People who are at least 18 years of age, who experienced an acute ischemic or hemorrhagic stroke (identified by MRI/CT), and who suffer from gait ataxia are eligible, if they score at least 1 point on the items gait, stance, trunk or heel-shin-slide of the Scale for the Assessment and Rating of Ataxia (SARA; Schmitz-Hübsch et al., 2006) and 47 points or less on the Berg Balance Scale (BBS; Stevenson, 2001).

Exclusion criteria are a modified Rankin Scale (mRS) score of 5 or more (Rankin, 1957), a previous stroke leading to persistent debilitating neurological deficits defined as an mRS score of 3 or more, a comorbidity that limits active study participation (e.g., life expectancy of <3 months, alcohol or drug abuse), physical or mental conditions that would not allow safe participation in the study or would influence the assessment of outcomes (e.g., dementia, cardiac insufficiency, severe aphasia, etc.), or pregnancy.

Participants are randomly allocated into one of two groups. Group allocation uses a 1:1 ratio to either control group (CG) or intervention group (IG) utilizing stratified blocked randomization and hidden block sizes. The allocation sequence is generated using an online randomization software (Sealed Envelope Ltd., London, UK; Sealed Envelope Ltd., 2022). Stratification variables are age (≤ 70 years, >70 years) and gender (male, female).

Allocation concealment is conducted, as an independent researcher in charge for the randomization does not disclose patients' group allocation until after the baseline assessment (T0). To effectively conceal the randomization sequence, sequentially numbered, opaque sealed envelopes are used. After T0 assessment, department staff who are independent of assessment and intervention assign the patient to the appropriate group using the sealed envelopes. The staff then informs the therapist responsible for providing the assigned intervention.

Outcome assessors are blinded to patients' group assignment. Patients are informed that an examination of the effects of two distinct physiotherapy exercise programs is performed. However, they remain unaware of the precise details of the interventions of both groups and the study's rationale. Patients are encouraged not to talk to the assessor about the exercises. The planned time for regular unblinding is after study completion.

Coordination exercises (IG) are part of ataxia-specific neurorehabilitation and focus on high repetition while allowing rather than suppressing body sway associated to ataxia (Brötz et al., 2007). Coordination exercises focus on repetitive trunk and limb movements, dynamic balance, and gait training, including training of protective movements and prevention of falling (Synofzik et al., 2013; Figure 1). The following briefly outlines the main aspects and principles of coordination exercises:

Considering stroke rehabilitation in general, exercises follow the principles of motor learning, focusing on high repetition of exercises (Kwakkel et al., 2023). Bearing in mind cerebellar pathology, resulting in difficulties in motor learning (Hatakenaka et al., 2012), a high number of repetitions seems to be of special importance for patients with cerebellar lesions to compensate for difficulties in trial-and-error learning (Kelly and Shanley, 2016).

Regarding the number of repetitions and variety of exercises, a choice of 3 different exercises is recommended, each having at least 10 repetitions (per side), with a common functional goal (e.g., sit-to-stand) and their combination into one set. In a 45-min session, 3–4 sets can be performed.

Shaping is achieved by transitioning from rather static to dynamic balance exercises, from slow to fast movements (e.g., foot tapping exercises), and from single joint to complex multi-joint exercises (Ilg et al., 2009). The threshold for progressing exercises is determined by the patient's ability to resolve fixation mechanisms, with assistance from the therapist if needed. Resolving fixation mechanisms is addressed using a combination of hands-on facilitation paired with external feedback or solely through feedback, depending on the stage of motor learning. Feedback is provided by the therapist through verbal cues like “relax your shoulders” and as positive feedback on successful movements. Throughout the training, patients are guided to become aware of fixation mechanisms and learn to resolve them independently. By resolving fixation mechanisms free joint mobility is achieved. To work toward functional goals, for instance, if the objective is to ambulate from the bed to the toilette without any aid, the initial exercises may begin in a seated position and progress through activities such as sit-to-stand, standing, stepping, and eventually, walking exercises. Throughout this process, the therapist always ensures that the functional value and purpose of each exercise are clearly communicated. In order to keep the exercises challenging and at the individual performance limit until the end, i.e., when patients are already able to perform exercises on uneven ground, coordination exercises can be performed using trampolines (Giagazoglou et al., 2015). These exercises still follow the same principles.

