at clinical discharge and at 3, 6, 9 and 12 months post discharge, thus providing a time series per patient [9].
In a number of stroke patients with a moderate to severely affected arm and hand, moderate to severe grades of spasticity (Modified Ashworth Scale (MAS) scores +1 to 3) may occur already in an early, sub-acute phase post-stroke. This may seriously hinder AHF and AHSP treatment, leading to a slowing down of the patient's functional recovery. The problems caused by spasticity can occur at the level of impairment (e.g. restricted joint range of movement, pain and involuntary movement) and at the level of activity and participation (i.e. inability to perform daily activities, and limitations in taking up societal roles like work, family roles and leisure activities) [10].
There is a myriad of literature on the benefits of reducing spasticity on rehabilitation training effects regarding AHF and AHSP in chronic stroke patients [11][12][13][14][15][16][17][18]. However, thus far little evidence is available on the added-value of early spasticity reduction during rehabilitation training on the improvement of AHF and AHSP in sub-acute stroke patients. In a large study (n>300), Shaw et al. [19] recruited stroke patients in the sub-acute phase, but this study could not demonstrate improved active function. Foley et al. [15] reported small benefits as to improve passive AHF. Baker et al found modest but significant improvements on active AHF, although evidence quality was reported as being low [20]. A number of studies explored the added value of armhand rehabilitation immediately after the injection of botulinum toxin. Prazeres et al. [21] and Wolf et al. [22] did not find botulinum toxin plus rehabilitation to be superior to placebo plus rehabilitation with respect to AHF improvement. Takekawa et al. [23] and Devier et al. [24] reported improved AHF in stroke patients with a moderately to mildly impaired arm-hand who received botulinum toxin in combination with a tailored arm-hand rehabilitation program. Reducing the influence of the spastic component demands a holistic multidisciplinary approach to combine spasticity management and rehabilitation to optimize the likelihood of positive treatment effects [12,[25][26][27][28][29][30]. In most cases, a delay between spasticity reduction and improvement of AHF exists, suggesting that motor relearning continues as muscle tone is returning to baseline. The optimal types (modalities, therapy approaches, settings) and intensities of therapy for improving activity (active and passive function) in adults with post-stroke spasticity, in the short and longer term, however, are still unclear [31].
To date, evidence for the optimal type and intensity of multidisciplinary rehabilitation programs following focal spasticity management are based on expert opinion only. The present study will investigate the added-value of reduction of early signs of spasticity in the sub-acute phase after stroke on arm-hand rehabilitation treatment outcome involving a well-described rehabilitation intervention ('treatment-as-usual'), i.e. CARAS. This reduction of early signs of spasticity will be done using abobotulinum toxin A. It is assumed that reduction of spasticity in the upper extremity at an early, sub-acute, stage after stroke may enable the patient to: a) exercise more independently at an earlier stage; and b) exercise in a wider variety of therapy conditions also featuring more exercise challenges.
This, in turn, may lead to better treatment outcome at the ICF (International Classification of Functioning, activity and participation) [8] function level and activity level.

Aim
The aim of the present study is to investigate the added-value of reduction of early signs of spasticity in the upper extremity on improving functional arm-hand skill performance in sub-acute stroke patients with either a severely or moderately affected arm-hand (Utrechtse Arm-hand Test (UAT) [7] scores 1-3) and moderate to severe grades of spasticity, i.e. Modified Ashworth Scale (MAS) [32] score +1 to3. Therapy-as-usual, involving a regular, well-documented, concise arm-hand rehabilitation treatment (called CARAS) [3] will be provided during each patient's rehabilitation program.

