Introduction In recent years, client-centred and task-oriented training have emerged as important methods in rehabilitation including the rehabilitation of persons with spinal cord injury (SCI). The task-oriented intervention focuses on resolving, reducing and preventing impairment, developing effective task-specific strategies and adapting functional goal-oriented strategies to changing environmental conditions. Currently, task-specific training is mainly achieved by constant practice (i.e. repeating the task without variation) and is reported to improve performance of the trained task, but to have a negative impact on untrained tasks. Practice variability, however, is reported being advantageous to transfer training results into daily live. This paradox poses challenges in clinical practice, where task-specific training is essential to deliver client-centred training in order to focus on the patients' specific individual goals, but practice variability is important to be able to transfer the learned task into daily practice. Questions regarding the effective elements within rehabilitative interventions and the exact mechanisms behind the cerebral changes they may induce, remain. These questions require further research, for which ultra-high field fMRI techniques will be used. Furthermore, compensation of muscle function loss (i.e. the development of new muscle synergies) plays an important role in the improvement of skill performance in cervical SCI. Surface EMG allows to study the changes in muscle coordination, parallel to the changes at cerebral level.

Aims This study aims to

1. investigate which basic neural mechanisms of motor learning underlie functional recovery of arm hand skilled performance during client-centred task-oriented training of the upper limb in patients with cervical spinal cord injury and
2. investigate the contribution of 'practice variability' in contrast to 'constant practice' on arm-hand skilled performance, motor control and neural changes.

Study design This study features a multiple single-case experimental design (A-B-C design) with baseline (phase A) (6 weeks), intervention B (phase B) (3 weeks) and intervention C (phase C) (3 weeks). Intervention B will contain the 'practice variability' component. Intervention C will feature the 'constant practice' component. The order of phase B and C will be randomly assigned to participants. Four measurements during baseline and after each intervention phase (B and C) will be performed, thus producing a time series, per measure, for each patient. Also, meta-analyses on the pooled single-case data will be performed.

Setting/population Six patients with a cervical SCI (complete and incomplete) will be recruited from the Spinal cord unit of Adelante Rehabilitation Centre in the (sub)acute phase.

Intervention After therapy as usual (intervention A), the Task-oriented Client-centred Upper Extremity Skill Training (ToCUEST) module (Spooren et al., 2011) will be provided. In this program individual goals will be extracted using the Canadian Occupational Performance Measure(COPM) and the training program is based on a task-analysis and uses principles of training physiology and motor learning. Intervention B will consist of the ToCUEST program, including the component 'practice variability' (ToCUEST variability). Intervention C will consist of a modified ToCUEST program in which the component 'practice variability' will be replaced by 'constant practice' (ToCUEST constant) in order to evaluate the contribution of these components. Intervention A' will be therapy as usual.

Measurements Measurements will be taken at the level of activities (arm hand skilled performance, i.e. Van Lieshout Test, Spinal Cord Independence Measure, Goal Attainment Scale) and body function (Upper Extremity Motor Score, Graded Redefined Assessment of Strength Sensibility and Prehension Test, Surface EMG), and at cerebral level, i.e. neural activity changes (by means of ultra-high field fMRI).

The fMRI measurements will be taken before and after each intervention B and C in 4 patients (2 with complete and 2 with incomplete lesion).

Data-analyses Baseline data stability and any baseline trends regarding the outcome measures at body function level and activity level will be assessed. To control for, e.g., spontaneous recovery effects, baseline data (phase A) will be used in a computer-based detrending model. For the meta-analyses, mean data per outcome measure, per phase, per subject will be pooled and subsequently analysed using non-parametric statistics, i.e. Friedman analyses and Wilcoxon tests.

Inclusion Criteria:

• in)complete C-SCI at level C3-T1; age between 18 and 70 years; patients are able to sit in a wheelchair for at least for 2 consecutive hours; presence of problems with specific arm-hand skilled performance

Exclusion Criteria:

• opinion of an independent rehabilitation physician that additional neurological, orthopaedic or rheumatologic diseases strongly interfere with Activities of Daily Living (ADL) functioning and arm hand skilled performance; inability to perform arm hand skilled performance measurements