Further details are reported according to the Template for Intervention Description and Replication (TIDieR; Hoffmann et al., 2014; Supplementary Data Sheet 1). Exercises of the IG are summarized in a handout, which is provided in English (Supplementary Data Sheet 2) and German language (Supplementary Data Sheet 3).

Standard physiotherapy (CG) includes conventional post-stroke rehabilitation exercises. Exercise types were evaluated using a questionnaire for the pilot trial (Meier et al., 2021), confirmed for this RCT in the three study centers and corresponding with standard physiotherapy according to guidelines for stroke rehabilitation (Winstein et al., 2016; Intercollegiate Stroke Working Party, 2023). Exercises in CG can be categorized into four different groups:

The central principle of the CG is to keep goals and exercises in therapy relevant to everyday life. Therefore, the exercises are designed to target specific activities involving intricate movements such as transitioning from lying to standing, dressing oneself, or preparing a cup of coffee. While each repetition may take longer compared to the IG due to the complex sequence of activities involved, the exercises should still strive for a high number of repetitions in line with general stroke rehabilitation guidelines. The repetition count may vary depending on factors such as task duration, allowing for variability within the therapy session to cater to the patient's performance levels. This variability aims to maintain a level of challenge and interest for the patient. Additionally, considering the patient's condition is crucial; for instance, acute stroke patients may experience dizziness with repeated changes in body positions. Similar to the IG, exercise difficulty is adapted to the patient's limit of performance, but patients in the CG can receive support from the therapist or an assistive device to be able to perform the exercise. Another aspect in this group is to emphasize balance training, stability training and strengthening of the trunk.

Further information about the exercises in this group are reported using the TIDieR (Supplementary Data Sheet 4). The exercise handout for the CG is provided in English (Supplementary Data Sheet 5) and German language (Supplementary Data Sheet 6).

In addition to the supervised practice, patients in both groups are asked to practice independently 5 times per week, for 15 min throughout the 4-week intervention period (total 20x). The exercises for each day are provided during the supervised sessions according to their group assignment and level of performance.

Upon completing supervised treatments, both IG and CG patients are given a personalized home exercise program as a handout. This program includes a variety of exercises, tailored to each patient. Patients have the flexibility to choose from these exercises for their independent practice, which is recommended 5 times per week, each session lasting 15 min, over an 8-week period until the last visit (T2 assessment), totaling 40 sessions (Table 1). If IG patients have safely practiced with trampolines during supervised treatment, they are provided with a trampoline for continued exercise performance.

Both interventions follow the principles of motor learning; however, there are distinct differences between them. In the IG, exercises typically involve more repetitions but less variability in practice due to the exercise nature. Conversely, in the CG, there is more practice variability with fewer repetitions. The IG emphasizes allowing the body to sway and solving fixation mechanisms, which is not a central focus in the CG. Patients in the CG concentrate on enhancing stability through trunk exercises while in the IG, this aspect is not as prominent. The CG places a stronger emphasis on training for activities of daily living compared to the IG. Both interventions include walking training, but the approach differs. In the CG, walking training is scheduled regularly regardless of support levels, whereas in the IG, gait training is preferred at times when it can be executed without fixation mechanisms, therapist assistance, and with a focus on coordination.

To document independent practice sessions and facilitate adherence, patients are provided with a checklist. To enhance adherence further, patients undergo semi-structured phone call interviews every 2 weeks. These interviews encompass questions on the exercise frequency, experience and any issues encountered during the exercises, including aspects such as the exercise environment and motivation. Additionally, the interviews cover questions regarding completion of adherence and falling checklist (Table 2).