The general research question is:
To what extent does reduction of spasticity in the shoulder, arm and hand muscles, adjuvant to a Concise Arm-hand Rehabilitation Approach in Stroke (CARAS), improve functional arm-hand skill performance in sub-acute post-stroke patients with a moderately to severely affected arm-hand (UAT scores 1-3) and moderate to severe grades of spasticity?
This general research question is subdivided in to three specific sub-questions, to be analyzed and answered using 3 design approaches. Each design approach is explained in-depth below.
• Sub-question 1 (necessitating a single case experimental design): Which relation exists between the time the spasticity reducing treatment was started in the sub-acute phase after stroke in subject X and any changes in the time series' trend regarding each patient's armhand function and arm-hand skill performance?
• Sub-question 2 (necessitating a single arm group design): To what extent does the rate of improvement as to arm-hand function and arm-hand skill performance change after baseline as a result of the spasticity reducing treatment in sub-acute stroke patients?
• Sub-question 3 (necessitating a non-randomized double arm group design): To what extent do sub-acute stroke patients receiving spasticityreducing therapy adjunct to CARAS improve more as to their arm hand function and arm-hand skill performance than patients who only received CARAS (i.e. patients from the AMUSE study [9])?

Design & duration
The current interventional study gauges the added-value of early post-stroke spasticity reduction in the upper extremity during 'therapy as usual' , on AHF and AHSP levels in sub-acute stroke patients using three methodological approaches, i.e.: i) A (multiple baseline) single case experimental design [34] involving 10 individuals.
ii) A meta-analysis of the data of all single cases (single arm group design).
iii) Non-randomized double arm group design, i.e. a case-matched control design in which each patient receiving early post-stroke spasticity reduction treatment will be matched (according to arm-hand capacity, spasticity level, age, gender, UAT score) to a case from the AMUSE study [9].
After patient's eligibility screening and written informed consent, both primary and secondary outcome measures will be taken at baseline, followed by measurements at 1-wk intervals during the 2 x 6 weeks (total 12 wks, i.e. CARAS training episode 1 and CARAS training, episode 2 CARAS treatment and at 2-wks intervals during the ensuing 3 months follow up, resulting in a time series per patient per outcome measure. (Figure 1). Blinding of participants, therapists or data analysts as to the intervention provided is not possible in this study.