While a multi-disciplinary rehabilitation regimen may encompass speech therapy and occupational therapy, these are only included if deemed necessary for the patient. There are no additional one-on-one physiotherapy sessions beyond the trial intervention. Other rehabilitation modalities offered in the rehabilitation facilities may involve robotic interventions or group therapies with an exercise therapy focus (endurance training, strength training, Nordic walking, etc.) or an occupational therapy focus (crafting, gardening, writing training, etc.). The selection of additional therapies for study participants may vary based on individual patient needs, to deliver tailored care.

The primary outcome is balance as assessed using the BBS (Berg et al., 1989). The BBS is considered the gold standard for evaluating balance in people after stroke (Blum and Korner-Bitensky, 2008). It comprises 14 items, and a Likert scale is used ranging from 0 to 4, depending on the patient's performance (Table 3).

Secondary outcomes are static balance, as assessed using posturography (Tymo Balance Plate, Tyromotion, Austria), dynamic balance during walking, measured using the Functional Gait Assessment (FGA; Thieme et al., 2009), and trunk control, as assessed using the Trunk Control Test (TCT; Collin and Wade, 1990). Additionally, concerns about falling are assessed through the Falls Efficacy Scale International (FES-I; Yardley et al., 2005; Kempen et al., 2007), and risk of falling is determined according to the Functional Reach Test (FRT; Duncan et al., 1990, 1992). The number of falls is documented as well.

Further secondary outcomes include walking ability, evaluated using the Functional Ambulation Categories (FAC; Collen et al., 1990), functional mobility, as measured by the Timed Up and Go Test (TUG; Podsiadlo and Richardson, 1991), health-related quality of life, assessed utilizing the EuroQol questionnaire EQ-5D-3L (Rabin and de Charro, 2001; Greiner et al., 2005), severity of ataxic symptoms, as determined by the Scale for the Assessment and Rating of Ataxia (SARA), and patients' independence in daily life, measured using the Scores of Independence for Neurologic and Geriatric Rehabilitation (SINGER; Funke et al., 2009; Gerdes et al., 2012; Table 3).

Due to the time-point of study recruitment being a susceptible one for diminished physical and mental resilience (i.e., in the acute phase after stroke), FGA, EQ-5D-3L, and SINGER are not performed at T0.

Another secondary outcome for patients with cerebellar infarction is neuronal plasticity occurring over time, measured using magnetic resonance imaging (MRI) at T0 and T2. In addition to structural MRI, functional MRI is performed, including two motor paradigms: ankle dorsiflexion (Figure 3) and a balance task (Figure 4). As it is not possible to physically execute the balance task in the scanner, action observation plus motor imagery (AOMI) is performed while watching a video of a person doing the balance task. A combination of MI and video was chosen, as studies indicate a better activation of balance relevant brain areas than with MI or video alone (Taube et al., 2015; Ruffieux et al., 2018). MI capability is assessed using the Kinesthetic and Visual Imagery Questionnaire (KVIQ-10; Malouin et al., 2007). Mental chronometry, referring to the temporal coupling between real and simulated movement, is measured with a 6-Meter Walk Test (Lam et al., 2010). Assessments are validated to evaluate quality of MI (Schuster et al., 2012; Ma et al., 2022) and temporal consistency in stroke patients (Malouin et al., 2008; Malouin and Richards, 2010; Oostra et al., 2015; De Bartolo et al., 2020). After undergoing both assessments at T0, patients receive an MI training session before each MRI examination.

Any adverse events (AE or serious AE) are recorded systematically, evaluated by authorized study personnel, and reported in accordance with responsible ethics committee requirements.

During the intervention phase (weeks 1–4), the occurrence of all adverse events is recorded by the responsible study center staff and assessed by a member of the medical team. The occurrence of falling during home exercises (weeks 5–12) is recorded by the patient as an adverse event of special interest, given that it serves as a secondary outcome within the study.