Generic explanation of the study designs used:
i) Multiple baseline single case experimental design: In single case experimental design studies time series, consisting of many sequential observations (measurements), are recorded for each individual subject. After a baseline period an experimental stimulus or intervention is applied. In the subsequent time phase the effect of this experimental stimulus or intervention on primary and secondary outcome measures relative to baseline values is investigated. The length of the baseline phase may be constant between participants (e.g. measurements may be taken weekly for 5 wks) or may vary between participants (e.g. in subject A baseline measurements may be taken weekly for 4 wks, whereas for subject B baseline measurements may be taken weekly for 5 wks, and for subject C for 6 wks). The latter approach, featuring different starting moments of the experimental stimulus or intervention across participants in a study, is called "multiple baselines across subjects".
Using a 'multiple baseline' design in single case experimental research, as is done in the present study, strengthens the design [34], because it reduces the chance of any changes in the outcome parameters after the experimental stimulus of intervention being attributable to other factors than the experimental stimulus or intervention applied. In other words: If changes in outcome parameters occur ONLY AFTER the application of the experimental stimulus or intervention, the chance of this change being caused by a chance factor is reduced. We can (statistically) relate the 'position on the time axis' of the occurrence of the intervention to the 'position on the time axis' of the occurrence of any effect (outcome measure time series trend changes). In Figure 2 an example of a multiple baseline design across subjects is depicted.
Since the subjects in the proposed study are patients in the subacute phase after stroke, it is likely that their performance will improve due to e.g. spontaneous recovery and the therapy they receive, especially in the early phase post-stroke. On the other hand, the occurrence of spasticity in this sub-acute phase may slow down or   hamper improvement in voluntary movements. The baseline data (i.e. the baseline time series representing changes in the outcome measures during baseline phase) thus may reflect effects of spontaneous recovery and/or effects of 'therapy as usual' (in our case CARAS). These baseline data may show a trend (e.g. towards gradual improvement of armhand parameter outcome over time, or e.g. a non-improvement of arm-hand parameter outcome due to spasticity starting to develop). If, at some point, a spasticity-reducing treatment is applied, the trend of arm-hand parameter outcome observed in the prior phase (baseline phase) may change (the patient may become proficient more rapidly regarding outcome measure X or Y). An example of this concept is given in Figure 3.
ii) Meta-analysis of the data of all single cases (single arm group design) First, each time series of each subject will be 'de-trended' for any baseline trends, similar to the approach reported by Franck et al. [35]. This will result in a 'detrended' time series for each subject, data of which are rendered mutually independent by this de-trending technique. The latter is important in view of the ensuing statistics to be used. Next, for each subject, mean values of the residuals from the baseline phase and the post-intervention phase will be calculated. Subsequently, for each of the (two) phases, these data will be pooled across subjects, after which non-parametric statistical analyses will be performed on the pooled dataset.
iii) Case-matched control design (non-randomised double arm group design) In this design, for each of the participants in the present study, a case-matched control subject from the database of the AMUSE study will be sought. All stroke patients suffering from arm-hand problems, who are admitted to Adelante rehabilitation centre, receive CARAS treatment.
In the (previously performed) AMUSE study 89 sub-acute stroke patients who entered Adelante rehabilitation centre and who suffered from arm-hand problems due to the stroke, were monitored during their rehabilitation phase and up to 12 months post-discharge [9].
As to the frequency of measurements/assessment, in the AMUSE study arm-hand assessments at ICF function level and at activity level were performed at baseline (i.e. before arm-hand treatment started), at 6 wks into the CARAS arm-hand training program, immediately after the CARAS training program (i.e. at 12 wks post-baseline), at clinical discharge, and at 3, 6, 9 and 12 months post-discharge. Measures at ICF participation level were done at baseline, clinical discharge and at 12 months post-discharge. In order to compare data from the AMUSE study and the present study measurement, we will use data of the present study that are collected at the same time points during the AMUSE study. In Figure 4 an overview of the 'synchronous' measurement dates for both the AMUSE study and the present study are depicted ( Figure 4).
The AMUSE study and the present single case experimental design study share 5 similar measurement dates. The time at which baseline measurements are performed in the AMUSE study is identical to the time at which the first baseline measurement of the present study is performed (Figure 4).
Similarly, after the initial 6 weeks of CARAS training an arm-hand assessment has been performed in the AMUSE study (T6w or Tmid-CARAS). This measurement time is identical to one of the present study measurement dates. This also applies to the AMUSE measurement date immediately after CARAS has finished, i.e. after 12 weeks of CARAS training (T12w or T_end-of-CARAS).

• AMUSE: clinical discharge
In the AMUSE study arm-hand assessments are also performed at clinical discharge (Tcd). Again, this measurement time is identical to one of the present study measurement dates. (Figure 4, fourth vertical dotted line.) • Last measurement date of the present study In the AMUSE study no exact measurement date as counterpart of the last measurement of the present study is available. However, a good approximation of data from the AMUSE study at a similar time point may be performed by linearly interpolating the AMUSE data using data from Tcd and T3m, taking into account the time difference between Tcd and the time at which the last measurement of the present study was taken. (Figure 4, X-mark.)

Setting
All patients in the present study will be identified among the sub-acute stoke patient population of Adelante rehabilitation centre in Hoensbroek, the Netherlands. The present study will be performed at Adelante in Hoensbroek. All patients who, in the past, have participated in the AMUSE study (anonymous database data of which will be used in the case-matched control design part of the present study) were patients from Adelante rehabilitation centre.

Study population
The present study focusses on adult sub-acute stroke patients with either a severely or moderately affected arm-hand (Utrecht Arm-hand Test (UAT) [7] scores 1-3) and moderate to severe grades of spasticity, i.e. Modified Ashworth Scale (MAS) score +1 to 3. These patients will be recruited from the patient population of the department of brain injury rehabilitation of the Adelante rehabilitation centre in Hoensbroek, The Netherlands. This study will include 10 patients who develop (early signs of) spasticity in the upper extremity during the sub-acute phase after stroke.
Stroke patients who have a severe paretic arm and hand (UAT scores 1-3) at admission to the rehabilitation centre will be asked to participate (by letter from their rehabilitation physician (KR)) as soon as possible after admission to the rehabilitation centre. After giving written informed consent, collected by the researcher (JAF), measurements will start according to the protocol described.
Patients, who develop early signs of spasticity in the arm and/or hand, i.e. within 5 weeks after start of arm-hand treatment (CARAS), will remain in the study. In patients who have a severe paretic arm and hand (UAT scores 1-3) at admission to the rehabilitation centre, but who do not develop early signs of spasticity within 5 wks after start of arm-hand treatment (thereby not being in the target group), ABoNt-A

Skill performance
ABoNt-A=application of ABoNt-A. CARAS-1=CARAS training episode 1. CARAS-2=CARAS training episode 2. T=Measurement moment at 3, 6, 9 and 12 months after clinical discharge.  measurements to be used in the study will cease. Their rehabilitation treatment will follow 'therapy-as-usual' as will their regular therapyrelated clinimetrics. Any research data of the latter patient group recorded for the sole purpose of the research will be discarded/erased.

Inclusion criteria:
In order to be eligible to participate in this study, a subject must meet all of the following criteria: -Age>=18 yrs -Supratentorial stroke, i.e. arteria cerebri media infarction.
-Sub acute phase after stroke, i.e. between 2 wks and 3 months post-stroke.
-Functional disabling spasticity in the upper extremity: Modified Ashworth Scale (MAS) scores +1 to 3 (developing within 5 weeks after the start of CARAS).
-Eligible to participate in CARAS for a period of 12 wks.
-Being able to understand the questionnaires and measurement instructions.
As to functional disabling spasticity in the upper extremity, patients developing spasticity in the early sub-acute phase after stroke (i.e. within 5 wks after the start of CARAS) with a Modified Ashworth Scale (MAS) score of +1 to 3 will continue to participate in the study.
Exclusion criteria: A potential subject who meets any of the following criteria will be excluded from participation in this study: -Severe non-stroke related co-morbidity that may interfere with arm-hand function.
-Additional complaints that may interfere with the execution of the measurements.
-No informed consent.

Sample size calculation
As to the Single Case Experimental Design (SCED), by definition these studies involve 1 person. Despite this, a sample size calculation is needed given the group-wise analyses of the second and third part of the present study, i.e. the meta-analysis of the data of all single cases (single arm group design) and the case-matched control design (nonrandomized double arm group design).
The sample size calculation is based on data from the AMUSE cohort study, i.e. on the improvement on the main outcome measure used in the present study (ARAT) as observed in patients with an UAT scores of 1-3 who did (Group A) or did not (Group B) develop signs of spasticity in the upper extremity within the AMUSE study. In the present study the ARAT is the primary outcome measure. Mean improvement on the ARAT after 12 weeks of CARAS was 17.3 points (sd: 12.2) for Group A and 38.3 points (sd: 19.2) for group B. Given a two-sample method, a double-sided statistical test, a power of 80% and an alpha of 0.05, a loss to follow-up of 10%, 10 participants per group are needed.
It is expected from clinical experience, our experiences with the AMUSE cohort study, and recent influx numbers of patients in Adelante rehabilitation centre that the number of 10 patients meeting the inclusion criteria within a reasonable time span (i.e. 16 months) is well-feasible. The data of 10 case-matched control subjects will be extracted from the AMUSE database. Therefore, in total, data of 20 patients will be used.

Interventions CARAS:
Arm-hand rehabilitation treatment will be provided according to the CARAS approach as described by Franck et al. [3]. In Adelante rehabilitation centre and multiple other rehabilitation centres in the Netherlands, CARAS is 'therapy-as-usual' . After a standard initial clinical assessment, patients will receive arm-hand rehabilitation treatment for 2 x 6 weeks. Based on the UAT scores, patients are allocated to one of three training programs, i.e.: • CARAS program 1: severely impaired AHF subgroup (UAT=0-1).
Program 1 is titled "taking care and prevention". It is designed for stroke survivors who, due to the severity of the stroke, are not able to use their affected arm and hand for skill performance in daily life situations (non-functional arm-hand). Programs 2 and 3 are high intensity, taskoriented arm-hand performance training programs in which patients learn to integrate their affected arm and hand in daily occupations to optimize their overall functional abilities in daily situations. In this part a distinction is made between persons who have a moderately affected arm and hand, i.e. those who are able to use their affected arm and hand for passive and active stabilisation tasks, like fixating bread while making a sandwich; and persons with a mildly affected arm and hand, who are able to use their affected arm and hand instantaneously in daily situations.
CARAS is the standard therapy (therapy-as-usual) provided by physiotherapists and occupational therapists to all stroke patients with arm-hand problems who are admitted to Adelante rehabilitation centre for treatment. This means that the decision to apply CARAS is taken before inclusion of the patient in the study.
The present study focusses on patients with initial UAT scores of 1-3.

Spasticity-reducing treatment:
In order to reduce spasticity, abobotulinum toxin A (ABoNt-A) will be administered once, i.e. directly following the sequence of baseline measurements in each patient who is developing (early signs of) spasticity in the upper extremity.
In the present study abobotulinum toxin A (ABoNt-A) is a noninvestigational product. abobotulinum toxin selectively targets cholinergic nerve endings via binding to ecto-acceptors to block acetylcholine release at the neuromuscular junction thereby abolishing the motor end-plate potential. This causes prolonged muscle weakness. Original nerve terminals regain function 12 wks after ABoNt-A injection with full recovery of nerve-evoked muscle contraction [36]. In clinical practice reduced muscle tone is seen up to 24 wks in the subacute phase in patients with early post-stroke spasticity [37,38].
ABoNt-A will be administered by a senior rehabilitation physician at Adelante rehabilitation centre. Patients will receive ABoNt-A injected according to clinical judgement into the dominant spastic muscles of the arm and/or forearm. The total maximum dose for the upper limb will be 1000U [39]. ABoNt-A dosage for individual muscles will be in line with the dose ranges reported by Dashtipour et al. [40], and Gracies et al. [39]. Muscles will be identified using electro stimulation or echography according to the normal practice of the clinician.
The decision to use ABoNt-A is entirely based on clinical necessity, i.e. when a patient is developing spasticity in the upper limb muscles on the affected side (MAS score +1 to 3), as established by the rehabilitation physician. After the first finding of this increased MAS score, the decision to apply ABoNt-A will be taken by the rehabilitation physician and the application itself will be done within 1 wk.
Patients in the target group, i.e. those who have a severe paretic arm and hand (UAT scores 1-3) at admission to the rehabilitation centre, will be asked to participate as soon as possible after admission to the rehabilitation centre. After giving informed consent, measurements will start according to the protocol described. Therapy adherence will be monitored, based on regular clinimetric assessment, as part of the regular daily clinical practice.

Data collection & measures used
At the patient's entry into the study, i.e. after the eligibility screening and patient's written informed consent, the following sociodemographic variables will be recorded: age, gender, educational level and the patient's living situation. Furthermore, medical variables, obtained from medical files of the referring consultant in rehabilitation medicine, include: stroke type (haemorrhagic or ischemic), lesion site, time since stroke, paresis level, co-morbidity, hand dominance prior to the stroke, and arm-hand function status (UAT score). During the informed consent procedure, patients will be asked for permission to use this information of their medical file.
Data collection started on October 10 th , 2016 when the first patient was included. Currently, patients are being recruited and enrolled.
As to the primary outcome measures, improvement of the patients' functional arm-hand skill performance capacity will be gauged by using the Action Research Arm Test (ARAT) [41][42][43][44]. The ARAT is reliable, valid and sensitive to change in patients after stroke [42,[44][45][46]. It consists of four subtests comprising 16 grasp movements and three reaching movements. Items are scored on a 4-point scale, with a max score of 57.
-Bilateral arm activity monitoring [50,51]: As to actual armhand skill performance, bilateral activity monitors (3D accelerometry, AX3, Axivity Ltd) will be worn around both wrists for a period of 4 consecutive days.
-Fugl-Meyer Motor Assessment (FM) [44], gauging armhand function level. The upper extremity section of the FM is a reliable and valid test for measuring changes in armhand-function in stroke patients [52,53]. Its score ranges from 0 to 66.
During the first 12 wks measurements involving the ARAT, ABILHAND, FM, JAMAR, MI and MAS will be performed weekly. In the ensuing 3 months these measures will be administered every 2 wks. The accelerometry measurements will be performed once every 3 wks.

Data storage & safety
Data collected will be stored in an electronic trial master file on a secured network drive of the Adelante network. All data will be coded immediately during measurement. Coding will be done using a combination of numeric and alphanumeric characters, which are not related to the participant and cannot be used to trace/identify the participant. Non-coded data (e.g. participant's name) will be recorded separately. HAMS is the person who has sole access to the coding key. The non-coded data will solely be accessible to two persons, i.e. JAF and HAMS. The anonymised data will be accessible to JAF, RJEMS, and HAMS.
Four yrs after the project has finished all identifiable data will be destroyed, preventing any further link between the results and participants. All data will be destroyed and/or deleted after 15 yrs.

Data processing & statistical analysis
General: For graphical data presentation, time series per subject and boxplots (for grouped data) will be used. As to all data, any change over time of 5% or more will be considered a meaningful change.
From the accelerometer data mean 'activity counts' during waking h will be calculated for both hands. A 'count' is an occasion at which the accelerometer signal exceeds a predefined threshold, indicating(armhand) activity intensity. An example of identifying waking h (or 'uptime') is given in Figure 5.
For each 'uptime' the amount of time a patient used hands, no hands, his affected hand and his unaffected hand will be calculated. Similarly, the number of 'counts' at occasions where a patient used hands, no hands, his affected hand and his unaffected hand will be calculated. These data will each be averaged across the number of days the accelerometers were worn.
Handling of missing values: When 1 or 2 (temporally adjacent) value(s) are missing, these missing value(s) will be estimated by linear interpolation using the two valid adjacent values in the time series. In case of the final time series observation missing, the 'last-observationcarried-forward' principle will be used. In case of 3 or more consecutive missing values, the whole case (for the measure at hand) will be discarded.

First design procedure: Single case experimental design:
Statistics regarding ARAT, ABILHAND, FM, JAMAR and MI will encompass permutation tests involving the time series of each subject separately. The length of the baseline epoch will vary between 4-6 wks between subjects. This procedure will provide an answer to the question whether there is a relation between the time the spasticity reducing treatment was started in subject X and any changes in the time series' trend regarding each patient's arm-hand capacity (ARAT), each patient's arm-hand perceived performance (ABILHAND), as well as arm-hand function (FM), grip strength (JAMAR) and functional armhand strength (MI). (Figure 3) Here, data are analyzed for each subject separately. MAS results will be reported descriptively by plotting the time series of each subject.
Second design procedure: Meta-analysis of the data (single arm group design): Regarding ARAT, linear detrending for any baseline trends per subject, using a least squares method, to (partially) compensate for improvements caused by e.g. spontaneous recovery and/or other treatment received, will be calculated. The time series of these residuals (data of which thus having been rendered mutually independent) of the whole group, will be further processed and analysed.
Per subject, mean residuals data of ARAT will be calculated for the baseline phase (Phase A), for the treatment phase after application of the spasticity reducing therapy (Phase B), and for the post-treatment up to 3 months post-CARAS (Phase C), thus resulting in 3 mean values per subject per measure.
These mean residuals per phase will be statistically tested using non-parametric Kruskal-Wallis tests in a group design. Next, multiple comparison analysis, involving Mann-Whitney U-tests between phases A & C and phases B & C will be performed using a Bonferroni correction in order to avoid spurious false positive findings. This procedure will provide an answer to the question whether, at a group level, patients improved as to their arm-hand capacity (ARAT) faster after baseline as a result of the spasticity reducing treatment.
Data processing and statistical analysis for the ABILHAND, mean 'activity counts' , FM, JAMAR, and MI will be similar to the procedures described for the ARAT. MAS results will be reported descriptively.
Third design procedure: Case-matched control design (nonrandomized double arm group design): Finally, per subject, differences between baseline means and follow-up means will be calculated. The latter values will then be contrasted with similar values from a matched group of patients from the AMUSE study (see earlier). In this design, for each of the participants in the present study, a case-matched control subject from the database of the AMUSE study will be sought.
In order to compare data from the AMUSE study and data from the present study measurement dates from the present study that are synchronous to the measurement dates of the AMUSE study will be identified. In Figure 4 an overview of the 'synchronous' measurement dates for both the AMUSE study and the present study are depicted.
Each subject from the SCED study part will be matched with a person from the AMUSE study according to the following procedure: Upon entry in the SCED study (baseline phase) each patient will be assessed (identical to the assessment performed in the AMUSE study subjects) in which his/her UAT score will be identified. Based on this UAT score of the SCED study subject we will search our AMUSE database for persons with an equal UAT score. Next, a match will be sought regarding the age, gender and (baseline) ARAT scores. Matching cases for spasticity level will be done when subjects first show signs of spasticity occurring, i.e. within the first 5 weeks after the start of CARAS. As an indication: About 50% of patients who have suffered an arteria cerebri media infarction and who were admitted to Adelante rehabilitation centre developed a functional limiting form of spasticity within 3-5 wks after the start of the arm-hand training program (CARAS).
Statistical analyses of the primary outcome variable (ARAT) will include Mann-Whitney U-tests. This procedure will yield whether or not patients receiving spasticity-reducing therapy adjunct to CARAS have improved more as to their arm-hand capacity than patients who only received CARAS (i.e. patients form the AMUSE study).
Again, data processing and statistical analysis for the ABILHAND, mean 'activity counts' , FM, JAMAR, and MI will be similar to the procedures described for the ARAT. MAS results will be reported descriptively.

Data reporting
The investigators, without restrictions, will report the general, anonymous results of this study in scientific papers, and at international congresses. For all scientific reporting the guidelines

Auditing and annual progress report
The sponsor/investigator will submit a summary of the progress of the trial to the accredited Medical Ethics Committee (METC) once a year. Information will be provided on the date of inclusion of the first subject, numbers of subjects included and numbers of subjects that have completed the trial, serious adverse events/serious adverse reactions, other problems, and amendments. The METC exempted this study from instating a data monitoring committee. Furthermore, the sponsor/investigator adheres to Dutch law concerning any possible independent audit, at any time, by representatives of the competent national authorities.

Temporary halt and (prematurely) end of study report
The investigator/sponsor will notify the accredited METC of the end of the study within a period of 8 wks. The end of the study is defined as the last patient's last visit. The sponsor will notify the METC immediately of a temporary halt of the study, including the reason of such an action. In case the study is ended prematurely, the sponsor will notify the accredited METC within 15 days, including the reasons for the premature termination. Within one year after the end of the study, the investigator/sponsor will submit a final study report with the results of the study, including any publications/abstracts of the study, to the accredited METC.

Safety reporting
Temporary halt for reasons of subject safety: In accordance to section 10, sub-section 4, of the Dutch Person-related Research Act (WMO), the sponsor will suspend the study if there is sufficient ground that continuation of the study will jeopardise subject health or safety. The sponsor will notify the accredited METC without undue delay of a temporary halt including the reason for such an action. The study will be suspended pending a further positive decision by the accredited METC. The investigator will take care that all subjects are kept informed.
Adverse events (AEs): Adverse events are defined as any undesirable experience occurring to a subject during the study, whether or not considered related to CARAS and/or the application of abobotulinum toxin A. All adverse events reported spontaneously by the subject or observed by the investigator or his staff will be recorded.

Serious adverse events (SAEs):
A serious adverse event is any untoward medical occurrence or affects that: -Results in death.
-Is life threatening (at the time of the event).
-Requires hospitalisation or prolongation of existing inpatients' hospitalization.
-Results in persistent or significant disability or incapacity.
-Is a congenital anomaly or birth defect.
-Any other important medical event that did not result in any of the outcomes listed above due to medical or surgical intervention but could have been based upon appropriate judgment by the investigator.
An elective hospital admission will not be considered as a serious adverse event. The investigator will report all SAEs to the sponsor without undue delay after obtaining knowledge of the events. The sponsor will report the SAEs through the web portal Toetsing Online to the accredited METC that approved the protocol, within 7 days of first knowledge for SAEs that result in death or are life threatening followed by a period of maximum of 8 days to complete the initial preliminary report. All other SAEs will be reported within a period of maximum 15 days after the sponsor has first knowledge of the serious adverse events.
Follow-up of adverse events: All AEs will be followed until they have abated, or until a stable situation has been reached. Depending on the event, follow up may require additional tests or medical procedures as indicated, and/or referral to the general physician or a medical specialist. SAEs need to be reported till end of study within the Netherlands, as defined in the protocol.

Amendments
Amendments are changes made to the research after a favourable opinion by the accredited METC has been given. All amendments will be notified to the METC that gave a favourable opinion. All substantial amendments will be notified to the METC and to the competent authority. Non-substantial amendments will not be notified to the accredited METC and the competent authority, but will be recorded and filed by the sponsor.

Result and Discussion
Given the specific research (sub) questions, a number of issues have been taken into account in setting up this study protocol, leading to a combination of three methodological designs.
Firstly, we proposed time series analyses, i.e. a (multiple baseline) single case experimental design, in order to more individually control for spontaneous recovery and effects of therapy as usual in the subacute stage after stroke, which may vary considerably between patients. Modeling spontaneous recovery and effects of therapy as usual during baseline measurements has two advantages: a) the time series may be detrended for baseline trends, making any underlying mechanism to be investigated more pronounced, and b) the residuals obtained after the detrending procedure are mutually independent, thus making them eligible for statistical analyses aimed at the possible relation between the time the spasticity reducing treatment was started in the sub-acute phase after stroke in a single subject and any changes in the time series' trend regarding this patient's level of arm-hand function and arm-hand skill performance.
Secondly, pooling the averaged, baseline-detrended data of all subjects enables group-wise data analyses gauging the (average) rate of improvement in arm-hand performance.
Thirdly, in our previous single-armed prospective cohort study, i.e. the AMUSE study [9], we obtained longitudinal data on changes in arm-hand function and arm-hand skill performance in a large group of stroke patients' typically seen in daily clinical practice. By matching patients who have received a spasticity reducing treatment with patients from the aforementioned cohort study, we can assess the added value regarding the rate of improvement in arm-hand performance that may be attributable to the spasticity reducing treatment relative to 'therapy as usual only' in sub-acute stroke patients.
The results of this study will provide evidence on the added-value of reduction of early signs of spasticity in the upper extremity on improving functional arm-hand skill performance in sub-acute stroke patients with either a severely or moderately affected arm-hand and moderate to severe grades of spasticity.